When it comes to preventing infection in rheumatology patients, “vaccination is the best mode of infection protection” and works synergistically with masks and hand washing, according to Leonard H. Calabrese, DO.

“Patients with rheumatic diseases have increased morbidity and mortality [from infection] and a lot of risk factors, including age, comorbidities, cytopenias, and extra-articular disease immunosuppression,” he said in a virtual presentation at the annual Perspectives in Rheumatic Diseases held by Global Academy for Medical Education.

Unfortunately, vaccination uptake remains “much lower than we would like in this country,” he said. Notably, influenza vaccination remains well below the World Health Organization target of 75%, he said.
 

Influenza vaccination

Flu vaccination will be even more important this year in the context of the COVID-19 pandemic, said Dr. Calabrese, professor of medicine and the RJ Fasenmyer Chair of Clinical Immunology at the Cleveland Clinic in Ohio. “For everyone who comes in with a respiratory illness, we will have to figure out whether it is flu or COVID,” he emphasized.

The Centers for Disease Control and Prevention recommendations include a detailed special considerations section for patients with immunocompromising conditions; “the notes have everything you need to know” about advising rheumatology patients, most of whom can safely receive a flu vaccine, he said.

One concern that always comes up is whether an antibody response will be suppressed based on therapy, Dr. Calabrese noted. Two major drugs with the greatest ability to reduce response are methotrexate and rituximab, he said. His tip: “Withhold methotrexate for two doses following seasonal flu vaccination.” This advice stems from a series of “practice-changing” studies by Park et al. published in 2017, 2018, and 2019 that showed benefit in withholding methotrexate for two doses following vaccination.

In the past, high-dose trivalent flu vaccines have been more expensive, and not necessarily practice changing, with studies showing varying clinical effectiveness and cost-effectiveness, Dr. Calabrese said. This year, a high-dose quadrivalent vaccine should be available that showed a 24% improvement in protection from all strains of influenza, compared with the standard vaccine in a head-to-head, randomized, controlled trial, he noted.

“All patients in rheumatology practices should get a flu vaccine,” with a 2-week hold on methotrexate following vaccination, he advised, and those aged 65 years and older should receive the high-dose quadrivalent. Younger patients on immunosuppressive therapy also might be considered for the high-dose vaccine, he said.

Pneumococcal vaccination

Dr. Calabrese also emphasized the value of pneumococcal vaccines for rheumatology patients. “The mortality for invasive disease ranges from 5% to 32%, but patients with immunocompromising conditions are at increased risk.”

Dr. Calabrese added a note on safety: Patients with cryopyrin-associated periodic syndrome (CAPS), a rare hereditary inflammatory disorder with cutaneous, neurologic, ophthalmologic, and rheumatologic manifestations, may have severe local and systemic reactions to the 23-valent polysaccharide vaccine (PPSV23), he said.

However, immunization against pneumococcal disease is safe and effective for most patients with autoimmune and inflammatory disorders regardless of their current therapy, he said. As with influenza, the CDC’s vaccination recommendations provide details for special situations, including immunocompromised individuals, he noted.

Dr. Calabrese recommended the 13-valent pneumococcal conjugate vaccine (PCV13) as soon as possible for rheumatology patients who have never been vaccinated, with follow-up doses of the 23-valent polysaccharide vaccine (PPSV23) at least 8 weeks later, and a PPSV23 booster 5 years after the first PPSV23 dose.
 

Protecting against shingles

When it comes to managing the varicella zoster virus (VZV) in immunocompromised patients, “prevention is preferable to treatment, as our patients are particularly vulnerable because of age and declining immunity,” Dr. Calabrese said.

Prevention is important because “once herpes zoster develops, the available treatments, including antiviral therapy, do not prevent postherpetic neuralgia in all patients,” he emphasized. “The treatments are complicated and not always effective,” he added.

The complications of zoster are well known, but recent data show an increased risk of cardiovascular disease as well, Dr. Calabrese said. “All the more reason to protect rheumatology patients from incident zoster,” he said.

Currently, the nonlive recombinant subunit zoster vaccine (Shingrix) is the preferred option for VZV vaccination according to the CDC’s Advisory Committee on Immunization Practices, Dr. Calabrese said. The CDC initially recommended its use to prevent herpes zoster and related complications in all immunocompetent adults aged 50 years and older; in an update, a C-level recommendation extends to “all patients aged 50 with or without immunosuppressive illnesses regardless of previous Zostavax exposure,” Dr. Calabrese said. “All patients on or starting [Janus] kinase inhibitors, regardless of age, should be considered” to receive the herpes zoster vaccine, he noted.

In general, promoting vaccination for rheumatology patients and for all patients is a multipronged effort that might include reminders, rewards, education, and standing orders, Dr. Calabrese said. Clinicians must continue to educate patients not only by strongly recommending the appropriate vaccines, but dispelling myths about vaccination, addressing fears, and providing current and accurate information, he said.

Dr. Calabrese disclosed relationships with AbbVie, Bristol-Myers Squibb, Crescendo, Genentech, Gilead, GlaxoSmithKline, Janssen, Novartis, Pfizer, Sanofi-Regeneron, and UCB.

Global Academy for Medical Education and this news organization are owned by the same parent company.

When it comes to preventing infection in rheumatology patients, “vaccination is the best mode of infection protection” and works synergistically with masks and hand washing, according to Leonard H. Calabrese, DO.

“Patients with rheumatic diseases have increased morbidity and mortality [from infection] and a lot of risk factors, including age, comorbidities, cytopenias, and extra-articular disease immunosuppression,” he said in a virtual presentation at the annual Perspectives in Rheumatic Diseases held by Global Academy for Medical Education.

Unfortunately, vaccination uptake remains “much lower than we would like in this country,” he said. Notably, influenza vaccination remains well below the World Health Organization target of 75%, he said.
 

Influenza vaccination

Flu vaccination will be even more important this year in the context of the COVID-19 pandemic, said Dr. Calabrese, professor of medicine and the RJ Fasenmyer Chair of Clinical Immunology at the Cleveland Clinic in Ohio. “For everyone who comes in with a respiratory illness, we will have to figure out whether it is flu or COVID,” he emphasized.

The Centers for Disease Control and Prevention recommendations include a detailed special considerations section for patients with immunocompromising conditions; “the notes have everything you need to know” about advising rheumatology patients, most of whom can safely receive a flu vaccine, he said.

One concern that always comes up is whether an antibody response will be suppressed based on therapy, Dr. Calabrese noted. Two major drugs with the greatest ability to reduce response are methotrexate and rituximab, he said. His tip: “Withhold methotrexate for two doses following seasonal flu vaccination.” This advice stems from a series of “practice-changing” studies by Park et al. published in 2017, 2018, and 2019 that showed benefit in withholding methotrexate for two doses following vaccination.

In the past, high-dose trivalent flu vaccines have been more expensive, and not necessarily practice changing, with studies showing varying clinical effectiveness and cost-effectiveness, Dr. Calabrese said. This year, a high-dose quadrivalent vaccine should be available that showed a 24% improvement in protection from all strains of influenza, compared with the standard vaccine in a head-to-head, randomized, controlled trial, he noted.

“All patients in rheumatology practices should get a flu vaccine,” with a 2-week hold on methotrexate following vaccination, he advised, and those aged 65 years and older should receive the high-dose quadrivalent. Younger patients on immunosuppressive therapy also might be considered for the high-dose vaccine, he said.

Pneumococcal vaccination

Dr. Calabrese also emphasized the value of pneumococcal vaccines for rheumatology patients. “The mortality for invasive disease ranges from 5% to 32%, but patients with immunocompromising conditions are at increased risk.”

Dr. Calabrese added a note on safety: Patients with cryopyrin-associated periodic syndrome (CAPS), a rare hereditary inflammatory disorder with cutaneous, neurologic, ophthalmologic, and rheumatologic manifestations, may have severe local and systemic reactions to the 23-valent polysaccharide vaccine (PPSV23), he said.

However, immunization against pneumococcal disease is safe and effective for most patients with autoimmune and inflammatory disorders regardless of their current therapy, he said. As with influenza, the CDC’s vaccination recommendations provide details for special situations, including immunocompromised individuals, he noted.

Dr. Calabrese recommended the 13-valent pneumococcal conjugate vaccine (PCV13) as soon as possible for rheumatology patients who have never been vaccinated, with follow-up doses of the 23-valent polysaccharide vaccine (PPSV23) at least 8 weeks later, and a PPSV23 booster 5 years after the first PPSV23 dose.
 

Protecting against shingles

When it comes to managing the varicella zoster virus (VZV) in immunocompromised patients, “prevention is preferable to treatment, as our patients are particularly vulnerable because of age and declining immunity,” Dr. Calabrese said.

Prevention is important because “once herpes zoster develops, the available treatments, including antiviral therapy, do not prevent postherpetic neuralgia in all patients,” he emphasized. “The treatments are complicated and not always effective,” he added.

The complications of zoster are well known, but recent data show an increased risk of cardiovascular disease as well, Dr. Calabrese said. “All the more reason to protect rheumatology patients from incident zoster,” he said.

Currently, the nonlive recombinant subunit zoster vaccine (Shingrix) is the preferred option for VZV vaccination according to the CDC’s Advisory Committee on Immunization Practices, Dr. Calabrese said. The CDC initially recommended its use to prevent herpes zoster and related complications in all immunocompetent adults aged 50 years and older; in an update, a C-level recommendation extends to “all patients aged 50 with or without immunosuppressive illnesses regardless of previous Zostavax exposure,” Dr. Calabrese said. “All patients on or starting [Janus] kinase inhibitors, regardless of age, should be considered” to receive the herpes zoster vaccine, he noted.

In general, promoting vaccination for rheumatology patients and for all patients is a multipronged effort that might include reminders, rewards, education, and standing orders, Dr. Calabrese said. Clinicians must continue to educate patients not only by strongly recommending the appropriate vaccines, but dispelling myths about vaccination, addressing fears, and providing current and accurate information, he said.

Dr. Calabrese disclosed relationships with AbbVie, Bristol-Myers Squibb, Crescendo, Genentech, Gilead, GlaxoSmithKline, Janssen, Novartis, Pfizer, Sanofi-Regeneron, and UCB.

Global Academy for Medical Education and this news organization are owned by the same parent company.

When it comes to preventing infection in rheumatology patients, “vaccination is the best mode of infection protection” and works synergistically with masks and hand washing, according to Leonard H. Calabrese, DO.

“Patients with rheumatic diseases have increased morbidity and mortality [from infection] and a lot of risk factors, including age, comorbidities, cytopenias, and extra-articular disease immunosuppression,” he said in a virtual presentation at the annual Perspectives in Rheumatic Diseases held by Global Academy for Medical Education.

Unfortunately, vaccination uptake remains “much lower than we would like in this country,” he said. Notably, influenza vaccination remains well below the World Health Organization target of 75%, he said.
 

Influenza vaccination

Flu vaccination will be even more important this year in the context of the COVID-19 pandemic, said Dr. Calabrese, professor of medicine and the RJ Fasenmyer Chair of Clinical Immunology at the Cleveland Clinic in Ohio. “For everyone who comes in with a respiratory illness, we will have to figure out whether it is flu or COVID,” he emphasized.

The Centers for Disease Control and Prevention recommendations include a detailed special considerations section for patients with immunocompromising conditions; “the notes have everything you need to know” about advising rheumatology patients, most of whom can safely receive a flu vaccine, he said.

One concern that always comes up is whether an antibody response will be suppressed based on therapy, Dr. Calabrese noted. Two major drugs with the greatest ability to reduce response are methotrexate and rituximab, he said. His tip: “Withhold methotrexate for two doses following seasonal flu vaccination.” This advice stems from a series of “practice-changing” studies by Park et al. published in 2017, 2018, and 2019 that showed benefit in withholding methotrexate for two doses following vaccination.

In the past, high-dose trivalent flu vaccines have been more expensive, and not necessarily practice changing, with studies showing varying clinical effectiveness and cost-effectiveness, Dr. Calabrese said. This year, a high-dose quadrivalent vaccine should be available that showed a 24% improvement in protection from all strains of influenza, compared with the standard vaccine in a head-to-head, randomized, controlled trial, he noted.

“All patients in rheumatology practices should get a flu vaccine,” with a 2-week hold on methotrexate following vaccination, he advised, and those aged 65 years and older should receive the high-dose quadrivalent. Younger patients on immunosuppressive therapy also might be considered for the high-dose vaccine, he said.

Pneumococcal vaccination

Dr. Calabrese also emphasized the value of pneumococcal vaccines for rheumatology patients. “The mortality for invasive disease ranges from 5% to 32%, but patients with immunocompromising conditions are at increased risk.”

Dr. Calabrese added a note on safety: Patients with cryopyrin-associated periodic syndrome (CAPS), a rare hereditary inflammatory disorder with cutaneous, neurologic, ophthalmologic, and rheumatologic manifestations, may have severe local and systemic reactions to the 23-valent polysaccharide vaccine (PPSV23), he said.

However, immunization against pneumococcal disease is safe and effective for most patients with autoimmune and inflammatory disorders regardless of their current therapy, he said. As with influenza, the CDC’s vaccination recommendations provide details for special situations, including immunocompromised individuals, he noted.

Dr. Calabrese recommended the 13-valent pneumococcal conjugate vaccine (PCV13) as soon as possible for rheumatology patients who have never been vaccinated, with follow-up doses of the 23-valent polysaccharide vaccine (PPSV23) at least 8 weeks later, and a PPSV23 booster 5 years after the first PPSV23 dose.
 

Protecting against shingles

When it comes to managing the varicella zoster virus (VZV) in immunocompromised patients, “prevention is preferable to treatment, as our patients are particularly vulnerable because of age and declining immunity,” Dr. Calabrese said.

Prevention is important because “once herpes zoster develops, the available treatments, including antiviral therapy, do not prevent postherpetic neuralgia in all patients,” he emphasized. “The treatments are complicated and not always effective,” he added.

The complications of zoster are well known, but recent data show an increased risk of cardiovascular disease as well, Dr. Calabrese said. “All the more reason to protect rheumatology patients from incident zoster,” he said.

Currently, the nonlive recombinant subunit zoster vaccine (Shingrix) is the preferred option for VZV vaccination according to the CDC’s Advisory Committee on Immunization Practices, Dr. Calabrese said. The CDC initially recommended its use to prevent herpes zoster and related complications in all immunocompetent adults aged 50 years and older; in an update, a C-level recommendation extends to “all patients aged 50 with or without immunosuppressive illnesses regardless of previous Zostavax exposure,” Dr. Calabrese said. “All patients on or starting [Janus] kinase inhibitors, regardless of age, should be considered” to receive the herpes zoster vaccine, he noted.

In general, promoting vaccination for rheumatology patients and for all patients is a multipronged effort that might include reminders, rewards, education, and standing orders, Dr. Calabrese said. Clinicians must continue to educate patients not only by strongly recommending the appropriate vaccines, but dispelling myths about vaccination, addressing fears, and providing current and accurate information, he said.

Dr. Calabrese disclosed relationships with AbbVie, Bristol-Myers Squibb, Crescendo, Genentech, Gilead, GlaxoSmithKline, Janssen, Novartis, Pfizer, Sanofi-Regeneron, and UCB.

Global Academy for Medical Education and this news organization are owned by the same parent company.

Annually, more than 40 million noncardiac surgeries take place in the US,¹ with 1%-3% of patients experiencing a major adverse cardiovascular event (MACE) such as acute myocardial infarction (AMI) or cardiac arrest postoperatively.² Such patients are at markedly increased risk of both perioperative and long-term death.^2-5

Over the past 40 years, efforts to model the risk of cardiac complications after noncardiac surgery have examined relationships between preoperative risk factors and postoperative cardiovascular events. The resulting risk-stratification tools, such as the Lee Revised Cardiac Risk Index (RCRI), have been used to inform perioperative care, including strategies for risk factor management prior to surgery, testing for cardiac events after surgery, and decisions regarding postoperative disposition.⁶ However, tools used in practice have not incorporated intraoperative data on hemodynamics or medication administration in the transition to postoperative care, which is often provided by nonsurgical clinicians such as hospitalists. Presently, there is active debate about the optimal approach to postoperative evaluation and management of MACE, particularly with regard to indications for cardiac biomarker testing after surgery in patients without signs or symptoms of acute cardiac syndromes. The lack of consensus is reflected in differences among guidelines for postoperative cardiac biomarker testing across professional societies in Europe, Canada, and the United States.^7-9

In this study, we examined whether the addition of intraoperative data to preoperative data (together, perioperative data) improved prediction of MACE after noncardiac surgery when compared with RCRI. Additionally, to investigate how such a model could be applied in practice, we compared risk stratification based on our model to a published risk factor–based guideline algorithm for postoperative cardiac biomarker testing.⁷ In particular, we evaluated to what extent patients recommended for postoperative cardiac biomarkers under the risk factor–based guideline algorithm would be reclassified as low risk by the model using perioperative data. Conducting biomarker tests on these patients would potentially represent low-value care. We hypothesized that adding intraoperative data would (a) lead to improved prediction of MACE complications when compared with RCRI and (b) more effectively identify, compared with a risk factor–based guideline algorithm, patients for whom cardiac biomarker testing would or would not be clinically meaningful.

We followed the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) reporting guideline.¹⁰

Study Data

Baseline, preoperative, and intraoperative data were collected for patients undergoing surgery between January 2014 and April 2018 within the University of Pennsylvania Health System (UPHS) electronic health record (EHR), and these data were then integrated into a comprehensive perioperative dataset (data containing administrative, preoperative, intraoperative, and postoperative information related to surgeries) created through a collaboration with the Multicenter Perioperative Outcomes Group.¹¹ The University of Pennsylvania Institutional Review Board approved this study.

Study Population

Patients aged 18 years or older who underwent inpatient major noncardiac surgery across four tertiary academic medical centers within UPHS in Pennsylvania during the study period were included in the cohort (see Appendix for inclusion/exclusion criteria).^12,13 Noncardiac surgery was identified using primary Current Procedural Terminology (CPT) code specification ranges for noncardiac surgeries 10021-32999 and 34001-69990. The study sample was divided randomly into a training set (60%), validation (20%), and test set (20%),¹⁴ with similar rates of MACE in the resulting sets. We used a holdout test set for all final analyses to avoid overfitting during model selection.

Outcomes

The composite outcome used to develop the risk-stratification models was in-hospital MACE after major noncardiac surgery. Following prior literature, MACE was defined using billing codes for ST-elevation/non–ST-elevation myocardial infarction (STEMI/NSTEMI, ICD-9-CM 410.xx, ICD-10-CM I21.xx), cardiac arrest (ICD-9-CM 427.5, ICD-10-CM I46.x, I97.121), or all-cause in-hospital death.^2,15-17

Variables

Variables were selected from baseline administrative, preoperative clinical, and intraoperative clinical data sources (full list in Appendix). Baseline variables included demographics, insurance type, and Elixhauser comorbidities.^18,19 Preoperative variables included surgery type, laboratory results, and American Society of Anesthesiologists (ASA) Physical Status classification.²⁰ Intraoperative variables included vital signs, estimated blood loss, fluid administration, and vasopressor use. We winsorized outlier values and used multiple imputation to address missingness. Rates of missing data can be found in Appendix Table 1.

Risk-Stratification Models Used as Comparisons

Briefly, RCRI variables include the presence of high-risk surgery,²¹ comorbid cardiovascular diseases (ie, ischemic heart disease, congestive heart failure, and cerebrovascular disease), preoperative use of insulin, and elevated preoperative serum creatinine.⁶ RCRI uses the inputs to calculate a point score that equates to different risk strata and is based on a stepwise logistic regression model with postoperative cardiovascular complications as the dependent outcome variable. For this study, we implemented the weighted version of the RCRI algorithm and computed the point scores (Appendix).^6,7,22

We also applied a risk factor–based algorithm for postoperative cardiac biomarker testing published in 2017 by the Canadian Cardiovascular Society (CCS) guidelines to each patient in the study sample.⁷ Specifically, this algorithm recommends daily troponin surveillance for 48 to 72 hours after surgery among patients who have (1) an elevated NT-proBNP/BNP measurement or no NT-proBNP/BNP measurement before surgery, (2) have a Revised Cardiac Risk Index score of 1 or greater, (3) are aged 65 years and older, (4) are aged 45 to 64 years with significant cardiovascular disease undergoing elective surgery, or (5) are aged 18 to 64 years with significant cardiovascular disease undergoing semiurgent, urgent, or emergent surgery.

Statistical Analysis

We compared patient characteristics and outcomes between those who did and those who did not experience MACE during hospitalization. Chi-square tests were used to compare categorical variables and Mann Whitney tests were used to compare continuous variables.

To create the perioperative risk-stratification model based on baseline, preoperative, and intraoperative data, we used a logistic regression with elastic net selection using a dichotomous dependent variable indicating MACE and independent variables described earlier. This perioperative model was fit on the training set and the model coefficients were then applied to the patients in the test set. The area under the receiver operating characteristic curve (AUC) was reported and the outcomes were reported by predicted risk decile, with higher deciles indicating higher risk (ie, higher numbers of patients with MACE outcomes in higher deciles implied better risk stratification). Because predicted risk of postoperative MACE may not have been distributed evenly across deciles, we also examined the distribution of the predicted probability of MACE and examined the number of patients below thresholds of risk corresponding to 0.1% or less, 0.25% or less, 0.5% or less, and 1% or less. These thresholds were chosen because they were close to the overall rate of MACE within our cohort.

We tested for differences in predictive performance between the RCRI logistic regression model AUC and the perioperative model AUC using DeLong’s test.²³ Additionally, we illustrated differences between the perioperative and RCRI models’ performance in two ways by stratifying patients into deciles based on predicted risk. First, we compared rates of MACE and MACE component events by predicted decile of the perioperative and RCRI models. Second, we further classified patients as RCRI high or low risk (per RCRI score classification in which RCRI score of 1 or greater is high risk and RCRI score of 0 is low risk) and examined numbers of surgical cases and MACE complications within these categories stratified by perioperative model predicted decile.

To compare the perioperative model’s performance with that of a risk factor–based guideline algorithm, we classified patients according to CCS guidelines as high risk (those for whom the CCS guidelines algorithm would recommend postoperative troponin surveillance testing) and low risk (those for whom the CCS guidelines algorithm would not recommend surveillance testing). We also used a logistic regression to examine if the predicted risk from our model was independently associated with MACE above and beyond the testing recommendation of the CCS guidelines algorithm. This model used MACE as the dependent variable and model-predicted risk and a CCS guidelines–defined high-risk indicator as predictors. We computed the association between a 10 percentage–point increase in predicted risk on observed MACE outcome rates.²⁴

In sensitivity analyses, we used a random forest machine learning classifier to test an alternate model specification, used complete case analysis, varied RCRI thresholds, and limited to patients aged 50 years or older. We also varied the penalty parameter in the elastic net model and plotted AUC versus the number of variables included to examine parsimonious models. SAS v9.4 (SAS Institute Inc) was used for main analyses. Data preparations and sensitivity analysis were done in Python v3.6 with Pandas v0.24.2 and Scikit-learn v0.19.1.

Study Sample

Patients who underwent major noncardiac surgery in our sample (n = 72,909) were approximately a mean age of 56 years, 58% female, 66% of White race and 26% of Black race, and most likely to have received orthopedic surgery (33%) or general surgery (20%). Those who experienced MACE (n = 558; 0.77%) differed along several characteristics (Table 1). For example, those with MACE were older (mean age, 65.4 vs 55.4 years; P < .001) and less likely to be female (41.9% vs 58.3%; P < .001).

Model Performance After Intraoperative Data Inclusion

In the perioperative model combining preoperative and intraoperative data, 26 variables were included after elastic net selection (Appendix Table 2). Model discrimination in the test set of patients demonstrated an AUC of 0.88 (95% CI, 0.85-0.92; Figure). When examining outcome rates by predicted decile, the outcome rates of in-hospital MACE complications were higher in the highest decile than in the lowest decile, notably with 58 of 92 (63%) cases with MACE complications within the top decile of predicted risk (Table 2). The majority of patients had low predicted risk of MACE, with 5,309 (36.1%), 8,796 (59.7%), 11,335 (77.0%), and 12,972 (88.1%) below the risk thresholds of to 0.1%, 0.25%, 0.5%, and 1.0% respectively. The associated MACE rates were 0.04%, 0.10%, 0.17%, and 0.25% (average rate in sample was 0.63%) (Appendix Table 3).

Model Performance Comparisons

The perioperative model AUC of 0.88 was higher when compared with RCRI’s AUC of 0.79 (95% CI, 0.74-0.84; P < .001). The number of MACE complications was more concentrated in the top decile of predicted risk of the perioperative model than it was in that of the RCRI model (58 vs 43 of 92 events, respectively; 63% vs 47%; Table 2). Furthermore, there were fewer cases with MACE complications in the low-risk deciles (ie, deciles 1 to 5) of the perioperative model than in the those of the RCRI model. These relative differences were consistent for MACE component outcomes of STEMI/NSTEMI, cardiac arrest, and in-hospital death, as well.

There was substantial heterogeneity in the perioperative model predicted risk of patients classified as either RCRI low risk or high risk (ie, each category included patients with low and high predicted risk) categories (Table 3). Patients in the bottom (low-risk) five deciles of the perioperative model’s predicted risk who were in the RCRI model’s high-risk group were very unlikely to experience MACE complications (3 out of 722 cases; 0.42%). Furthermore, among those classified as low risk by the RCRI model but were in the top decile of the perioperative model’s predicted risk, the MACE complication rate was 3.5% (8 out of 229), which was 6 times the sample mean MACE complication rate.

The perioperative model identified more patients as low risk than did the CCS guidelines’ risk factor–based algorithm (Table 3). For example, 2,341 of the patients the CCS guidelines algorithm identified as high risk were in the bottom 50% of the perioperative model’s predicted risk for experiencing MACE (below a 0.18% chance of a MACE complication); only four of these patients (0.17%) actually experienced MACE. This indicates that the 2,341 of 7,597 (31%) high-risk patients identified as low risk in the perioperative model would have been recommended for postoperative troponin testing by CCS guidelines based on preoperative risk factors alone—but did not go on to experience a MACE. Regression results indicated that both CCS guidelines risk-factor classification and the perioperative model’s predicted risk were predictive of MACE outcomes. A change in the perioperative model’s predicted risk of 10 percentage points was associated with an increase in the probability of a MACE outcomes of 0.45 percentage points (95% CI, 0.35-0.55 percentage points; P < .001) and moving from CCS guidelines’ low- to high-risk categories was associated with an increased probability of MACE by 0.96 percentage points (95% CI, 0.75-1.16 percentage points; P < .001).

Results were consistent with the main analysis across all sensitivity analyses (Appendix Tables 4-7). Parsimonious models with variables as few as eight variables retained strong predictive power (AUC, 0.870; Appendix Figure 1 and Table 8).

In this study, the addition of intraoperative data improved risk stratification for MACE complications when compared with standard risk tools such as RCRI. This approach also outperformed a guidelines-based approach and identified additional patients at low risk of cardiovascular complications. This study has three main implications.

First, this study demonstrated the additional value of combining intraoperative data with preoperative data in risk prediction for postoperative cardiovascular events. The intraoperative data most strongly associated with MACE, which likely were responsible for the performance improvement, included administration of medications (eg, sodium bicarbonate or calcium chloride) and blood products (eg, platelets and packed red blood cells), vitals (ie, heart rate), and intraoperative procedures (ie, arterial line placement); all model variables and coefficients are reported in Appendix Table 9. The risk-stratification model using intraoperative clinical data outperformed validated standard models such as RCRI. While this model should not be used in causal inference and cannot be used to inform decisions about risk-benefit tradeoffs of undergoing surgery, its improved performance relative to prior models highlights the potential in using real-time data. Preliminary illustrative analysis demonstrated that parsimonious models with as few as eight variables perform well, whose implementation as risk scores in EHRs is likely straightforward (Appendix Table 8). This is particularly important for longitudinal care in the hospital, in which patients frequently are cared for by multiple clinical services and experience handoffs. For example, many orthopedic surgery patients with significant medical comorbidity are managed postoperatively by hospitalist physicians after initial surgical care.

Second, our study aligns well with the cardiac risk-stratification literature more broadly. For example, the patient characteristics and clinical variables most associated with cardiovascular complications were age, history of ischemic heart disease, American Society of Anesthesiologists physical status, use of intraoperative sodium bicarbonate or vasopressors, lowest intraoperative heart rate measured, and lowest intraoperative mean arterial pressure measured. While many of these variables overlap with those included in the RCRI model, others (such as American Society of Anesthesiologists physical status) are not included in RCRI but have been shown to be important in risk prediction in other studies using different data variables.^6,25,26

Third, we illustrated a clinical application of this model in identifying patients at low risk of cardiovascular complications, although benefit may extend to other patients as well. This is particularly germane to clinicians who frequently manage patients in the postsurgical or postprocedural setting. Moreover, the clinical relevance to these clinicians is underscored by the lack of consensus among professional societies across Europe, Canada, and the United States about which subgroups of patients undergoing noncardiac surgery should receive postoperative cardiac biomarker surveillance testing in the 48 to 72 hours after surgery.^6-9 This may be in part caused by differences in clinical objectives. For example, the CCS guidelines in part aim to detect myocardial injury after noncardiac surgery (MINS) up to 30 days after surgery, which may be more sensitive to myocardial injury but less strongly associated with outcomes like MACE. The results of this study suggest that adopting such risk factor–based testing would likely lead to additional testing of low risk patients, which may represent low value surveillance tests. For example, there were 2,257 patients without postoperative cardiac biomarker testing in our data who would have been categorized as high risk by risk factor guidelines and therefore recommended to receive at least one postoperative cardiac biomarker surveillance test but were classified as low-risk individuals using a predicted probability of MACE less than 0.18% per our perioperative risk stratification model (Appendix Table 4). If each of these patients received one troponin biomarker test, the associated cost increase would be $372,405 (using the $165 cost per test reported at our institution). These costs would multiply if daily surveillance troponin biomarker tests were ordered for 48 to 72 hours after surgery, as recommended by the risk factor–based testing guidelines. This would be a departure from testing among patients using clinician discretion that may avoid low-value testing.

Applying the perioperative model developed in this paper to clinical practice still requires several steps. The technical aspects of finding a parsimonious model that can be implemented in the EHR is likely quite straightforward. Our preliminary analysis illustrates that doing so will not require accessing large numbers of intraoperative variables. Perhaps more important steps include prospective validation of the safety, usability, and clinical benefit of such an algorithm-based risk score.²⁷

The study has several limitations. First, it was an observational study using EHR data subject to missingness and data quality issues that may have persisted despite our methods. Furthermore, EHR data is not generated randomly, and unmeasured variables observed by clinicians but not by researchers could confound the results. However, our approach used the statistical model to examine risk, not causal inference. Second, this is a single institution study and the availability of EHR data, as well as practice patterns, may vary at other institutions. Furthermore, it is possible that performance of the RCRI score, the model fitting RCRI classification of high vs low risk on the sample data, and our model’s performance may not generalize to other clinical settings. However, we utilized data from multiple hospitals within a health system with different surgery and anesthesia groups and providers, and a similar AUC was reported for RCRI in original validation study.⁶ Third, our follow up period was limited to the hospital setting and we do not capture longitudinal outcomes, such as 30-day MACE. This may impact the ability to risk stratify for other important longer-term outcomes, limit clinical utility, and hinder comparability to other studies. Fourth, results may vary for other important cardiovascular outcomes that may be more sensitive to myocardial injury, such as MINS. Fifth, we used a limited number of modeling strategies.

Addition of intraoperative data to preoperative data improves prediction of cardiovascular complications after noncardiac surgery. Improving the identification of patients at low risk for such complications could potentially be applied to reduce unnecessary postoperative cardiac biomarker testing after noncardiac surgery, but it will require further validation in prospective clinical settings.

Dr Navathe reports grants from the following entities: Hawaii Medical Service Association, Anthem Public Policy Institute, Commonwealth Fund, Oscar Health, Cigna Corporation, Robert Wood Johnson Foundation, Donaghue Foundation, Pennsylvania Department of Health, Ochsner Health System, United Healthcare, Blue Cross Blue Shield of NC, Blue Shield of CA; personal fees from the following: Navvis Healthcare, Agathos, Inc, Navahealth, YNHHSC/CORE, Maine Health Accountable Care Organization, Maine Department of Health and Human Services, National University Health System - Singapore, Ministry of Health - Singapore, Social Security Administration - France, Elsevier Press, Medicare Payment Advisory Commission, Cleveland Clinic, Embedded Healthcare; and other support from Integrated Services, Inc, outside of the submitted work. Dr Volpp reports grants from Humana during the conduct of the study; grants from Hawaii Medical Services Agency, Discovery (South Africa), Merck, Weight Watchers, and CVS outside of the submitted work; he has received consulting income from CVS and VALHealth and is a principal in VALHealth, a behavioral economics consulting firm. Dr Holmes receives funding from the Pennsylvania Department of Health, US Public Health Service, and the Cardiovascular Medicine Research and Education Foundation. All other authors declare no conflicts of interest.

2019 Academy Health Annual Research Meeting, Poster Abstract Presentation, June 2 to June 4, 2019, Washington, DC.

This project was funded, in part, under a grant with the Pennsylvania Department of Health. This research was independent from the funder. The funder had no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the article for publication. The department specifically disclaims responsibility for any analyses, interpretations, or conclusions.

Annually, more than 40 million noncardiac surgeries take place in the US,¹ with 1%-3% of patients experiencing a major adverse cardiovascular event (MACE) such as acute myocardial infarction (AMI) or cardiac arrest postoperatively.² Such patients are at markedly increased risk of both perioperative and long-term death.^2-5

Over the past 40 years, efforts to model the risk of cardiac complications after noncardiac surgery have examined relationships between preoperative risk factors and postoperative cardiovascular events. The resulting risk-stratification tools, such as the Lee Revised Cardiac Risk Index (RCRI), have been used to inform perioperative care, including strategies for risk factor management prior to surgery, testing for cardiac events after surgery, and decisions regarding postoperative disposition.⁶ However, tools used in practice have not incorporated intraoperative data on hemodynamics or medication administration in the transition to postoperative care, which is often provided by nonsurgical clinicians such as hospitalists. Presently, there is active debate about the optimal approach to postoperative evaluation and management of MACE, particularly with regard to indications for cardiac biomarker testing after surgery in patients without signs or symptoms of acute cardiac syndromes. The lack of consensus is reflected in differences among guidelines for postoperative cardiac biomarker testing across professional societies in Europe, Canada, and the United States.^7-9

In this study, we examined whether the addition of intraoperative data to preoperative data (together, perioperative data) improved prediction of MACE after noncardiac surgery when compared with RCRI. Additionally, to investigate how such a model could be applied in practice, we compared risk stratification based on our model to a published risk factor–based guideline algorithm for postoperative cardiac biomarker testing.⁷ In particular, we evaluated to what extent patients recommended for postoperative cardiac biomarkers under the risk factor–based guideline algorithm would be reclassified as low risk by the model using perioperative data. Conducting biomarker tests on these patients would potentially represent low-value care. We hypothesized that adding intraoperative data would (a) lead to improved prediction of MACE complications when compared with RCRI and (b) more effectively identify, compared with a risk factor–based guideline algorithm, patients for whom cardiac biomarker testing would or would not be clinically meaningful.

We followed the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) reporting guideline.¹⁰

Study Data

Baseline, preoperative, and intraoperative data were collected for patients undergoing surgery between January 2014 and April 2018 within the University of Pennsylvania Health System (UPHS) electronic health record (EHR), and these data were then integrated into a comprehensive perioperative dataset (data containing administrative, preoperative, intraoperative, and postoperative information related to surgeries) created through a collaboration with the Multicenter Perioperative Outcomes Group.¹¹ The University of Pennsylvania Institutional Review Board approved this study.

Study Population

Patients aged 18 years or older who underwent inpatient major noncardiac surgery across four tertiary academic medical centers within UPHS in Pennsylvania during the study period were included in the cohort (see Appendix for inclusion/exclusion criteria).^12,13 Noncardiac surgery was identified using primary Current Procedural Terminology (CPT) code specification ranges for noncardiac surgeries 10021-32999 and 34001-69990. The study sample was divided randomly into a training set (60%), validation (20%), and test set (20%),¹⁴ with similar rates of MACE in the resulting sets. We used a holdout test set for all final analyses to avoid overfitting during model selection.

Outcomes

The composite outcome used to develop the risk-stratification models was in-hospital MACE after major noncardiac surgery. Following prior literature, MACE was defined using billing codes for ST-elevation/non–ST-elevation myocardial infarction (STEMI/NSTEMI, ICD-9-CM 410.xx, ICD-10-CM I21.xx), cardiac arrest (ICD-9-CM 427.5, ICD-10-CM I46.x, I97.121), or all-cause in-hospital death.^2,15-17

Variables

Variables were selected from baseline administrative, preoperative clinical, and intraoperative clinical data sources (full list in Appendix). Baseline variables included demographics, insurance type, and Elixhauser comorbidities.^18,19 Preoperative variables included surgery type, laboratory results, and American Society of Anesthesiologists (ASA) Physical Status classification.²⁰ Intraoperative variables included vital signs, estimated blood loss, fluid administration, and vasopressor use. We winsorized outlier values and used multiple imputation to address missingness. Rates of missing data can be found in Appendix Table 1.

Risk-Stratification Models Used as Comparisons

Briefly, RCRI variables include the presence of high-risk surgery,²¹ comorbid cardiovascular diseases (ie, ischemic heart disease, congestive heart failure, and cerebrovascular disease), preoperative use of insulin, and elevated preoperative serum creatinine.⁶ RCRI uses the inputs to calculate a point score that equates to different risk strata and is based on a stepwise logistic regression model with postoperative cardiovascular complications as the dependent outcome variable. For this study, we implemented the weighted version of the RCRI algorithm and computed the point scores (Appendix).^6,7,22

We also applied a risk factor–based algorithm for postoperative cardiac biomarker testing published in 2017 by the Canadian Cardiovascular Society (CCS) guidelines to each patient in the study sample.⁷ Specifically, this algorithm recommends daily troponin surveillance for 48 to 72 hours after surgery among patients who have (1) an elevated NT-proBNP/BNP measurement or no NT-proBNP/BNP measurement before surgery, (2) have a Revised Cardiac Risk Index score of 1 or greater, (3) are aged 65 years and older, (4) are aged 45 to 64 years with significant cardiovascular disease undergoing elective surgery, or (5) are aged 18 to 64 years with significant cardiovascular disease undergoing semiurgent, urgent, or emergent surgery.

Statistical Analysis

We compared patient characteristics and outcomes between those who did and those who did not experience MACE during hospitalization. Chi-square tests were used to compare categorical variables and Mann Whitney tests were used to compare continuous variables.

To create the perioperative risk-stratification model based on baseline, preoperative, and intraoperative data, we used a logistic regression with elastic net selection using a dichotomous dependent variable indicating MACE and independent variables described earlier. This perioperative model was fit on the training set and the model coefficients were then applied to the patients in the test set. The area under the receiver operating characteristic curve (AUC) was reported and the outcomes were reported by predicted risk decile, with higher deciles indicating higher risk (ie, higher numbers of patients with MACE outcomes in higher deciles implied better risk stratification). Because predicted risk of postoperative MACE may not have been distributed evenly across deciles, we also examined the distribution of the predicted probability of MACE and examined the number of patients below thresholds of risk corresponding to 0.1% or less, 0.25% or less, 0.5% or less, and 1% or less. These thresholds were chosen because they were close to the overall rate of MACE within our cohort.

We tested for differences in predictive performance between the RCRI logistic regression model AUC and the perioperative model AUC using DeLong’s test.²³ Additionally, we illustrated differences between the perioperative and RCRI models’ performance in two ways by stratifying patients into deciles based on predicted risk. First, we compared rates of MACE and MACE component events by predicted decile of the perioperative and RCRI models. Second, we further classified patients as RCRI high or low risk (per RCRI score classification in which RCRI score of 1 or greater is high risk and RCRI score of 0 is low risk) and examined numbers of surgical cases and MACE complications within these categories stratified by perioperative model predicted decile.

To compare the perioperative model’s performance with that of a risk factor–based guideline algorithm, we classified patients according to CCS guidelines as high risk (those for whom the CCS guidelines algorithm would recommend postoperative troponin surveillance testing) and low risk (those for whom the CCS guidelines algorithm would not recommend surveillance testing). We also used a logistic regression to examine if the predicted risk from our model was independently associated with MACE above and beyond the testing recommendation of the CCS guidelines algorithm. This model used MACE as the dependent variable and model-predicted risk and a CCS guidelines–defined high-risk indicator as predictors. We computed the association between a 10 percentage–point increase in predicted risk on observed MACE outcome rates.²⁴

In sensitivity analyses, we used a random forest machine learning classifier to test an alternate model specification, used complete case analysis, varied RCRI thresholds, and limited to patients aged 50 years or older. We also varied the penalty parameter in the elastic net model and plotted AUC versus the number of variables included to examine parsimonious models. SAS v9.4 (SAS Institute Inc) was used for main analyses. Data preparations and sensitivity analysis were done in Python v3.6 with Pandas v0.24.2 and Scikit-learn v0.19.1.

Study Sample

Patients who underwent major noncardiac surgery in our sample (n = 72,909) were approximately a mean age of 56 years, 58% female, 66% of White race and 26% of Black race, and most likely to have received orthopedic surgery (33%) or general surgery (20%). Those who experienced MACE (n = 558; 0.77%) differed along several characteristics (Table 1). For example, those with MACE were older (mean age, 65.4 vs 55.4 years; P < .001) and less likely to be female (41.9% vs 58.3%; P < .001).

Model Performance After Intraoperative Data Inclusion

In the perioperative model combining preoperative and intraoperative data, 26 variables were included after elastic net selection (Appendix Table 2). Model discrimination in the test set of patients demonstrated an AUC of 0.88 (95% CI, 0.85-0.92; Figure). When examining outcome rates by predicted decile, the outcome rates of in-hospital MACE complications were higher in the highest decile than in the lowest decile, notably with 58 of 92 (63%) cases with MACE complications within the top decile of predicted risk (Table 2). The majority of patients had low predicted risk of MACE, with 5,309 (36.1%), 8,796 (59.7%), 11,335 (77.0%), and 12,972 (88.1%) below the risk thresholds of to 0.1%, 0.25%, 0.5%, and 1.0% respectively. The associated MACE rates were 0.04%, 0.10%, 0.17%, and 0.25% (average rate in sample was 0.63%) (Appendix Table 3).

Model Performance Comparisons

The perioperative model AUC of 0.88 was higher when compared with RCRI’s AUC of 0.79 (95% CI, 0.74-0.84; P < .001). The number of MACE complications was more concentrated in the top decile of predicted risk of the perioperative model than it was in that of the RCRI model (58 vs 43 of 92 events, respectively; 63% vs 47%; Table 2). Furthermore, there were fewer cases with MACE complications in the low-risk deciles (ie, deciles 1 to 5) of the perioperative model than in the those of the RCRI model. These relative differences were consistent for MACE component outcomes of STEMI/NSTEMI, cardiac arrest, and in-hospital death, as well.

There was substantial heterogeneity in the perioperative model predicted risk of patients classified as either RCRI low risk or high risk (ie, each category included patients with low and high predicted risk) categories (Table 3). Patients in the bottom (low-risk) five deciles of the perioperative model’s predicted risk who were in the RCRI model’s high-risk group were very unlikely to experience MACE complications (3 out of 722 cases; 0.42%). Furthermore, among those classified as low risk by the RCRI model but were in the top decile of the perioperative model’s predicted risk, the MACE complication rate was 3.5% (8 out of 229), which was 6 times the sample mean MACE complication rate.

The perioperative model identified more patients as low risk than did the CCS guidelines’ risk factor–based algorithm (Table 3). For example, 2,341 of the patients the CCS guidelines algorithm identified as high risk were in the bottom 50% of the perioperative model’s predicted risk for experiencing MACE (below a 0.18% chance of a MACE complication); only four of these patients (0.17%) actually experienced MACE. This indicates that the 2,341 of 7,597 (31%) high-risk patients identified as low risk in the perioperative model would have been recommended for postoperative troponin testing by CCS guidelines based on preoperative risk factors alone—but did not go on to experience a MACE. Regression results indicated that both CCS guidelines risk-factor classification and the perioperative model’s predicted risk were predictive of MACE outcomes. A change in the perioperative model’s predicted risk of 10 percentage points was associated with an increase in the probability of a MACE outcomes of 0.45 percentage points (95% CI, 0.35-0.55 percentage points; P < .001) and moving from CCS guidelines’ low- to high-risk categories was associated with an increased probability of MACE by 0.96 percentage points (95% CI, 0.75-1.16 percentage points; P < .001).

Results were consistent with the main analysis across all sensitivity analyses (Appendix Tables 4-7). Parsimonious models with variables as few as eight variables retained strong predictive power (AUC, 0.870; Appendix Figure 1 and Table 8).

In this study, the addition of intraoperative data improved risk stratification for MACE complications when compared with standard risk tools such as RCRI. This approach also outperformed a guidelines-based approach and identified additional patients at low risk of cardiovascular complications. This study has three main implications.

First, this study demonstrated the additional value of combining intraoperative data with preoperative data in risk prediction for postoperative cardiovascular events. The intraoperative data most strongly associated with MACE, which likely were responsible for the performance improvement, included administration of medications (eg, sodium bicarbonate or calcium chloride) and blood products (eg, platelets and packed red blood cells), vitals (ie, heart rate), and intraoperative procedures (ie, arterial line placement); all model variables and coefficients are reported in Appendix Table 9. The risk-stratification model using intraoperative clinical data outperformed validated standard models such as RCRI. While this model should not be used in causal inference and cannot be used to inform decisions about risk-benefit tradeoffs of undergoing surgery, its improved performance relative to prior models highlights the potential in using real-time data. Preliminary illustrative analysis demonstrated that parsimonious models with as few as eight variables perform well, whose implementation as risk scores in EHRs is likely straightforward (Appendix Table 8). This is particularly important for longitudinal care in the hospital, in which patients frequently are cared for by multiple clinical services and experience handoffs. For example, many orthopedic surgery patients with significant medical comorbidity are managed postoperatively by hospitalist physicians after initial surgical care.

Second, our study aligns well with the cardiac risk-stratification literature more broadly. For example, the patient characteristics and clinical variables most associated with cardiovascular complications were age, history of ischemic heart disease, American Society of Anesthesiologists physical status, use of intraoperative sodium bicarbonate or vasopressors, lowest intraoperative heart rate measured, and lowest intraoperative mean arterial pressure measured. While many of these variables overlap with those included in the RCRI model, others (such as American Society of Anesthesiologists physical status) are not included in RCRI but have been shown to be important in risk prediction in other studies using different data variables.^6,25,26

Third, we illustrated a clinical application of this model in identifying patients at low risk of cardiovascular complications, although benefit may extend to other patients as well. This is particularly germane to clinicians who frequently manage patients in the postsurgical or postprocedural setting. Moreover, the clinical relevance to these clinicians is underscored by the lack of consensus among professional societies across Europe, Canada, and the United States about which subgroups of patients undergoing noncardiac surgery should receive postoperative cardiac biomarker surveillance testing in the 48 to 72 hours after surgery.^6-9 This may be in part caused by differences in clinical objectives. For example, the CCS guidelines in part aim to detect myocardial injury after noncardiac surgery (MINS) up to 30 days after surgery, which may be more sensitive to myocardial injury but less strongly associated with outcomes like MACE. The results of this study suggest that adopting such risk factor–based testing would likely lead to additional testing of low risk patients, which may represent low value surveillance tests. For example, there were 2,257 patients without postoperative cardiac biomarker testing in our data who would have been categorized as high risk by risk factor guidelines and therefore recommended to receive at least one postoperative cardiac biomarker surveillance test but were classified as low-risk individuals using a predicted probability of MACE less than 0.18% per our perioperative risk stratification model (Appendix Table 4). If each of these patients received one troponin biomarker test, the associated cost increase would be $372,405 (using the $165 cost per test reported at our institution). These costs would multiply if daily surveillance troponin biomarker tests were ordered for 48 to 72 hours after surgery, as recommended by the risk factor–based testing guidelines. This would be a departure from testing among patients using clinician discretion that may avoid low-value testing.

Applying the perioperative model developed in this paper to clinical practice still requires several steps. The technical aspects of finding a parsimonious model that can be implemented in the EHR is likely quite straightforward. Our preliminary analysis illustrates that doing so will not require accessing large numbers of intraoperative variables. Perhaps more important steps include prospective validation of the safety, usability, and clinical benefit of such an algorithm-based risk score.²⁷

The study has several limitations. First, it was an observational study using EHR data subject to missingness and data quality issues that may have persisted despite our methods. Furthermore, EHR data is not generated randomly, and unmeasured variables observed by clinicians but not by researchers could confound the results. However, our approach used the statistical model to examine risk, not causal inference. Second, this is a single institution study and the availability of EHR data, as well as practice patterns, may vary at other institutions. Furthermore, it is possible that performance of the RCRI score, the model fitting RCRI classification of high vs low risk on the sample data, and our model’s performance may not generalize to other clinical settings. However, we utilized data from multiple hospitals within a health system with different surgery and anesthesia groups and providers, and a similar AUC was reported for RCRI in original validation study.⁶ Third, our follow up period was limited to the hospital setting and we do not capture longitudinal outcomes, such as 30-day MACE. This may impact the ability to risk stratify for other important longer-term outcomes, limit clinical utility, and hinder comparability to other studies. Fourth, results may vary for other important cardiovascular outcomes that may be more sensitive to myocardial injury, such as MINS. Fifth, we used a limited number of modeling strategies.

Addition of intraoperative data to preoperative data improves prediction of cardiovascular complications after noncardiac surgery. Improving the identification of patients at low risk for such complications could potentially be applied to reduce unnecessary postoperative cardiac biomarker testing after noncardiac surgery, but it will require further validation in prospective clinical settings.

Dr Navathe reports grants from the following entities: Hawaii Medical Service Association, Anthem Public Policy Institute, Commonwealth Fund, Oscar Health, Cigna Corporation, Robert Wood Johnson Foundation, Donaghue Foundation, Pennsylvania Department of Health, Ochsner Health System, United Healthcare, Blue Cross Blue Shield of NC, Blue Shield of CA; personal fees from the following: Navvis Healthcare, Agathos, Inc, Navahealth, YNHHSC/CORE, Maine Health Accountable Care Organization, Maine Department of Health and Human Services, National University Health System - Singapore, Ministry of Health - Singapore, Social Security Administration - France, Elsevier Press, Medicare Payment Advisory Commission, Cleveland Clinic, Embedded Healthcare; and other support from Integrated Services, Inc, outside of the submitted work. Dr Volpp reports grants from Humana during the conduct of the study; grants from Hawaii Medical Services Agency, Discovery (South Africa), Merck, Weight Watchers, and CVS outside of the submitted work; he has received consulting income from CVS and VALHealth and is a principal in VALHealth, a behavioral economics consulting firm. Dr Holmes receives funding from the Pennsylvania Department of Health, US Public Health Service, and the Cardiovascular Medicine Research and Education Foundation. All other authors declare no conflicts of interest.

2019 Academy Health Annual Research Meeting, Poster Abstract Presentation, June 2 to June 4, 2019, Washington, DC.

This project was funded, in part, under a grant with the Pennsylvania Department of Health. This research was independent from the funder. The funder had no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the article for publication. The department specifically disclaims responsibility for any analyses, interpretations, or conclusions.

Annually, more than 40 million noncardiac surgeries take place in the US,¹ with 1%-3% of patients experiencing a major adverse cardiovascular event (MACE) such as acute myocardial infarction (AMI) or cardiac arrest postoperatively.² Such patients are at markedly increased risk of both perioperative and long-term death.^2-5

Over the past 40 years, efforts to model the risk of cardiac complications after noncardiac surgery have examined relationships between preoperative risk factors and postoperative cardiovascular events. The resulting risk-stratification tools, such as the Lee Revised Cardiac Risk Index (RCRI), have been used to inform perioperative care, including strategies for risk factor management prior to surgery, testing for cardiac events after surgery, and decisions regarding postoperative disposition.⁶ However, tools used in practice have not incorporated intraoperative data on hemodynamics or medication administration in the transition to postoperative care, which is often provided by nonsurgical clinicians such as hospitalists. Presently, there is active debate about the optimal approach to postoperative evaluation and management of MACE, particularly with regard to indications for cardiac biomarker testing after surgery in patients without signs or symptoms of acute cardiac syndromes. The lack of consensus is reflected in differences among guidelines for postoperative cardiac biomarker testing across professional societies in Europe, Canada, and the United States.^7-9

In this study, we examined whether the addition of intraoperative data to preoperative data (together, perioperative data) improved prediction of MACE after noncardiac surgery when compared with RCRI. Additionally, to investigate how such a model could be applied in practice, we compared risk stratification based on our model to a published risk factor–based guideline algorithm for postoperative cardiac biomarker testing.⁷ In particular, we evaluated to what extent patients recommended for postoperative cardiac biomarkers under the risk factor–based guideline algorithm would be reclassified as low risk by the model using perioperative data. Conducting biomarker tests on these patients would potentially represent low-value care. We hypothesized that adding intraoperative data would (a) lead to improved prediction of MACE complications when compared with RCRI and (b) more effectively identify, compared with a risk factor–based guideline algorithm, patients for whom cardiac biomarker testing would or would not be clinically meaningful.

We followed the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) reporting guideline.¹⁰

Study Data

Baseline, preoperative, and intraoperative data were collected for patients undergoing surgery between January 2014 and April 2018 within the University of Pennsylvania Health System (UPHS) electronic health record (EHR), and these data were then integrated into a comprehensive perioperative dataset (data containing administrative, preoperative, intraoperative, and postoperative information related to surgeries) created through a collaboration with the Multicenter Perioperative Outcomes Group.¹¹ The University of Pennsylvania Institutional Review Board approved this study.

Study Population

Patients aged 18 years or older who underwent inpatient major noncardiac surgery across four tertiary academic medical centers within UPHS in Pennsylvania during the study period were included in the cohort (see Appendix for inclusion/exclusion criteria).^12,13 Noncardiac surgery was identified using primary Current Procedural Terminology (CPT) code specification ranges for noncardiac surgeries 10021-32999 and 34001-69990. The study sample was divided randomly into a training set (60%), validation (20%), and test set (20%),¹⁴ with similar rates of MACE in the resulting sets. We used a holdout test set for all final analyses to avoid overfitting during model selection.

Outcomes

The composite outcome used to develop the risk-stratification models was in-hospital MACE after major noncardiac surgery. Following prior literature, MACE was defined using billing codes for ST-elevation/non–ST-elevation myocardial infarction (STEMI/NSTEMI, ICD-9-CM 410.xx, ICD-10-CM I21.xx), cardiac arrest (ICD-9-CM 427.5, ICD-10-CM I46.x, I97.121), or all-cause in-hospital death.^2,15-17

Variables

Variables were selected from baseline administrative, preoperative clinical, and intraoperative clinical data sources (full list in Appendix). Baseline variables included demographics, insurance type, and Elixhauser comorbidities.^18,19 Preoperative variables included surgery type, laboratory results, and American Society of Anesthesiologists (ASA) Physical Status classification.²⁰ Intraoperative variables included vital signs, estimated blood loss, fluid administration, and vasopressor use. We winsorized outlier values and used multiple imputation to address missingness. Rates of missing data can be found in Appendix Table 1.

Risk-Stratification Models Used as Comparisons

Briefly, RCRI variables include the presence of high-risk surgery,²¹ comorbid cardiovascular diseases (ie, ischemic heart disease, congestive heart failure, and cerebrovascular disease), preoperative use of insulin, and elevated preoperative serum creatinine.⁶ RCRI uses the inputs to calculate a point score that equates to different risk strata and is based on a stepwise logistic regression model with postoperative cardiovascular complications as the dependent outcome variable. For this study, we implemented the weighted version of the RCRI algorithm and computed the point scores (Appendix).^6,7,22

We also applied a risk factor–based algorithm for postoperative cardiac biomarker testing published in 2017 by the Canadian Cardiovascular Society (CCS) guidelines to each patient in the study sample.⁷ Specifically, this algorithm recommends daily troponin surveillance for 48 to 72 hours after surgery among patients who have (1) an elevated NT-proBNP/BNP measurement or no NT-proBNP/BNP measurement before surgery, (2) have a Revised Cardiac Risk Index score of 1 or greater, (3) are aged 65 years and older, (4) are aged 45 to 64 years with significant cardiovascular disease undergoing elective surgery, or (5) are aged 18 to 64 years with significant cardiovascular disease undergoing semiurgent, urgent, or emergent surgery.

Statistical Analysis

We compared patient characteristics and outcomes between those who did and those who did not experience MACE during hospitalization. Chi-square tests were used to compare categorical variables and Mann Whitney tests were used to compare continuous variables.

To create the perioperative risk-stratification model based on baseline, preoperative, and intraoperative data, we used a logistic regression with elastic net selection using a dichotomous dependent variable indicating MACE and independent variables described earlier. This perioperative model was fit on the training set and the model coefficients were then applied to the patients in the test set. The area under the receiver operating characteristic curve (AUC) was reported and the outcomes were reported by predicted risk decile, with higher deciles indicating higher risk (ie, higher numbers of patients with MACE outcomes in higher deciles implied better risk stratification). Because predicted risk of postoperative MACE may not have been distributed evenly across deciles, we also examined the distribution of the predicted probability of MACE and examined the number of patients below thresholds of risk corresponding to 0.1% or less, 0.25% or less, 0.5% or less, and 1% or less. These thresholds were chosen because they were close to the overall rate of MACE within our cohort.

We tested for differences in predictive performance between the RCRI logistic regression model AUC and the perioperative model AUC using DeLong’s test.²³ Additionally, we illustrated differences between the perioperative and RCRI models’ performance in two ways by stratifying patients into deciles based on predicted risk. First, we compared rates of MACE and MACE component events by predicted decile of the perioperative and RCRI models. Second, we further classified patients as RCRI high or low risk (per RCRI score classification in which RCRI score of 1 or greater is high risk and RCRI score of 0 is low risk) and examined numbers of surgical cases and MACE complications within these categories stratified by perioperative model predicted decile.

To compare the perioperative model’s performance with that of a risk factor–based guideline algorithm, we classified patients according to CCS guidelines as high risk (those for whom the CCS guidelines algorithm would recommend postoperative troponin surveillance testing) and low risk (those for whom the CCS guidelines algorithm would not recommend surveillance testing). We also used a logistic regression to examine if the predicted risk from our model was independently associated with MACE above and beyond the testing recommendation of the CCS guidelines algorithm. This model used MACE as the dependent variable and model-predicted risk and a CCS guidelines–defined high-risk indicator as predictors. We computed the association between a 10 percentage–point increase in predicted risk on observed MACE outcome rates.²⁴

In sensitivity analyses, we used a random forest machine learning classifier to test an alternate model specification, used complete case analysis, varied RCRI thresholds, and limited to patients aged 50 years or older. We also varied the penalty parameter in the elastic net model and plotted AUC versus the number of variables included to examine parsimonious models. SAS v9.4 (SAS Institute Inc) was used for main analyses. Data preparations and sensitivity analysis were done in Python v3.6 with Pandas v0.24.2 and Scikit-learn v0.19.1.

Study Sample

Patients who underwent major noncardiac surgery in our sample (n = 72,909) were approximately a mean age of 56 years, 58% female, 66% of White race and 26% of Black race, and most likely to have received orthopedic surgery (33%) or general surgery (20%). Those who experienced MACE (n = 558; 0.77%) differed along several characteristics (Table 1). For example, those with MACE were older (mean age, 65.4 vs 55.4 years; P < .001) and less likely to be female (41.9% vs 58.3%; P < .001).

Model Performance After Intraoperative Data Inclusion

In the perioperative model combining preoperative and intraoperative data, 26 variables were included after elastic net selection (Appendix Table 2). Model discrimination in the test set of patients demonstrated an AUC of 0.88 (95% CI, 0.85-0.92; Figure). When examining outcome rates by predicted decile, the outcome rates of in-hospital MACE complications were higher in the highest decile than in the lowest decile, notably with 58 of 92 (63%) cases with MACE complications within the top decile of predicted risk (Table 2). The majority of patients had low predicted risk of MACE, with 5,309 (36.1%), 8,796 (59.7%), 11,335 (77.0%), and 12,972 (88.1%) below the risk thresholds of to 0.1%, 0.25%, 0.5%, and 1.0% respectively. The associated MACE rates were 0.04%, 0.10%, 0.17%, and 0.25% (average rate in sample was 0.63%) (Appendix Table 3).

Model Performance Comparisons

The perioperative model AUC of 0.88 was higher when compared with RCRI’s AUC of 0.79 (95% CI, 0.74-0.84; P < .001). The number of MACE complications was more concentrated in the top decile of predicted risk of the perioperative model than it was in that of the RCRI model (58 vs 43 of 92 events, respectively; 63% vs 47%; Table 2). Furthermore, there were fewer cases with MACE complications in the low-risk deciles (ie, deciles 1 to 5) of the perioperative model than in the those of the RCRI model. These relative differences were consistent for MACE component outcomes of STEMI/NSTEMI, cardiac arrest, and in-hospital death, as well.

There was substantial heterogeneity in the perioperative model predicted risk of patients classified as either RCRI low risk or high risk (ie, each category included patients with low and high predicted risk) categories (Table 3). Patients in the bottom (low-risk) five deciles of the perioperative model’s predicted risk who were in the RCRI model’s high-risk group were very unlikely to experience MACE complications (3 out of 722 cases; 0.42%). Furthermore, among those classified as low risk by the RCRI model but were in the top decile of the perioperative model’s predicted risk, the MACE complication rate was 3.5% (8 out of 229), which was 6 times the sample mean MACE complication rate.

The perioperative model identified more patients as low risk than did the CCS guidelines’ risk factor–based algorithm (Table 3). For example, 2,341 of the patients the CCS guidelines algorithm identified as high risk were in the bottom 50% of the perioperative model’s predicted risk for experiencing MACE (below a 0.18% chance of a MACE complication); only four of these patients (0.17%) actually experienced MACE. This indicates that the 2,341 of 7,597 (31%) high-risk patients identified as low risk in the perioperative model would have been recommended for postoperative troponin testing by CCS guidelines based on preoperative risk factors alone—but did not go on to experience a MACE. Regression results indicated that both CCS guidelines risk-factor classification and the perioperative model’s predicted risk were predictive of MACE outcomes. A change in the perioperative model’s predicted risk of 10 percentage points was associated with an increase in the probability of a MACE outcomes of 0.45 percentage points (95% CI, 0.35-0.55 percentage points; P < .001) and moving from CCS guidelines’ low- to high-risk categories was associated with an increased probability of MACE by 0.96 percentage points (95% CI, 0.75-1.16 percentage points; P < .001).

Results were consistent with the main analysis across all sensitivity analyses (Appendix Tables 4-7). Parsimonious models with variables as few as eight variables retained strong predictive power (AUC, 0.870; Appendix Figure 1 and Table 8).

In this study, the addition of intraoperative data improved risk stratification for MACE complications when compared with standard risk tools such as RCRI. This approach also outperformed a guidelines-based approach and identified additional patients at low risk of cardiovascular complications. This study has three main implications.

First, this study demonstrated the additional value of combining intraoperative data with preoperative data in risk prediction for postoperative cardiovascular events. The intraoperative data most strongly associated with MACE, which likely were responsible for the performance improvement, included administration of medications (eg, sodium bicarbonate or calcium chloride) and blood products (eg, platelets and packed red blood cells), vitals (ie, heart rate), and intraoperative procedures (ie, arterial line placement); all model variables and coefficients are reported in Appendix Table 9. The risk-stratification model using intraoperative clinical data outperformed validated standard models such as RCRI. While this model should not be used in causal inference and cannot be used to inform decisions about risk-benefit tradeoffs of undergoing surgery, its improved performance relative to prior models highlights the potential in using real-time data. Preliminary illustrative analysis demonstrated that parsimonious models with as few as eight variables perform well, whose implementation as risk scores in EHRs is likely straightforward (Appendix Table 8). This is particularly important for longitudinal care in the hospital, in which patients frequently are cared for by multiple clinical services and experience handoffs. For example, many orthopedic surgery patients with significant medical comorbidity are managed postoperatively by hospitalist physicians after initial surgical care.

Second, our study aligns well with the cardiac risk-stratification literature more broadly. For example, the patient characteristics and clinical variables most associated with cardiovascular complications were age, history of ischemic heart disease, American Society of Anesthesiologists physical status, use of intraoperative sodium bicarbonate or vasopressors, lowest intraoperative heart rate measured, and lowest intraoperative mean arterial pressure measured. While many of these variables overlap with those included in the RCRI model, others (such as American Society of Anesthesiologists physical status) are not included in RCRI but have been shown to be important in risk prediction in other studies using different data variables.^6,25,26

Third, we illustrated a clinical application of this model in identifying patients at low risk of cardiovascular complications, although benefit may extend to other patients as well. This is particularly germane to clinicians who frequently manage patients in the postsurgical or postprocedural setting. Moreover, the clinical relevance to these clinicians is underscored by the lack of consensus among professional societies across Europe, Canada, and the United States about which subgroups of patients undergoing noncardiac surgery should receive postoperative cardiac biomarker surveillance testing in the 48 to 72 hours after surgery.^6-9 This may be in part caused by differences in clinical objectives. For example, the CCS guidelines in part aim to detect myocardial injury after noncardiac surgery (MINS) up to 30 days after surgery, which may be more sensitive to myocardial injury but less strongly associated with outcomes like MACE. The results of this study suggest that adopting such risk factor–based testing would likely lead to additional testing of low risk patients, which may represent low value surveillance tests. For example, there were 2,257 patients without postoperative cardiac biomarker testing in our data who would have been categorized as high risk by risk factor guidelines and therefore recommended to receive at least one postoperative cardiac biomarker surveillance test but were classified as low-risk individuals using a predicted probability of MACE less than 0.18% per our perioperative risk stratification model (Appendix Table 4). If each of these patients received one troponin biomarker test, the associated cost increase would be $372,405 (using the $165 cost per test reported at our institution). These costs would multiply if daily surveillance troponin biomarker tests were ordered for 48 to 72 hours after surgery, as recommended by the risk factor–based testing guidelines. This would be a departure from testing among patients using clinician discretion that may avoid low-value testing.

Applying the perioperative model developed in this paper to clinical practice still requires several steps. The technical aspects of finding a parsimonious model that can be implemented in the EHR is likely quite straightforward. Our preliminary analysis illustrates that doing so will not require accessing large numbers of intraoperative variables. Perhaps more important steps include prospective validation of the safety, usability, and clinical benefit of such an algorithm-based risk score.²⁷

The study has several limitations. First, it was an observational study using EHR data subject to missingness and data quality issues that may have persisted despite our methods. Furthermore, EHR data is not generated randomly, and unmeasured variables observed by clinicians but not by researchers could confound the results. However, our approach used the statistical model to examine risk, not causal inference. Second, this is a single institution study and the availability of EHR data, as well as practice patterns, may vary at other institutions. Furthermore, it is possible that performance of the RCRI score, the model fitting RCRI classification of high vs low risk on the sample data, and our model’s performance may not generalize to other clinical settings. However, we utilized data from multiple hospitals within a health system with different surgery and anesthesia groups and providers, and a similar AUC was reported for RCRI in original validation study.⁶ Third, our follow up period was limited to the hospital setting and we do not capture longitudinal outcomes, such as 30-day MACE. This may impact the ability to risk stratify for other important longer-term outcomes, limit clinical utility, and hinder comparability to other studies. Fourth, results may vary for other important cardiovascular outcomes that may be more sensitive to myocardial injury, such as MINS. Fifth, we used a limited number of modeling strategies.

Addition of intraoperative data to preoperative data improves prediction of cardiovascular complications after noncardiac surgery. Improving the identification of patients at low risk for such complications could potentially be applied to reduce unnecessary postoperative cardiac biomarker testing after noncardiac surgery, but it will require further validation in prospective clinical settings.

Dr Navathe reports grants from the following entities: Hawaii Medical Service Association, Anthem Public Policy Institute, Commonwealth Fund, Oscar Health, Cigna Corporation, Robert Wood Johnson Foundation, Donaghue Foundation, Pennsylvania Department of Health, Ochsner Health System, United Healthcare, Blue Cross Blue Shield of NC, Blue Shield of CA; personal fees from the following: Navvis Healthcare, Agathos, Inc, Navahealth, YNHHSC/CORE, Maine Health Accountable Care Organization, Maine Department of Health and Human Services, National University Health System - Singapore, Ministry of Health - Singapore, Social Security Administration - France, Elsevier Press, Medicare Payment Advisory Commission, Cleveland Clinic, Embedded Healthcare; and other support from Integrated Services, Inc, outside of the submitted work. Dr Volpp reports grants from Humana during the conduct of the study; grants from Hawaii Medical Services Agency, Discovery (South Africa), Merck, Weight Watchers, and CVS outside of the submitted work; he has received consulting income from CVS and VALHealth and is a principal in VALHealth, a behavioral economics consulting firm. Dr Holmes receives funding from the Pennsylvania Department of Health, US Public Health Service, and the Cardiovascular Medicine Research and Education Foundation. All other authors declare no conflicts of interest.

2019 Academy Health Annual Research Meeting, Poster Abstract Presentation, June 2 to June 4, 2019, Washington, DC.

This project was funded, in part, under a grant with the Pennsylvania Department of Health. This research was independent from the funder. The funder had no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the article for publication. The department specifically disclaims responsibility for any analyses, interpretations, or conclusions.

In recent years, mounting evidence has emerged questioning the practice of using prolonged intravenous antibiotic therapy to treat certain serious bacterial infections in children, including complicated appendicitis, osteomyelitis, and complicated pneumonia. Historically, treatment of these conditions was often completed intravenously after hospital discharge using peripherally inserted central catheters (PICCs). Line infections, clots, mechanical problems, and general discomfort complicate PICCs, which led to their removal in more than 20% of children in one study.¹ Oral antibiotics avoid these complications and are less burdensome to families.² Recently, a series of multicenter studies showed no difference in outcomes between oral and postdischarge intravenous antibiotic therapy (PD-IV) for complicated appendicitis, osteomyelitis, and complicated pneumonia.^3-5

Despite a growing body of evidence suggesting that oral therapy ought to be the default treatment strategy rather than PD-IV, the extent to which practices have changed is unknown. In this study, we measured national trends in PD-IV use and variation by hospital for complicated appendicitis, osteomyelitis, and complicated pneumonia.

We performed a retrospective cohort study of children discharged from hospitals that contributed data to the Pediatric Health Information System (PHIS) database from January 2000 through December 2018. PHIS is an administrative database of children’s hospitals managed by the Children’s Hospital Association (Lenexa, Kansas) and contains deidentified patient-­level demographic data, discharge diagnosis and procedure codes, and detailed billing information, including medical supply charges.

The cohorts were defined using International Classification of Diseases, 9th and 10th Revisions (ICD-9 and ICD-10) discharge diagnosis and procedure codes. Patients admitted through September 2015 were identified using ICD-9 codes and patients admitted from October 2015 through December 2018 were identified using ICD-10 codes. The Centers for Medicaid & Medicare Services crosswalk was used to align ICD-9 and ICD-10 codes.⁶ Inclusion and exclusion criteria identifying cohorts of children hospitalized for complicated appendicitis, osteomyelitis, or complicated pneumonia were based on prior studies using the PHIS database.^3-5 These studies augmented the PHIS administrative dataset with local chart review to identify patients from 2009-2012 with the following inclusion and exclusion criteria: Patients with complicated appendicitis were defined by a diagnosis code for acute appendicitis and a procedure code for appendectomy, with postoperative length of stay lasting between 3 and 7 days. Patients with osteomyelitis had a diagnosis code of acute or unspecified osteomyelitis with a hospital length of stay between 2 and 14 days. Patients with complicated pneumonia were defined by a diagnosis code for both pneumonia and pleural effusion with one of these as the primary diagnosis. Patients were excluded if they were older than 18 years or if they were younger than 2 months for osteomyelitis and complicated pneumonia or younger than 3 years for appendicitis. For all three conditions, children with a complex chronic condition⁷ were excluded. Only the index encounter meeting inclusion and exclusion criteria for each patient was included. PD-IV therapy was defined using procedure codes and hospital charges during the index hospitalization. This definition for PD-IV therapy has been validated among children with complicated pneumonia, demonstrating positive and negative predictive values for PICC exposure of 85% and 99%, respectively.⁸

Trends in the percentage of patients receiving PD-IV were adjusted for age, race, insurance type, intensive care unit days, and hospital-level case mix index with use of Poisson regression. Calculated risk ratios represent the change in PD-IV across the entire 19-year study period for each condition (as opposed to an annual rate of change). An inflection point for each condition was identified using piecewise linear regression in which the line slope has one value up to a point in time and a second value after that point. The transition point is determined by maximizing model fit.

Some hospitals were added to the database throughout the time period and therefore did not have data for all years of the study. To account for the possibility of a group of high– or low–PD-IV use hospitals entering the cohort and biasing the overall trend, we performed a sensitivity analysis restricted to hospitals continuously contributing data to PHIS every year between 2004 (when a majority of hospitals joined PHIS) and 2018. Significance testing for individual hospital trends was conducted among continuously contributing hospitals, with each hospital tested in the above Poisson model independently.

For the most recent year of 2018, we reported the distribution of adjusted percentages of PD-IV at the individual hospital level. Only hospitals with at least five patients for a given condition are included in the percent PD-IV calculations for 2018. To examine the extent to which an individual hospital might be a low– or high–PD-IV user across conditions, we divided hospitals into quartiles based on PD-IV use for each condition in 2017-2018 and calculated the percent of hospitals in the lowest- and highest-use quartiles for all three conditions. All statistics were performed using Stata 15 (StataCorp).

Among 52 hospitals over a 19-year study period, there were 60,575 hospitalizations for complicated appendicitis, 24,753 hospitalizations for osteomyelitis, and 13,700 hospitalizations for complicated pneumonia. From 2000 to 2018, PD-IV decreased from 13% to 2% (RR, 0.15; 95% CI, 0.14-0.16) for complicated appendicitis, from 61% to 22% (RR, 0.41; 95% CI, 0.39-0.43) for osteomyelitis, and from 29% to 19% (RR, 0.63; 95% CI, 0.58-0.69) for complicated pneumonia (Figure 1). The inflection points occurred in 2009 for complicated appendicitis, 2009 for complicated pneumonia, and 2010 for osteomyelitis. The sensitivity analysis included 31 hospitals that contributed data to PHIS for every year between 2004-2018 and revealed similar findings for all three conditions: Complicated appendicitis had an RR of 0.15 (95% CI, 0.14-0.17), osteomyelitis had an RR of 0.34 (95% CI, 0.32-0.36), and complicated pneumonia had an RR of 0.55 (95% CI, 0.49-0.61). Most individual hospitals decreased PD-IV use (complicated appendicitis: 21 decreased, 8 no change, 2 increased; osteomyelitis: 25 decreased, 6 no change; complicated pneumonia: 14 decreased, 16 no change, 1 increased). While overall decreases in PD-IV were observed for all three conditions, considerable variation remained in 2018 for use of PD-IV (Figure 2), particularly for osteomyelitis (median, 18%; interquartile range [IQR] 9%-40%) and complicated pneumonia (median, 13%; IQR, 3%-30%). In 2017-2018, 1 out of 52 hospitals was in the lowest PD-IV–use quartile for all three conditions, and three hospitals were in the highest-use quartile for all three conditions.

Over a 19-year period, we observed a national decline in use of PD-IV for three serious and common bacterial infections. The decline in PD-IV is notable given that it has occurred largely in the absence of nationally coordinated guidelines or improvement efforts. Despite the overall declines, substantial variation in the use of PD-IV for these conditions persists across children’s hospitals.

The observed decrease in PD-IV use is a natural example of deimplementation, the abandonment of medical practices found to be harmful or ineffective.⁹ What is most compelling about the deimplementation of PD-IV for these infectious conditions is the seemingly organic motivation that propelled it. Studies of physician practice patterns for interventions that have undergone evidence reversals demonstrate that physicians might readily implement new interventions with an early evidence base but be less willing to deimplement them when more definitive evidence later questions their efficacy.¹⁰ Therefore, concerted improvement efforts backed by national guidelines are often needed to reduce the use of a widely accepted medical practice. For example, as evidence questioning the efficacy of steroid use in bronchiolitis mounted,¹¹ bronchiolitis guidelines recommended against steroid use¹² and a national quality improvement effort led to reductions in exposure to steroids among patients hospitalized with bronchiolitis.¹³ Complicated intra-abdominal infection guidelines acknowledge oral antibiotic therapy as an option,¹⁴ but no such national guidelines or improvement projects exist for osteomyelitis or complicated pneumonia PD-IV.

What is it about PD-IV for complicated appendicitis, osteomyelitis, and complicated pneumonia that fostered the observed organic deimplementation? Our findings that few hospitals were in the top or bottom quartile of PD-IV across all three conditions suggest that the impetus to decrease PD-IV was not likely the product of a broad hospital-wide practice shift. Most deimplementation frameworks suggest that successful deimplementation must be supported by high-quality evidence that the intervention is not only ineffective, but also harmful.¹⁵ In this case, the inflection point for osteomyelitis occurred in 2009, the same year that the first large multicenter study suggesting efficacy and decreased complications of early oral therapy for osteomyelitis was published.¹⁶ A direct link between a publication and inflection points for complicated pneumonia and appendicitis is less clear. It is possible that growth of the field of pediatric hospital medicine,¹⁷ with a stated emphasis on healthcare value,¹⁸ played a role. Greater understanding of the drivers and barriers to deimplementation in this and similar contexts will be important.

Our study has some important limitations. While inclusion and exclusion criteria were consistent over the study period, practice patterns (ie, length of stay in uncomplicated patients) change and could alter the case-mix of patients over time. Additionally, the PHIS database largely comprises children’s hospitals, and the trends we observed in PD-IV may not generalize to community settings.

The degree of deimplementation of PD-IV observed across children’s hospitals is impressive, but opportunity for further improvement likely remains. We found that marked hospital-­level variation in use of PD-IV still exists, with some hospitals almost never using PD-IV and others using it for most patients. While the ideal amount of PD-IV is probably not zero, a portion of the observed variation likely represents overuse of PD-IV. To reduce costs and complications associated with antibiotic therapy, national guidelines and a targeted national improvement collaborative may be necessary to achieve further reductions in PD-IV.

In recent years, mounting evidence has emerged questioning the practice of using prolonged intravenous antibiotic therapy to treat certain serious bacterial infections in children, including complicated appendicitis, osteomyelitis, and complicated pneumonia. Historically, treatment of these conditions was often completed intravenously after hospital discharge using peripherally inserted central catheters (PICCs). Line infections, clots, mechanical problems, and general discomfort complicate PICCs, which led to their removal in more than 20% of children in one study.¹ Oral antibiotics avoid these complications and are less burdensome to families.² Recently, a series of multicenter studies showed no difference in outcomes between oral and postdischarge intravenous antibiotic therapy (PD-IV) for complicated appendicitis, osteomyelitis, and complicated pneumonia.^3-5

Despite a growing body of evidence suggesting that oral therapy ought to be the default treatment strategy rather than PD-IV, the extent to which practices have changed is unknown. In this study, we measured national trends in PD-IV use and variation by hospital for complicated appendicitis, osteomyelitis, and complicated pneumonia.

We performed a retrospective cohort study of children discharged from hospitals that contributed data to the Pediatric Health Information System (PHIS) database from January 2000 through December 2018. PHIS is an administrative database of children’s hospitals managed by the Children’s Hospital Association (Lenexa, Kansas) and contains deidentified patient-­level demographic data, discharge diagnosis and procedure codes, and detailed billing information, including medical supply charges.

The cohorts were defined using International Classification of Diseases, 9th and 10th Revisions (ICD-9 and ICD-10) discharge diagnosis and procedure codes. Patients admitted through September 2015 were identified using ICD-9 codes and patients admitted from October 2015 through December 2018 were identified using ICD-10 codes. The Centers for Medicaid & Medicare Services crosswalk was used to align ICD-9 and ICD-10 codes.⁶ Inclusion and exclusion criteria identifying cohorts of children hospitalized for complicated appendicitis, osteomyelitis, or complicated pneumonia were based on prior studies using the PHIS database.^3-5 These studies augmented the PHIS administrative dataset with local chart review to identify patients from 2009-2012 with the following inclusion and exclusion criteria: Patients with complicated appendicitis were defined by a diagnosis code for acute appendicitis and a procedure code for appendectomy, with postoperative length of stay lasting between 3 and 7 days. Patients with osteomyelitis had a diagnosis code of acute or unspecified osteomyelitis with a hospital length of stay between 2 and 14 days. Patients with complicated pneumonia were defined by a diagnosis code for both pneumonia and pleural effusion with one of these as the primary diagnosis. Patients were excluded if they were older than 18 years or if they were younger than 2 months for osteomyelitis and complicated pneumonia or younger than 3 years for appendicitis. For all three conditions, children with a complex chronic condition⁷ were excluded. Only the index encounter meeting inclusion and exclusion criteria for each patient was included. PD-IV therapy was defined using procedure codes and hospital charges during the index hospitalization. This definition for PD-IV therapy has been validated among children with complicated pneumonia, demonstrating positive and negative predictive values for PICC exposure of 85% and 99%, respectively.⁸

Trends in the percentage of patients receiving PD-IV were adjusted for age, race, insurance type, intensive care unit days, and hospital-level case mix index with use of Poisson regression. Calculated risk ratios represent the change in PD-IV across the entire 19-year study period for each condition (as opposed to an annual rate of change). An inflection point for each condition was identified using piecewise linear regression in which the line slope has one value up to a point in time and a second value after that point. The transition point is determined by maximizing model fit.

Some hospitals were added to the database throughout the time period and therefore did not have data for all years of the study. To account for the possibility of a group of high– or low–PD-IV use hospitals entering the cohort and biasing the overall trend, we performed a sensitivity analysis restricted to hospitals continuously contributing data to PHIS every year between 2004 (when a majority of hospitals joined PHIS) and 2018. Significance testing for individual hospital trends was conducted among continuously contributing hospitals, with each hospital tested in the above Poisson model independently.

For the most recent year of 2018, we reported the distribution of adjusted percentages of PD-IV at the individual hospital level. Only hospitals with at least five patients for a given condition are included in the percent PD-IV calculations for 2018. To examine the extent to which an individual hospital might be a low– or high–PD-IV user across conditions, we divided hospitals into quartiles based on PD-IV use for each condition in 2017-2018 and calculated the percent of hospitals in the lowest- and highest-use quartiles for all three conditions. All statistics were performed using Stata 15 (StataCorp).

Among 52 hospitals over a 19-year study period, there were 60,575 hospitalizations for complicated appendicitis, 24,753 hospitalizations for osteomyelitis, and 13,700 hospitalizations for complicated pneumonia. From 2000 to 2018, PD-IV decreased from 13% to 2% (RR, 0.15; 95% CI, 0.14-0.16) for complicated appendicitis, from 61% to 22% (RR, 0.41; 95% CI, 0.39-0.43) for osteomyelitis, and from 29% to 19% (RR, 0.63; 95% CI, 0.58-0.69) for complicated pneumonia (Figure 1). The inflection points occurred in 2009 for complicated appendicitis, 2009 for complicated pneumonia, and 2010 for osteomyelitis. The sensitivity analysis included 31 hospitals that contributed data to PHIS for every year between 2004-2018 and revealed similar findings for all three conditions: Complicated appendicitis had an RR of 0.15 (95% CI, 0.14-0.17), osteomyelitis had an RR of 0.34 (95% CI, 0.32-0.36), and complicated pneumonia had an RR of 0.55 (95% CI, 0.49-0.61). Most individual hospitals decreased PD-IV use (complicated appendicitis: 21 decreased, 8 no change, 2 increased; osteomyelitis: 25 decreased, 6 no change; complicated pneumonia: 14 decreased, 16 no change, 1 increased). While overall decreases in PD-IV were observed for all three conditions, considerable variation remained in 2018 for use of PD-IV (Figure 2), particularly for osteomyelitis (median, 18%; interquartile range [IQR] 9%-40%) and complicated pneumonia (median, 13%; IQR, 3%-30%). In 2017-2018, 1 out of 52 hospitals was in the lowest PD-IV–use quartile for all three conditions, and three hospitals were in the highest-use quartile for all three conditions.

Over a 19-year period, we observed a national decline in use of PD-IV for three serious and common bacterial infections. The decline in PD-IV is notable given that it has occurred largely in the absence of nationally coordinated guidelines or improvement efforts. Despite the overall declines, substantial variation in the use of PD-IV for these conditions persists across children’s hospitals.

The observed decrease in PD-IV use is a natural example of deimplementation, the abandonment of medical practices found to be harmful or ineffective.⁹ What is most compelling about the deimplementation of PD-IV for these infectious conditions is the seemingly organic motivation that propelled it. Studies of physician practice patterns for interventions that have undergone evidence reversals demonstrate that physicians might readily implement new interventions with an early evidence base but be less willing to deimplement them when more definitive evidence later questions their efficacy.¹⁰ Therefore, concerted improvement efforts backed by national guidelines are often needed to reduce the use of a widely accepted medical practice. For example, as evidence questioning the efficacy of steroid use in bronchiolitis mounted,¹¹ bronchiolitis guidelines recommended against steroid use¹² and a national quality improvement effort led to reductions in exposure to steroids among patients hospitalized with bronchiolitis.¹³ Complicated intra-abdominal infection guidelines acknowledge oral antibiotic therapy as an option,¹⁴ but no such national guidelines or improvement projects exist for osteomyelitis or complicated pneumonia PD-IV.

What is it about PD-IV for complicated appendicitis, osteomyelitis, and complicated pneumonia that fostered the observed organic deimplementation? Our findings that few hospitals were in the top or bottom quartile of PD-IV across all three conditions suggest that the impetus to decrease PD-IV was not likely the product of a broad hospital-wide practice shift. Most deimplementation frameworks suggest that successful deimplementation must be supported by high-quality evidence that the intervention is not only ineffective, but also harmful.¹⁵ In this case, the inflection point for osteomyelitis occurred in 2009, the same year that the first large multicenter study suggesting efficacy and decreased complications of early oral therapy for osteomyelitis was published.¹⁶ A direct link between a publication and inflection points for complicated pneumonia and appendicitis is less clear. It is possible that growth of the field of pediatric hospital medicine,¹⁷ with a stated emphasis on healthcare value,¹⁸ played a role. Greater understanding of the drivers and barriers to deimplementation in this and similar contexts will be important.

Our study has some important limitations. While inclusion and exclusion criteria were consistent over the study period, practice patterns (ie, length of stay in uncomplicated patients) change and could alter the case-mix of patients over time. Additionally, the PHIS database largely comprises children’s hospitals, and the trends we observed in PD-IV may not generalize to community settings.

The degree of deimplementation of PD-IV observed across children’s hospitals is impressive, but opportunity for further improvement likely remains. We found that marked hospital-­level variation in use of PD-IV still exists, with some hospitals almost never using PD-IV and others using it for most patients. While the ideal amount of PD-IV is probably not zero, a portion of the observed variation likely represents overuse of PD-IV. To reduce costs and complications associated with antibiotic therapy, national guidelines and a targeted national improvement collaborative may be necessary to achieve further reductions in PD-IV.

In recent years, mounting evidence has emerged questioning the practice of using prolonged intravenous antibiotic therapy to treat certain serious bacterial infections in children, including complicated appendicitis, osteomyelitis, and complicated pneumonia. Historically, treatment of these conditions was often completed intravenously after hospital discharge using peripherally inserted central catheters (PICCs). Line infections, clots, mechanical problems, and general discomfort complicate PICCs, which led to their removal in more than 20% of children in one study.¹ Oral antibiotics avoid these complications and are less burdensome to families.² Recently, a series of multicenter studies showed no difference in outcomes between oral and postdischarge intravenous antibiotic therapy (PD-IV) for complicated appendicitis, osteomyelitis, and complicated pneumonia.^3-5

Despite a growing body of evidence suggesting that oral therapy ought to be the default treatment strategy rather than PD-IV, the extent to which practices have changed is unknown. In this study, we measured national trends in PD-IV use and variation by hospital for complicated appendicitis, osteomyelitis, and complicated pneumonia.

We performed a retrospective cohort study of children discharged from hospitals that contributed data to the Pediatric Health Information System (PHIS) database from January 2000 through December 2018. PHIS is an administrative database of children’s hospitals managed by the Children’s Hospital Association (Lenexa, Kansas) and contains deidentified patient-­level demographic data, discharge diagnosis and procedure codes, and detailed billing information, including medical supply charges.

The cohorts were defined using International Classification of Diseases, 9th and 10th Revisions (ICD-9 and ICD-10) discharge diagnosis and procedure codes. Patients admitted through September 2015 were identified using ICD-9 codes and patients admitted from October 2015 through December 2018 were identified using ICD-10 codes. The Centers for Medicaid & Medicare Services crosswalk was used to align ICD-9 and ICD-10 codes.⁶ Inclusion and exclusion criteria identifying cohorts of children hospitalized for complicated appendicitis, osteomyelitis, or complicated pneumonia were based on prior studies using the PHIS database.^3-5 These studies augmented the PHIS administrative dataset with local chart review to identify patients from 2009-2012 with the following inclusion and exclusion criteria: Patients with complicated appendicitis were defined by a diagnosis code for acute appendicitis and a procedure code for appendectomy, with postoperative length of stay lasting between 3 and 7 days. Patients with osteomyelitis had a diagnosis code of acute or unspecified osteomyelitis with a hospital length of stay between 2 and 14 days. Patients with complicated pneumonia were defined by a diagnosis code for both pneumonia and pleural effusion with one of these as the primary diagnosis. Patients were excluded if they were older than 18 years or if they were younger than 2 months for osteomyelitis and complicated pneumonia or younger than 3 years for appendicitis. For all three conditions, children with a complex chronic condition⁷ were excluded. Only the index encounter meeting inclusion and exclusion criteria for each patient was included. PD-IV therapy was defined using procedure codes and hospital charges during the index hospitalization. This definition for PD-IV therapy has been validated among children with complicated pneumonia, demonstrating positive and negative predictive values for PICC exposure of 85% and 99%, respectively.⁸

Trends in the percentage of patients receiving PD-IV were adjusted for age, race, insurance type, intensive care unit days, and hospital-level case mix index with use of Poisson regression. Calculated risk ratios represent the change in PD-IV across the entire 19-year study period for each condition (as opposed to an annual rate of change). An inflection point for each condition was identified using piecewise linear regression in which the line slope has one value up to a point in time and a second value after that point. The transition point is determined by maximizing model fit.

Some hospitals were added to the database throughout the time period and therefore did not have data for all years of the study. To account for the possibility of a group of high– or low–PD-IV use hospitals entering the cohort and biasing the overall trend, we performed a sensitivity analysis restricted to hospitals continuously contributing data to PHIS every year between 2004 (when a majority of hospitals joined PHIS) and 2018. Significance testing for individual hospital trends was conducted among continuously contributing hospitals, with each hospital tested in the above Poisson model independently.

For the most recent year of 2018, we reported the distribution of adjusted percentages of PD-IV at the individual hospital level. Only hospitals with at least five patients for a given condition are included in the percent PD-IV calculations for 2018. To examine the extent to which an individual hospital might be a low– or high–PD-IV user across conditions, we divided hospitals into quartiles based on PD-IV use for each condition in 2017-2018 and calculated the percent of hospitals in the lowest- and highest-use quartiles for all three conditions. All statistics were performed using Stata 15 (StataCorp).

Among 52 hospitals over a 19-year study period, there were 60,575 hospitalizations for complicated appendicitis, 24,753 hospitalizations for osteomyelitis, and 13,700 hospitalizations for complicated pneumonia. From 2000 to 2018, PD-IV decreased from 13% to 2% (RR, 0.15; 95% CI, 0.14-0.16) for complicated appendicitis, from 61% to 22% (RR, 0.41; 95% CI, 0.39-0.43) for osteomyelitis, and from 29% to 19% (RR, 0.63; 95% CI, 0.58-0.69) for complicated pneumonia (Figure 1). The inflection points occurred in 2009 for complicated appendicitis, 2009 for complicated pneumonia, and 2010 for osteomyelitis. The sensitivity analysis included 31 hospitals that contributed data to PHIS for every year between 2004-2018 and revealed similar findings for all three conditions: Complicated appendicitis had an RR of 0.15 (95% CI, 0.14-0.17), osteomyelitis had an RR of 0.34 (95% CI, 0.32-0.36), and complicated pneumonia had an RR of 0.55 (95% CI, 0.49-0.61). Most individual hospitals decreased PD-IV use (complicated appendicitis: 21 decreased, 8 no change, 2 increased; osteomyelitis: 25 decreased, 6 no change; complicated pneumonia: 14 decreased, 16 no change, 1 increased). While overall decreases in PD-IV were observed for all three conditions, considerable variation remained in 2018 for use of PD-IV (Figure 2), particularly for osteomyelitis (median, 18%; interquartile range [IQR] 9%-40%) and complicated pneumonia (median, 13%; IQR, 3%-30%). In 2017-2018, 1 out of 52 hospitals was in the lowest PD-IV–use quartile for all three conditions, and three hospitals were in the highest-use quartile for all three conditions.

Over a 19-year period, we observed a national decline in use of PD-IV for three serious and common bacterial infections. The decline in PD-IV is notable given that it has occurred largely in the absence of nationally coordinated guidelines or improvement efforts. Despite the overall declines, substantial variation in the use of PD-IV for these conditions persists across children’s hospitals.

The observed decrease in PD-IV use is a natural example of deimplementation, the abandonment of medical practices found to be harmful or ineffective.⁹ What is most compelling about the deimplementation of PD-IV for these infectious conditions is the seemingly organic motivation that propelled it. Studies of physician practice patterns for interventions that have undergone evidence reversals demonstrate that physicians might readily implement new interventions with an early evidence base but be less willing to deimplement them when more definitive evidence later questions their efficacy.¹⁰ Therefore, concerted improvement efforts backed by national guidelines are often needed to reduce the use of a widely accepted medical practice. For example, as evidence questioning the efficacy of steroid use in bronchiolitis mounted,¹¹ bronchiolitis guidelines recommended against steroid use¹² and a national quality improvement effort led to reductions in exposure to steroids among patients hospitalized with bronchiolitis.¹³ Complicated intra-abdominal infection guidelines acknowledge oral antibiotic therapy as an option,¹⁴ but no such national guidelines or improvement projects exist for osteomyelitis or complicated pneumonia PD-IV.

What is it about PD-IV for complicated appendicitis, osteomyelitis, and complicated pneumonia that fostered the observed organic deimplementation? Our findings that few hospitals were in the top or bottom quartile of PD-IV across all three conditions suggest that the impetus to decrease PD-IV was not likely the product of a broad hospital-wide practice shift. Most deimplementation frameworks suggest that successful deimplementation must be supported by high-quality evidence that the intervention is not only ineffective, but also harmful.¹⁵ In this case, the inflection point for osteomyelitis occurred in 2009, the same year that the first large multicenter study suggesting efficacy and decreased complications of early oral therapy for osteomyelitis was published.¹⁶ A direct link between a publication and inflection points for complicated pneumonia and appendicitis is less clear. It is possible that growth of the field of pediatric hospital medicine,¹⁷ with a stated emphasis on healthcare value,¹⁸ played a role. Greater understanding of the drivers and barriers to deimplementation in this and similar contexts will be important.

Our study has some important limitations. While inclusion and exclusion criteria were consistent over the study period, practice patterns (ie, length of stay in uncomplicated patients) change and could alter the case-mix of patients over time. Additionally, the PHIS database largely comprises children’s hospitals, and the trends we observed in PD-IV may not generalize to community settings.

The degree of deimplementation of PD-IV observed across children’s hospitals is impressive, but opportunity for further improvement likely remains. We found that marked hospital-­level variation in use of PD-IV still exists, with some hospitals almost never using PD-IV and others using it for most patients. While the ideal amount of PD-IV is probably not zero, a portion of the observed variation likely represents overuse of PD-IV. To reduce costs and complications associated with antibiotic therapy, national guidelines and a targeted national improvement collaborative may be necessary to achieve further reductions in PD-IV.

Healthcare quality is defined as the extent to which healthcare services result in desired outcomes.¹ Quality of care depends on how the healthcare system’s various components, including healthcare practitioners, interact to meet each patient’s needs.² These components can be shaped to achieve desired outcomes through rules, incentives, and other approaches, but influencing the behaviors of each component, such as the performance of hospitalists, requires defining goals for performance and implementing measurement approaches to assess progress toward these goals.

One set of principles to define goals for quality and guide assessment of desired behaviors is the multidimensional STEEEP framework. This framework, created by the Institute of Medicine, identifies six domains of quality: Safe, Timely, Effective, Efficient, Equitable, and Patient Centered.² Briefly, “Safe” means avoiding injuries to patients, “Timely” means reducing waits and delays in care, “Effective” means providing care based on evidence, “Efficient” means avoiding waste, “Equitable” means ensuring quality does not vary based on personal characteristics such as race and gender, and “Patient Centered” means providing care that is responsive to patients’ values and preferences. The STEEEP domains are not coequal; rather, they ensure that quality is considered broadly, while avoiding errors such as measuring only an intervention’s impact on effectiveness but not assessing its impact on multiple domains of quality, such as how patient centered, efficient (cost effective), or equitable the resulting care is.

Based on our review of the literature, a multidimensional framework like STEEEP has not been used in defining and assessing the quality of individual hospitalists’ performance. Some quality metrics at the hospital level impact several dimensions simultaneously, such as door to balloon time for acute myocardial infarction, which measures effectiveness and timeliness of care. Programs like pay-for-performance, Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS), and the Merit-Based Incentive Payment System (MIPS) have tied reimbursement to assessments aligned with several STEEEP domains at both individual and institutional levels but lack a holistic approach to quality.^3-6 The every-­other-year State of Hospital Medicine Report, the most widely used description of individual hospitalist performance, reports group-level performance including relative value units and whether groups are accountable for measures of quality such as performance on core measures, timely documentation, and “citizenship” (eg, committee participation or academic work).⁷ While these are useful benchmarks, the report focuses on performance at the group level. Concurrently, several academic groups have described more complete dashboards or scorecards to assess individual hospitalist performance, primarily designed to facilitate comparison across hospitalist groups or to incentivize overall group performance.^8-10 However, these efforts are not guided by an overarching framework and are structured after traditional academic models with components related to teaching and scholarship, which may not translate to nonacademic environments. Finally, the Core Competencies for Hospital Medicine outlines some goals for hospitalist performance but does not speak to specific measurement approaches.¹¹

Overall, assessing individual hospitalist performance is hindered by lack of consensus on important concepts to measure, a limited number of valid measures, and challenges in data collection such as resource limitations and feasibility. Developing and refining measures grounded in the STEEEP framework may provide a more comprehensive assessment of hospitalist quality and identify approaches to improve overall health outcomes. Comparative data could help individual hospitalists improve performance; leaders of hospitalist groups could use this data to guide faculty development and advancement as they ensure quality care at the individual, group, and system levels.

To better inform quality measurement of individual hospitalists, we sought to identify existing publications on individual hospitalist quality. Our goal was to define the published literature about quality measurement at the individual hospitalist level, relate these publications to domains of quality defined by the STEEEP framework, and identify directions for assessment or further research that could affect the overall quality of care.

We conducted a scoping review following methods outlined by Arksey and O’Malley¹² and Tricco.¹³ The goal of a scoping review is to map the extent of research within a specific field. This methodology is well suited to characterizing the existing research related to the quality of hospitalist care at the individual level. A protocol for the scoping review was not registered.

Evidence Search

A systematic search for published, English-language literature on hospitalist care was conducted in Medline (Ovid; 1946 - June 4, 2019) on June 5, 2019. The search used a combination of keywords and controlled vocabulary for the concept of hospitalists or hospital medicine. The search strategy used in this review is described in the Appendix. In addition, a hand search of reference lists of articles was used to discover publications not identified in the database searches.

Study Selection

All references were uploaded to Covidence systematic review software (www.covidence.org; Covidence), and duplicates were removed. Four reviewers (A.D., B.C., L.H., R.Q.) conducted title and abstract, as well as full-text, review to identify studies that measured differences in the performance of hospitalists at the individual level. Any disagreements among reviewers were resolved by consensus. Articles included both adult and pediatric populations. Articles that focused on group-level outcomes could be included if nonpooled data at the individual level was also reported. Studies were excluded if they did not focus on individual quality of care indicators or were not published in English.

Data Charting and Synthesis

We extracted the following information using a standardized data collection form: author, title, year of publication, study design, intervention, and outcome measures. Original manuscripts were accessed as needed to supplement analysis. Critical appraisal of individual studies was not conducted in this review because the goal of this review was to analyze which quality indicators have been studied and how they were measured. Articles were then coded for their alignment to the STEEEP framework by two reviewers (AD and BC). After initial coding was conducted, the reviewers met to consolidate codes and resolve any disagreement by consensus. The results of the analysis were summarized in both text and tabular format with studies grouped by focus of assessment with each one’s methods of assessment listed.

Results of the search strategy are shown in the Figure. The search retrieved a total of 2,363 references of which 113 were duplicates, leaving 2,250 to be screened. After title and abstract and full-text screening, 42 studies were included in the review. The final 42 studies were coded for alignment with the STEEEP framework. The Table displays the focus of assessment and methods of assessment within each STEEEP domain.

Eighteen studies were coded into a single domain while the rest were coded into at least two domains. The domain Patient Centered was coded as having the most studies (n = 23), followed by the domain of Safe (n = 15). Timely, Effective, and Efficient domains had 11, 9, and 12 studies, respectively. No studies were coded into the domain of Equitable.

Safe

Nearly all studies coded into the Safe domain focused on transitions of care. These included transfers into a hospital from other hospitals,¹⁴ transitions of care to cross-covering providers^15,16 and new primary providers,¹⁷ and transition out from the acute care setting.^18-28 Measures of hospital discharge included measures of both processes^18-22 and outcomes.^23-27 Methods of assessment varied from use of trained observers or scorers to surveys of individuals and colleagues about performance. Though a few leveraged informatics,^22,27 all approaches relied on human interaction, and none were automated.

Timely

All studies coded into the Timely domain were coded into at least one other domain. For example, Anderson et al looked at how hospitalists communicated about potential life-limiting illness at the time of hospital admission and the subsequent effects on plans of care²⁹; this was coded as both Timely and Patient Centered. Likewise, another group of studies centered on application of evidence-based guidelines, such as giving antibiotics within a certain time interval for sepsis and were coded as both Timely and Effective. Another set of authors described dashboards or scorecards that captured a number of group-level metrics of processes of care that span STEEEP domains and may be applicable to individuals, including Fox et al for pediatrics⁸ and Hwa et al for an adult academic hospitalist group.⁹ Methods of assessment varied widely across studies and included observations in the clinical environment,^28,30,31 performance in simulations,³² and surveys about performance.^22-26 A handful of approaches were more automated and made use of informatics^8,9,22 or data collected for other health system purposes.^8,9

Effective

Effectiveness was most often assessed through adherence to consensus and evidence-based guidelines. Examples included processes of care related to sepsis, venous thromboembolism prophylaxis, COPD, heart failure, pediatric asthma, and antibiotic appropriateness.^8,9,23,32-36 Through the review, multiple other studies that included group-level measures of effectiveness for a variety of health conditions were excluded because data on individual-level variation were not reported. Methods of assessment included expert review of cases or discharge summaries, compliance with core measures, performance in simulation, and self-assessment on practice behaviors. Other than those efforts aligned with institutional data collection, most approaches were resource intensive.

Efficient

As with those in the Timely domain, most studies coded into the Efficient domain were coded into at least one other domain. One exception measured unnecessary daily lab work and both showed provider-level variation and demonstrated improvement in quality based on an intervention.³⁷ Another paper coded into the Effective domain evaluated adherence to components of the Choosing Wisely® recommendations.³⁴ In addition to these two studies focusing on cost efficacy, other studies coded to this domain assessed concepts such as ensuring more efficient care from other providers by optimizing transitions of care^15-17 and clarifying patients’ goals for care.³⁸ Although integrating insurer information into care plans is emphasized in the Core Competencies of Hospital Medicine,¹¹ this concept was not represented in any of the identified articles. Methods of assessment varied and mostly relied on observation of behaviors or survey of providers. Several approaches were more automated or used Medicare claims data to assess the efficiency of individual providers relative to peers.^34,37,39

Equitable

Among the studies reviewed, none were coded into the Equitable domain despite care of vulnerable populations being identified as a core competency of hospital medicine.⁴⁰

Patient Centered

Studies coded to the Patient Centered domain assessed hospitalist performance through ratings of patient satisfaction,^8,9,41-44 rating of communication between hospitalists and patients,^{19-21,29,45-51} identification of patient preferences,^38,52 outcomes of patient-centered care activities,^27,28 and peer ratings.^53,54 Authors applied several theoretical constructs to these assessments including shared decision-making,⁵⁰ etiquette-based medicine,^47,48 empathetic responsiveness,⁴⁵ agreement about the goals of care between the patient and healthcare team members,⁵² and lapses in professionalism.⁵³ Studies often crossed STEEEP domains, such as those assessing quality of discharge information provided to patients, which were coded as both Safe and Patient Centered.^19-21 In addition to coded or observed performance in the clinical setting, studies in this domain also used patient ratings as a method of assessment.^{8,9,28,41-44,49,50} Only a few of these approaches aligned with existing performance measures of health systems and were more automated.^8,9

This scoping review of performance data for individual hospitalists coded to the STEEEP framework identified robust areas in the published literature, as well as opportunities to develop new approaches or refine existing measures. Transitions of care, both intrahospital and at discharge, and adherence to evidence-based guidelines are areas for which current research has created a foundation for care that is Safe, Timely, Effective, and Efficient. The Patient Centered domain also has several measures described, though the conceptual underpinnings are heterogeneous, and consensus appears necessary to compare performance across groups. No studies were coded to the Equitable domain. Across domains, approaches to measurement varied in resource intensity from simple ones, like integrating existing data collected by hospitals, to more complex ones, like shadowing physicians or coding interactions.

Methods of assessment coded into the Safe domain focused on communication and, less so, patient outcomes around transitions of care. Transitions of care that were evaluated included transfer of patients into a new facility, sign-out to new physicians for both cross-cover responsibilities and for newly assuming the role of primary attending, and discharge from the hospital. Most measures rated the quality of communication, although several^23-27 examined patient outcomes. Approaches that survey individuals downstream from a transition of care^15,17,24-26 may be the simplest and most feasible approach to implement in the future but, as described to date, do not include all transitions of care and may miss patient outcomes. Important core competencies for hospital medicine under the Safe domain that were not identified in this review include areas such as diagnostic error, hospital-acquired infections, error reporting, and medication safety.¹¹ These are potential areas for future measure development.

The assessments in many studies were coded across more than one domain; for example, measures of the application of evidence-based guidelines were coded into domains of Effective, Timely, Efficient, and others. Applying the six domains of the STEEEP framework revealed the multidimensional outcomes of hospitalist work and could guide more meaningful quality assessments of individual hospitalist performance. For example, assessing adherence to evidence-based guidelines, as well as consideration of the Core Competencies of Hospital Medicine and recommendations of the Choosing Wisely® campaign, are promising areas for measurement and may align with existing hospital metrics. Notably, several reviewed studies measured group-level adherence to guidelines but were excluded because they did not examine variation at the individual level. Future measures based on evidence-based guidelines could center on the Effective domain while also integrating assessment of domains such as Efficient, Timely, and Patient Centered and, in so doing, provide a richer assessment of the diverse aspects of quality.

Several other approaches in the domains of Timely, Effective, and Efficient were described only in a few studies yet deserve consideration for further development. Two time-­motion studies^30,31 were coded into the domains of Timely and Efficient and would be cumbersome in regular practice but, with advances in wearable technology and electronic health records, could become more feasible in the future. Another approach used Medicare payment data to detect provider-level variation.³⁹ Potentially, “big data” could be analyzed in other ways to compare the performance of individual hospitalists.

The lack of studies coded into the Equitable domain may seem surprising, but the Institute for Healthcare Improvement identifies Equitable as the “forgotten aim” of the STEEEP framework. This organization has developed a guide for health care organizations to promote equitable care.⁵⁵ While this guide focuses mostly on organizational-level actions, some are focused on individual providers, such as training in implicit bias. Future research should seek to identify disparities in care by individual providers and develop interventions to address any discovered gaps.

The “Patient Centered” domain was the most frequently coded and had the most heterogeneous underpinnings for assessment. Studies varied widely in terminology and conceptual foundations. The field would benefit from future work to identify how “Patient Centered” care might be more clearly conceptualized, guided by comparative studies among different assessment approaches to define those most valid and feasible.

The overarching goal for measuring individual hospitalist quality should be to improve the delivery of patient care in a supportive and formative way. To further this goal, adding or expanding on metrics identified in this article may provide a more complete description of performance. As a future direction, groups should consider partnering with one another to define measurement approaches, collaborate with existing data sources, and even share deidentified individual data to establish performance benchmarks at the individual and group levels.

While this study used broad search terms to support completeness, the search process could have missed important studies. Grey literature, non–English language studies, and industry reports were not included in this review. Groups may also be using other assessments of individual hospitalist performance that are not published in the peer-reviewed literature. Coding of study assessments was achieved through consensus reconciliation; other coders might have classified studies differently.

This scoping review describes the peer-reviewed literature of individual hospitalist performance and is the first to link it to the STEEEP quality framework. Assessments of transitions of care, evidence-based care, and cost-effective care are exemplars in the published literature. Patient-centered care is well studied but assessed in a heterogeneous fashion. Assessments of equity in care are notably absent. The STEEEP framework provides a model to structure assessment of individual performance. Future research should build on this framework to define meaningful assessment approaches that are actionable and improve the welfare of our patients and our system.

The authors have nothing to disclose.

Healthcare quality is defined as the extent to which healthcare services result in desired outcomes.¹ Quality of care depends on how the healthcare system’s various components, including healthcare practitioners, interact to meet each patient’s needs.² These components can be shaped to achieve desired outcomes through rules, incentives, and other approaches, but influencing the behaviors of each component, such as the performance of hospitalists, requires defining goals for performance and implementing measurement approaches to assess progress toward these goals.

One set of principles to define goals for quality and guide assessment of desired behaviors is the multidimensional STEEEP framework. This framework, created by the Institute of Medicine, identifies six domains of quality: Safe, Timely, Effective, Efficient, Equitable, and Patient Centered.² Briefly, “Safe” means avoiding injuries to patients, “Timely” means reducing waits and delays in care, “Effective” means providing care based on evidence, “Efficient” means avoiding waste, “Equitable” means ensuring quality does not vary based on personal characteristics such as race and gender, and “Patient Centered” means providing care that is responsive to patients’ values and preferences. The STEEEP domains are not coequal; rather, they ensure that quality is considered broadly, while avoiding errors such as measuring only an intervention’s impact on effectiveness but not assessing its impact on multiple domains of quality, such as how patient centered, efficient (cost effective), or equitable the resulting care is.

Based on our review of the literature, a multidimensional framework like STEEEP has not been used in defining and assessing the quality of individual hospitalists’ performance. Some quality metrics at the hospital level impact several dimensions simultaneously, such as door to balloon time for acute myocardial infarction, which measures effectiveness and timeliness of care. Programs like pay-for-performance, Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS), and the Merit-Based Incentive Payment System (MIPS) have tied reimbursement to assessments aligned with several STEEEP domains at both individual and institutional levels but lack a holistic approach to quality.^3-6 The every-­other-year State of Hospital Medicine Report, the most widely used description of individual hospitalist performance, reports group-level performance including relative value units and whether groups are accountable for measures of quality such as performance on core measures, timely documentation, and “citizenship” (eg, committee participation or academic work).⁷ While these are useful benchmarks, the report focuses on performance at the group level. Concurrently, several academic groups have described more complete dashboards or scorecards to assess individual hospitalist performance, primarily designed to facilitate comparison across hospitalist groups or to incentivize overall group performance.^8-10 However, these efforts are not guided by an overarching framework and are structured after traditional academic models with components related to teaching and scholarship, which may not translate to nonacademic environments. Finally, the Core Competencies for Hospital Medicine outlines some goals for hospitalist performance but does not speak to specific measurement approaches.¹¹

Overall, assessing individual hospitalist performance is hindered by lack of consensus on important concepts to measure, a limited number of valid measures, and challenges in data collection such as resource limitations and feasibility. Developing and refining measures grounded in the STEEEP framework may provide a more comprehensive assessment of hospitalist quality and identify approaches to improve overall health outcomes. Comparative data could help individual hospitalists improve performance; leaders of hospitalist groups could use this data to guide faculty development and advancement as they ensure quality care at the individual, group, and system levels.

To better inform quality measurement of individual hospitalists, we sought to identify existing publications on individual hospitalist quality. Our goal was to define the published literature about quality measurement at the individual hospitalist level, relate these publications to domains of quality defined by the STEEEP framework, and identify directions for assessment or further research that could affect the overall quality of care.

We conducted a scoping review following methods outlined by Arksey and O’Malley¹² and Tricco.¹³ The goal of a scoping review is to map the extent of research within a specific field. This methodology is well suited to characterizing the existing research related to the quality of hospitalist care at the individual level. A protocol for the scoping review was not registered.

Evidence Search

A systematic search for published, English-language literature on hospitalist care was conducted in Medline (Ovid; 1946 - June 4, 2019) on June 5, 2019. The search used a combination of keywords and controlled vocabulary for the concept of hospitalists or hospital medicine. The search strategy used in this review is described in the Appendix. In addition, a hand search of reference lists of articles was used to discover publications not identified in the database searches.

Study Selection

All references were uploaded to Covidence systematic review software (www.covidence.org; Covidence), and duplicates were removed. Four reviewers (A.D., B.C., L.H., R.Q.) conducted title and abstract, as well as full-text, review to identify studies that measured differences in the performance of hospitalists at the individual level. Any disagreements among reviewers were resolved by consensus. Articles included both adult and pediatric populations. Articles that focused on group-level outcomes could be included if nonpooled data at the individual level was also reported. Studies were excluded if they did not focus on individual quality of care indicators or were not published in English.

Data Charting and Synthesis

We extracted the following information using a standardized data collection form: author, title, year of publication, study design, intervention, and outcome measures. Original manuscripts were accessed as needed to supplement analysis. Critical appraisal of individual studies was not conducted in this review because the goal of this review was to analyze which quality indicators have been studied and how they were measured. Articles were then coded for their alignment to the STEEEP framework by two reviewers (AD and BC). After initial coding was conducted, the reviewers met to consolidate codes and resolve any disagreement by consensus. The results of the analysis were summarized in both text and tabular format with studies grouped by focus of assessment with each one’s methods of assessment listed.

Results of the search strategy are shown in the Figure. The search retrieved a total of 2,363 references of which 113 were duplicates, leaving 2,250 to be screened. After title and abstract and full-text screening, 42 studies were included in the review. The final 42 studies were coded for alignment with the STEEEP framework. The Table displays the focus of assessment and methods of assessment within each STEEEP domain.

Eighteen studies were coded into a single domain while the rest were coded into at least two domains. The domain Patient Centered was coded as having the most studies (n = 23), followed by the domain of Safe (n = 15). Timely, Effective, and Efficient domains had 11, 9, and 12 studies, respectively. No studies were coded into the domain of Equitable.

Safe

Nearly all studies coded into the Safe domain focused on transitions of care. These included transfers into a hospital from other hospitals,¹⁴ transitions of care to cross-covering providers^15,16 and new primary providers,¹⁷ and transition out from the acute care setting.^18-28 Measures of hospital discharge included measures of both processes^18-22 and outcomes.^23-27 Methods of assessment varied from use of trained observers or scorers to surveys of individuals and colleagues about performance. Though a few leveraged informatics,^22,27 all approaches relied on human interaction, and none were automated.

Timely

All studies coded into the Timely domain were coded into at least one other domain. For example, Anderson et al looked at how hospitalists communicated about potential life-limiting illness at the time of hospital admission and the subsequent effects on plans of care²⁹; this was coded as both Timely and Patient Centered. Likewise, another group of studies centered on application of evidence-based guidelines, such as giving antibiotics within a certain time interval for sepsis and were coded as both Timely and Effective. Another set of authors described dashboards or scorecards that captured a number of group-level metrics of processes of care that span STEEEP domains and may be applicable to individuals, including Fox et al for pediatrics⁸ and Hwa et al for an adult academic hospitalist group.⁹ Methods of assessment varied widely across studies and included observations in the clinical environment,^28,30,31 performance in simulations,³² and surveys about performance.^22-26 A handful of approaches were more automated and made use of informatics^8,9,22 or data collected for other health system purposes.^8,9

Effective

Effectiveness was most often assessed through adherence to consensus and evidence-based guidelines. Examples included processes of care related to sepsis, venous thromboembolism prophylaxis, COPD, heart failure, pediatric asthma, and antibiotic appropriateness.^8,9,23,32-36 Through the review, multiple other studies that included group-level measures of effectiveness for a variety of health conditions were excluded because data on individual-level variation were not reported. Methods of assessment included expert review of cases or discharge summaries, compliance with core measures, performance in simulation, and self-assessment on practice behaviors. Other than those efforts aligned with institutional data collection, most approaches were resource intensive.

Efficient

As with those in the Timely domain, most studies coded into the Efficient domain were coded into at least one other domain. One exception measured unnecessary daily lab work and both showed provider-level variation and demonstrated improvement in quality based on an intervention.³⁷ Another paper coded into the Effective domain evaluated adherence to components of the Choosing Wisely® recommendations.³⁴ In addition to these two studies focusing on cost efficacy, other studies coded to this domain assessed concepts such as ensuring more efficient care from other providers by optimizing transitions of care^15-17 and clarifying patients’ goals for care.³⁸ Although integrating insurer information into care plans is emphasized in the Core Competencies of Hospital Medicine,¹¹ this concept was not represented in any of the identified articles. Methods of assessment varied and mostly relied on observation of behaviors or survey of providers. Several approaches were more automated or used Medicare claims data to assess the efficiency of individual providers relative to peers.^34,37,39

Equitable

Among the studies reviewed, none were coded into the Equitable domain despite care of vulnerable populations being identified as a core competency of hospital medicine.⁴⁰

Patient Centered

Studies coded to the Patient Centered domain assessed hospitalist performance through ratings of patient satisfaction,^8,9,41-44 rating of communication between hospitalists and patients,^{19-21,29,45-51} identification of patient preferences,^38,52 outcomes of patient-centered care activities,^27,28 and peer ratings.^53,54 Authors applied several theoretical constructs to these assessments including shared decision-making,⁵⁰ etiquette-based medicine,^47,48 empathetic responsiveness,⁴⁵ agreement about the goals of care between the patient and healthcare team members,⁵² and lapses in professionalism.⁵³ Studies often crossed STEEEP domains, such as those assessing quality of discharge information provided to patients, which were coded as both Safe and Patient Centered.^19-21 In addition to coded or observed performance in the clinical setting, studies in this domain also used patient ratings as a method of assessment.^{8,9,28,41-44,49,50} Only a few of these approaches aligned with existing performance measures of health systems and were more automated.^8,9

This scoping review of performance data for individual hospitalists coded to the STEEEP framework identified robust areas in the published literature, as well as opportunities to develop new approaches or refine existing measures. Transitions of care, both intrahospital and at discharge, and adherence to evidence-based guidelines are areas for which current research has created a foundation for care that is Safe, Timely, Effective, and Efficient. The Patient Centered domain also has several measures described, though the conceptual underpinnings are heterogeneous, and consensus appears necessary to compare performance across groups. No studies were coded to the Equitable domain. Across domains, approaches to measurement varied in resource intensity from simple ones, like integrating existing data collected by hospitals, to more complex ones, like shadowing physicians or coding interactions.

Methods of assessment coded into the Safe domain focused on communication and, less so, patient outcomes around transitions of care. Transitions of care that were evaluated included transfer of patients into a new facility, sign-out to new physicians for both cross-cover responsibilities and for newly assuming the role of primary attending, and discharge from the hospital. Most measures rated the quality of communication, although several^23-27 examined patient outcomes. Approaches that survey individuals downstream from a transition of care^15,17,24-26 may be the simplest and most feasible approach to implement in the future but, as described to date, do not include all transitions of care and may miss patient outcomes. Important core competencies for hospital medicine under the Safe domain that were not identified in this review include areas such as diagnostic error, hospital-acquired infections, error reporting, and medication safety.¹¹ These are potential areas for future measure development.

The assessments in many studies were coded across more than one domain; for example, measures of the application of evidence-based guidelines were coded into domains of Effective, Timely, Efficient, and others. Applying the six domains of the STEEEP framework revealed the multidimensional outcomes of hospitalist work and could guide more meaningful quality assessments of individual hospitalist performance. For example, assessing adherence to evidence-based guidelines, as well as consideration of the Core Competencies of Hospital Medicine and recommendations of the Choosing Wisely® campaign, are promising areas for measurement and may align with existing hospital metrics. Notably, several reviewed studies measured group-level adherence to guidelines but were excluded because they did not examine variation at the individual level. Future measures based on evidence-based guidelines could center on the Effective domain while also integrating assessment of domains such as Efficient, Timely, and Patient Centered and, in so doing, provide a richer assessment of the diverse aspects of quality.

Several other approaches in the domains of Timely, Effective, and Efficient were described only in a few studies yet deserve consideration for further development. Two time-­motion studies^30,31 were coded into the domains of Timely and Efficient and would be cumbersome in regular practice but, with advances in wearable technology and electronic health records, could become more feasible in the future. Another approach used Medicare payment data to detect provider-level variation.³⁹ Potentially, “big data” could be analyzed in other ways to compare the performance of individual hospitalists.

The lack of studies coded into the Equitable domain may seem surprising, but the Institute for Healthcare Improvement identifies Equitable as the “forgotten aim” of the STEEEP framework. This organization has developed a guide for health care organizations to promote equitable care.⁵⁵ While this guide focuses mostly on organizational-level actions, some are focused on individual providers, such as training in implicit bias. Future research should seek to identify disparities in care by individual providers and develop interventions to address any discovered gaps.

The “Patient Centered” domain was the most frequently coded and had the most heterogeneous underpinnings for assessment. Studies varied widely in terminology and conceptual foundations. The field would benefit from future work to identify how “Patient Centered” care might be more clearly conceptualized, guided by comparative studies among different assessment approaches to define those most valid and feasible.

The overarching goal for measuring individual hospitalist quality should be to improve the delivery of patient care in a supportive and formative way. To further this goal, adding or expanding on metrics identified in this article may provide a more complete description of performance. As a future direction, groups should consider partnering with one another to define measurement approaches, collaborate with existing data sources, and even share deidentified individual data to establish performance benchmarks at the individual and group levels.

While this study used broad search terms to support completeness, the search process could have missed important studies. Grey literature, non–English language studies, and industry reports were not included in this review. Groups may also be using other assessments of individual hospitalist performance that are not published in the peer-reviewed literature. Coding of study assessments was achieved through consensus reconciliation; other coders might have classified studies differently.

This scoping review describes the peer-reviewed literature of individual hospitalist performance and is the first to link it to the STEEEP quality framework. Assessments of transitions of care, evidence-based care, and cost-effective care are exemplars in the published literature. Patient-centered care is well studied but assessed in a heterogeneous fashion. Assessments of equity in care are notably absent. The STEEEP framework provides a model to structure assessment of individual performance. Future research should build on this framework to define meaningful assessment approaches that are actionable and improve the welfare of our patients and our system.

The authors have nothing to disclose.

Healthcare quality is defined as the extent to which healthcare services result in desired outcomes.¹ Quality of care depends on how the healthcare system’s various components, including healthcare practitioners, interact to meet each patient’s needs.² These components can be shaped to achieve desired outcomes through rules, incentives, and other approaches, but influencing the behaviors of each component, such as the performance of hospitalists, requires defining goals for performance and implementing measurement approaches to assess progress toward these goals.

One set of principles to define goals for quality and guide assessment of desired behaviors is the multidimensional STEEEP framework. This framework, created by the Institute of Medicine, identifies six domains of quality: Safe, Timely, Effective, Efficient, Equitable, and Patient Centered.² Briefly, “Safe” means avoiding injuries to patients, “Timely” means reducing waits and delays in care, “Effective” means providing care based on evidence, “Efficient” means avoiding waste, “Equitable” means ensuring quality does not vary based on personal characteristics such as race and gender, and “Patient Centered” means providing care that is responsive to patients’ values and preferences. The STEEEP domains are not coequal; rather, they ensure that quality is considered broadly, while avoiding errors such as measuring only an intervention’s impact on effectiveness but not assessing its impact on multiple domains of quality, such as how patient centered, efficient (cost effective), or equitable the resulting care is.

Based on our review of the literature, a multidimensional framework like STEEEP has not been used in defining and assessing the quality of individual hospitalists’ performance. Some quality metrics at the hospital level impact several dimensions simultaneously, such as door to balloon time for acute myocardial infarction, which measures effectiveness and timeliness of care. Programs like pay-for-performance, Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS), and the Merit-Based Incentive Payment System (MIPS) have tied reimbursement to assessments aligned with several STEEEP domains at both individual and institutional levels but lack a holistic approach to quality.^3-6 The every-­other-year State of Hospital Medicine Report, the most widely used description of individual hospitalist performance, reports group-level performance including relative value units and whether groups are accountable for measures of quality such as performance on core measures, timely documentation, and “citizenship” (eg, committee participation or academic work).⁷ While these are useful benchmarks, the report focuses on performance at the group level. Concurrently, several academic groups have described more complete dashboards or scorecards to assess individual hospitalist performance, primarily designed to facilitate comparison across hospitalist groups or to incentivize overall group performance.^8-10 However, these efforts are not guided by an overarching framework and are structured after traditional academic models with components related to teaching and scholarship, which may not translate to nonacademic environments. Finally, the Core Competencies for Hospital Medicine outlines some goals for hospitalist performance but does not speak to specific measurement approaches.¹¹

Overall, assessing individual hospitalist performance is hindered by lack of consensus on important concepts to measure, a limited number of valid measures, and challenges in data collection such as resource limitations and feasibility. Developing and refining measures grounded in the STEEEP framework may provide a more comprehensive assessment of hospitalist quality and identify approaches to improve overall health outcomes. Comparative data could help individual hospitalists improve performance; leaders of hospitalist groups could use this data to guide faculty development and advancement as they ensure quality care at the individual, group, and system levels.

To better inform quality measurement of individual hospitalists, we sought to identify existing publications on individual hospitalist quality. Our goal was to define the published literature about quality measurement at the individual hospitalist level, relate these publications to domains of quality defined by the STEEEP framework, and identify directions for assessment or further research that could affect the overall quality of care.

We conducted a scoping review following methods outlined by Arksey and O’Malley¹² and Tricco.¹³ The goal of a scoping review is to map the extent of research within a specific field. This methodology is well suited to characterizing the existing research related to the quality of hospitalist care at the individual level. A protocol for the scoping review was not registered.

Evidence Search

A systematic search for published, English-language literature on hospitalist care was conducted in Medline (Ovid; 1946 - June 4, 2019) on June 5, 2019. The search used a combination of keywords and controlled vocabulary for the concept of hospitalists or hospital medicine. The search strategy used in this review is described in the Appendix. In addition, a hand search of reference lists of articles was used to discover publications not identified in the database searches.

Study Selection

All references were uploaded to Covidence systematic review software (www.covidence.org; Covidence), and duplicates were removed. Four reviewers (A.D., B.C., L.H., R.Q.) conducted title and abstract, as well as full-text, review to identify studies that measured differences in the performance of hospitalists at the individual level. Any disagreements among reviewers were resolved by consensus. Articles included both adult and pediatric populations. Articles that focused on group-level outcomes could be included if nonpooled data at the individual level was also reported. Studies were excluded if they did not focus on individual quality of care indicators or were not published in English.

Data Charting and Synthesis

We extracted the following information using a standardized data collection form: author, title, year of publication, study design, intervention, and outcome measures. Original manuscripts were accessed as needed to supplement analysis. Critical appraisal of individual studies was not conducted in this review because the goal of this review was to analyze which quality indicators have been studied and how they were measured. Articles were then coded for their alignment to the STEEEP framework by two reviewers (AD and BC). After initial coding was conducted, the reviewers met to consolidate codes and resolve any disagreement by consensus. The results of the analysis were summarized in both text and tabular format with studies grouped by focus of assessment with each one’s methods of assessment listed.

Results of the search strategy are shown in the Figure. The search retrieved a total of 2,363 references of which 113 were duplicates, leaving 2,250 to be screened. After title and abstract and full-text screening, 42 studies were included in the review. The final 42 studies were coded for alignment with the STEEEP framework. The Table displays the focus of assessment and methods of assessment within each STEEEP domain.

Eighteen studies were coded into a single domain while the rest were coded into at least two domains. The domain Patient Centered was coded as having the most studies (n = 23), followed by the domain of Safe (n = 15). Timely, Effective, and Efficient domains had 11, 9, and 12 studies, respectively. No studies were coded into the domain of Equitable.

Safe

Nearly all studies coded into the Safe domain focused on transitions of care. These included transfers into a hospital from other hospitals,¹⁴ transitions of care to cross-covering providers^15,16 and new primary providers,¹⁷ and transition out from the acute care setting.^18-28 Measures of hospital discharge included measures of both processes^18-22 and outcomes.^23-27 Methods of assessment varied from use of trained observers or scorers to surveys of individuals and colleagues about performance. Though a few leveraged informatics,^22,27 all approaches relied on human interaction, and none were automated.

Timely

All studies coded into the Timely domain were coded into at least one other domain. For example, Anderson et al looked at how hospitalists communicated about potential life-limiting illness at the time of hospital admission and the subsequent effects on plans of care²⁹; this was coded as both Timely and Patient Centered. Likewise, another group of studies centered on application of evidence-based guidelines, such as giving antibiotics within a certain time interval for sepsis and were coded as both Timely and Effective. Another set of authors described dashboards or scorecards that captured a number of group-level metrics of processes of care that span STEEEP domains and may be applicable to individuals, including Fox et al for pediatrics⁸ and Hwa et al for an adult academic hospitalist group.⁹ Methods of assessment varied widely across studies and included observations in the clinical environment,^28,30,31 performance in simulations,³² and surveys about performance.^22-26 A handful of approaches were more automated and made use of informatics^8,9,22 or data collected for other health system purposes.^8,9

Effective

Effectiveness was most often assessed through adherence to consensus and evidence-based guidelines. Examples included processes of care related to sepsis, venous thromboembolism prophylaxis, COPD, heart failure, pediatric asthma, and antibiotic appropriateness.^8,9,23,32-36 Through the review, multiple other studies that included group-level measures of effectiveness for a variety of health conditions were excluded because data on individual-level variation were not reported. Methods of assessment included expert review of cases or discharge summaries, compliance with core measures, performance in simulation, and self-assessment on practice behaviors. Other than those efforts aligned with institutional data collection, most approaches were resource intensive.

Efficient

As with those in the Timely domain, most studies coded into the Efficient domain were coded into at least one other domain. One exception measured unnecessary daily lab work and both showed provider-level variation and demonstrated improvement in quality based on an intervention.³⁷ Another paper coded into the Effective domain evaluated adherence to components of the Choosing Wisely® recommendations.³⁴ In addition to these two studies focusing on cost efficacy, other studies coded to this domain assessed concepts such as ensuring more efficient care from other providers by optimizing transitions of care^15-17 and clarifying patients’ goals for care.³⁸ Although integrating insurer information into care plans is emphasized in the Core Competencies of Hospital Medicine,¹¹ this concept was not represented in any of the identified articles. Methods of assessment varied and mostly relied on observation of behaviors or survey of providers. Several approaches were more automated or used Medicare claims data to assess the efficiency of individual providers relative to peers.^34,37,39

Equitable

Among the studies reviewed, none were coded into the Equitable domain despite care of vulnerable populations being identified as a core competency of hospital medicine.⁴⁰

Patient Centered

Studies coded to the Patient Centered domain assessed hospitalist performance through ratings of patient satisfaction,^8,9,41-44 rating of communication between hospitalists and patients,^{19-21,29,45-51} identification of patient preferences,^38,52 outcomes of patient-centered care activities,^27,28 and peer ratings.^53,54 Authors applied several theoretical constructs to these assessments including shared decision-making,⁵⁰ etiquette-based medicine,^47,48 empathetic responsiveness,⁴⁵ agreement about the goals of care between the patient and healthcare team members,⁵² and lapses in professionalism.⁵³ Studies often crossed STEEEP domains, such as those assessing quality of discharge information provided to patients, which were coded as both Safe and Patient Centered.^19-21 In addition to coded or observed performance in the clinical setting, studies in this domain also used patient ratings as a method of assessment.^{8,9,28,41-44,49,50} Only a few of these approaches aligned with existing performance measures of health systems and were more automated.^8,9

This scoping review of performance data for individual hospitalists coded to the STEEEP framework identified robust areas in the published literature, as well as opportunities to develop new approaches or refine existing measures. Transitions of care, both intrahospital and at discharge, and adherence to evidence-based guidelines are areas for which current research has created a foundation for care that is Safe, Timely, Effective, and Efficient. The Patient Centered domain also has several measures described, though the conceptual underpinnings are heterogeneous, and consensus appears necessary to compare performance across groups. No studies were coded to the Equitable domain. Across domains, approaches to measurement varied in resource intensity from simple ones, like integrating existing data collected by hospitals, to more complex ones, like shadowing physicians or coding interactions.

Methods of assessment coded into the Safe domain focused on communication and, less so, patient outcomes around transitions of care. Transitions of care that were evaluated included transfer of patients into a new facility, sign-out to new physicians for both cross-cover responsibilities and for newly assuming the role of primary attending, and discharge from the hospital. Most measures rated the quality of communication, although several^23-27 examined patient outcomes. Approaches that survey individuals downstream from a transition of care^15,17,24-26 may be the simplest and most feasible approach to implement in the future but, as described to date, do not include all transitions of care and may miss patient outcomes. Important core competencies for hospital medicine under the Safe domain that were not identified in this review include areas such as diagnostic error, hospital-acquired infections, error reporting, and medication safety.¹¹ These are potential areas for future measure development.

The assessments in many studies were coded across more than one domain; for example, measures of the application of evidence-based guidelines were coded into domains of Effective, Timely, Efficient, and others. Applying the six domains of the STEEEP framework revealed the multidimensional outcomes of hospitalist work and could guide more meaningful quality assessments of individual hospitalist performance. For example, assessing adherence to evidence-based guidelines, as well as consideration of the Core Competencies of Hospital Medicine and recommendations of the Choosing Wisely® campaign, are promising areas for measurement and may align with existing hospital metrics. Notably, several reviewed studies measured group-level adherence to guidelines but were excluded because they did not examine variation at the individual level. Future measures based on evidence-based guidelines could center on the Effective domain while also integrating assessment of domains such as Efficient, Timely, and Patient Centered and, in so doing, provide a richer assessment of the diverse aspects of quality.

Several other approaches in the domains of Timely, Effective, and Efficient were described only in a few studies yet deserve consideration for further development. Two time-­motion studies^30,31 were coded into the domains of Timely and Efficient and would be cumbersome in regular practice but, with advances in wearable technology and electronic health records, could become more feasible in the future. Another approach used Medicare payment data to detect provider-level variation.³⁹ Potentially, “big data” could be analyzed in other ways to compare the performance of individual hospitalists.

The lack of studies coded into the Equitable domain may seem surprising, but the Institute for Healthcare Improvement identifies Equitable as the “forgotten aim” of the STEEEP framework. This organization has developed a guide for health care organizations to promote equitable care.⁵⁵ While this guide focuses mostly on organizational-level actions, some are focused on individual providers, such as training in implicit bias. Future research should seek to identify disparities in care by individual providers and develop interventions to address any discovered gaps.

The “Patient Centered” domain was the most frequently coded and had the most heterogeneous underpinnings for assessment. Studies varied widely in terminology and conceptual foundations. The field would benefit from future work to identify how “Patient Centered” care might be more clearly conceptualized, guided by comparative studies among different assessment approaches to define those most valid and feasible.

The overarching goal for measuring individual hospitalist quality should be to improve the delivery of patient care in a supportive and formative way. To further this goal, adding or expanding on metrics identified in this article may provide a more complete description of performance. As a future direction, groups should consider partnering with one another to define measurement approaches, collaborate with existing data sources, and even share deidentified individual data to establish performance benchmarks at the individual and group levels.

While this study used broad search terms to support completeness, the search process could have missed important studies. Grey literature, non–English language studies, and industry reports were not included in this review. Groups may also be using other assessments of individual hospitalist performance that are not published in the peer-reviewed literature. Coding of study assessments was achieved through consensus reconciliation; other coders might have classified studies differently.

This scoping review describes the peer-reviewed literature of individual hospitalist performance and is the first to link it to the STEEEP quality framework. Assessments of transitions of care, evidence-based care, and cost-effective care are exemplars in the published literature. Patient-centered care is well studied but assessed in a heterogeneous fashion. Assessments of equity in care are notably absent. The STEEEP framework provides a model to structure assessment of individual performance. Future research should build on this framework to define meaningful assessment approaches that are actionable and improve the welfare of our patients and our system.

The authors have nothing to disclose.

Inspired by the ABIM Foundation’s Choosing Wisely^® campaign, the “Things We Do for No Reason^™” (TWDFNR) series reviews practices that have become common parts of hospital care but may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent clear-cut conclusions or clinical practice standards but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion.

The hospitalist admits a 52-year-old man with alcoholic cirrhosis for tense ascites and altered mentation. Home medications include furosemide, spironolactone, lactulose, and rifaximin, but his family notes he ran out last week. Although afebrile and hemodynamically stable, the patient’s coagulopathy, with an international normalized ratio (INR) of 2.3, and thrombocytopenia, with a platelet count of 37,000/μL, worries the hospitalist. The hospitalist wonders whether to transfuse fresh frozen plasma (FFP) and platelets prior to diagnostic paracentesis to reduce the risk of procedural bleeding.

Many patients undergoing paracentesis have severe liver disease and present with both thrombocytopenia and elevated INRs. While platelet count and INR serve as surrogate markers for bleeding risk in many settings, clinicians often extrapolate this concept to patients with cirrhosis. Many hospitalists routinely check INR and platelet count and administer FFP and platelets prior to diagnostic or therapeutic paracentesis to mitigate procedure-related bleeding risk. Some medical resources recommend this practice,¹ while case reports and personal experiences with bleeding in these patients create availability bias that influences perception of bleeding risk.² One recent study of patients with decompensated cirrhosis presenting to a US tertiary care center found that, of those receiving large-­volume paracentesis, 22.2% received prophylactic FFP and 17.3% received prophylactic platelets before paracentesis.³

Advances in our understanding of coagulation in cirrhosis demonstrate neither INR nor platelet count accurately predict bleeding risk in this population. Additionally, evidence demonstrates the overall safety of paracentesis in cirrhosis—even in the presence of high INR and thrombocytopenia—and the lack of benefit from prophylactic transfusions with FFP or platelets.

Substantial evidence in patients with cirrhosis demonstrates that changes in coagulation and platelet function confer a “balanced coagulopathy” in which patients oscillate between hyper- and hypocoagulable states. In a cirrhotic liver, hepatic synthetic dysfunction results in a complex milieu through reduced production and plasma concentrations of both pro- and anticoagulant factors that can lead to either bleeding or clotting.⁴ This “rebalancing” makes prothrombin time (PT) and INR unreliable indicators of bleeding or clotting risk. Similarly, in patients with cirrhosis, thrombocytopenia does not necessarily reflect impaired clotting ability. These patients experience an increase in production of von Willebrand Factor, which may compensate for low platelet counts by producing stronger platelet adhesion to collagen.⁴ Unfortunately, we currently lack a reliable test or risk score to assess true bleeding risk in patients with cirrhosis.

Observational studies support these laboratory findings. Large case series consistently demonstrate no association between INR or platelet counts and bleeding risk in either diagnostic or therapeutic paracentesis, including large-volume paracentesis (See Appendix for a list of recent representative studies).^5-10 Moreover, prophylactic transfusion of FFP or platelets does not significantly reduce bleeding risk.

In a 1991 study by McVay et al, the researchers examined bleeding outcomes of 441 paracenteses performed on hospitalized patients.¹¹ Among patients who did not receive FFP prior to paracentesis, only one required a transfusion for procedure-­related bleeding, an event rate of 0.25%. This single patient had a normal platelet count and an elevated PT to the same extent as 261 others who underwent paracentesis without complication. In a pooled analysis that included 391 paracenteses and 207 thoracenteses, the authors concluded neither PT nor platelet level predicted bleeding risk. Similarly, the largest published case series on this topic examined 4,729 paracenteses over a decade on a liver unit and found low rates of major bleeding (0.19%).⁹ Furthermore, preprocedure INR or platelet count did not correlate with bleeding risk. The authors did not report preprocedure transfusion rates, but they noted transfusions occurred only “occasionally.”

Subsequent observational studies have consistently revealed low bleeding risks even in settings of high coagulopathy prevalence. Grabau et al reviewed all large-volume paracenteses performed in a gastroenterology clinic over 7 years.¹⁰ In over 1,100 procedures, no major bleeding events occurred despite 27% of patients having INR greater than 2.0 and 54% having platelet counts less than 50,000/μL. Kurup et al examined bleeding risk among 304 procedures performed on patients with platelet counts less than 50,000/μL referred to radiology for ultrasound-guided paracentesis.⁷ Three bleeding events occurred, an overall event rate of 0.99%. They also found no association between preprocedure platelet count and bleeding risk.

In addition to observational data, one randomized, controlled trial evaluated the effects of FFP and platelet administration on bleeding risk among 60 patients with cirrhosis undergoing invasive procedures, including 19 paracenteses.⁶ Enrollment criteria included INR greater than 1.8 and/or platelet count less than 50,000/μL. One hundred percent of patients randomized to the usual care control arm received platelets or FFP as compared to 17% in the thromboelastography (TEG)–guided transfusion strategy arm. TEG assesses the viscoelastic properties of evolving clot formation in whole blood. Only one patient, a patient in the control arm who received FFP, developed procedure-related bleeding. Although receiving many fewer transfusions, the TEG-guided group experienced no bleeding.

In the presence of multiple studies demonstrating lack of benefit from FFP and platelet transfusion, guidelines published by the American Association for the Study of Liver Disease (AASLD), the American Gastroenterological Association (AGA), and the Society of Interventional Radiology (SIR) acknowledge the inaccuracy of platelet count and INR in predicting bleeding risk.^12-14 Both AASLD and AGA recommend against routine transfusion of platelets and FFP prior to paracentesis.^12,13 SIR guidelines from 2019 recommend against using an INR threshold for low-risk procedures like paracentesis and lowered their recommended platelet transfusion threshold from less than 50,000/μL to less than 20,000/μL.¹⁴ While we have limited safety data for paracentesis in patients with very low platelet counts, Kurup et al observed no bleeding events in the 19 patients in their cohort with platelets less than 20,000/μL undergoing ultrasound-guided paracentesis.⁷

In addition to lack of proven benefit, preprocedure transfusion exposes patients to objective risk. Transfusion-­related acute lung injury and transfusion-associated circulatory overload develop at a rate of 0.48 and 3.8 per 100,000 components transfused, respectively.¹⁵ FFP transfusions also risk anaphylactic reactions with incidence ranging from 1:18,000 to 1:172,000.¹⁶ Platelets carry additional risk of bacterial contamination and resultant sepsis estimated at 1:5,000 to 1:8,000 per unit.¹⁷ Volume expansion from transfusions may contribute to portal hypertension and increase risk of variceal bleeding in decompensated liver disease.

Finally, FFP and platelet transfusions carry a significant cost. Rowley et al estimated eliminating preprocedure transfusions over 2 years and 3,116 paracenteses saved their institution $816,000.⁵ Furthermore, checking and correcting INR and thrombocytopenia can lead to procedural delay. Studies have demonstrated increased mortality from delaying paracentesis.¹⁸

While most patients undergoing paracentesis have cirrhosis, patients without cirrhosis also undergo this procedure. Although several cited studies examined paracentesis among all-comers with ascites, our recommendations specifically apply to patients with ascites from cirrhosis.

Furthermore, although no paracentesis data in patients with severe coagulopathy (INR >2.5 or platelet count <20,000/μL) suggest periprocedural transfusion helps, we also lack data to prove it does not help.

Current recommendations from the AASLD suggest correcting coagulopathy in patients with clinically evident disseminated intravascular coagulation or hyperfibrinolysis prior to procedures.¹² While no clear guidance related to paracentesis exists on when to assess for these entities, we recommend evaluating for them only when the clinical situation otherwise merits doing so and not solely for the purpose of screening prior to paracentesis. Measuring fibrinogen before paracentesis to predict bleeding risk is an emerging concept, but it cannot be routinely recommended at this time.¹³ Other factors that may play an important role in bleeding risk—ultrasound guidance, operator experience, and ability to avoid epigastric vessels and collateral veins—are beyond the scope of this article.

Given that laboratory evaluations like INR and platelet count cannot predict which patients with cirrhosis will experience major bleeding complications after paracentesis and given that routinely transfusing FFP or platelets does not confer benefit and may cause serious harm, providers should avoid measuring INR or platelet count to prepare for paracentesis. Likewise, providers should avoid routinely transfusing FFP and platelets prior to paracentesis even in the presence of abnormal laboratory values because such values do not accurately reflect bleeding risk in patients with cirrhosis. Perform clinically indicated paracentesis without the delays that accompany unnecessary laboratory evaluations or transfusions.

Keep the following in mind with patients presenting with ascites from cirrhosis:

• Do not routinely use platelet count or INR when preparing for paracentesis, whether diagnostic or therapeutic, because no evidence-based “cutoff” for safe performance of paracentesis exists.
• Do not routinely transfuse FFP or platelets for prophylaxis prior to paracentesis in patients with cirrhosis.
• Reserve preprocedure transfusion of FFP or platelets for patients with disseminated intravascular coagulation, hyperfibrinolysis, or other indications for transfusion unrelated to procedural prophylaxis.

Case series representing diverse institutional experiences with thousands of patients consistently demonstrate that bleeding after paracentesis is rare (<1%), mortality from bleeding occurs very infrequently, and neither INR nor platelet counts predict bleeding risk during paracentesis in cirrhosis. These studies demonstrate that abandoning routine correction of coagulopathy does not lead to worse outcomes, can avoid potentially significant transfusion-related adverse events, and can save scarce resources.

Returning to our clinical scenario, the hospitalist should not transfuse FFP or platelets and should not delay the diagnostic paracentesis.

Do you think this is a low-value practice? Is this truly a “Thing We Do for No Reason™”? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other “Things We Do for No Reason™” topics by emailing [email protected].

The authors wish to acknowledge James Burton, MD, H Raymond Tahhan, MD, John Hess, MD, MPH, and Terry Gernsheimer, MD, for directing the authors to useful references cited in the manuscript.

Inspired by the ABIM Foundation’s Choosing Wisely^® campaign, the “Things We Do for No Reason^™” (TWDFNR) series reviews practices that have become common parts of hospital care but may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent clear-cut conclusions or clinical practice standards but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion.

The hospitalist admits a 52-year-old man with alcoholic cirrhosis for tense ascites and altered mentation. Home medications include furosemide, spironolactone, lactulose, and rifaximin, but his family notes he ran out last week. Although afebrile and hemodynamically stable, the patient’s coagulopathy, with an international normalized ratio (INR) of 2.3, and thrombocytopenia, with a platelet count of 37,000/μL, worries the hospitalist. The hospitalist wonders whether to transfuse fresh frozen plasma (FFP) and platelets prior to diagnostic paracentesis to reduce the risk of procedural bleeding.

Many patients undergoing paracentesis have severe liver disease and present with both thrombocytopenia and elevated INRs. While platelet count and INR serve as surrogate markers for bleeding risk in many settings, clinicians often extrapolate this concept to patients with cirrhosis. Many hospitalists routinely check INR and platelet count and administer FFP and platelets prior to diagnostic or therapeutic paracentesis to mitigate procedure-related bleeding risk. Some medical resources recommend this practice,¹ while case reports and personal experiences with bleeding in these patients create availability bias that influences perception of bleeding risk.² One recent study of patients with decompensated cirrhosis presenting to a US tertiary care center found that, of those receiving large-­volume paracentesis, 22.2% received prophylactic FFP and 17.3% received prophylactic platelets before paracentesis.³

Advances in our understanding of coagulation in cirrhosis demonstrate neither INR nor platelet count accurately predict bleeding risk in this population. Additionally, evidence demonstrates the overall safety of paracentesis in cirrhosis—even in the presence of high INR and thrombocytopenia—and the lack of benefit from prophylactic transfusions with FFP or platelets.

Substantial evidence in patients with cirrhosis demonstrates that changes in coagulation and platelet function confer a “balanced coagulopathy” in which patients oscillate between hyper- and hypocoagulable states. In a cirrhotic liver, hepatic synthetic dysfunction results in a complex milieu through reduced production and plasma concentrations of both pro- and anticoagulant factors that can lead to either bleeding or clotting.⁴ This “rebalancing” makes prothrombin time (PT) and INR unreliable indicators of bleeding or clotting risk. Similarly, in patients with cirrhosis, thrombocytopenia does not necessarily reflect impaired clotting ability. These patients experience an increase in production of von Willebrand Factor, which may compensate for low platelet counts by producing stronger platelet adhesion to collagen.⁴ Unfortunately, we currently lack a reliable test or risk score to assess true bleeding risk in patients with cirrhosis.

Observational studies support these laboratory findings. Large case series consistently demonstrate no association between INR or platelet counts and bleeding risk in either diagnostic or therapeutic paracentesis, including large-volume paracentesis (See Appendix for a list of recent representative studies).^5-10 Moreover, prophylactic transfusion of FFP or platelets does not significantly reduce bleeding risk.

In a 1991 study by McVay et al, the researchers examined bleeding outcomes of 441 paracenteses performed on hospitalized patients.¹¹ Among patients who did not receive FFP prior to paracentesis, only one required a transfusion for procedure-­related bleeding, an event rate of 0.25%. This single patient had a normal platelet count and an elevated PT to the same extent as 261 others who underwent paracentesis without complication. In a pooled analysis that included 391 paracenteses and 207 thoracenteses, the authors concluded neither PT nor platelet level predicted bleeding risk. Similarly, the largest published case series on this topic examined 4,729 paracenteses over a decade on a liver unit and found low rates of major bleeding (0.19%).⁹ Furthermore, preprocedure INR or platelet count did not correlate with bleeding risk. The authors did not report preprocedure transfusion rates, but they noted transfusions occurred only “occasionally.”

Subsequent observational studies have consistently revealed low bleeding risks even in settings of high coagulopathy prevalence. Grabau et al reviewed all large-volume paracenteses performed in a gastroenterology clinic over 7 years.¹⁰ In over 1,100 procedures, no major bleeding events occurred despite 27% of patients having INR greater than 2.0 and 54% having platelet counts less than 50,000/μL. Kurup et al examined bleeding risk among 304 procedures performed on patients with platelet counts less than 50,000/μL referred to radiology for ultrasound-guided paracentesis.⁷ Three bleeding events occurred, an overall event rate of 0.99%. They also found no association between preprocedure platelet count and bleeding risk.

In addition to observational data, one randomized, controlled trial evaluated the effects of FFP and platelet administration on bleeding risk among 60 patients with cirrhosis undergoing invasive procedures, including 19 paracenteses.⁶ Enrollment criteria included INR greater than 1.8 and/or platelet count less than 50,000/μL. One hundred percent of patients randomized to the usual care control arm received platelets or FFP as compared to 17% in the thromboelastography (TEG)–guided transfusion strategy arm. TEG assesses the viscoelastic properties of evolving clot formation in whole blood. Only one patient, a patient in the control arm who received FFP, developed procedure-related bleeding. Although receiving many fewer transfusions, the TEG-guided group experienced no bleeding.

In the presence of multiple studies demonstrating lack of benefit from FFP and platelet transfusion, guidelines published by the American Association for the Study of Liver Disease (AASLD), the American Gastroenterological Association (AGA), and the Society of Interventional Radiology (SIR) acknowledge the inaccuracy of platelet count and INR in predicting bleeding risk.^12-14 Both AASLD and AGA recommend against routine transfusion of platelets and FFP prior to paracentesis.^12,13 SIR guidelines from 2019 recommend against using an INR threshold for low-risk procedures like paracentesis and lowered their recommended platelet transfusion threshold from less than 50,000/μL to less than 20,000/μL.¹⁴ While we have limited safety data for paracentesis in patients with very low platelet counts, Kurup et al observed no bleeding events in the 19 patients in their cohort with platelets less than 20,000/μL undergoing ultrasound-guided paracentesis.⁷

In addition to lack of proven benefit, preprocedure transfusion exposes patients to objective risk. Transfusion-­related acute lung injury and transfusion-associated circulatory overload develop at a rate of 0.48 and 3.8 per 100,000 components transfused, respectively.¹⁵ FFP transfusions also risk anaphylactic reactions with incidence ranging from 1:18,000 to 1:172,000.¹⁶ Platelets carry additional risk of bacterial contamination and resultant sepsis estimated at 1:5,000 to 1:8,000 per unit.¹⁷ Volume expansion from transfusions may contribute to portal hypertension and increase risk of variceal bleeding in decompensated liver disease.

Finally, FFP and platelet transfusions carry a significant cost. Rowley et al estimated eliminating preprocedure transfusions over 2 years and 3,116 paracenteses saved their institution $816,000.⁵ Furthermore, checking and correcting INR and thrombocytopenia can lead to procedural delay. Studies have demonstrated increased mortality from delaying paracentesis.¹⁸

While most patients undergoing paracentesis have cirrhosis, patients without cirrhosis also undergo this procedure. Although several cited studies examined paracentesis among all-comers with ascites, our recommendations specifically apply to patients with ascites from cirrhosis.

Furthermore, although no paracentesis data in patients with severe coagulopathy (INR >2.5 or platelet count <20,000/μL) suggest periprocedural transfusion helps, we also lack data to prove it does not help.

Current recommendations from the AASLD suggest correcting coagulopathy in patients with clinically evident disseminated intravascular coagulation or hyperfibrinolysis prior to procedures.¹² While no clear guidance related to paracentesis exists on when to assess for these entities, we recommend evaluating for them only when the clinical situation otherwise merits doing so and not solely for the purpose of screening prior to paracentesis. Measuring fibrinogen before paracentesis to predict bleeding risk is an emerging concept, but it cannot be routinely recommended at this time.¹³ Other factors that may play an important role in bleeding risk—ultrasound guidance, operator experience, and ability to avoid epigastric vessels and collateral veins—are beyond the scope of this article.

Given that laboratory evaluations like INR and platelet count cannot predict which patients with cirrhosis will experience major bleeding complications after paracentesis and given that routinely transfusing FFP or platelets does not confer benefit and may cause serious harm, providers should avoid measuring INR or platelet count to prepare for paracentesis. Likewise, providers should avoid routinely transfusing FFP and platelets prior to paracentesis even in the presence of abnormal laboratory values because such values do not accurately reflect bleeding risk in patients with cirrhosis. Perform clinically indicated paracentesis without the delays that accompany unnecessary laboratory evaluations or transfusions.

Keep the following in mind with patients presenting with ascites from cirrhosis:

• Do not routinely use platelet count or INR when preparing for paracentesis, whether diagnostic or therapeutic, because no evidence-based “cutoff” for safe performance of paracentesis exists.
• Do not routinely transfuse FFP or platelets for prophylaxis prior to paracentesis in patients with cirrhosis.
• Reserve preprocedure transfusion of FFP or platelets for patients with disseminated intravascular coagulation, hyperfibrinolysis, or other indications for transfusion unrelated to procedural prophylaxis.

Case series representing diverse institutional experiences with thousands of patients consistently demonstrate that bleeding after paracentesis is rare (<1%), mortality from bleeding occurs very infrequently, and neither INR nor platelet counts predict bleeding risk during paracentesis in cirrhosis. These studies demonstrate that abandoning routine correction of coagulopathy does not lead to worse outcomes, can avoid potentially significant transfusion-related adverse events, and can save scarce resources.

Returning to our clinical scenario, the hospitalist should not transfuse FFP or platelets and should not delay the diagnostic paracentesis.

Do you think this is a low-value practice? Is this truly a “Thing We Do for No Reason™”? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other “Things We Do for No Reason™” topics by emailing [email protected].

The authors wish to acknowledge James Burton, MD, H Raymond Tahhan, MD, John Hess, MD, MPH, and Terry Gernsheimer, MD, for directing the authors to useful references cited in the manuscript.

Inspired by the ABIM Foundation’s Choosing Wisely^® campaign, the “Things We Do for No Reason^™” (TWDFNR) series reviews practices that have become common parts of hospital care but may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent clear-cut conclusions or clinical practice standards but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion.

The hospitalist admits a 52-year-old man with alcoholic cirrhosis for tense ascites and altered mentation. Home medications include furosemide, spironolactone, lactulose, and rifaximin, but his family notes he ran out last week. Although afebrile and hemodynamically stable, the patient’s coagulopathy, with an international normalized ratio (INR) of 2.3, and thrombocytopenia, with a platelet count of 37,000/μL, worries the hospitalist. The hospitalist wonders whether to transfuse fresh frozen plasma (FFP) and platelets prior to diagnostic paracentesis to reduce the risk of procedural bleeding.

Many patients undergoing paracentesis have severe liver disease and present with both thrombocytopenia and elevated INRs. While platelet count and INR serve as surrogate markers for bleeding risk in many settings, clinicians often extrapolate this concept to patients with cirrhosis. Many hospitalists routinely check INR and platelet count and administer FFP and platelets prior to diagnostic or therapeutic paracentesis to mitigate procedure-related bleeding risk. Some medical resources recommend this practice,¹ while case reports and personal experiences with bleeding in these patients create availability bias that influences perception of bleeding risk.² One recent study of patients with decompensated cirrhosis presenting to a US tertiary care center found that, of those receiving large-­volume paracentesis, 22.2% received prophylactic FFP and 17.3% received prophylactic platelets before paracentesis.³

Advances in our understanding of coagulation in cirrhosis demonstrate neither INR nor platelet count accurately predict bleeding risk in this population. Additionally, evidence demonstrates the overall safety of paracentesis in cirrhosis—even in the presence of high INR and thrombocytopenia—and the lack of benefit from prophylactic transfusions with FFP or platelets.

Substantial evidence in patients with cirrhosis demonstrates that changes in coagulation and platelet function confer a “balanced coagulopathy” in which patients oscillate between hyper- and hypocoagulable states. In a cirrhotic liver, hepatic synthetic dysfunction results in a complex milieu through reduced production and plasma concentrations of both pro- and anticoagulant factors that can lead to either bleeding or clotting.⁴ This “rebalancing” makes prothrombin time (PT) and INR unreliable indicators of bleeding or clotting risk. Similarly, in patients with cirrhosis, thrombocytopenia does not necessarily reflect impaired clotting ability. These patients experience an increase in production of von Willebrand Factor, which may compensate for low platelet counts by producing stronger platelet adhesion to collagen.⁴ Unfortunately, we currently lack a reliable test or risk score to assess true bleeding risk in patients with cirrhosis.

Observational studies support these laboratory findings. Large case series consistently demonstrate no association between INR or platelet counts and bleeding risk in either diagnostic or therapeutic paracentesis, including large-volume paracentesis (See Appendix for a list of recent representative studies).^5-10 Moreover, prophylactic transfusion of FFP or platelets does not significantly reduce bleeding risk.

In a 1991 study by McVay et al, the researchers examined bleeding outcomes of 441 paracenteses performed on hospitalized patients.¹¹ Among patients who did not receive FFP prior to paracentesis, only one required a transfusion for procedure-­related bleeding, an event rate of 0.25%. This single patient had a normal platelet count and an elevated PT to the same extent as 261 others who underwent paracentesis without complication. In a pooled analysis that included 391 paracenteses and 207 thoracenteses, the authors concluded neither PT nor platelet level predicted bleeding risk. Similarly, the largest published case series on this topic examined 4,729 paracenteses over a decade on a liver unit and found low rates of major bleeding (0.19%).⁹ Furthermore, preprocedure INR or platelet count did not correlate with bleeding risk. The authors did not report preprocedure transfusion rates, but they noted transfusions occurred only “occasionally.”

Subsequent observational studies have consistently revealed low bleeding risks even in settings of high coagulopathy prevalence. Grabau et al reviewed all large-volume paracenteses performed in a gastroenterology clinic over 7 years.¹⁰ In over 1,100 procedures, no major bleeding events occurred despite 27% of patients having INR greater than 2.0 and 54% having platelet counts less than 50,000/μL. Kurup et al examined bleeding risk among 304 procedures performed on patients with platelet counts less than 50,000/μL referred to radiology for ultrasound-guided paracentesis.⁷ Three bleeding events occurred, an overall event rate of 0.99%. They also found no association between preprocedure platelet count and bleeding risk.

In addition to observational data, one randomized, controlled trial evaluated the effects of FFP and platelet administration on bleeding risk among 60 patients with cirrhosis undergoing invasive procedures, including 19 paracenteses.⁶ Enrollment criteria included INR greater than 1.8 and/or platelet count less than 50,000/μL. One hundred percent of patients randomized to the usual care control arm received platelets or FFP as compared to 17% in the thromboelastography (TEG)–guided transfusion strategy arm. TEG assesses the viscoelastic properties of evolving clot formation in whole blood. Only one patient, a patient in the control arm who received FFP, developed procedure-related bleeding. Although receiving many fewer transfusions, the TEG-guided group experienced no bleeding.

In the presence of multiple studies demonstrating lack of benefit from FFP and platelet transfusion, guidelines published by the American Association for the Study of Liver Disease (AASLD), the American Gastroenterological Association (AGA), and the Society of Interventional Radiology (SIR) acknowledge the inaccuracy of platelet count and INR in predicting bleeding risk.^12-14 Both AASLD and AGA recommend against routine transfusion of platelets and FFP prior to paracentesis.^12,13 SIR guidelines from 2019 recommend against using an INR threshold for low-risk procedures like paracentesis and lowered their recommended platelet transfusion threshold from less than 50,000/μL to less than 20,000/μL.¹⁴ While we have limited safety data for paracentesis in patients with very low platelet counts, Kurup et al observed no bleeding events in the 19 patients in their cohort with platelets less than 20,000/μL undergoing ultrasound-guided paracentesis.⁷

In addition to lack of proven benefit, preprocedure transfusion exposes patients to objective risk. Transfusion-­related acute lung injury and transfusion-associated circulatory overload develop at a rate of 0.48 and 3.8 per 100,000 components transfused, respectively.¹⁵ FFP transfusions also risk anaphylactic reactions with incidence ranging from 1:18,000 to 1:172,000.¹⁶ Platelets carry additional risk of bacterial contamination and resultant sepsis estimated at 1:5,000 to 1:8,000 per unit.¹⁷ Volume expansion from transfusions may contribute to portal hypertension and increase risk of variceal bleeding in decompensated liver disease.

Finally, FFP and platelet transfusions carry a significant cost. Rowley et al estimated eliminating preprocedure transfusions over 2 years and 3,116 paracenteses saved their institution $816,000.⁵ Furthermore, checking and correcting INR and thrombocytopenia can lead to procedural delay. Studies have demonstrated increased mortality from delaying paracentesis.¹⁸

While most patients undergoing paracentesis have cirrhosis, patients without cirrhosis also undergo this procedure. Although several cited studies examined paracentesis among all-comers with ascites, our recommendations specifically apply to patients with ascites from cirrhosis.

Furthermore, although no paracentesis data in patients with severe coagulopathy (INR >2.5 or platelet count <20,000/μL) suggest periprocedural transfusion helps, we also lack data to prove it does not help.

Current recommendations from the AASLD suggest correcting coagulopathy in patients with clinically evident disseminated intravascular coagulation or hyperfibrinolysis prior to procedures.¹² While no clear guidance related to paracentesis exists on when to assess for these entities, we recommend evaluating for them only when the clinical situation otherwise merits doing so and not solely for the purpose of screening prior to paracentesis. Measuring fibrinogen before paracentesis to predict bleeding risk is an emerging concept, but it cannot be routinely recommended at this time.¹³ Other factors that may play an important role in bleeding risk—ultrasound guidance, operator experience, and ability to avoid epigastric vessels and collateral veins—are beyond the scope of this article.

Given that laboratory evaluations like INR and platelet count cannot predict which patients with cirrhosis will experience major bleeding complications after paracentesis and given that routinely transfusing FFP or platelets does not confer benefit and may cause serious harm, providers should avoid measuring INR or platelet count to prepare for paracentesis. Likewise, providers should avoid routinely transfusing FFP and platelets prior to paracentesis even in the presence of abnormal laboratory values because such values do not accurately reflect bleeding risk in patients with cirrhosis. Perform clinically indicated paracentesis without the delays that accompany unnecessary laboratory evaluations or transfusions.

Keep the following in mind with patients presenting with ascites from cirrhosis:

• Do not routinely use platelet count or INR when preparing for paracentesis, whether diagnostic or therapeutic, because no evidence-based “cutoff” for safe performance of paracentesis exists.
• Do not routinely transfuse FFP or platelets for prophylaxis prior to paracentesis in patients with cirrhosis.
• Reserve preprocedure transfusion of FFP or platelets for patients with disseminated intravascular coagulation, hyperfibrinolysis, or other indications for transfusion unrelated to procedural prophylaxis.

Case series representing diverse institutional experiences with thousands of patients consistently demonstrate that bleeding after paracentesis is rare (<1%), mortality from bleeding occurs very infrequently, and neither INR nor platelet counts predict bleeding risk during paracentesis in cirrhosis. These studies demonstrate that abandoning routine correction of coagulopathy does not lead to worse outcomes, can avoid potentially significant transfusion-related adverse events, and can save scarce resources.

Returning to our clinical scenario, the hospitalist should not transfuse FFP or platelets and should not delay the diagnostic paracentesis.

Do you think this is a low-value practice? Is this truly a “Thing We Do for No Reason™”? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other “Things We Do for No Reason™” topics by emailing [email protected].

The authors wish to acknowledge James Burton, MD, H Raymond Tahhan, MD, John Hess, MD, MPH, and Terry Gernsheimer, MD, for directing the authors to useful references cited in the manuscript.

Inspired by the ABIM Foundation’s Choosing Wisely^® campaign, the “Things We Do for No Reason^™” (TWDFNR) series reviews practices that have become common parts of hospital care but may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent  clear-cut conclusions or clinical practice standards but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion.

An 88-year-old woman with a history of dementia presents to the emergency room with new-onset dyspnea following 2 days of a self-limited gastrointestinal illness associated with nausea, vomiting, and diarrhea. After noting a new supplemental oxygen requirement of 4 L and a temperature of 38.6 °C, the hospitalist’s exam finds an edentulous patient with bibasilar lung crackles and a nontender abdomen. Taking into account her elevated white blood cell count and chest radiograph with right greater than left bibasilar opacities, the admitting hospitalist diagnoses aspiration pneumonia (AP) and specifically selects an antibiotic regimen with anaerobic coverage.

Aspiration, the inhalation of oropharyngeal or gastric materials into the lung, takes one of the following three forms: (1) “microaspiration,” wherein a small number of virulent organisms from oropharynx gains entry into the alveoli, (2) “macroaspiration,” wherein a large volume of typically less virulent organisms gains entry into the airways, or (3) a combination of the two. Hospitalists may struggle to distinguish unwitnessed macroaspiration causing AP from other typical causes of pneumonia, such as community-acquired pneumonia (CAP) or hospital-acquired pneumonia (HAP).¹ A hospitalist should suspect macroaspiration—the most common cause of AP—in patients with risk factors such as dysphagia, diminished cough reflex or impaired swallowing, and infiltrates in the dependent bronchopulmonary segments, or of course, in cases of witnessed aspiration.²

Moreover, hospitalists must differentiate AP, an infectious entity, from aspiration pneumonitis, a noninfectious entity caused by macroaspiration of mostly sterile gastric content. Aspiration pneumonitis presents with acute lung injury within hours of an aspiration event, whereas AP entails a gradual onset of symptoms and signs of pneumonia.² Although aspiration pneumonitis can present dramatically with hypoxemia and pulmonary edema and may evolve into AP, patients do not initially benefit from empiric antibiotics.¹

In the 1970s, several studies of patients who were presumed to have AP because of risk factors for macroaspiration, such as alcohol use disorder, illicit drug use, and seizure disorder, identified anaerobes as major etiologic pathogens. These studies reported the presence of putrid sputum and obtained samples through invasive methods (eg, transtracheal aspirates, thoracentesis, and blood cultures).^3,4 Many of the patients studied had radiographic findings of pleuropulmonary disease. For example, in the study by Bartlett et al, 70% of patients had radiographic evidence of abscess or pulmonary necrosis. These findings led to the assumption that anaerobes play a significant role in all cases of aspiration-related pulmonary syndromes. Because anaerobic bacteria live in the gingival sulcus, with an especially high burden in dental plaques, their role as a potential pathogen in AP may seem logical.⁵ Given the backdrop of those concerns, Kioka et al found that providers treated 90% of presumed AP patients in the intensive care unit with antibiotics that have anaerobic activity despite only 30% meeting the criteria for anaerobic coverage.⁶

In contrast to the population of patients with AP described from the 1970s, we now diagnose AP more frequently in nursing home residents, the elderly with cognitive impairment, and those with tube feed dependence, dysphagia, or gastrointestinal motility disorders.¹ Concurrent with this change in the epidemiology of AP, we have witnessed a shift in recovered bacteria from anaerobes to aerobes in recent studies.^7,8 In an intensive care unit study from 1999, respiratory tract organisms of patients with suspected aspiration mirrored those of patients with CAP or HAP.⁹ In a systematic review of eight observational studies that included studies from 1993 to 2014 and involved elderly patients with uncomplicated AP, only two out of eight studies demonstrated the presence of anaerobes in respiratory cultures. Even in those two studies, anaerobic bacteria frequently coexisted with aerobes. The majority of organisms in all eight studies consisted of aerobic gram-positives, gram-negatives, or both.¹⁰

A study by El-Solh et al most frequently isolated pathogenic aerobic gram-negative bacteria (49% of cases), followed by anaerobic bacteria (16%), among institutionalized elderly patients with severe AP diagnosed by clinical features. In that same study, most anaerobes coexisted with aerobic gram-negative bacteria, and the clinical illness promptly resolved in the absence of specific anaerobic coverage.¹¹ AP can be successfully treated without anaerobic coverage due to a variety of factors: the insignificant role of anaerobes in the pathogenesis of uncomplicated AP, lower severity of illness in the absence of abscesses or pulmonary necrosis (uncomplicated), and altered local redox-potential from the elimination of aerobic pathogens, which effectively also treats anaerobes.¹ Moreover, anaerobes possess generally less virulence in comparison with aerobes. AP from these organisms typically requires risk for excessive oral growth (eg, periodontal disease) and macroaspiration of a large number of organisms.⁵

There are also potential harms associated with the unnecessary treatment of anaerobic bacteria. Since anaerobes account for the majority of the bacteria present in the bowel, targeting anaerobes can result in gut dysbiosis.¹ Moreover, a prospective study showed an increase in the incidence of vancomycin-resistant enterococci and antibiotic-resistant gram-negative bacteria associated with the empiric use of antibiotics with anaerobic activity.¹² Finally, a systematic review detailed the high incidence of Clostridioides difficile infections among patients receiving clindamycin and carbapenems.¹³

Despite the predominance of aerobic organisms in the respiratory tract specimens of patients diagnosed with AP in the current era, situations still exist that require treatment of anaerobes. These include necrotizing pneumonia, empyema, or lung abscess.² Additionally, patients with severe periodontal disease may harbor anaerobic bacteria such as Bacteroides species, Peptostreptococcus species, and Actinomyces israelii.⁵ When we suspect macroaspiration leading to AP, patients with severe periodontal disease may benefit from anaerobic coverage. Putrid sputum generation may indicate the presence of anaerobic organisms that produce the characteristic foul odor of short-chain volatile fatty acids observed in patients with lung abscess or empyema.² It often takes about 8 to 14 days after an aspiration event for lung cavitation or empyema to develop.¹⁴ Therefore, a longer duration of illness or putrid sputum production may signal a significant concurrent burden of anaerobes. The 2019 official guidelines of the American Thoracic Society and Infectious Disease Society of America recommend adding anaerobic coverage to CAP only when empyema or lung abscess is suspected (conditional recommendation, very low quality of evidence).¹⁵

When you suspect AP in a patient, categorize it as either community or hospital acquired based on risk factors similar to CAP or HAP. For patients with witnessed macroaspiration or in patients with substantial macroaspiration risk factors, perform a radiologic evaluation and a thorough oral examination to evaluate for poor dentition, gingival disease (marked redness, tendency to bleed, ulceration), and tongue coating. For patients presenting from the community with suspected AP without complications, treat with the standard therapy (without additional anaerobic coverage) for CAP. Provide empiric anaerobic coverage for complicated AP (eg, lung abscess, necrosis, or empyema) or for macroaspiration in the setting of severe periodontal disease, putrid sputum, or longer duration of illness. Similarly, treat hospital-acquired AP as HAP (Figure).

When prescribing anaerobic coverage of AP, use combination drugs that include a ß-lactamase inhibitor (eg, ampicillin-sulbactam), clindamycin (either alone or in combination with ß-lactams), or moxifloxacin.¹ Most anaerobes have ß-lactamase or cephalosporinase activity, which renders penicillin and cephalosporins ineffective. Despite its potential side effects, such as C difficile infection, treating with clindamycin has the benefit of a relatively low cost and its association with lower rates of methicillin-resistant Staphylococcus aureus emergence after treatment.¹⁶ Piperacillin-tazobactam and carbapenems also have excellent anaerobic coverage, but we should reserve them for more severe and complicated cases of AP given their extensive antibacterial activity and concern for the emergence of resistance.⁸ Although well known and used for decades for its activity against clinically important anaerobes, avoid metronidazole due to its reduced cure rate in lung abscess caused by microaerophilic streptococci of the oral cavity.¹⁷ Due to a lack of evidence, we do not recommend the use of metronidazole in lung infections.

• Empirically treat most suspected cases of AP with regimens similar to the standard antibiotics for CAP and HAP. In the absence of specific risk factors for anaerobic infections, do not routinely provide anaerobic coverage.
• Provide anaerobic coverage empirically for AP associated with macroaspiration in the setting of severe periodontal disease, putrid sputum, or longer duration of illness.
• Provide anaerobic coverage in AP with evidence of necrotizing pneumonia, empyema, or lung abscess.

Current evidence does not support routine anaerobic coverage of AP in the absence of identifiable risk factors for an anaerobic lung infection.

In consideration of the clinical case, importantly, she has no periodontal disease and no evidence for necrotizing pneumonia, empyema, or lung abscess radiographically. For these reasons, select an empiric antibiotic regime that targets CAP organisms predominantly and forgo additional anaerobic coverage.

Do you think this is a low-value practice? Is this truly a “Thing We Do for No Reason ^™ ”? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other “Things We Do for No Reason ^™ ” topics by emailing [email protected].

The authors have no conflicts of interest relevant to this article.

Inspired by the ABIM Foundation’s Choosing Wisely^® campaign, the “Things We Do for No Reason^™” (TWDFNR) series reviews practices that have become common parts of hospital care but may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent  clear-cut conclusions or clinical practice standards but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion.

An 88-year-old woman with a history of dementia presents to the emergency room with new-onset dyspnea following 2 days of a self-limited gastrointestinal illness associated with nausea, vomiting, and diarrhea. After noting a new supplemental oxygen requirement of 4 L and a temperature of 38.6 °C, the hospitalist’s exam finds an edentulous patient with bibasilar lung crackles and a nontender abdomen. Taking into account her elevated white blood cell count and chest radiograph with right greater than left bibasilar opacities, the admitting hospitalist diagnoses aspiration pneumonia (AP) and specifically selects an antibiotic regimen with anaerobic coverage.

Aspiration, the inhalation of oropharyngeal or gastric materials into the lung, takes one of the following three forms: (1) “microaspiration,” wherein a small number of virulent organisms from oropharynx gains entry into the alveoli, (2) “macroaspiration,” wherein a large volume of typically less virulent organisms gains entry into the airways, or (3) a combination of the two. Hospitalists may struggle to distinguish unwitnessed macroaspiration causing AP from other typical causes of pneumonia, such as community-acquired pneumonia (CAP) or hospital-acquired pneumonia (HAP).¹ A hospitalist should suspect macroaspiration—the most common cause of AP—in patients with risk factors such as dysphagia, diminished cough reflex or impaired swallowing, and infiltrates in the dependent bronchopulmonary segments, or of course, in cases of witnessed aspiration.²

Moreover, hospitalists must differentiate AP, an infectious entity, from aspiration pneumonitis, a noninfectious entity caused by macroaspiration of mostly sterile gastric content. Aspiration pneumonitis presents with acute lung injury within hours of an aspiration event, whereas AP entails a gradual onset of symptoms and signs of pneumonia.² Although aspiration pneumonitis can present dramatically with hypoxemia and pulmonary edema and may evolve into AP, patients do not initially benefit from empiric antibiotics.¹

In the 1970s, several studies of patients who were presumed to have AP because of risk factors for macroaspiration, such as alcohol use disorder, illicit drug use, and seizure disorder, identified anaerobes as major etiologic pathogens. These studies reported the presence of putrid sputum and obtained samples through invasive methods (eg, transtracheal aspirates, thoracentesis, and blood cultures).^3,4 Many of the patients studied had radiographic findings of pleuropulmonary disease. For example, in the study by Bartlett et al, 70% of patients had radiographic evidence of abscess or pulmonary necrosis. These findings led to the assumption that anaerobes play a significant role in all cases of aspiration-related pulmonary syndromes. Because anaerobic bacteria live in the gingival sulcus, with an especially high burden in dental plaques, their role as a potential pathogen in AP may seem logical.⁵ Given the backdrop of those concerns, Kioka et al found that providers treated 90% of presumed AP patients in the intensive care unit with antibiotics that have anaerobic activity despite only 30% meeting the criteria for anaerobic coverage.⁶

In contrast to the population of patients with AP described from the 1970s, we now diagnose AP more frequently in nursing home residents, the elderly with cognitive impairment, and those with tube feed dependence, dysphagia, or gastrointestinal motility disorders.¹ Concurrent with this change in the epidemiology of AP, we have witnessed a shift in recovered bacteria from anaerobes to aerobes in recent studies.^7,8 In an intensive care unit study from 1999, respiratory tract organisms of patients with suspected aspiration mirrored those of patients with CAP or HAP.⁹ In a systematic review of eight observational studies that included studies from 1993 to 2014 and involved elderly patients with uncomplicated AP, only two out of eight studies demonstrated the presence of anaerobes in respiratory cultures. Even in those two studies, anaerobic bacteria frequently coexisted with aerobes. The majority of organisms in all eight studies consisted of aerobic gram-positives, gram-negatives, or both.¹⁰

A study by El-Solh et al most frequently isolated pathogenic aerobic gram-negative bacteria (49% of cases), followed by anaerobic bacteria (16%), among institutionalized elderly patients with severe AP diagnosed by clinical features. In that same study, most anaerobes coexisted with aerobic gram-negative bacteria, and the clinical illness promptly resolved in the absence of specific anaerobic coverage.¹¹ AP can be successfully treated without anaerobic coverage due to a variety of factors: the insignificant role of anaerobes in the pathogenesis of uncomplicated AP, lower severity of illness in the absence of abscesses or pulmonary necrosis (uncomplicated), and altered local redox-potential from the elimination of aerobic pathogens, which effectively also treats anaerobes.¹ Moreover, anaerobes possess generally less virulence in comparison with aerobes. AP from these organisms typically requires risk for excessive oral growth (eg, periodontal disease) and macroaspiration of a large number of organisms.⁵

There are also potential harms associated with the unnecessary treatment of anaerobic bacteria. Since anaerobes account for the majority of the bacteria present in the bowel, targeting anaerobes can result in gut dysbiosis.¹ Moreover, a prospective study showed an increase in the incidence of vancomycin-resistant enterococci and antibiotic-resistant gram-negative bacteria associated with the empiric use of antibiotics with anaerobic activity.¹² Finally, a systematic review detailed the high incidence of Clostridioides difficile infections among patients receiving clindamycin and carbapenems.¹³

Despite the predominance of aerobic organisms in the respiratory tract specimens of patients diagnosed with AP in the current era, situations still exist that require treatment of anaerobes. These include necrotizing pneumonia, empyema, or lung abscess.² Additionally, patients with severe periodontal disease may harbor anaerobic bacteria such as Bacteroides species, Peptostreptococcus species, and Actinomyces israelii.⁵ When we suspect macroaspiration leading to AP, patients with severe periodontal disease may benefit from anaerobic coverage. Putrid sputum generation may indicate the presence of anaerobic organisms that produce the characteristic foul odor of short-chain volatile fatty acids observed in patients with lung abscess or empyema.² It often takes about 8 to 14 days after an aspiration event for lung cavitation or empyema to develop.¹⁴ Therefore, a longer duration of illness or putrid sputum production may signal a significant concurrent burden of anaerobes. The 2019 official guidelines of the American Thoracic Society and Infectious Disease Society of America recommend adding anaerobic coverage to CAP only when empyema or lung abscess is suspected (conditional recommendation, very low quality of evidence).¹⁵

When you suspect AP in a patient, categorize it as either community or hospital acquired based on risk factors similar to CAP or HAP. For patients with witnessed macroaspiration or in patients with substantial macroaspiration risk factors, perform a radiologic evaluation and a thorough oral examination to evaluate for poor dentition, gingival disease (marked redness, tendency to bleed, ulceration), and tongue coating. For patients presenting from the community with suspected AP without complications, treat with the standard therapy (without additional anaerobic coverage) for CAP. Provide empiric anaerobic coverage for complicated AP (eg, lung abscess, necrosis, or empyema) or for macroaspiration in the setting of severe periodontal disease, putrid sputum, or longer duration of illness. Similarly, treat hospital-acquired AP as HAP (Figure).

When prescribing anaerobic coverage of AP, use combination drugs that include a ß-lactamase inhibitor (eg, ampicillin-sulbactam), clindamycin (either alone or in combination with ß-lactams), or moxifloxacin.¹ Most anaerobes have ß-lactamase or cephalosporinase activity, which renders penicillin and cephalosporins ineffective. Despite its potential side effects, such as C difficile infection, treating with clindamycin has the benefit of a relatively low cost and its association with lower rates of methicillin-resistant Staphylococcus aureus emergence after treatment.¹⁶ Piperacillin-tazobactam and carbapenems also have excellent anaerobic coverage, but we should reserve them for more severe and complicated cases of AP given their extensive antibacterial activity and concern for the emergence of resistance.⁸ Although well known and used for decades for its activity against clinically important anaerobes, avoid metronidazole due to its reduced cure rate in lung abscess caused by microaerophilic streptococci of the oral cavity.¹⁷ Due to a lack of evidence, we do not recommend the use of metronidazole in lung infections.

• Empirically treat most suspected cases of AP with regimens similar to the standard antibiotics for CAP and HAP. In the absence of specific risk factors for anaerobic infections, do not routinely provide anaerobic coverage.
• Provide anaerobic coverage empirically for AP associated with macroaspiration in the setting of severe periodontal disease, putrid sputum, or longer duration of illness.
• Provide anaerobic coverage in AP with evidence of necrotizing pneumonia, empyema, or lung abscess.

Current evidence does not support routine anaerobic coverage of AP in the absence of identifiable risk factors for an anaerobic lung infection.

In consideration of the clinical case, importantly, she has no periodontal disease and no evidence for necrotizing pneumonia, empyema, or lung abscess radiographically. For these reasons, select an empiric antibiotic regime that targets CAP organisms predominantly and forgo additional anaerobic coverage.

Do you think this is a low-value practice? Is this truly a “Thing We Do for No Reason ^™ ”? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other “Things We Do for No Reason ^™ ” topics by emailing [email protected].

The authors have no conflicts of interest relevant to this article.

Inspired by the ABIM Foundation’s Choosing Wisely^® campaign, the “Things We Do for No Reason^™” (TWDFNR) series reviews practices that have become common parts of hospital care but may provide little value to our patients. Practices reviewed in the TWDFNR series do not represent  clear-cut conclusions or clinical practice standards but are meant as a starting place for research and active discussions among hospitalists and patients. We invite you to be part of that discussion.

An 88-year-old woman with a history of dementia presents to the emergency room with new-onset dyspnea following 2 days of a self-limited gastrointestinal illness associated with nausea, vomiting, and diarrhea. After noting a new supplemental oxygen requirement of 4 L and a temperature of 38.6 °C, the hospitalist’s exam finds an edentulous patient with bibasilar lung crackles and a nontender abdomen. Taking into account her elevated white blood cell count and chest radiograph with right greater than left bibasilar opacities, the admitting hospitalist diagnoses aspiration pneumonia (AP) and specifically selects an antibiotic regimen with anaerobic coverage.

Aspiration, the inhalation of oropharyngeal or gastric materials into the lung, takes one of the following three forms: (1) “microaspiration,” wherein a small number of virulent organisms from oropharynx gains entry into the alveoli, (2) “macroaspiration,” wherein a large volume of typically less virulent organisms gains entry into the airways, or (3) a combination of the two. Hospitalists may struggle to distinguish unwitnessed macroaspiration causing AP from other typical causes of pneumonia, such as community-acquired pneumonia (CAP) or hospital-acquired pneumonia (HAP).¹ A hospitalist should suspect macroaspiration—the most common cause of AP—in patients with risk factors such as dysphagia, diminished cough reflex or impaired swallowing, and infiltrates in the dependent bronchopulmonary segments, or of course, in cases of witnessed aspiration.²

Moreover, hospitalists must differentiate AP, an infectious entity, from aspiration pneumonitis, a noninfectious entity caused by macroaspiration of mostly sterile gastric content. Aspiration pneumonitis presents with acute lung injury within hours of an aspiration event, whereas AP entails a gradual onset of symptoms and signs of pneumonia.² Although aspiration pneumonitis can present dramatically with hypoxemia and pulmonary edema and may evolve into AP, patients do not initially benefit from empiric antibiotics.¹

In the 1970s, several studies of patients who were presumed to have AP because of risk factors for macroaspiration, such as alcohol use disorder, illicit drug use, and seizure disorder, identified anaerobes as major etiologic pathogens. These studies reported the presence of putrid sputum and obtained samples through invasive methods (eg, transtracheal aspirates, thoracentesis, and blood cultures).^3,4 Many of the patients studied had radiographic findings of pleuropulmonary disease. For example, in the study by Bartlett et al, 70% of patients had radiographic evidence of abscess or pulmonary necrosis. These findings led to the assumption that anaerobes play a significant role in all cases of aspiration-related pulmonary syndromes. Because anaerobic bacteria live in the gingival sulcus, with an especially high burden in dental plaques, their role as a potential pathogen in AP may seem logical.⁵ Given the backdrop of those concerns, Kioka et al found that providers treated 90% of presumed AP patients in the intensive care unit with antibiotics that have anaerobic activity despite only 30% meeting the criteria for anaerobic coverage.⁶

In contrast to the population of patients with AP described from the 1970s, we now diagnose AP more frequently in nursing home residents, the elderly with cognitive impairment, and those with tube feed dependence, dysphagia, or gastrointestinal motility disorders.¹ Concurrent with this change in the epidemiology of AP, we have witnessed a shift in recovered bacteria from anaerobes to aerobes in recent studies.^7,8 In an intensive care unit study from 1999, respiratory tract organisms of patients with suspected aspiration mirrored those of patients with CAP or HAP.⁹ In a systematic review of eight observational studies that included studies from 1993 to 2014 and involved elderly patients with uncomplicated AP, only two out of eight studies demonstrated the presence of anaerobes in respiratory cultures. Even in those two studies, anaerobic bacteria frequently coexisted with aerobes. The majority of organisms in all eight studies consisted of aerobic gram-positives, gram-negatives, or both.¹⁰

A study by El-Solh et al most frequently isolated pathogenic aerobic gram-negative bacteria (49% of cases), followed by anaerobic bacteria (16%), among institutionalized elderly patients with severe AP diagnosed by clinical features. In that same study, most anaerobes coexisted with aerobic gram-negative bacteria, and the clinical illness promptly resolved in the absence of specific anaerobic coverage.¹¹ AP can be successfully treated without anaerobic coverage due to a variety of factors: the insignificant role of anaerobes in the pathogenesis of uncomplicated AP, lower severity of illness in the absence of abscesses or pulmonary necrosis (uncomplicated), and altered local redox-potential from the elimination of aerobic pathogens, which effectively also treats anaerobes.¹ Moreover, anaerobes possess generally less virulence in comparison with aerobes. AP from these organisms typically requires risk for excessive oral growth (eg, periodontal disease) and macroaspiration of a large number of organisms.⁵

There are also potential harms associated with the unnecessary treatment of anaerobic bacteria. Since anaerobes account for the majority of the bacteria present in the bowel, targeting anaerobes can result in gut dysbiosis.¹ Moreover, a prospective study showed an increase in the incidence of vancomycin-resistant enterococci and antibiotic-resistant gram-negative bacteria associated with the empiric use of antibiotics with anaerobic activity.¹² Finally, a systematic review detailed the high incidence of Clostridioides difficile infections among patients receiving clindamycin and carbapenems.¹³

Despite the predominance of aerobic organisms in the respiratory tract specimens of patients diagnosed with AP in the current era, situations still exist that require treatment of anaerobes. These include necrotizing pneumonia, empyema, or lung abscess.² Additionally, patients with severe periodontal disease may harbor anaerobic bacteria such as Bacteroides species, Peptostreptococcus species, and Actinomyces israelii.⁵ When we suspect macroaspiration leading to AP, patients with severe periodontal disease may benefit from anaerobic coverage. Putrid sputum generation may indicate the presence of anaerobic organisms that produce the characteristic foul odor of short-chain volatile fatty acids observed in patients with lung abscess or empyema.² It often takes about 8 to 14 days after an aspiration event for lung cavitation or empyema to develop.¹⁴ Therefore, a longer duration of illness or putrid sputum production may signal a significant concurrent burden of anaerobes. The 2019 official guidelines of the American Thoracic Society and Infectious Disease Society of America recommend adding anaerobic coverage to CAP only when empyema or lung abscess is suspected (conditional recommendation, very low quality of evidence).¹⁵

When you suspect AP in a patient, categorize it as either community or hospital acquired based on risk factors similar to CAP or HAP. For patients with witnessed macroaspiration or in patients with substantial macroaspiration risk factors, perform a radiologic evaluation and a thorough oral examination to evaluate for poor dentition, gingival disease (marked redness, tendency to bleed, ulceration), and tongue coating. For patients presenting from the community with suspected AP without complications, treat with the standard therapy (without additional anaerobic coverage) for CAP. Provide empiric anaerobic coverage for complicated AP (eg, lung abscess, necrosis, or empyema) or for macroaspiration in the setting of severe periodontal disease, putrid sputum, or longer duration of illness. Similarly, treat hospital-acquired AP as HAP (Figure).

When prescribing anaerobic coverage of AP, use combination drugs that include a ß-lactamase inhibitor (eg, ampicillin-sulbactam), clindamycin (either alone or in combination with ß-lactams), or moxifloxacin.¹ Most anaerobes have ß-lactamase or cephalosporinase activity, which renders penicillin and cephalosporins ineffective. Despite its potential side effects, such as C difficile infection, treating with clindamycin has the benefit of a relatively low cost and its association with lower rates of methicillin-resistant Staphylococcus aureus emergence after treatment.¹⁶ Piperacillin-tazobactam and carbapenems also have excellent anaerobic coverage, but we should reserve them for more severe and complicated cases of AP given their extensive antibacterial activity and concern for the emergence of resistance.⁸ Although well known and used for decades for its activity against clinically important anaerobes, avoid metronidazole due to its reduced cure rate in lung abscess caused by microaerophilic streptococci of the oral cavity.¹⁷ Due to a lack of evidence, we do not recommend the use of metronidazole in lung infections.

• Empirically treat most suspected cases of AP with regimens similar to the standard antibiotics for CAP and HAP. In the absence of specific risk factors for anaerobic infections, do not routinely provide anaerobic coverage.
• Provide anaerobic coverage empirically for AP associated with macroaspiration in the setting of severe periodontal disease, putrid sputum, or longer duration of illness.
• Provide anaerobic coverage in AP with evidence of necrotizing pneumonia, empyema, or lung abscess.

Current evidence does not support routine anaerobic coverage of AP in the absence of identifiable risk factors for an anaerobic lung infection.

In consideration of the clinical case, importantly, she has no periodontal disease and no evidence for necrotizing pneumonia, empyema, or lung abscess radiographically. For these reasons, select an empiric antibiotic regime that targets CAP organisms predominantly and forgo additional anaerobic coverage.

Do you think this is a low-value practice? Is this truly a “Thing We Do for No Reason ^™ ”? Share what you do in your practice and join in the conversation online by retweeting it on Twitter (#TWDFNR) and liking it on Facebook. We invite you to propose ideas for other “Things We Do for No Reason ^™ ” topics by emailing [email protected].

The authors have no conflicts of interest relevant to this article.

A 64-year-old man presented with a 2-month history of a nonproductive cough, weight loss, and subjective fevers. He had no chest pain, hemoptysis, or shortness of breath. He also described worsening anorexia and a 15-pound weight loss over the previous 3 months. He had no arthralgias, myalgias, abdominal pain, nausea, emesis, or diarrhea.

Two weeks prior to his presentation, he was diagnosed with pneumonia and given a 5-day course of azithromycin. His symptoms did not improve, so he presented to the emergency room. 

He had not been seen regularly by a physician in decades and had no known medical conditions. He did not take any medications. He immigrated from China 3 years prior and lived with his wife in California. He had a 30 pack-year smoking history. He drank a shot glass of liquor daily and denied any drug use.

Weight loss might result from inflammatory disorders like cancer or noninflammatory causes such as decreased oral intake (eg, diminished appetite) or malabsorption (eg, celiac disease). However, his fevers suggest inflammation, which usually reflects an underlying infection, cancer, or autoimmune process. While chronic cough typically results from upper airway cough syndrome (allergic or nonallergic rhinitis), gastroesophageal reflux disease, or asthma, it can also point to pathology of the lung, which may be intrinsic (bronchiectasis) or extrinsic (mediastinal mass). The duration of 2 months makes a typical infectious process like pneumococcal pneumonia unlikely. Atypical infections such as tuberculosis, melioidosis, and talaromycosis are possible given his immigration from East Asia, and coccidioidomycosis given his residence in California. He might have undiagnosed medical conditions, such as diabetes, that could be relevant to his current presentation and classify him as immunocompromised. His smoking history prompts consideration of lung cancer.

His temperature was 36.5 ^oC, heart rate 70 beats per minute, blood pressure 118/66 mm Hg, respiratory rate 16 breaths per minute, oxygen saturation 98% on room air, and body mass index 23 kg/m². He was in no acute distress. The findings from the cardiac, lung, abdominal, and neurological exams were normal.

Skin examination found a fixed, symmetric, 5-cm, firm nodule at top of sternum (Figure 1A). In addition, he had two 1-cm, mobile, firm, subcutaneous nodules, one on his anterior left chest and another underneath his right axilla. He also had two 2-cm, erythematous, tender nodules on his left anterior forearm and a 1-cm nodule with a central black plug on the dorsal surface of his right hand (Figure 1B). He did not have any edema.

The white blood cell count was 10,500/mm³ (42% neutrophils, 37% lymphocytes, 16.4% monocytes, and 2.9% eosinophils), hemoglobin was 12.2 g/dL with a mean corpuscular volume of 91 fL, and the platelet count was 441,000/mm³. Basic metabolic panel, aminotransferase, bilirubin, and alkaline phosphatase were within reference ranges. Serum albumin was 3.1 g/dL. Serum total protein was elevated at 8.8 g/dL. Serum calcium was 9.0 mg/dL. Urinalysis results were normal.

The slightly low albumin, mildly elevated platelet count, monocytosis, and normocytic anemia suggest inflammation, although monocytosis might represent a hematologic malignancy like chronic myelomonocytic leukemia (CMML). His subjective fevers and weight loss further corroborate underlying inflammation. What is driving the inflammation? There are two localizing findings: cough and nodular skin lesions.

His lack of dyspnea and normal oxygen saturation, respiratory rate, and lung exam make an extrapulmonary cause of cough such as lymphadenopathy or mediastinal infection possible. The number of nodular skin lesions, wide-spread distribution, and appearance (eg, erythematous, tender) point to either a primary cutaneous disease with systemic manifestations (eg, cutaneous lymphoma) or a systemic disease with cutaneous features (eg, sarcoidosis).

Three categories—inflammatory, infectious, and neoplastic—account for most nodular skin lesions. Usually microscopic evaluation is necessary for definitive diagnosis, though epidemiology, associated symptoms, and characteristics of the nodules help prioritize the differential diagnosis. Tender nodules might reflect a panniculitis; erythema nodosum is the most common type, and while this classically develops on the anterior shins, it may also occur on the forearm. His immigration from China prompts consideration of tuberculosis and cutaneous leishmaniasis. Coccidioidomycosis can lead to inflammation and nodular skin lesions. Other infections such as nontuberculous mycobacteria, nocardiosis, and cryptococcosis may cause disseminated infection with pulmonary and skin manifestations. His smoking puts him at risk of lung cancer, which rarely results in metastatic subcutaneous infiltrates.

A chest radiograph demonstrated a prominent density in the right paratracheal region of the mediastinum with adjacent streaky opacities. A computed tomography scan of the chest with intravenous contrast demonstrated centrilobular emphysematous changes and revealed a 2.6 × 4.7-cm necrotic mass in the anterior chest wall with erosion into the manubrium, a 3.8 × 2.1-cm centrally necrotic soft-tissue mass in the right hilum, a 5-mm left upper-lobe noncalcified solid pulmonary nodule, and prominent subcarinal, paratracheal, hilar, and bilateral supraclavicular lymphadenopathy (Figure 2). 

Flow cytometry of the peripheral blood did not demonstrate a lymphoproliferative disorder. Blood smear demonstrated normal red blood cell, white blood cell, and platelet morphology. HIV antibody was negative. Hemoglobin A1c was 6.1%. Smear microscopy for acid-fast bacilli (AFB) was negative and sputum AFB samples were sent for culture. Bacterial, fungal, and AFB blood cultures were collected and pending. 

Causes of necrotizing pneumonia include liquid (eg, lymphoma) and solid (eg, squamous cell carcinoma) cancers, infections, and noninfectious inflammatory processes such as granulomatosis with polyangiitis (GPA). Given his subacute presentation and extrapulmonary cutaneous manifestations, consideration of mycobacteria, fungi (eg, Coccidioides, Aspergillus, and Cryptococcus), and filamentous bacteria (eg, Nocardia and Actinomyces) is prioritized among the myriad of infections that can cause a lung cavity. His smoking history and centrilobular emphysematous changes are highly suggestive of chronic obstructive pulmonary disease, which puts him at increased risk of bacterial colonization and recurrent pulmonary infections. Tuberculosis is still possible despite three negative AFB-sputa smears given the sensitivity of smear microscopy (with three specimens) is roughly 70% in an immunocompetent host.

The lymphadenopathy likely reflects spread from the necrotic lung mass. The frequency of non-Hodgkin lymphoma increases with age. The results of the peripheral flow cytometry do not exclude the possibility of an aggressive lymphoma with pulmonary and cutaneous manifestations.

The erosive property of the chest wall mass makes an autoimmune process like GPA unlikely. An aggressive and disseminated infection or cancer is most likely. A pathologic process that originated in the lung and then spread to the lymph nodes and skin is more likely than a disorder which started in the skin. It would be unlikely for a primary cutaneous disorder to cause such a well-defined necrotic lung mass. Lung cancer rarely metastasizes to the skin and, instead, preferentially involves the chest. Ultimately, ascertaining what the patient experienced first (ie, respiratory or cutaneous symptoms) will determine where the pathology originated.

Computed tomography scan of the abdomen and pelvis with intravenous contrast demonstrated multiple ill-­defined lytic lesions in the pelvis, including a 12-mm lesion of the left sacral ala and multiple subcentimeter lesions in the medial left iliac bone and superior right acetabulum. In addition, there were two 1-cm, rim-enhancing, hypodense nodules in the subcutaneous fat of the right flank at the level of L5 and the left lower quadrant, respectively. There was also a 2.2 × 1.9-cm faintly rim-enhancing hypodensity within the left iliopsoas muscle belly.

These imaging findings further corroborate a widely metastatic process probably originating in the lung and spreading to the lymph nodes, skin, muscles, and bones. The characterization of lesions as lytic as opposed to blastic is less helpful because many diseases can cause both. It does prompt consideration of multiple myeloma; however, multiple myeloma less commonly manifests with extramedullary plasmacytomas and is less likely given his normal renal function and calcium level. Bone lesions lessen the likelihood of GPA, and his necrotic lung mass makes sarcoidosis unlikely. Atypical infections and cancers are the prime suspect of his multisystemic disease.

There are no data yet to suggest a weakened immune system, which would increase his risk for atypical infections. His chronic lung disease, identified on imaging, is a risk factor for nocardiosis. This gram-positive, weakly acid-fast bacterium can involve any organ, although lung, brain, and skin are most commonly involved. Disseminated nocardiosis can result from a pulmonary or cutaneous site of origin. Mycobacteria; Actinomyces; dimorphic fungi like Histoplasma, Coccidioides, and Blastomyces; and molds such as Aspergillus can also cause disseminated disease with pulmonary, cutaneous, and musculoskeletal manifestations.

While metastases to muscle itself are rare, they can occur with primary lung cancers. Primary lung cancer with extrapulmonary features is feasible. Squamous cell lung cancer is the most likely to cavitate, although it rarely spreads to the skin. An aggressive lymphoma like diffuse large B-cell lymphoma or cutaneous T-cell lymphoma (higher occurrence in Asians) might also explain his constellation of findings. If culture data remain negative, then biopsy of the chest wall mass might be the safest and highest-yield target.

On hospital day 2, the patient developed new-onset severe neck pain. Magnetic resonance imaging of the cervical, thoracic, and lumbar spine revealed multilevel, bony, lytic lesions with notable cortical breakthrough of the C2 and C3 vertebrae into the prevertebral space, as well as epidural extension and paraspinal soft-tissue extension of the thoracic and lumbar vertebral lesions (Figure 3). 

On hospital day 3, the patient reported increased tenderness in his skin nodules with one on his left forearm spontaneously draining purulent fluid. Repeat complete blood count demonstrated a white blood cell count of 12,600/mm³ (45% neutrophils, 43% lymphocytes, 8.4% monocytes, and 4.3% eosinophils), hemoglobin of 16 g/dL, and platelet count of 355,000/mm³.

The erosion into the manubrium and cortical destruction of the cervical spine attests to the aggressiveness of the underlying disease process. Noncutaneous lymphoma and lung cancer are unlikely to have such prominent skin findings; the visceral pathology, necrotizing lung mass, and bone lesions make cutaneous lymphoma less likely. At this point, a disseminated infectious process is most likely. Leading considerations based on his emigration from China and residence in California are tuberculosis and coccidioidomycosis, respectively. Tuberculous spondylitis most commonly involves the lower thoracic and upper lumbar region, and less commonly the cervical spine. His three negative AFB sputa samples further reduce its posttest probability. Ultimately microbiologic data are needed to distinguish between a disseminated fungal process, like coccidioidomycosis, or tuberculosis.

Given the concern for malignancy, a fine needle aspiration of the left supraclavicular lymph node was pursued. This revealed fungal microorganisms morphologically compatible with Coccidioides spp. with a background of necrotizing granulomas and acute inflammation. Fungal blood cultures grew Coccidioides immitis. AFB blood cultures were discontinued due to overgrowth of mold. The Coccidioides immitis antibody immunodiffusion titer was positive at 1:256. 

During the remainder of the hospitalization, the patient was treated with oral fluconazole 800 mg daily. The patient underwent surgical debridement of the manubrium. In addition, given the concern for cervical spine instability, neurosurgery recommended follow-up with interval imaging. Since his discharge from the hospital, the patient continues to take oral fluconazole with resolution of his cutaneous lesions and respiratory symptoms. His titers have incrementally decreased from 1:256 to 1:16 after 8 months of treatment. 

This elderly gentleman from China presented with subacute symptoms and was found to have numerous cutaneous nodules, lymphadenopathy, and diffuse osseous lesions. This multisystem illness posed a diagnostic challenge, forcing our discussant to search for a disease process that could lead to such varied findings. Ultimately, epidemiologic and clinical clues suggested a diagnosis of disseminated coccidioidomycosis, which was later confirmed on lymph node biopsy.

Coccidioides species are important fungal pathogens in the Western Hemisphere. This organism exhibits dimorphism, existing as mycelia (with arthroconidia) in soil and spherules in tissues. Coccidioides spp are endemic to the Southwestern United States, particularly California’s central valley and parts of Arizona; it additionally remains an important pathogen in Mexico, Central America, and South America.¹ Newer epidemiologic studies have raised concerns that the incidence of coccidioidomycosis is increasing and that its geographic range may be more extensive than previously appreciated, with it now being found as far north as Washington state.² 

Coccidioidal infection can take several forms. One-half to two-thirds of infections may be asymptomatic.³ Clinically significant infections can include an acute self-limiting respiratory illness, pulmonary nodules and cavities, chronic fibrocavitary pneumonia, and infections with extrapulmonary dissemination. Early respiratory infection is often indistinguishable from typical community-acquired pneumonia (10%-15% of pneumonia in endemic areas) but can be associated with certain suggestive features, such as erythema nodosum, erythema multiforme, prominent arthralgias (ie, “desert rheumatism”), and a peripheral eosinophilia.^4,5 

Extrapulmonary dissemination is rare and most commonly associated with immunocompromising states.⁶ However, individuals of African or Filipino ancestry also appear to be at increased risk for disseminated disease, which led to a California court decision that excluded African American inmates from state prisons located in Coccidioides endemic areas.⁷ The most common sites of extrapulmonary dissemination include the skin and soft tissues, bones and joints, and the central nervous system (CNS).⁶ CNS disease has a predilection to manifest as a chronic basilar meningitis, most often complicated by hydrocephalus, vasculitic infarction, and spinal arachnoiditis.⁸

Cutaneous manifestations of coccidioidomycosis can occur as immunologic phenomenon associated with pulmonary disease or represent skin and soft tissue foci of disseminated infection.⁹ In primary pulmonary infection, skin findings can range from a nonspecific exanthem to erythema nodosum and erythema multiforme, which are thought to represent hypersensitivity responses. In contrast, Coccidioides spp can infect the skin either through direct inoculation (as in primary cutaneous coccidioidomycosis) or via hematogenous dissemination.^9,10 A variety of lesions have been described, with painless nodules being the most frequently encountered morphotype in one study.^11,12 On histopathologic examination, these lesions often have features of granulomatous dermatitis, eosinophilic infiltration, gummatous necrosis, microabscesses, or perivascular inflammation.¹³

Another common and highly morbid site of extrapulmonary dissemination is the musculoskeletal system. Bone and joint coccidioidomycosis most frequently affect the axial skeleton, although peripheral skeletal structures and joints can also be involved.^6,12 Vertebral coccidioidomycosis is associated with significant morbidity. A study describing the magnetic resonance imaging findings of patients with vertebral coccidioidomycosis found that Coccidioides spp appeared to have a predilection for the thoracic vertebrae (in up to 80% of the study’s cohort).¹⁴ Skip lesions with noncontiguously involved vertebrae occurred in roughly half of patients, highlighting the usefulness of imaging the total spine in suspected cases. 

The diagnosis of coccidioidomycosis is often established through serologic testing or by isolation of Coccidioides spp. on histopathology or culture. Obtaining sputum or tissue may be difficult, so clinicians often rely on noninvasive diagnostic tests such as coccidioidal antigen and serologies by enzyme immunoassays, immunodiffusion, and complement fixation. Enzyme immunoassays IgM and IgG results are positive early in the disease process and need to be confirmed with immunodiffusion or complement fixation testing. Complement fixation IgG is additionally useful to monitor disease activity over time and can help inform risk of disseminated disease.¹⁵ The gold standard of diagnosis of disseminated coccidioidomycosis infection remains histopathologic confirmation either by direct visualization of a spherule or growth in fungal cultures.¹⁶ Polymerase chain reaction testing of sputum samples is an emerging diagnostic technique that has been found to have similar sensitivity rates to fungal culture.¹⁷

Treatment decisions in coccidioidomycosis are complex and vary by site of infection, immune status of the host, and extent of disease.¹⁶ While uncomplicated primary pulmonary infections can often be managed with observation alone, prolonged medical therapy with azole antifungals is often recommended for complicated pulmonary infections, symptomatic cavitary disease, and virtually all forms of extrapulmonary disease. Intravenous liposomal amphotericin is often used as initial therapy in immunosuppressed individuals, pregnant women, and those with extensive disease. CNS disease represents a particularly challenging treatment scenario and requires lifelong azole therapy.^8,16 

The patient in this case initially presented with vague inflammatory symptoms, with each aliquot revealing further evidence of a metastatic disease process. Such multisystem presentations are diagnostically challenging and force clinicians to reach for some feature around which to build their differential diagnosis. It is with this in mind that we are often taught to “localize the lesion” in order to focus our search for a unifying diagnosis. Yet, in this case, the sheer number of disease foci ultimately helped the discussant to narrow the range of diagnostic possibilities because only a limited number of conditions could present with such widespread, multisystem manifestations. Therefore, this case serves as a reminder that, sometimes in clinical reasoning, “more is less.”

• Coccidioidomycosis is a fungal infection that can present with pulmonary or extrapulmonary disease. Risk of extrapulmonary dissemination is greatest among immunocompromised individuals and those of African or Filipino ancestry.^3,7
• The most common sites of extrapulmonary dissemination include the skin and soft tissues, bones and joints, and the CNS.⁶
• While serologic testing can be diagnostically useful, the gold standard for diagnosis of disseminated coccidioidomycosis infection remains histopathologic confirmation with direct visualization of a spherule or growth in fungal cultures.¹⁶

A 64-year-old man presented with a 2-month history of a nonproductive cough, weight loss, and subjective fevers. He had no chest pain, hemoptysis, or shortness of breath. He also described worsening anorexia and a 15-pound weight loss over the previous 3 months. He had no arthralgias, myalgias, abdominal pain, nausea, emesis, or diarrhea.

Two weeks prior to his presentation, he was diagnosed with pneumonia and given a 5-day course of azithromycin. His symptoms did not improve, so he presented to the emergency room. 

He had not been seen regularly by a physician in decades and had no known medical conditions. He did not take any medications. He immigrated from China 3 years prior and lived with his wife in California. He had a 30 pack-year smoking history. He drank a shot glass of liquor daily and denied any drug use.

Weight loss might result from inflammatory disorders like cancer or noninflammatory causes such as decreased oral intake (eg, diminished appetite) or malabsorption (eg, celiac disease). However, his fevers suggest inflammation, which usually reflects an underlying infection, cancer, or autoimmune process. While chronic cough typically results from upper airway cough syndrome (allergic or nonallergic rhinitis), gastroesophageal reflux disease, or asthma, it can also point to pathology of the lung, which may be intrinsic (bronchiectasis) or extrinsic (mediastinal mass). The duration of 2 months makes a typical infectious process like pneumococcal pneumonia unlikely. Atypical infections such as tuberculosis, melioidosis, and talaromycosis are possible given his immigration from East Asia, and coccidioidomycosis given his residence in California. He might have undiagnosed medical conditions, such as diabetes, that could be relevant to his current presentation and classify him as immunocompromised. His smoking history prompts consideration of lung cancer.

His temperature was 36.5 ^oC, heart rate 70 beats per minute, blood pressure 118/66 mm Hg, respiratory rate 16 breaths per minute, oxygen saturation 98% on room air, and body mass index 23 kg/m². He was in no acute distress. The findings from the cardiac, lung, abdominal, and neurological exams were normal.

Skin examination found a fixed, symmetric, 5-cm, firm nodule at top of sternum (Figure 1A). In addition, he had two 1-cm, mobile, firm, subcutaneous nodules, one on his anterior left chest and another underneath his right axilla. He also had two 2-cm, erythematous, tender nodules on his left anterior forearm and a 1-cm nodule with a central black plug on the dorsal surface of his right hand (Figure 1B). He did not have any edema.

The white blood cell count was 10,500/mm³ (42% neutrophils, 37% lymphocytes, 16.4% monocytes, and 2.9% eosinophils), hemoglobin was 12.2 g/dL with a mean corpuscular volume of 91 fL, and the platelet count was 441,000/mm³. Basic metabolic panel, aminotransferase, bilirubin, and alkaline phosphatase were within reference ranges. Serum albumin was 3.1 g/dL. Serum total protein was elevated at 8.8 g/dL. Serum calcium was 9.0 mg/dL. Urinalysis results were normal.

The slightly low albumin, mildly elevated platelet count, monocytosis, and normocytic anemia suggest inflammation, although monocytosis might represent a hematologic malignancy like chronic myelomonocytic leukemia (CMML). His subjective fevers and weight loss further corroborate underlying inflammation. What is driving the inflammation? There are two localizing findings: cough and nodular skin lesions.

His lack of dyspnea and normal oxygen saturation, respiratory rate, and lung exam make an extrapulmonary cause of cough such as lymphadenopathy or mediastinal infection possible. The number of nodular skin lesions, wide-spread distribution, and appearance (eg, erythematous, tender) point to either a primary cutaneous disease with systemic manifestations (eg, cutaneous lymphoma) or a systemic disease with cutaneous features (eg, sarcoidosis).

Three categories—inflammatory, infectious, and neoplastic—account for most nodular skin lesions. Usually microscopic evaluation is necessary for definitive diagnosis, though epidemiology, associated symptoms, and characteristics of the nodules help prioritize the differential diagnosis. Tender nodules might reflect a panniculitis; erythema nodosum is the most common type, and while this classically develops on the anterior shins, it may also occur on the forearm. His immigration from China prompts consideration of tuberculosis and cutaneous leishmaniasis. Coccidioidomycosis can lead to inflammation and nodular skin lesions. Other infections such as nontuberculous mycobacteria, nocardiosis, and cryptococcosis may cause disseminated infection with pulmonary and skin manifestations. His smoking puts him at risk of lung cancer, which rarely results in metastatic subcutaneous infiltrates.

A chest radiograph demonstrated a prominent density in the right paratracheal region of the mediastinum with adjacent streaky opacities. A computed tomography scan of the chest with intravenous contrast demonstrated centrilobular emphysematous changes and revealed a 2.6 × 4.7-cm necrotic mass in the anterior chest wall with erosion into the manubrium, a 3.8 × 2.1-cm centrally necrotic soft-tissue mass in the right hilum, a 5-mm left upper-lobe noncalcified solid pulmonary nodule, and prominent subcarinal, paratracheal, hilar, and bilateral supraclavicular lymphadenopathy (Figure 2). 

Flow cytometry of the peripheral blood did not demonstrate a lymphoproliferative disorder. Blood smear demonstrated normal red blood cell, white blood cell, and platelet morphology. HIV antibody was negative. Hemoglobin A1c was 6.1%. Smear microscopy for acid-fast bacilli (AFB) was negative and sputum AFB samples were sent for culture. Bacterial, fungal, and AFB blood cultures were collected and pending. 

Causes of necrotizing pneumonia include liquid (eg, lymphoma) and solid (eg, squamous cell carcinoma) cancers, infections, and noninfectious inflammatory processes such as granulomatosis with polyangiitis (GPA). Given his subacute presentation and extrapulmonary cutaneous manifestations, consideration of mycobacteria, fungi (eg, Coccidioides, Aspergillus, and Cryptococcus), and filamentous bacteria (eg, Nocardia and Actinomyces) is prioritized among the myriad of infections that can cause a lung cavity. His smoking history and centrilobular emphysematous changes are highly suggestive of chronic obstructive pulmonary disease, which puts him at increased risk of bacterial colonization and recurrent pulmonary infections. Tuberculosis is still possible despite three negative AFB-sputa smears given the sensitivity of smear microscopy (with three specimens) is roughly 70% in an immunocompetent host.

The lymphadenopathy likely reflects spread from the necrotic lung mass. The frequency of non-Hodgkin lymphoma increases with age. The results of the peripheral flow cytometry do not exclude the possibility of an aggressive lymphoma with pulmonary and cutaneous manifestations.

The erosive property of the chest wall mass makes an autoimmune process like GPA unlikely. An aggressive and disseminated infection or cancer is most likely. A pathologic process that originated in the lung and then spread to the lymph nodes and skin is more likely than a disorder which started in the skin. It would be unlikely for a primary cutaneous disorder to cause such a well-defined necrotic lung mass. Lung cancer rarely metastasizes to the skin and, instead, preferentially involves the chest. Ultimately, ascertaining what the patient experienced first (ie, respiratory or cutaneous symptoms) will determine where the pathology originated.

Computed tomography scan of the abdomen and pelvis with intravenous contrast demonstrated multiple ill-­defined lytic lesions in the pelvis, including a 12-mm lesion of the left sacral ala and multiple subcentimeter lesions in the medial left iliac bone and superior right acetabulum. In addition, there were two 1-cm, rim-enhancing, hypodense nodules in the subcutaneous fat of the right flank at the level of L5 and the left lower quadrant, respectively. There was also a 2.2 × 1.9-cm faintly rim-enhancing hypodensity within the left iliopsoas muscle belly.

These imaging findings further corroborate a widely metastatic process probably originating in the lung and spreading to the lymph nodes, skin, muscles, and bones. The characterization of lesions as lytic as opposed to blastic is less helpful because many diseases can cause both. It does prompt consideration of multiple myeloma; however, multiple myeloma less commonly manifests with extramedullary plasmacytomas and is less likely given his normal renal function and calcium level. Bone lesions lessen the likelihood of GPA, and his necrotic lung mass makes sarcoidosis unlikely. Atypical infections and cancers are the prime suspect of his multisystemic disease.

There are no data yet to suggest a weakened immune system, which would increase his risk for atypical infections. His chronic lung disease, identified on imaging, is a risk factor for nocardiosis. This gram-positive, weakly acid-fast bacterium can involve any organ, although lung, brain, and skin are most commonly involved. Disseminated nocardiosis can result from a pulmonary or cutaneous site of origin. Mycobacteria; Actinomyces; dimorphic fungi like Histoplasma, Coccidioides, and Blastomyces; and molds such as Aspergillus can also cause disseminated disease with pulmonary, cutaneous, and musculoskeletal manifestations.

While metastases to muscle itself are rare, they can occur with primary lung cancers. Primary lung cancer with extrapulmonary features is feasible. Squamous cell lung cancer is the most likely to cavitate, although it rarely spreads to the skin. An aggressive lymphoma like diffuse large B-cell lymphoma or cutaneous T-cell lymphoma (higher occurrence in Asians) might also explain his constellation of findings. If culture data remain negative, then biopsy of the chest wall mass might be the safest and highest-yield target.

On hospital day 2, the patient developed new-onset severe neck pain. Magnetic resonance imaging of the cervical, thoracic, and lumbar spine revealed multilevel, bony, lytic lesions with notable cortical breakthrough of the C2 and C3 vertebrae into the prevertebral space, as well as epidural extension and paraspinal soft-tissue extension of the thoracic and lumbar vertebral lesions (Figure 3). 

On hospital day 3, the patient reported increased tenderness in his skin nodules with one on his left forearm spontaneously draining purulent fluid. Repeat complete blood count demonstrated a white blood cell count of 12,600/mm³ (45% neutrophils, 43% lymphocytes, 8.4% monocytes, and 4.3% eosinophils), hemoglobin of 16 g/dL, and platelet count of 355,000/mm³.

The erosion into the manubrium and cortical destruction of the cervical spine attests to the aggressiveness of the underlying disease process. Noncutaneous lymphoma and lung cancer are unlikely to have such prominent skin findings; the visceral pathology, necrotizing lung mass, and bone lesions make cutaneous lymphoma less likely. At this point, a disseminated infectious process is most likely. Leading considerations based on his emigration from China and residence in California are tuberculosis and coccidioidomycosis, respectively. Tuberculous spondylitis most commonly involves the lower thoracic and upper lumbar region, and less commonly the cervical spine. His three negative AFB sputa samples further reduce its posttest probability. Ultimately microbiologic data are needed to distinguish between a disseminated fungal process, like coccidioidomycosis, or tuberculosis.

Given the concern for malignancy, a fine needle aspiration of the left supraclavicular lymph node was pursued. This revealed fungal microorganisms morphologically compatible with Coccidioides spp. with a background of necrotizing granulomas and acute inflammation. Fungal blood cultures grew Coccidioides immitis. AFB blood cultures were discontinued due to overgrowth of mold. The Coccidioides immitis antibody immunodiffusion titer was positive at 1:256. 

During the remainder of the hospitalization, the patient was treated with oral fluconazole 800 mg daily. The patient underwent surgical debridement of the manubrium. In addition, given the concern for cervical spine instability, neurosurgery recommended follow-up with interval imaging. Since his discharge from the hospital, the patient continues to take oral fluconazole with resolution of his cutaneous lesions and respiratory symptoms. His titers have incrementally decreased from 1:256 to 1:16 after 8 months of treatment. 

This elderly gentleman from China presented with subacute symptoms and was found to have numerous cutaneous nodules, lymphadenopathy, and diffuse osseous lesions. This multisystem illness posed a diagnostic challenge, forcing our discussant to search for a disease process that could lead to such varied findings. Ultimately, epidemiologic and clinical clues suggested a diagnosis of disseminated coccidioidomycosis, which was later confirmed on lymph node biopsy.

Coccidioides species are important fungal pathogens in the Western Hemisphere. This organism exhibits dimorphism, existing as mycelia (with arthroconidia) in soil and spherules in tissues. Coccidioides spp are endemic to the Southwestern United States, particularly California’s central valley and parts of Arizona; it additionally remains an important pathogen in Mexico, Central America, and South America.¹ Newer epidemiologic studies have raised concerns that the incidence of coccidioidomycosis is increasing and that its geographic range may be more extensive than previously appreciated, with it now being found as far north as Washington state.² 

Coccidioidal infection can take several forms. One-half to two-thirds of infections may be asymptomatic.³ Clinically significant infections can include an acute self-limiting respiratory illness, pulmonary nodules and cavities, chronic fibrocavitary pneumonia, and infections with extrapulmonary dissemination. Early respiratory infection is often indistinguishable from typical community-acquired pneumonia (10%-15% of pneumonia in endemic areas) but can be associated with certain suggestive features, such as erythema nodosum, erythema multiforme, prominent arthralgias (ie, “desert rheumatism”), and a peripheral eosinophilia.^4,5 

Extrapulmonary dissemination is rare and most commonly associated with immunocompromising states.⁶ However, individuals of African or Filipino ancestry also appear to be at increased risk for disseminated disease, which led to a California court decision that excluded African American inmates from state prisons located in Coccidioides endemic areas.⁷ The most common sites of extrapulmonary dissemination include the skin and soft tissues, bones and joints, and the central nervous system (CNS).⁶ CNS disease has a predilection to manifest as a chronic basilar meningitis, most often complicated by hydrocephalus, vasculitic infarction, and spinal arachnoiditis.⁸

Cutaneous manifestations of coccidioidomycosis can occur as immunologic phenomenon associated with pulmonary disease or represent skin and soft tissue foci of disseminated infection.⁹ In primary pulmonary infection, skin findings can range from a nonspecific exanthem to erythema nodosum and erythema multiforme, which are thought to represent hypersensitivity responses. In contrast, Coccidioides spp can infect the skin either through direct inoculation (as in primary cutaneous coccidioidomycosis) or via hematogenous dissemination.^9,10 A variety of lesions have been described, with painless nodules being the most frequently encountered morphotype in one study.^11,12 On histopathologic examination, these lesions often have features of granulomatous dermatitis, eosinophilic infiltration, gummatous necrosis, microabscesses, or perivascular inflammation.¹³

Another common and highly morbid site of extrapulmonary dissemination is the musculoskeletal system. Bone and joint coccidioidomycosis most frequently affect the axial skeleton, although peripheral skeletal structures and joints can also be involved.^6,12 Vertebral coccidioidomycosis is associated with significant morbidity. A study describing the magnetic resonance imaging findings of patients with vertebral coccidioidomycosis found that Coccidioides spp appeared to have a predilection for the thoracic vertebrae (in up to 80% of the study’s cohort).¹⁴ Skip lesions with noncontiguously involved vertebrae occurred in roughly half of patients, highlighting the usefulness of imaging the total spine in suspected cases. 

The diagnosis of coccidioidomycosis is often established through serologic testing or by isolation of Coccidioides spp. on histopathology or culture. Obtaining sputum or tissue may be difficult, so clinicians often rely on noninvasive diagnostic tests such as coccidioidal antigen and serologies by enzyme immunoassays, immunodiffusion, and complement fixation. Enzyme immunoassays IgM and IgG results are positive early in the disease process and need to be confirmed with immunodiffusion or complement fixation testing. Complement fixation IgG is additionally useful to monitor disease activity over time and can help inform risk of disseminated disease.¹⁵ The gold standard of diagnosis of disseminated coccidioidomycosis infection remains histopathologic confirmation either by direct visualization of a spherule or growth in fungal cultures.¹⁶ Polymerase chain reaction testing of sputum samples is an emerging diagnostic technique that has been found to have similar sensitivity rates to fungal culture.¹⁷

Treatment decisions in coccidioidomycosis are complex and vary by site of infection, immune status of the host, and extent of disease.¹⁶ While uncomplicated primary pulmonary infections can often be managed with observation alone, prolonged medical therapy with azole antifungals is often recommended for complicated pulmonary infections, symptomatic cavitary disease, and virtually all forms of extrapulmonary disease. Intravenous liposomal amphotericin is often used as initial therapy in immunosuppressed individuals, pregnant women, and those with extensive disease. CNS disease represents a particularly challenging treatment scenario and requires lifelong azole therapy.^8,16 

The patient in this case initially presented with vague inflammatory symptoms, with each aliquot revealing further evidence of a metastatic disease process. Such multisystem presentations are diagnostically challenging and force clinicians to reach for some feature around which to build their differential diagnosis. It is with this in mind that we are often taught to “localize the lesion” in order to focus our search for a unifying diagnosis. Yet, in this case, the sheer number of disease foci ultimately helped the discussant to narrow the range of diagnostic possibilities because only a limited number of conditions could present with such widespread, multisystem manifestations. Therefore, this case serves as a reminder that, sometimes in clinical reasoning, “more is less.”

• Coccidioidomycosis is a fungal infection that can present with pulmonary or extrapulmonary disease. Risk of extrapulmonary dissemination is greatest among immunocompromised individuals and those of African or Filipino ancestry.^3,7
• The most common sites of extrapulmonary dissemination include the skin and soft tissues, bones and joints, and the CNS.⁶
• While serologic testing can be diagnostically useful, the gold standard for diagnosis of disseminated coccidioidomycosis infection remains histopathologic confirmation with direct visualization of a spherule or growth in fungal cultures.¹⁶

A 64-year-old man presented with a 2-month history of a nonproductive cough, weight loss, and subjective fevers. He had no chest pain, hemoptysis, or shortness of breath. He also described worsening anorexia and a 15-pound weight loss over the previous 3 months. He had no arthralgias, myalgias, abdominal pain, nausea, emesis, or diarrhea.

Two weeks prior to his presentation, he was diagnosed with pneumonia and given a 5-day course of azithromycin. His symptoms did not improve, so he presented to the emergency room. 

He had not been seen regularly by a physician in decades and had no known medical conditions. He did not take any medications. He immigrated from China 3 years prior and lived with his wife in California. He had a 30 pack-year smoking history. He drank a shot glass of liquor daily and denied any drug use.

Weight loss might result from inflammatory disorders like cancer or noninflammatory causes such as decreased oral intake (eg, diminished appetite) or malabsorption (eg, celiac disease). However, his fevers suggest inflammation, which usually reflects an underlying infection, cancer, or autoimmune process. While chronic cough typically results from upper airway cough syndrome (allergic or nonallergic rhinitis), gastroesophageal reflux disease, or asthma, it can also point to pathology of the lung, which may be intrinsic (bronchiectasis) or extrinsic (mediastinal mass). The duration of 2 months makes a typical infectious process like pneumococcal pneumonia unlikely. Atypical infections such as tuberculosis, melioidosis, and talaromycosis are possible given his immigration from East Asia, and coccidioidomycosis given his residence in California. He might have undiagnosed medical conditions, such as diabetes, that could be relevant to his current presentation and classify him as immunocompromised. His smoking history prompts consideration of lung cancer.

His temperature was 36.5 ^oC, heart rate 70 beats per minute, blood pressure 118/66 mm Hg, respiratory rate 16 breaths per minute, oxygen saturation 98% on room air, and body mass index 23 kg/m². He was in no acute distress. The findings from the cardiac, lung, abdominal, and neurological exams were normal.

Skin examination found a fixed, symmetric, 5-cm, firm nodule at top of sternum (Figure 1A). In addition, he had two 1-cm, mobile, firm, subcutaneous nodules, one on his anterior left chest and another underneath his right axilla. He also had two 2-cm, erythematous, tender nodules on his left anterior forearm and a 1-cm nodule with a central black plug on the dorsal surface of his right hand (Figure 1B). He did not have any edema.

The white blood cell count was 10,500/mm³ (42% neutrophils, 37% lymphocytes, 16.4% monocytes, and 2.9% eosinophils), hemoglobin was 12.2 g/dL with a mean corpuscular volume of 91 fL, and the platelet count was 441,000/mm³. Basic metabolic panel, aminotransferase, bilirubin, and alkaline phosphatase were within reference ranges. Serum albumin was 3.1 g/dL. Serum total protein was elevated at 8.8 g/dL. Serum calcium was 9.0 mg/dL. Urinalysis results were normal.

The slightly low albumin, mildly elevated platelet count, monocytosis, and normocytic anemia suggest inflammation, although monocytosis might represent a hematologic malignancy like chronic myelomonocytic leukemia (CMML). His subjective fevers and weight loss further corroborate underlying inflammation. What is driving the inflammation? There are two localizing findings: cough and nodular skin lesions.

His lack of dyspnea and normal oxygen saturation, respiratory rate, and lung exam make an extrapulmonary cause of cough such as lymphadenopathy or mediastinal infection possible. The number of nodular skin lesions, wide-spread distribution, and appearance (eg, erythematous, tender) point to either a primary cutaneous disease with systemic manifestations (eg, cutaneous lymphoma) or a systemic disease with cutaneous features (eg, sarcoidosis).

Three categories—inflammatory, infectious, and neoplastic—account for most nodular skin lesions. Usually microscopic evaluation is necessary for definitive diagnosis, though epidemiology, associated symptoms, and characteristics of the nodules help prioritize the differential diagnosis. Tender nodules might reflect a panniculitis; erythema nodosum is the most common type, and while this classically develops on the anterior shins, it may also occur on the forearm. His immigration from China prompts consideration of tuberculosis and cutaneous leishmaniasis. Coccidioidomycosis can lead to inflammation and nodular skin lesions. Other infections such as nontuberculous mycobacteria, nocardiosis, and cryptococcosis may cause disseminated infection with pulmonary and skin manifestations. His smoking puts him at risk of lung cancer, which rarely results in metastatic subcutaneous infiltrates.

A chest radiograph demonstrated a prominent density in the right paratracheal region of the mediastinum with adjacent streaky opacities. A computed tomography scan of the chest with intravenous contrast demonstrated centrilobular emphysematous changes and revealed a 2.6 × 4.7-cm necrotic mass in the anterior chest wall with erosion into the manubrium, a 3.8 × 2.1-cm centrally necrotic soft-tissue mass in the right hilum, a 5-mm left upper-lobe noncalcified solid pulmonary nodule, and prominent subcarinal, paratracheal, hilar, and bilateral supraclavicular lymphadenopathy (Figure 2). 

Flow cytometry of the peripheral blood did not demonstrate a lymphoproliferative disorder. Blood smear demonstrated normal red blood cell, white blood cell, and platelet morphology. HIV antibody was negative. Hemoglobin A1c was 6.1%. Smear microscopy for acid-fast bacilli (AFB) was negative and sputum AFB samples were sent for culture. Bacterial, fungal, and AFB blood cultures were collected and pending. 

Causes of necrotizing pneumonia include liquid (eg, lymphoma) and solid (eg, squamous cell carcinoma) cancers, infections, and noninfectious inflammatory processes such as granulomatosis with polyangiitis (GPA). Given his subacute presentation and extrapulmonary cutaneous manifestations, consideration of mycobacteria, fungi (eg, Coccidioides, Aspergillus, and Cryptococcus), and filamentous bacteria (eg, Nocardia and Actinomyces) is prioritized among the myriad of infections that can cause a lung cavity. His smoking history and centrilobular emphysematous changes are highly suggestive of chronic obstructive pulmonary disease, which puts him at increased risk of bacterial colonization and recurrent pulmonary infections. Tuberculosis is still possible despite three negative AFB-sputa smears given the sensitivity of smear microscopy (with three specimens) is roughly 70% in an immunocompetent host.

The lymphadenopathy likely reflects spread from the necrotic lung mass. The frequency of non-Hodgkin lymphoma increases with age. The results of the peripheral flow cytometry do not exclude the possibility of an aggressive lymphoma with pulmonary and cutaneous manifestations.

The erosive property of the chest wall mass makes an autoimmune process like GPA unlikely. An aggressive and disseminated infection or cancer is most likely. A pathologic process that originated in the lung and then spread to the lymph nodes and skin is more likely than a disorder which started in the skin. It would be unlikely for a primary cutaneous disorder to cause such a well-defined necrotic lung mass. Lung cancer rarely metastasizes to the skin and, instead, preferentially involves the chest. Ultimately, ascertaining what the patient experienced first (ie, respiratory or cutaneous symptoms) will determine where the pathology originated.

Computed tomography scan of the abdomen and pelvis with intravenous contrast demonstrated multiple ill-­defined lytic lesions in the pelvis, including a 12-mm lesion of the left sacral ala and multiple subcentimeter lesions in the medial left iliac bone and superior right acetabulum. In addition, there were two 1-cm, rim-enhancing, hypodense nodules in the subcutaneous fat of the right flank at the level of L5 and the left lower quadrant, respectively. There was also a 2.2 × 1.9-cm faintly rim-enhancing hypodensity within the left iliopsoas muscle belly.

These imaging findings further corroborate a widely metastatic process probably originating in the lung and spreading to the lymph nodes, skin, muscles, and bones. The characterization of lesions as lytic as opposed to blastic is less helpful because many diseases can cause both. It does prompt consideration of multiple myeloma; however, multiple myeloma less commonly manifests with extramedullary plasmacytomas and is less likely given his normal renal function and calcium level. Bone lesions lessen the likelihood of GPA, and his necrotic lung mass makes sarcoidosis unlikely. Atypical infections and cancers are the prime suspect of his multisystemic disease.

There are no data yet to suggest a weakened immune system, which would increase his risk for atypical infections. His chronic lung disease, identified on imaging, is a risk factor for nocardiosis. This gram-positive, weakly acid-fast bacterium can involve any organ, although lung, brain, and skin are most commonly involved. Disseminated nocardiosis can result from a pulmonary or cutaneous site of origin. Mycobacteria; Actinomyces; dimorphic fungi like Histoplasma, Coccidioides, and Blastomyces; and molds such as Aspergillus can also cause disseminated disease with pulmonary, cutaneous, and musculoskeletal manifestations.

While metastases to muscle itself are rare, they can occur with primary lung cancers. Primary lung cancer with extrapulmonary features is feasible. Squamous cell lung cancer is the most likely to cavitate, although it rarely spreads to the skin. An aggressive lymphoma like diffuse large B-cell lymphoma or cutaneous T-cell lymphoma (higher occurrence in Asians) might also explain his constellation of findings. If culture data remain negative, then biopsy of the chest wall mass might be the safest and highest-yield target.

On hospital day 2, the patient developed new-onset severe neck pain. Magnetic resonance imaging of the cervical, thoracic, and lumbar spine revealed multilevel, bony, lytic lesions with notable cortical breakthrough of the C2 and C3 vertebrae into the prevertebral space, as well as epidural extension and paraspinal soft-tissue extension of the thoracic and lumbar vertebral lesions (Figure 3). 

On hospital day 3, the patient reported increased tenderness in his skin nodules with one on his left forearm spontaneously draining purulent fluid. Repeat complete blood count demonstrated a white blood cell count of 12,600/mm³ (45% neutrophils, 43% lymphocytes, 8.4% monocytes, and 4.3% eosinophils), hemoglobin of 16 g/dL, and platelet count of 355,000/mm³.

The erosion into the manubrium and cortical destruction of the cervical spine attests to the aggressiveness of the underlying disease process. Noncutaneous lymphoma and lung cancer are unlikely to have such prominent skin findings; the visceral pathology, necrotizing lung mass, and bone lesions make cutaneous lymphoma less likely. At this point, a disseminated infectious process is most likely. Leading considerations based on his emigration from China and residence in California are tuberculosis and coccidioidomycosis, respectively. Tuberculous spondylitis most commonly involves the lower thoracic and upper lumbar region, and less commonly the cervical spine. His three negative AFB sputa samples further reduce its posttest probability. Ultimately microbiologic data are needed to distinguish between a disseminated fungal process, like coccidioidomycosis, or tuberculosis.

Given the concern for malignancy, a fine needle aspiration of the left supraclavicular lymph node was pursued. This revealed fungal microorganisms morphologically compatible with Coccidioides spp. with a background of necrotizing granulomas and acute inflammation. Fungal blood cultures grew Coccidioides immitis. AFB blood cultures were discontinued due to overgrowth of mold. The Coccidioides immitis antibody immunodiffusion titer was positive at 1:256. 

During the remainder of the hospitalization, the patient was treated with oral fluconazole 800 mg daily. The patient underwent surgical debridement of the manubrium. In addition, given the concern for cervical spine instability, neurosurgery recommended follow-up with interval imaging. Since his discharge from the hospital, the patient continues to take oral fluconazole with resolution of his cutaneous lesions and respiratory symptoms. His titers have incrementally decreased from 1:256 to 1:16 after 8 months of treatment. 

This elderly gentleman from China presented with subacute symptoms and was found to have numerous cutaneous nodules, lymphadenopathy, and diffuse osseous lesions. This multisystem illness posed a diagnostic challenge, forcing our discussant to search for a disease process that could lead to such varied findings. Ultimately, epidemiologic and clinical clues suggested a diagnosis of disseminated coccidioidomycosis, which was later confirmed on lymph node biopsy.

Coccidioides species are important fungal pathogens in the Western Hemisphere. This organism exhibits dimorphism, existing as mycelia (with arthroconidia) in soil and spherules in tissues. Coccidioides spp are endemic to the Southwestern United States, particularly California’s central valley and parts of Arizona; it additionally remains an important pathogen in Mexico, Central America, and South America.¹ Newer epidemiologic studies have raised concerns that the incidence of coccidioidomycosis is increasing and that its geographic range may be more extensive than previously appreciated, with it now being found as far north as Washington state.² 

Coccidioidal infection can take several forms. One-half to two-thirds of infections may be asymptomatic.³ Clinically significant infections can include an acute self-limiting respiratory illness, pulmonary nodules and cavities, chronic fibrocavitary pneumonia, and infections with extrapulmonary dissemination. Early respiratory infection is often indistinguishable from typical community-acquired pneumonia (10%-15% of pneumonia in endemic areas) but can be associated with certain suggestive features, such as erythema nodosum, erythema multiforme, prominent arthralgias (ie, “desert rheumatism”), and a peripheral eosinophilia.^4,5 

Extrapulmonary dissemination is rare and most commonly associated with immunocompromising states.⁶ However, individuals of African or Filipino ancestry also appear to be at increased risk for disseminated disease, which led to a California court decision that excluded African American inmates from state prisons located in Coccidioides endemic areas.⁷ The most common sites of extrapulmonary dissemination include the skin and soft tissues, bones and joints, and the central nervous system (CNS).⁶ CNS disease has a predilection to manifest as a chronic basilar meningitis, most often complicated by hydrocephalus, vasculitic infarction, and spinal arachnoiditis.⁸

Cutaneous manifestations of coccidioidomycosis can occur as immunologic phenomenon associated with pulmonary disease or represent skin and soft tissue foci of disseminated infection.⁹ In primary pulmonary infection, skin findings can range from a nonspecific exanthem to erythema nodosum and erythema multiforme, which are thought to represent hypersensitivity responses. In contrast, Coccidioides spp can infect the skin either through direct inoculation (as in primary cutaneous coccidioidomycosis) or via hematogenous dissemination.^9,10 A variety of lesions have been described, with painless nodules being the most frequently encountered morphotype in one study.^11,12 On histopathologic examination, these lesions often have features of granulomatous dermatitis, eosinophilic infiltration, gummatous necrosis, microabscesses, or perivascular inflammation.¹³

Another common and highly morbid site of extrapulmonary dissemination is the musculoskeletal system. Bone and joint coccidioidomycosis most frequently affect the axial skeleton, although peripheral skeletal structures and joints can also be involved.^6,12 Vertebral coccidioidomycosis is associated with significant morbidity. A study describing the magnetic resonance imaging findings of patients with vertebral coccidioidomycosis found that Coccidioides spp appeared to have a predilection for the thoracic vertebrae (in up to 80% of the study’s cohort).¹⁴ Skip lesions with noncontiguously involved vertebrae occurred in roughly half of patients, highlighting the usefulness of imaging the total spine in suspected cases. 

The diagnosis of coccidioidomycosis is often established through serologic testing or by isolation of Coccidioides spp. on histopathology or culture. Obtaining sputum or tissue may be difficult, so clinicians often rely on noninvasive diagnostic tests such as coccidioidal antigen and serologies by enzyme immunoassays, immunodiffusion, and complement fixation. Enzyme immunoassays IgM and IgG results are positive early in the disease process and need to be confirmed with immunodiffusion or complement fixation testing. Complement fixation IgG is additionally useful to monitor disease activity over time and can help inform risk of disseminated disease.¹⁵ The gold standard of diagnosis of disseminated coccidioidomycosis infection remains histopathologic confirmation either by direct visualization of a spherule or growth in fungal cultures.¹⁶ Polymerase chain reaction testing of sputum samples is an emerging diagnostic technique that has been found to have similar sensitivity rates to fungal culture.¹⁷

Treatment decisions in coccidioidomycosis are complex and vary by site of infection, immune status of the host, and extent of disease.¹⁶ While uncomplicated primary pulmonary infections can often be managed with observation alone, prolonged medical therapy with azole antifungals is often recommended for complicated pulmonary infections, symptomatic cavitary disease, and virtually all forms of extrapulmonary disease. Intravenous liposomal amphotericin is often used as initial therapy in immunosuppressed individuals, pregnant women, and those with extensive disease. CNS disease represents a particularly challenging treatment scenario and requires lifelong azole therapy.^8,16 

The patient in this case initially presented with vague inflammatory symptoms, with each aliquot revealing further evidence of a metastatic disease process. Such multisystem presentations are diagnostically challenging and force clinicians to reach for some feature around which to build their differential diagnosis. It is with this in mind that we are often taught to “localize the lesion” in order to focus our search for a unifying diagnosis. Yet, in this case, the sheer number of disease foci ultimately helped the discussant to narrow the range of diagnostic possibilities because only a limited number of conditions could present with such widespread, multisystem manifestations. Therefore, this case serves as a reminder that, sometimes in clinical reasoning, “more is less.”

• Coccidioidomycosis is a fungal infection that can present with pulmonary or extrapulmonary disease. Risk of extrapulmonary dissemination is greatest among immunocompromised individuals and those of African or Filipino ancestry.^3,7
• The most common sites of extrapulmonary dissemination include the skin and soft tissues, bones and joints, and the CNS.⁶
• While serologic testing can be diagnostically useful, the gold standard for diagnosis of disseminated coccidioidomycosis infection remains histopathologic confirmation with direct visualization of a spherule or growth in fungal cultures.¹⁶

The spread of SARS-CoV-2, the pathogen behind the COVID-19 pandemic, has converged with an unrelenting addiction epidemic. These combined crises will have profound effects on people with substance use disorders (SUD) and people in recovery. Hospitals—which were already hit hard by the addiction epidemic—are the last line of defense in the COVID-19 pandemic. Hospitalists have an important role in balancing the effects of these intersecting, synergistic crises.

People with SUD are disproportionately affected by major medical illnesses, including infections such as hepatitis C, HIV, and cardiovascular, pulmonary, and liver diseases.¹ They also experience high rates of hospitalization due to drug-related infections, injury, and overdose.² People with SUD commonly have intersecting vulnerabilities that may affect their healthcare experience and health outcomes, including housing and food insecurity, mental illness, and experiences of racism, incarceration, and other trauma. They may also harbor mistrust of healthcare providers because of previous negative encounters and discrimination with health systems.³ These vulnerabilities increase risks for COVID-19 morbidity and mortality.^4,5 The COVID-19 pandemic may drive increases in use and harms from SUD among patients who already have an SUD, with widespread job loss, insurance loss,⁶ anxiety, and social isolation on the rise. We may also see increases in return to use among people in recovery or new substance use among those without a history of SUD.

The intersecting crises of SUD and COVID-19 are important for people with SUD and for public health. In this perspective, we describe how the COVID-19 pandemic has affected people with SUD and share practical resources for hospital providers to improve care for people with SUD during the pandemic and beyond.

Mistrust of Hospitals and Healthcare Providers

Fear of stigmatization is an ongoing problem for people with SUD, who often experience discrimination in hospitals and, as a result, may avoid hospital care.⁷ Much of this stigma is based on the false but persistent belief—widespread even among healthcare providers—that addiction is the result of bad choices and limited willpower; however, the science is clear that addiction is a disorder with neurobiological, genetic, and environmental underpinnings.³ These attitudes are likely to be amplified during COVID-19, as patients and providers experience higher levels of stress.

Increased Risks of Substance Use

Typically, people who use drugs are counseled to use with others nearby so that they might administer naloxone or call 911 in the event of an overdose.⁸ With physical distancing, people may be more likely to use alone. COVID-19 also introduces uncertainty into the drug supply chain through changes in drug production and trafficking.⁹ Further, access to alcohol may be limited as liquor stores close and public transportation becomes less available. As has been shown in other complex emergencies (such as social, political, economic, and environmental disasters), these barriers to obtaining substances may increase risks for withdrawal, for needing to exchange sex for money or drugs, for sharing syringes or drug preparation equipment,¹⁰ or for consuming other available sources of substances, like rubbing alcohol or hand sanitizer. COVID-19 may also increase risk for depression, anxiety, social isolation, and suicidality, all of which increase risk for return to use and overdose.

Changes to the Treatment Milieu

Many of the resources and services that people who use substances rely on to keep safe may be disrupted by COVID-19. Social distancing—the cornerstone of mitigating COVID-19 spread—may be challenging among people with SUD. Though federal regulations around methadone dispensing and buprenorphine prescribing have loosened in response to the pandemic,¹¹ individuals in treatment may still be required to provide urine drug screens or be physically present to receive methadone doses, sometimes daily and in crowded waiting rooms.

Recovery support groups such as Alcoholics Anonymous (AA) and Self-Management and Recovery Training (SMART) provide social connection and are the foundation of many people’s recovery. While many in-person meetings have rapidly transformed to online and telephone support, they remain inaccessible to the most marginalized members of communities: people without smart phones, computers, or internet. This digital shift may also disproportionately affect older adults, people with limited English proficiency, and people with low technological literacy. Limits for other resources, such as syringe service programs, community centers, food pantries, housing shelters, and other places that people depend on for clean water, food, showers, soap, and safer spaces to use, may limit services or close altogether; those that remain open may see an unprecedented rise in need for services as millions of Americans file for unemployment. For many, anxiety about the pandemic, unemployment, financial strain, increased isolation, family stressors, illness, and community losses can lead to enormous personal distress and trigger return to use; loss of a recovery network may further exacerbate this.

Intersectionality of SUD and Other Structural Inequities

Many of the inequities that increase people’s risk for undertreated SUD also increase risk for COVID-19 infection, including racism,¹² poverty, and homelessness.⁴ “Stay home and stay safe” is not an option for people who are unsheltered or whose homes are unsafe because of risks of physical, sexual, or emotional violence. Poverty commonly forces people to live in crowded communal apartments or shelters, rely on public transportation, wait in long lines at food pantries, and continue to work, even if unwell. Many shelters have had to reduce the number of people they serve to reduce crowding and support social distancing, which further compounds risks of unstable housing. Unfortunately, the same structural inequities that exacerbate SUD worsen the COVID-19 crisis.¹³

The intersecting vulnerabilities of SUD and COVID-19 heighten an already urgent need to address SUD among hospitalized patients.¹⁴ While COVID-19 may increase harms of substance use, it may also increase people’s readiness to engage in treatment given changes to the drug supply and patient’s concerns about health risks. As such, it is even more critical to make treatment readily accessible and support harm reduction. Hospitalists can take important, actionable steps for patients with SUD—many of which are good general practices¹⁴ (Appendix Table).

Hospitalists should do the following:

1. Identify and treat acute withdrawal.¹⁵

2. Manage acute pain, including providing high-dose opioids if needed.¹⁶ Both practices (1 and 2) are evidence-based, can promote patient’s trust in providers,¹⁷ and can help avoid patients leaving against medical advice (AMA). Leaving AMA can lead to poor individual health and further threaten public health if patients leave with undiagnosed or unmanaged COVID-19 infection.

3. Encourage their hospitals to provide patients with tablets or other means to communicate with family, friends, and recovery supports via videolink, and refer patients to virtual peer support and recovery meetings during hospitalization.¹⁸ These practices may further support patients in tolerating hospitalization and prevent AMA discharge.

4. Initiate medication for addiction during admission and refer to addictions treatment after discharge. COVID-19–related regulatory changes such as expanded telehealth buprenorphine options and fewer daily dosing requirements for methadone may make this easier. Further, hospitalists should offer medication for alcohol and tobacco use disorders,¹⁵ especially given heightened possibility of unhealthy alcohol use and the respiratory complications associated with both tobacco and COVID-19.

5. Assess mental health and suicide risks¹⁹ given their association with social isolation, job loss, and financial insecurity.

6. Discuss relapse prevention among people in recovery.

7. Assess overdose risk and promote harm reduction.¹⁹ Specifically, this may include counseling patients to avoid sharing smoking supplies to avoid COVID-19 transmission, identifying places to access clean syringes, prescribing naloxone,²⁰ and providing supports so that, if patients need to use alone, they can do so more safely.²¹

8. Consider high-risk transitions that may be exacerbated by COVID-19. COVID-19 may make safe discharge plans among people experiencing homelessness very challenging. Some communities are rapidly repurposing existing spaces or building new ones to care for people without a safe place to recover after acute hospitalization, yet many communities have no such resources. Hospital teams should consider the possibility that community services and SUD treatment resources may change rapidly during the pandemic. Hospitals can maintain updated resource lists and consider partnering with state and local health departments to improve safe care for people experiencing homelessness or lacking basic services.

COVID-19 is putting enormous strain on many US hospitals. Hospital-based addictions care is under resourced in the best of times,¹⁴ and while some hospitals have addiction consult services, many do not. To what degree hospitalists and hospital teams can address anything beyond COVID-19 emergencies will vary based on settings and resources. Furthermore, we recognize that who performs various activities will depend on individual hospital’s resources and practices. Addiction consult services, if available, can play a critical role, as can hospital social workers and care managers, nurses, residents, students, and other members of the healthcare team.

Finally, though COVID-19 adds tremendous stress to hospitals, permanent improvements in SUD treatment systems such as telephone visits for buprenorphine or eased methadone restrictions may emerge that could reduce barriers to hospital-based addictions care.¹¹ Leveraging these changes now may help hospital providers to better support patients long-term.

Hospitalization can be a challenging time for patients with SUD and for the hospital teams who care for them. These tensions are exacerbated by the COVID-19 pandemic, yet hospitalists play a critical role in addressing the converging crises of SUD and COVID-19. Providing comprehensive, compassionate, evidence-based care for hospitalized patients with SUD is important for both individual and community health during COVID-19.

The authors would like to thank Alisa Patten for help preparing this manuscript.

The authors have no conflicts of interest to disclose.

Dr King received grant support from the National Institutes of Health (UG1DA015815) and the National Institute on Drug Abuse (R01DA037441). Dr Snyder received a Public Health Institute grant payable to her institution.

The spread of SARS-CoV-2, the pathogen behind the COVID-19 pandemic, has converged with an unrelenting addiction epidemic. These combined crises will have profound effects on people with substance use disorders (SUD) and people in recovery. Hospitals—which were already hit hard by the addiction epidemic—are the last line of defense in the COVID-19 pandemic. Hospitalists have an important role in balancing the effects of these intersecting, synergistic crises.

People with SUD are disproportionately affected by major medical illnesses, including infections such as hepatitis C, HIV, and cardiovascular, pulmonary, and liver diseases.¹ They also experience high rates of hospitalization due to drug-related infections, injury, and overdose.² People with SUD commonly have intersecting vulnerabilities that may affect their healthcare experience and health outcomes, including housing and food insecurity, mental illness, and experiences of racism, incarceration, and other trauma. They may also harbor mistrust of healthcare providers because of previous negative encounters and discrimination with health systems.³ These vulnerabilities increase risks for COVID-19 morbidity and mortality.^4,5 The COVID-19 pandemic may drive increases in use and harms from SUD among patients who already have an SUD, with widespread job loss, insurance loss,⁶ anxiety, and social isolation on the rise. We may also see increases in return to use among people in recovery or new substance use among those without a history of SUD.

The intersecting crises of SUD and COVID-19 are important for people with SUD and for public health. In this perspective, we describe how the COVID-19 pandemic has affected people with SUD and share practical resources for hospital providers to improve care for people with SUD during the pandemic and beyond.

Mistrust of Hospitals and Healthcare Providers

Fear of stigmatization is an ongoing problem for people with SUD, who often experience discrimination in hospitals and, as a result, may avoid hospital care.⁷ Much of this stigma is based on the false but persistent belief—widespread even among healthcare providers—that addiction is the result of bad choices and limited willpower; however, the science is clear that addiction is a disorder with neurobiological, genetic, and environmental underpinnings.³ These attitudes are likely to be amplified during COVID-19, as patients and providers experience higher levels of stress.

Increased Risks of Substance Use

Typically, people who use drugs are counseled to use with others nearby so that they might administer naloxone or call 911 in the event of an overdose.⁸ With physical distancing, people may be more likely to use alone. COVID-19 also introduces uncertainty into the drug supply chain through changes in drug production and trafficking.⁹ Further, access to alcohol may be limited as liquor stores close and public transportation becomes less available. As has been shown in other complex emergencies (such as social, political, economic, and environmental disasters), these barriers to obtaining substances may increase risks for withdrawal, for needing to exchange sex for money or drugs, for sharing syringes or drug preparation equipment,¹⁰ or for consuming other available sources of substances, like rubbing alcohol or hand sanitizer. COVID-19 may also increase risk for depression, anxiety, social isolation, and suicidality, all of which increase risk for return to use and overdose.

Changes to the Treatment Milieu

Many of the resources and services that people who use substances rely on to keep safe may be disrupted by COVID-19. Social distancing—the cornerstone of mitigating COVID-19 spread—may be challenging among people with SUD. Though federal regulations around methadone dispensing and buprenorphine prescribing have loosened in response to the pandemic,¹¹ individuals in treatment may still be required to provide urine drug screens or be physically present to receive methadone doses, sometimes daily and in crowded waiting rooms.

Recovery support groups such as Alcoholics Anonymous (AA) and Self-Management and Recovery Training (SMART) provide social connection and are the foundation of many people’s recovery. While many in-person meetings have rapidly transformed to online and telephone support, they remain inaccessible to the most marginalized members of communities: people without smart phones, computers, or internet. This digital shift may also disproportionately affect older adults, people with limited English proficiency, and people with low technological literacy. Limits for other resources, such as syringe service programs, community centers, food pantries, housing shelters, and other places that people depend on for clean water, food, showers, soap, and safer spaces to use, may limit services or close altogether; those that remain open may see an unprecedented rise in need for services as millions of Americans file for unemployment. For many, anxiety about the pandemic, unemployment, financial strain, increased isolation, family stressors, illness, and community losses can lead to enormous personal distress and trigger return to use; loss of a recovery network may further exacerbate this.

Intersectionality of SUD and Other Structural Inequities

Many of the inequities that increase people’s risk for undertreated SUD also increase risk for COVID-19 infection, including racism,¹² poverty, and homelessness.⁴ “Stay home and stay safe” is not an option for people who are unsheltered or whose homes are unsafe because of risks of physical, sexual, or emotional violence. Poverty commonly forces people to live in crowded communal apartments or shelters, rely on public transportation, wait in long lines at food pantries, and continue to work, even if unwell. Many shelters have had to reduce the number of people they serve to reduce crowding and support social distancing, which further compounds risks of unstable housing. Unfortunately, the same structural inequities that exacerbate SUD worsen the COVID-19 crisis.¹³

The intersecting vulnerabilities of SUD and COVID-19 heighten an already urgent need to address SUD among hospitalized patients.¹⁴ While COVID-19 may increase harms of substance use, it may also increase people’s readiness to engage in treatment given changes to the drug supply and patient’s concerns about health risks. As such, it is even more critical to make treatment readily accessible and support harm reduction. Hospitalists can take important, actionable steps for patients with SUD—many of which are good general practices¹⁴ (Appendix Table).

Hospitalists should do the following:

1. Identify and treat acute withdrawal.¹⁵

2. Manage acute pain, including providing high-dose opioids if needed.¹⁶ Both practices (1 and 2) are evidence-based, can promote patient’s trust in providers,¹⁷ and can help avoid patients leaving against medical advice (AMA). Leaving AMA can lead to poor individual health and further threaten public health if patients leave with undiagnosed or unmanaged COVID-19 infection.

3. Encourage their hospitals to provide patients with tablets or other means to communicate with family, friends, and recovery supports via videolink, and refer patients to virtual peer support and recovery meetings during hospitalization.¹⁸ These practices may further support patients in tolerating hospitalization and prevent AMA discharge.

4. Initiate medication for addiction during admission and refer to addictions treatment after discharge. COVID-19–related regulatory changes such as expanded telehealth buprenorphine options and fewer daily dosing requirements for methadone may make this easier. Further, hospitalists should offer medication for alcohol and tobacco use disorders,¹⁵ especially given heightened possibility of unhealthy alcohol use and the respiratory complications associated with both tobacco and COVID-19.

5. Assess mental health and suicide risks¹⁹ given their association with social isolation, job loss, and financial insecurity.

6. Discuss relapse prevention among people in recovery.

7. Assess overdose risk and promote harm reduction.¹⁹ Specifically, this may include counseling patients to avoid sharing smoking supplies to avoid COVID-19 transmission, identifying places to access clean syringes, prescribing naloxone,²⁰ and providing supports so that, if patients need to use alone, they can do so more safely.²¹

8. Consider high-risk transitions that may be exacerbated by COVID-19. COVID-19 may make safe discharge plans among people experiencing homelessness very challenging. Some communities are rapidly repurposing existing spaces or building new ones to care for people without a safe place to recover after acute hospitalization, yet many communities have no such resources. Hospital teams should consider the possibility that community services and SUD treatment resources may change rapidly during the pandemic. Hospitals can maintain updated resource lists and consider partnering with state and local health departments to improve safe care for people experiencing homelessness or lacking basic services.

COVID-19 is putting enormous strain on many US hospitals. Hospital-based addictions care is under resourced in the best of times,¹⁴ and while some hospitals have addiction consult services, many do not. To what degree hospitalists and hospital teams can address anything beyond COVID-19 emergencies will vary based on settings and resources. Furthermore, we recognize that who performs various activities will depend on individual hospital’s resources and practices. Addiction consult services, if available, can play a critical role, as can hospital social workers and care managers, nurses, residents, students, and other members of the healthcare team.

Finally, though COVID-19 adds tremendous stress to hospitals, permanent improvements in SUD treatment systems such as telephone visits for buprenorphine or eased methadone restrictions may emerge that could reduce barriers to hospital-based addictions care.¹¹ Leveraging these changes now may help hospital providers to better support patients long-term.

Hospitalization can be a challenging time for patients with SUD and for the hospital teams who care for them. These tensions are exacerbated by the COVID-19 pandemic, yet hospitalists play a critical role in addressing the converging crises of SUD and COVID-19. Providing comprehensive, compassionate, evidence-based care for hospitalized patients with SUD is important for both individual and community health during COVID-19.

The authors would like to thank Alisa Patten for help preparing this manuscript.

The authors have no conflicts of interest to disclose.

Dr King received grant support from the National Institutes of Health (UG1DA015815) and the National Institute on Drug Abuse (R01DA037441). Dr Snyder received a Public Health Institute grant payable to her institution.

The spread of SARS-CoV-2, the pathogen behind the COVID-19 pandemic, has converged with an unrelenting addiction epidemic. These combined crises will have profound effects on people with substance use disorders (SUD) and people in recovery. Hospitals—which were already hit hard by the addiction epidemic—are the last line of defense in the COVID-19 pandemic. Hospitalists have an important role in balancing the effects of these intersecting, synergistic crises.

People with SUD are disproportionately affected by major medical illnesses, including infections such as hepatitis C, HIV, and cardiovascular, pulmonary, and liver diseases.¹ They also experience high rates of hospitalization due to drug-related infections, injury, and overdose.² People with SUD commonly have intersecting vulnerabilities that may affect their healthcare experience and health outcomes, including housing and food insecurity, mental illness, and experiences of racism, incarceration, and other trauma. They may also harbor mistrust of healthcare providers because of previous negative encounters and discrimination with health systems.³ These vulnerabilities increase risks for COVID-19 morbidity and mortality.^4,5 The COVID-19 pandemic may drive increases in use and harms from SUD among patients who already have an SUD, with widespread job loss, insurance loss,⁶ anxiety, and social isolation on the rise. We may also see increases in return to use among people in recovery or new substance use among those without a history of SUD.

The intersecting crises of SUD and COVID-19 are important for people with SUD and for public health. In this perspective, we describe how the COVID-19 pandemic has affected people with SUD and share practical resources for hospital providers to improve care for people with SUD during the pandemic and beyond.

Mistrust of Hospitals and Healthcare Providers

Fear of stigmatization is an ongoing problem for people with SUD, who often experience discrimination in hospitals and, as a result, may avoid hospital care.⁷ Much of this stigma is based on the false but persistent belief—widespread even among healthcare providers—that addiction is the result of bad choices and limited willpower; however, the science is clear that addiction is a disorder with neurobiological, genetic, and environmental underpinnings.³ These attitudes are likely to be amplified during COVID-19, as patients and providers experience higher levels of stress.

Increased Risks of Substance Use

Typically, people who use drugs are counseled to use with others nearby so that they might administer naloxone or call 911 in the event of an overdose.⁸ With physical distancing, people may be more likely to use alone. COVID-19 also introduces uncertainty into the drug supply chain through changes in drug production and trafficking.⁹ Further, access to alcohol may be limited as liquor stores close and public transportation becomes less available. As has been shown in other complex emergencies (such as social, political, economic, and environmental disasters), these barriers to obtaining substances may increase risks for withdrawal, for needing to exchange sex for money or drugs, for sharing syringes or drug preparation equipment,¹⁰ or for consuming other available sources of substances, like rubbing alcohol or hand sanitizer. COVID-19 may also increase risk for depression, anxiety, social isolation, and suicidality, all of which increase risk for return to use and overdose.

Changes to the Treatment Milieu

Many of the resources and services that people who use substances rely on to keep safe may be disrupted by COVID-19. Social distancing—the cornerstone of mitigating COVID-19 spread—may be challenging among people with SUD. Though federal regulations around methadone dispensing and buprenorphine prescribing have loosened in response to the pandemic,¹¹ individuals in treatment may still be required to provide urine drug screens or be physically present to receive methadone doses, sometimes daily and in crowded waiting rooms.

Recovery support groups such as Alcoholics Anonymous (AA) and Self-Management and Recovery Training (SMART) provide social connection and are the foundation of many people’s recovery. While many in-person meetings have rapidly transformed to online and telephone support, they remain inaccessible to the most marginalized members of communities: people without smart phones, computers, or internet. This digital shift may also disproportionately affect older adults, people with limited English proficiency, and people with low technological literacy. Limits for other resources, such as syringe service programs, community centers, food pantries, housing shelters, and other places that people depend on for clean water, food, showers, soap, and safer spaces to use, may limit services or close altogether; those that remain open may see an unprecedented rise in need for services as millions of Americans file for unemployment. For many, anxiety about the pandemic, unemployment, financial strain, increased isolation, family stressors, illness, and community losses can lead to enormous personal distress and trigger return to use; loss of a recovery network may further exacerbate this.

Intersectionality of SUD and Other Structural Inequities

Many of the inequities that increase people’s risk for undertreated SUD also increase risk for COVID-19 infection, including racism,¹² poverty, and homelessness.⁴ “Stay home and stay safe” is not an option for people who are unsheltered or whose homes are unsafe because of risks of physical, sexual, or emotional violence. Poverty commonly forces people to live in crowded communal apartments or shelters, rely on public transportation, wait in long lines at food pantries, and continue to work, even if unwell. Many shelters have had to reduce the number of people they serve to reduce crowding and support social distancing, which further compounds risks of unstable housing. Unfortunately, the same structural inequities that exacerbate SUD worsen the COVID-19 crisis.¹³

The intersecting vulnerabilities of SUD and COVID-19 heighten an already urgent need to address SUD among hospitalized patients.¹⁴ While COVID-19 may increase harms of substance use, it may also increase people’s readiness to engage in treatment given changes to the drug supply and patient’s concerns about health risks. As such, it is even more critical to make treatment readily accessible and support harm reduction. Hospitalists can take important, actionable steps for patients with SUD—many of which are good general practices¹⁴ (Appendix Table).

Hospitalists should do the following:

1. Identify and treat acute withdrawal.¹⁵

2. Manage acute pain, including providing high-dose opioids if needed.¹⁶ Both practices (1 and 2) are evidence-based, can promote patient’s trust in providers,¹⁷ and can help avoid patients leaving against medical advice (AMA). Leaving AMA can lead to poor individual health and further threaten public health if patients leave with undiagnosed or unmanaged COVID-19 infection.

3. Encourage their hospitals to provide patients with tablets or other means to communicate with family, friends, and recovery supports via videolink, and refer patients to virtual peer support and recovery meetings during hospitalization.¹⁸ These practices may further support patients in tolerating hospitalization and prevent AMA discharge.

4. Initiate medication for addiction during admission and refer to addictions treatment after discharge. COVID-19–related regulatory changes such as expanded telehealth buprenorphine options and fewer daily dosing requirements for methadone may make this easier. Further, hospitalists should offer medication for alcohol and tobacco use disorders,¹⁵ especially given heightened possibility of unhealthy alcohol use and the respiratory complications associated with both tobacco and COVID-19.

5. Assess mental health and suicide risks¹⁹ given their association with social isolation, job loss, and financial insecurity.

6. Discuss relapse prevention among people in recovery.

7. Assess overdose risk and promote harm reduction.¹⁹ Specifically, this may include counseling patients to avoid sharing smoking supplies to avoid COVID-19 transmission, identifying places to access clean syringes, prescribing naloxone,²⁰ and providing supports so that, if patients need to use alone, they can do so more safely.²¹

8. Consider high-risk transitions that may be exacerbated by COVID-19. COVID-19 may make safe discharge plans among people experiencing homelessness very challenging. Some communities are rapidly repurposing existing spaces or building new ones to care for people without a safe place to recover after acute hospitalization, yet many communities have no such resources. Hospital teams should consider the possibility that community services and SUD treatment resources may change rapidly during the pandemic. Hospitals can maintain updated resource lists and consider partnering with state and local health departments to improve safe care for people experiencing homelessness or lacking basic services.

COVID-19 is putting enormous strain on many US hospitals. Hospital-based addictions care is under resourced in the best of times,¹⁴ and while some hospitals have addiction consult services, many do not. To what degree hospitalists and hospital teams can address anything beyond COVID-19 emergencies will vary based on settings and resources. Furthermore, we recognize that who performs various activities will depend on individual hospital’s resources and practices. Addiction consult services, if available, can play a critical role, as can hospital social workers and care managers, nurses, residents, students, and other members of the healthcare team.

Finally, though COVID-19 adds tremendous stress to hospitals, permanent improvements in SUD treatment systems such as telephone visits for buprenorphine or eased methadone restrictions may emerge that could reduce barriers to hospital-based addictions care.¹¹ Leveraging these changes now may help hospital providers to better support patients long-term.

Hospitalization can be a challenging time for patients with SUD and for the hospital teams who care for them. These tensions are exacerbated by the COVID-19 pandemic, yet hospitalists play a critical role in addressing the converging crises of SUD and COVID-19. Providing comprehensive, compassionate, evidence-based care for hospitalized patients with SUD is important for both individual and community health during COVID-19.

The authors would like to thank Alisa Patten for help preparing this manuscript.

The authors have no conflicts of interest to disclose.

Dr King received grant support from the National Institutes of Health (UG1DA015815) and the National Institute on Drug Abuse (R01DA037441). Dr Snyder received a Public Health Institute grant payable to her institution.

Public health emergencies may require shifting from conventional to contingency and ultimately to crisis standards of care, which prompts consideration of needs and resources across hospital systems.^1,2 Within conventional care contexts, institutions have their usual resources including supplies, staff, and space and are able to provide a usual standard of care to patients. As institutions anticipate shortages in an emergency, they may enter a contingency state. In this state, the institution begins to plan for shortages, often by finding alternative uses of supplies, staff, and space that are functionally equivalent but still aiming to conserve resources such as rescheduling elective procedures and using alternative but functionally equivalent personal protective equipment. Still, during this state, institutions are able to provide the usual standard of care.

Under crisis standards of care, resources have reached a level of scarcity or circumstances are such that they do not permit normal operations. In this state, institutions may not be able to meet the usual standard of care. Instead, institutions are expected to provide care that is sufficient given available resources and circumstances. How to utilize scarce resources, however, invokes consideration of the ethical duties of institutions. Despite the likelihood of entering crisis standards of care (CSCs) in the current COVID-19 pandemic, limited ethical guidance exists regarding how institutions should relate to each other in a crisis. Relevant moral duties during conventional, contingency, and CSCs include duties of rescue, fidelity, solidarity, and justice. As CSCs develop, these duties require limiting elective procedures and instituting triage in certain circumstances, but how this relates to coordination among hospitals is unclear.

We argue that the primary role of pediatric institutions during the COVID-19 pandemic under CSCs is increasing system capacity by regionalization of pediatric care. Under regionalization of care, children’s hospitals that serve as local/regional referral centers would preferentially take all pediatric patients in the region, including those who might normally be admitted to a primarily adult hospital, thereby increasing availability of beds and resources at primarily adult facilities. This maximizes the expertise and resources of pediatric institutions and avoids unnecessary harm to all patients by mitigating shortages before any hospital faces conditions in which they need to invoke triage procedures. General hospitals should transfer pediatric patients to pediatric institutions and should consider transfer of patients and/or resources between regional institutions, which helps avoid triage conditions until all accessible resources are in use.

Institutions are prominent moral actors with duties to patients extending beyond those of providers.³

The duty to treat includes two subsidiary duties. First, the duty of rescue has a special role in emergencies, requiring providers to intervene with those helpless without assistance.^4-6 For children’s hospitals, this means providing care for children in the region who cannot receive needed care elsewhere. Second, the duty of fidelity requires promoting patients’ good, including giving precedence to patients with established treatment relationships.⁷

Institutions also have a duty of solidarity.⁸ Institutions must recognize they are bound together to care for the broader community and should work in tandem.⁹ Solidarity encompasses the duty of stewardship—responsibly using resources to mitigate shortages; this duty sometimes requires subsuming patient, provider, or institutional needs for overall community benefit.

Finally, institutions have duties of justice,² to provide fair and equitable care with transparency and trustworthiness. Justice requires that institutions ensure shifting to CSCs does not disfavor already disadvantaged groups.¹⁰

Public health emergencies strain health care resources in ways that hinder providing usual standards of care. Public health ethics guide healthcare systems during contingency or CSCs in ethically grounded approaches to mitigate shortages and allocate resources.^1,2 We consider how duties evolve from conventional care to CSCs, with a focus on actions to meet institutional duties under changing circumstances.

Conventional Care

Ordinarily, institutions provide usual standards of care, which follow typical operations. Interactions between institutions and providers rely on basic ethical principles, including primacy of patient welfare, autonomy, and social justice. A degree of redundancy allows institutions to meet duties of rescue, fidelity, solidarity, and justice even with increased demand. The duty to treat is primary but requires balancing duties to rescue with fiduciary duties. Thus, if the institution were near capacity and a decision is needed about which patient to accept in transfer, avoiding irreversible harm to a previously unknown patient who could not receive adequate care in the community should take precedence over accepting an established patient who could receive adequate care elsewhere. If neither patient could receive adequate care elsewhere, the patient known to the children’s hospital should be accepted, under the duty of fidelity. Fidelity also requires that patients currently admitted continue to receive treatment. Justice requires fair and equitable treatment of patients, without consideration of morally irrelevant features (eg, race or immigration status).

Contingency Care

Contingency care begins when a public health emergency introduces strains on hospital resources.^1,2,11 As long as typical or alternative resources last, adaptations in care have minimal effects on quality, and the duties of rescue, fidelity, and justice mirror conventional care; however, operations begin to shift to recognize greater duties of solidarity. In the COVID-19 pandemic, given their missions to provide specialized care for children, pediatric hospitals can meet their duty to treat by accepting patients who might otherwise receive care elsewhere. Children’s hospitals should consider accepting any child for which they have capacity to help decompress other systems (eg, liberating beds for more adults at other institutions). Children’s hospitals should also continue to preferentially admit children requiring tertiary care (eg, neonates requiring subspecialty surgery), which respects the duty of rescue.

The duty of solidarity supports strategic sharing and stewarding of resources, including personal protective equipment, ventilators, and staff. Strategies might include postponing elective procedures, repurposing facilities, or limiting staff entering isolation rooms; such alterations to standard care require careful discussions with providers to anticipate negative consequences, ensure safe practices, and plan for reassessment.

The duty of rescue requires maintaining ability to care for patients who cannot receive adequate care elsewhere. Institutions can meet this duty by reserving a small number of intensive care and general beds to care for patients needing emergent specialty care.

Crisis Standards of Care

Under CSCs, resources are insufficient to maintain usual standards of care and mitigation attempts no longer suffice. Scarcity demands greater duties of solidarity, reducing attention to some individuals to promote the community good. To meet duties of solidarity, institutions should prepare for triage after exhausting efforts to preserve system resources.

During a pandemic such as COVID-19 that primarily affects adults, pediatric resources should be consolidated by transferring children to regional pediatric facilities. Without transfer, children who present to primarily adult facilities, where resources are more strained given the higher burden of disease in adults, may otherwise be subject to triaging of scarce resources at the adult facility. But, no child should have care determined by any hospital’s triage system if any pediatric bed is available within a region, and if pediatric resources are regionalized, children will be less likely to face triage at primarily adult facilities unless the entire system has reached capacity. In addition to regionalization, children’s hospitals may also face requests to accept adult patients or share equipment and/or staff with adult facilities; when these actions do not compromise the capacity of the pediatric institution to provide care to children, institutions should consider them.¹² However, pediatric institutions can best meet the duty of solidarity by expanding regional capacity through freeing up resources in general hospitals, including beds, ventilators, and staffing usable for adults, preventing all hospitals from needing to triage. If triage is necessary because the entire system has reached capacity, triage should also take place at children’s hospitals, in respect of solidarity, to optimize this community resource.

Under CSCs, significant practice variation in triage policies may occur. Regional institutions may individually employ triage policies during crisis standards of care and deny critical care resources to some individuals who might receive them in noncrisis times, when there isn’t such scarcity. Minimizing denials across a region requires collaboration between centers to ensure solidarity. Processes should be fair and equitable. Justice entails ensuring consistency in allocation criteria, with differences prioritizing those least well off. Triage teams in a region should use consistent, aligned processes so that similarly situated patients have equitable access to resources and care across centers. However, triaging pediatric and adult patients together could disadvantage children (eg, priority given to health workers); moreover, illness severity measures for infants/children differ from those applicable to adults, which makes equivalent scoring for allocation challenging.¹³ Some resources are specific to pediatric or adult care. Therefore, it may be necessary to separate pediatric and adult allocation processes.

Triage criteria must not discriminate based on morally irrelevant criteria, such as sex, race/ethnicity, or disability.¹ Institutions using “objective” scoring systems for morbidity and mortality should acknowledge that these systems could disadvantage marginalized populations with higher rates of chronic conditions resulting from systemic inequities.

A commitment to justice mandates that no patient should be triaged if the required resources (eg, ventilators) are available at a regional hospital and transfer is feasible. Transfer should occur across all regional hospitals, not just partners within hospital networks. Facilitating transfers requires institutions to engage in close communication. If no centralized external system exists, a group of individuals with knowledge of inpatient resources—but without direct care duties—should provide coordination.

Because CSCs are so different from conventional standards, institutions should collect data on regionalization and triage protocols. Recognition of inequitable outcomes may necessitate changing scoring criteria or reveal disproportionate burdens on vulnerable populations.

To maintain public trust and promote justice, institutions must be transparent regarding triage policies and procedures for CSCs. These should be available for public review, revised with public input, and readily available once finalized.

Despite the ethical justification for regional coordination of care and resources, there are multiple barriers to implementation. Providers and families may hesitate to disturb continuity of care at medical homes. Organizations may have financial disincentives to transfer long-term patients to new institutions. Openness with patients and families regarding the temporary nature of transfers and plans to return to their usual care may help. Granting temporary privileges at recipient institutions for providers to continue seeing their patients may lessen discontinuity. Solidarity in public health emergencies requires all institutions to compromise their own interests to some degree.

Similarly, barriers in achieving consistency across institutional triage policies may arise. Allocation strategies embody multiple values, for example, regarding quality of life or contributions of essential workers. Resolution of these value differences may prove difficult.

In the current COVID-19 pandemic, an ethical approach to CSCs necessitates coordination to align available resources at the regional, rather than institutional, level to avoid triage at individual institutions. Pediatric regionalization of care is the first step in freeing up system capacity for adults. Solidarity rises in importance, but must be balanced by duties of rescue, fidelity, and justice so that pediatric institutions continue to care for children with urgent needs requiring pediatric expertise.

Dr Paquette reported funding under the Pediatric Critical Care and Trauma Scientist Development Program, NICHD K12HDO47349 and NICHD Loan Repayment Program L40 HD089260. Dr Derrington is a director at large for the American Society of Bioethics and Medical Humanities and had travel expenses reimbursed for the annual conference in 2019. Dr Michelson has received funding from the National Palliative Care Research Center and is a consultant on a National Institutes of Health study that are unrelated to this work. Dr Michelson is also involved in unrelated work supported by the National Alliance for Grieving Children. All other authors declared they have nothing to disclose.

Public health emergencies may require shifting from conventional to contingency and ultimately to crisis standards of care, which prompts consideration of needs and resources across hospital systems.^1,2 Within conventional care contexts, institutions have their usual resources including supplies, staff, and space and are able to provide a usual standard of care to patients. As institutions anticipate shortages in an emergency, they may enter a contingency state. In this state, the institution begins to plan for shortages, often by finding alternative uses of supplies, staff, and space that are functionally equivalent but still aiming to conserve resources such as rescheduling elective procedures and using alternative but functionally equivalent personal protective equipment. Still, during this state, institutions are able to provide the usual standard of care.

Under crisis standards of care, resources have reached a level of scarcity or circumstances are such that they do not permit normal operations. In this state, institutions may not be able to meet the usual standard of care. Instead, institutions are expected to provide care that is sufficient given available resources and circumstances. How to utilize scarce resources, however, invokes consideration of the ethical duties of institutions. Despite the likelihood of entering crisis standards of care (CSCs) in the current COVID-19 pandemic, limited ethical guidance exists regarding how institutions should relate to each other in a crisis. Relevant moral duties during conventional, contingency, and CSCs include duties of rescue, fidelity, solidarity, and justice. As CSCs develop, these duties require limiting elective procedures and instituting triage in certain circumstances, but how this relates to coordination among hospitals is unclear.

We argue that the primary role of pediatric institutions during the COVID-19 pandemic under CSCs is increasing system capacity by regionalization of pediatric care. Under regionalization of care, children’s hospitals that serve as local/regional referral centers would preferentially take all pediatric patients in the region, including those who might normally be admitted to a primarily adult hospital, thereby increasing availability of beds and resources at primarily adult facilities. This maximizes the expertise and resources of pediatric institutions and avoids unnecessary harm to all patients by mitigating shortages before any hospital faces conditions in which they need to invoke triage procedures. General hospitals should transfer pediatric patients to pediatric institutions and should consider transfer of patients and/or resources between regional institutions, which helps avoid triage conditions until all accessible resources are in use.

Institutions are prominent moral actors with duties to patients extending beyond those of providers.³

The duty to treat includes two subsidiary duties. First, the duty of rescue has a special role in emergencies, requiring providers to intervene with those helpless without assistance.^4-6 For children’s hospitals, this means providing care for children in the region who cannot receive needed care elsewhere. Second, the duty of fidelity requires promoting patients’ good, including giving precedence to patients with established treatment relationships.⁷

Institutions also have a duty of solidarity.⁸ Institutions must recognize they are bound together to care for the broader community and should work in tandem.⁹ Solidarity encompasses the duty of stewardship—responsibly using resources to mitigate shortages; this duty sometimes requires subsuming patient, provider, or institutional needs for overall community benefit.

Finally, institutions have duties of justice,² to provide fair and equitable care with transparency and trustworthiness. Justice requires that institutions ensure shifting to CSCs does not disfavor already disadvantaged groups.¹⁰

Public health emergencies strain health care resources in ways that hinder providing usual standards of care. Public health ethics guide healthcare systems during contingency or CSCs in ethically grounded approaches to mitigate shortages and allocate resources.^1,2 We consider how duties evolve from conventional care to CSCs, with a focus on actions to meet institutional duties under changing circumstances.

Conventional Care

Ordinarily, institutions provide usual standards of care, which follow typical operations. Interactions between institutions and providers rely on basic ethical principles, including primacy of patient welfare, autonomy, and social justice. A degree of redundancy allows institutions to meet duties of rescue, fidelity, solidarity, and justice even with increased demand. The duty to treat is primary but requires balancing duties to rescue with fiduciary duties. Thus, if the institution were near capacity and a decision is needed about which patient to accept in transfer, avoiding irreversible harm to a previously unknown patient who could not receive adequate care in the community should take precedence over accepting an established patient who could receive adequate care elsewhere. If neither patient could receive adequate care elsewhere, the patient known to the children’s hospital should be accepted, under the duty of fidelity. Fidelity also requires that patients currently admitted continue to receive treatment. Justice requires fair and equitable treatment of patients, without consideration of morally irrelevant features (eg, race or immigration status).

Contingency Care

Contingency care begins when a public health emergency introduces strains on hospital resources.^1,2,11 As long as typical or alternative resources last, adaptations in care have minimal effects on quality, and the duties of rescue, fidelity, and justice mirror conventional care; however, operations begin to shift to recognize greater duties of solidarity. In the COVID-19 pandemic, given their missions to provide specialized care for children, pediatric hospitals can meet their duty to treat by accepting patients who might otherwise receive care elsewhere. Children’s hospitals should consider accepting any child for which they have capacity to help decompress other systems (eg, liberating beds for more adults at other institutions). Children’s hospitals should also continue to preferentially admit children requiring tertiary care (eg, neonates requiring subspecialty surgery), which respects the duty of rescue.

The duty of solidarity supports strategic sharing and stewarding of resources, including personal protective equipment, ventilators, and staff. Strategies might include postponing elective procedures, repurposing facilities, or limiting staff entering isolation rooms; such alterations to standard care require careful discussions with providers to anticipate negative consequences, ensure safe practices, and plan for reassessment.

The duty of rescue requires maintaining ability to care for patients who cannot receive adequate care elsewhere. Institutions can meet this duty by reserving a small number of intensive care and general beds to care for patients needing emergent specialty care.

Crisis Standards of Care

Under CSCs, resources are insufficient to maintain usual standards of care and mitigation attempts no longer suffice. Scarcity demands greater duties of solidarity, reducing attention to some individuals to promote the community good. To meet duties of solidarity, institutions should prepare for triage after exhausting efforts to preserve system resources.

During a pandemic such as COVID-19 that primarily affects adults, pediatric resources should be consolidated by transferring children to regional pediatric facilities. Without transfer, children who present to primarily adult facilities, where resources are more strained given the higher burden of disease in adults, may otherwise be subject to triaging of scarce resources at the adult facility. But, no child should have care determined by any hospital’s triage system if any pediatric bed is available within a region, and if pediatric resources are regionalized, children will be less likely to face triage at primarily adult facilities unless the entire system has reached capacity. In addition to regionalization, children’s hospitals may also face requests to accept adult patients or share equipment and/or staff with adult facilities; when these actions do not compromise the capacity of the pediatric institution to provide care to children, institutions should consider them.¹² However, pediatric institutions can best meet the duty of solidarity by expanding regional capacity through freeing up resources in general hospitals, including beds, ventilators, and staffing usable for adults, preventing all hospitals from needing to triage. If triage is necessary because the entire system has reached capacity, triage should also take place at children’s hospitals, in respect of solidarity, to optimize this community resource.

Under CSCs, significant practice variation in triage policies may occur. Regional institutions may individually employ triage policies during crisis standards of care and deny critical care resources to some individuals who might receive them in noncrisis times, when there isn’t such scarcity. Minimizing denials across a region requires collaboration between centers to ensure solidarity. Processes should be fair and equitable. Justice entails ensuring consistency in allocation criteria, with differences prioritizing those least well off. Triage teams in a region should use consistent, aligned processes so that similarly situated patients have equitable access to resources and care across centers. However, triaging pediatric and adult patients together could disadvantage children (eg, priority given to health workers); moreover, illness severity measures for infants/children differ from those applicable to adults, which makes equivalent scoring for allocation challenging.¹³ Some resources are specific to pediatric or adult care. Therefore, it may be necessary to separate pediatric and adult allocation processes.

Triage criteria must not discriminate based on morally irrelevant criteria, such as sex, race/ethnicity, or disability.¹ Institutions using “objective” scoring systems for morbidity and mortality should acknowledge that these systems could disadvantage marginalized populations with higher rates of chronic conditions resulting from systemic inequities.

A commitment to justice mandates that no patient should be triaged if the required resources (eg, ventilators) are available at a regional hospital and transfer is feasible. Transfer should occur across all regional hospitals, not just partners within hospital networks. Facilitating transfers requires institutions to engage in close communication. If no centralized external system exists, a group of individuals with knowledge of inpatient resources—but without direct care duties—should provide coordination.

Because CSCs are so different from conventional standards, institutions should collect data on regionalization and triage protocols. Recognition of inequitable outcomes may necessitate changing scoring criteria or reveal disproportionate burdens on vulnerable populations.

To maintain public trust and promote justice, institutions must be transparent regarding triage policies and procedures for CSCs. These should be available for public review, revised with public input, and readily available once finalized.

Despite the ethical justification for regional coordination of care and resources, there are multiple barriers to implementation. Providers and families may hesitate to disturb continuity of care at medical homes. Organizations may have financial disincentives to transfer long-term patients to new institutions. Openness with patients and families regarding the temporary nature of transfers and plans to return to their usual care may help. Granting temporary privileges at recipient institutions for providers to continue seeing their patients may lessen discontinuity. Solidarity in public health emergencies requires all institutions to compromise their own interests to some degree.

Similarly, barriers in achieving consistency across institutional triage policies may arise. Allocation strategies embody multiple values, for example, regarding quality of life or contributions of essential workers. Resolution of these value differences may prove difficult.

In the current COVID-19 pandemic, an ethical approach to CSCs necessitates coordination to align available resources at the regional, rather than institutional, level to avoid triage at individual institutions. Pediatric regionalization of care is the first step in freeing up system capacity for adults. Solidarity rises in importance, but must be balanced by duties of rescue, fidelity, and justice so that pediatric institutions continue to care for children with urgent needs requiring pediatric expertise.

Dr Paquette reported funding under the Pediatric Critical Care and Trauma Scientist Development Program, NICHD K12HDO47349 and NICHD Loan Repayment Program L40 HD089260. Dr Derrington is a director at large for the American Society of Bioethics and Medical Humanities and had travel expenses reimbursed for the annual conference in 2019. Dr Michelson has received funding from the National Palliative Care Research Center and is a consultant on a National Institutes of Health study that are unrelated to this work. Dr Michelson is also involved in unrelated work supported by the National Alliance for Grieving Children. All other authors declared they have nothing to disclose.

Public health emergencies may require shifting from conventional to contingency and ultimately to crisis standards of care, which prompts consideration of needs and resources across hospital systems.^1,2 Within conventional care contexts, institutions have their usual resources including supplies, staff, and space and are able to provide a usual standard of care to patients. As institutions anticipate shortages in an emergency, they may enter a contingency state. In this state, the institution begins to plan for shortages, often by finding alternative uses of supplies, staff, and space that are functionally equivalent but still aiming to conserve resources such as rescheduling elective procedures and using alternative but functionally equivalent personal protective equipment. Still, during this state, institutions are able to provide the usual standard of care.

Under crisis standards of care, resources have reached a level of scarcity or circumstances are such that they do not permit normal operations. In this state, institutions may not be able to meet the usual standard of care. Instead, institutions are expected to provide care that is sufficient given available resources and circumstances. How to utilize scarce resources, however, invokes consideration of the ethical duties of institutions. Despite the likelihood of entering crisis standards of care (CSCs) in the current COVID-19 pandemic, limited ethical guidance exists regarding how institutions should relate to each other in a crisis. Relevant moral duties during conventional, contingency, and CSCs include duties of rescue, fidelity, solidarity, and justice. As CSCs develop, these duties require limiting elective procedures and instituting triage in certain circumstances, but how this relates to coordination among hospitals is unclear.

We argue that the primary role of pediatric institutions during the COVID-19 pandemic under CSCs is increasing system capacity by regionalization of pediatric care. Under regionalization of care, children’s hospitals that serve as local/regional referral centers would preferentially take all pediatric patients in the region, including those who might normally be admitted to a primarily adult hospital, thereby increasing availability of beds and resources at primarily adult facilities. This maximizes the expertise and resources of pediatric institutions and avoids unnecessary harm to all patients by mitigating shortages before any hospital faces conditions in which they need to invoke triage procedures. General hospitals should transfer pediatric patients to pediatric institutions and should consider transfer of patients and/or resources between regional institutions, which helps avoid triage conditions until all accessible resources are in use.

Institutions are prominent moral actors with duties to patients extending beyond those of providers.³

The duty to treat includes two subsidiary duties. First, the duty of rescue has a special role in emergencies, requiring providers to intervene with those helpless without assistance.^4-6 For children’s hospitals, this means providing care for children in the region who cannot receive needed care elsewhere. Second, the duty of fidelity requires promoting patients’ good, including giving precedence to patients with established treatment relationships.⁷

Institutions also have a duty of solidarity.⁸ Institutions must recognize they are bound together to care for the broader community and should work in tandem.⁹ Solidarity encompasses the duty of stewardship—responsibly using resources to mitigate shortages; this duty sometimes requires subsuming patient, provider, or institutional needs for overall community benefit.

Finally, institutions have duties of justice,² to provide fair and equitable care with transparency and trustworthiness. Justice requires that institutions ensure shifting to CSCs does not disfavor already disadvantaged groups.¹⁰

Public health emergencies strain health care resources in ways that hinder providing usual standards of care. Public health ethics guide healthcare systems during contingency or CSCs in ethically grounded approaches to mitigate shortages and allocate resources.^1,2 We consider how duties evolve from conventional care to CSCs, with a focus on actions to meet institutional duties under changing circumstances.

Conventional Care

Ordinarily, institutions provide usual standards of care, which follow typical operations. Interactions between institutions and providers rely on basic ethical principles, including primacy of patient welfare, autonomy, and social justice. A degree of redundancy allows institutions to meet duties of rescue, fidelity, solidarity, and justice even with increased demand. The duty to treat is primary but requires balancing duties to rescue with fiduciary duties. Thus, if the institution were near capacity and a decision is needed about which patient to accept in transfer, avoiding irreversible harm to a previously unknown patient who could not receive adequate care in the community should take precedence over accepting an established patient who could receive adequate care elsewhere. If neither patient could receive adequate care elsewhere, the patient known to the children’s hospital should be accepted, under the duty of fidelity. Fidelity also requires that patients currently admitted continue to receive treatment. Justice requires fair and equitable treatment of patients, without consideration of morally irrelevant features (eg, race or immigration status).

Contingency Care

Contingency care begins when a public health emergency introduces strains on hospital resources.^1,2,11 As long as typical or alternative resources last, adaptations in care have minimal effects on quality, and the duties of rescue, fidelity, and justice mirror conventional care; however, operations begin to shift to recognize greater duties of solidarity. In the COVID-19 pandemic, given their missions to provide specialized care for children, pediatric hospitals can meet their duty to treat by accepting patients who might otherwise receive care elsewhere. Children’s hospitals should consider accepting any child for which they have capacity to help decompress other systems (eg, liberating beds for more adults at other institutions). Children’s hospitals should also continue to preferentially admit children requiring tertiary care (eg, neonates requiring subspecialty surgery), which respects the duty of rescue.

The duty of solidarity supports strategic sharing and stewarding of resources, including personal protective equipment, ventilators, and staff. Strategies might include postponing elective procedures, repurposing facilities, or limiting staff entering isolation rooms; such alterations to standard care require careful discussions with providers to anticipate negative consequences, ensure safe practices, and plan for reassessment.

The duty of rescue requires maintaining ability to care for patients who cannot receive adequate care elsewhere. Institutions can meet this duty by reserving a small number of intensive care and general beds to care for patients needing emergent specialty care.

Crisis Standards of Care

Under CSCs, resources are insufficient to maintain usual standards of care and mitigation attempts no longer suffice. Scarcity demands greater duties of solidarity, reducing attention to some individuals to promote the community good. To meet duties of solidarity, institutions should prepare for triage after exhausting efforts to preserve system resources.

During a pandemic such as COVID-19 that primarily affects adults, pediatric resources should be consolidated by transferring children to regional pediatric facilities. Without transfer, children who present to primarily adult facilities, where resources are more strained given the higher burden of disease in adults, may otherwise be subject to triaging of scarce resources at the adult facility. But, no child should have care determined by any hospital’s triage system if any pediatric bed is available within a region, and if pediatric resources are regionalized, children will be less likely to face triage at primarily adult facilities unless the entire system has reached capacity. In addition to regionalization, children’s hospitals may also face requests to accept adult patients or share equipment and/or staff with adult facilities; when these actions do not compromise the capacity of the pediatric institution to provide care to children, institutions should consider them.¹² However, pediatric institutions can best meet the duty of solidarity by expanding regional capacity through freeing up resources in general hospitals, including beds, ventilators, and staffing usable for adults, preventing all hospitals from needing to triage. If triage is necessary because the entire system has reached capacity, triage should also take place at children’s hospitals, in respect of solidarity, to optimize this community resource.

Under CSCs, significant practice variation in triage policies may occur. Regional institutions may individually employ triage policies during crisis standards of care and deny critical care resources to some individuals who might receive them in noncrisis times, when there isn’t such scarcity. Minimizing denials across a region requires collaboration between centers to ensure solidarity. Processes should be fair and equitable. Justice entails ensuring consistency in allocation criteria, with differences prioritizing those least well off. Triage teams in a region should use consistent, aligned processes so that similarly situated patients have equitable access to resources and care across centers. However, triaging pediatric and adult patients together could disadvantage children (eg, priority given to health workers); moreover, illness severity measures for infants/children differ from those applicable to adults, which makes equivalent scoring for allocation challenging.¹³ Some resources are specific to pediatric or adult care. Therefore, it may be necessary to separate pediatric and adult allocation processes.

Triage criteria must not discriminate based on morally irrelevant criteria, such as sex, race/ethnicity, or disability.¹ Institutions using “objective” scoring systems for morbidity and mortality should acknowledge that these systems could disadvantage marginalized populations with higher rates of chronic conditions resulting from systemic inequities.

A commitment to justice mandates that no patient should be triaged if the required resources (eg, ventilators) are available at a regional hospital and transfer is feasible. Transfer should occur across all regional hospitals, not just partners within hospital networks. Facilitating transfers requires institutions to engage in close communication. If no centralized external system exists, a group of individuals with knowledge of inpatient resources—but without direct care duties—should provide coordination.

Because CSCs are so different from conventional standards, institutions should collect data on regionalization and triage protocols. Recognition of inequitable outcomes may necessitate changing scoring criteria or reveal disproportionate burdens on vulnerable populations.

To maintain public trust and promote justice, institutions must be transparent regarding triage policies and procedures for CSCs. These should be available for public review, revised with public input, and readily available once finalized.

Despite the ethical justification for regional coordination of care and resources, there are multiple barriers to implementation. Providers and families may hesitate to disturb continuity of care at medical homes. Organizations may have financial disincentives to transfer long-term patients to new institutions. Openness with patients and families regarding the temporary nature of transfers and plans to return to their usual care may help. Granting temporary privileges at recipient institutions for providers to continue seeing their patients may lessen discontinuity. Solidarity in public health emergencies requires all institutions to compromise their own interests to some degree.

Similarly, barriers in achieving consistency across institutional triage policies may arise. Allocation strategies embody multiple values, for example, regarding quality of life or contributions of essential workers. Resolution of these value differences may prove difficult.

In the current COVID-19 pandemic, an ethical approach to CSCs necessitates coordination to align available resources at the regional, rather than institutional, level to avoid triage at individual institutions. Pediatric regionalization of care is the first step in freeing up system capacity for adults. Solidarity rises in importance, but must be balanced by duties of rescue, fidelity, and justice so that pediatric institutions continue to care for children with urgent needs requiring pediatric expertise.

Dr Paquette reported funding under the Pediatric Critical Care and Trauma Scientist Development Program, NICHD K12HDO47349 and NICHD Loan Repayment Program L40 HD089260. Dr Derrington is a director at large for the American Society of Bioethics and Medical Humanities and had travel expenses reimbursed for the annual conference in 2019. Dr Michelson has received funding from the National Palliative Care Research Center and is a consultant on a National Institutes of Health study that are unrelated to this work. Dr Michelson is also involved in unrelated work supported by the National Alliance for Grieving Children. All other authors declared they have nothing to disclose.

“I have a rare dermatologic disorder. In medical school, I read a case report about treatment for my disorder. I was surprised to read my history and shocked to see my childhood face staring back at me in the figures section. The case report was written when I was a child and my parents had signed a consent form that stated my case and images could be used for ‘educational purposes.’ My parents were not notified that my images and case were published. While surprised and shocked to read my history and see images of myself in a medical journal, I trusted my privacy was protected because the journal would only be read by medical professionals. Fast-forward to today, I do not know how comfortable I would feel if my images were shared on social media, with the potential to reach viewers outside of the medical community. If I were a parent, I would feel even more uncomfortable with reading my child’s case on social media, let alone viewing an image of my child.”

—A.K.

Social media has become ingrained in our society, including many facets of our professional life. According to a 2019 report from the Pew Research Center, 73% of Americans use social media.¹ The PricewaterhouseCoopers Health Institute found 90% of physicians use social media personally, and 65% use it professionally.²

As the Pediatric Hospital Medicine Conference Social Media Cochairs (2015-2019), we managed official profiles on Twitter, Facebook, and Instagram. We also crafted and executed the conference’s social media strategy. During that time, we witnessed a substantial increase in the presence of physicians on social media with little available guidance on best practices. Here, we discuss patient privacy challenges with social media as well as solutions to address them.

In 2011, Greyson et al surveyed executive directors of all medical and osteopathic boards in the United States for online professionalism violations.³ Online violations of patient confidentiality were reported by over 55% of the 48 boards that responded. Of those, 10% reported more than three violations of patient confidentiality, and no actions were initially taken in 25% of violations. While these violations were not specific to social media, they highlight online patient confidentiality breaches are occurring, even if they are not being disciplined.

Several organizations, including the American Medical Association (AMA), the American Academy of Pediatrics (AAP), and the American College of Physicians (ACP) have developed social media guidelines.^4-6 However, these guidelines are not always followed. Fanti Silva and Colleoni studied surgeons and surgical trainees at a university hospital and found that social media guidelines were unknown to 100% of medical students, 85% of residents, and 78% of attendings.⁷ They also found that 53% of medical students, 86% of residents, and 32% of attendings were sharing patient information on social media despite hospitals’ privacy policies.

Social media provides forums for physicians to discuss cases and share experiences in hopes of educating others. These posts may include images or videos. Unfortunately, sharing specific clinical information or improperly deidentifying images may lead to the unintentional identification of patients.⁸ Some information may not be protected by the US Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule, and may lead to patient identification when shared.⁹ Despite disguising or omitting demographics, encounter information, or unique characteristics of the presentation, some physicians—not the posting physician—believe patients may still be able to identify their cases.⁸

Physicians who try to be mindful of patient privacy concerns face challenges with social media platforms themselves. For example, Facebook allows users to create Closed Groups (CGs) in which the group’s “administrators” can grant “admission” to users wishing to join the conversation (eg, Physician Moms Group). These groups are left to govern themselves and comply only with Facebook’s safety standards. The Society of Gastrointestinal and Endoscopic Surgeons used Facebook’s CGs to create a forum for education, consultation, and collaboration for society members. Group administrators grant admittance only after group members have agreed to HIPAA compliance. Group members may then share deidentified images and videos when discussing cases.¹⁰ However, Facebook’s Terms of Service states the company has “a non-exclusive, transferable, sub-licensable, royalty-free, worldwide license to host, use, distribute, modify, run, copy, publicly perform or display, translate, and create derivative works” of the content based on the privacy settings of the individual posting the content.¹¹ Therefore, these CGs may create a false sense of security because many members may assume the content of the CGs are private. Twitter’s Terms of Service are similar to Facebook’s, but state that users should have “obtained, all rights, licenses, consents, permissions, power and/or authority necessary to grant the rights . . . for any Content that is posted.”¹² If a patient’s deidentified story is posted on Twitter, the posting physician may be violating Twitter’s Terms of Service by not obtaining the patient’s consent/permission or explicitly stating so in their tweet.

In light of the challenges faced when posting medical cases on social media, we propose several solutions that the medical community should adopt to mitigate and limit any potential breaches to patient privacy. These are summarized in the Table.

Medical Education

Many medical students and residents are active on social media. However, not all are formally educated on appropriate engagement online and social media etiquette. A recent article from the Association of American Medical Colleges (AAMC) highlights how this “curriculum” is missing from many medical schools and residency programs.¹³ There are plenty of resources outlining how to maintain professionalism on social media in a general sense, but maintaining patient privacy usually is not concretely explored. Consequently, many programs are left to individually provide this education without firm guidance on best practices. We propose that governing organizations for medical education such as the AAMC and Accreditation Council for Graduate Medical Education have formal requirements, guidelines, and example curriculum on educating trainees on best practices for social media activity.

Health Organization Consent Forms

Healthcare organizations have a responsibility to protect patient privacy. We propose that healthcare organizations should develop independent social media consent forms that address sharing of images, videos, and cases. This separate social media consent form would allow patients/guardians to discuss whether they want their information shared. Some organizations have taken this step and developed consent forms for sharing deidentified posts on HIPAA-compliant CGs.¹⁰ However, it is still far from standard of practice for a healthcare organization to develop a separate consent form addressing the educational uses of sharing cases on social media. The Federation of State Medical Board’s (FSMB) Social Media and Electronic Communications policy endorses obtaining “express written consent” from patients.¹⁴ The policy states that “the physician must adequately explain the risks . . . for consent to be fully informed.” The FSMB policy also reminds readers that any social media post is permanent, even after it has been deleted.

Professional Organizations

Many professional organizations have acknowledged the growing role of social media in the professional lives of medical providers and have adopted policy statements and guidelines to address social media use. However, these guidelines are quite variable. All professional organizations should take the time to clarify and discuss the nuances of patient privacy on social media in their guidelines. For example, the American College of Obstetrics and Gynecology statement warns members that “any public communication about work-­related clinical events may violate . . . privacy” and posting of deidentified general events “may be traced, through public vital statistics data, to a specific patient or hospital” directly violating HIPAA.¹⁵ In comparison, the AAP and ACP’s social media guidelines and toolkits fall short when discussing how to maintain patient privacy specifically. Within these toolkits and guidelines, there is no explicit guidance or discussion about maintaining patient privacy with the use of case examples or best practices.^5,6 As physicians on social media, we should be aware of these variable policy statements and guidelines from our professional organizations. Even further, as active members of our professional organizations, we should call on them to update their guidelines to increase details regarding the nuances of patient privacy.

When a case is posted on social media, it should be the posting physician’s responsibility to clearly state in the initial post that consent was obtained. To simplify the process, we propose the use of the hashtag, #ConsentObtained, to easily identify that assurances were made to protect the patient. Moreover, we encourage our physician colleagues to remind others to explicitly state if consent was obtained if it is not mentioned. The AMA’s code of ethics states that if physicians read posts that they feel are unprofessional, then those physicians “have a responsibility to bring that content to the attention of the individual, so that he or she can remove it and/or take other appropriate actions.”⁴ Therefore, we encourage all readers of social media posts to ensure that posts include #ConsentObtained or otherwise clearly state that patient permission was obtained. If the hashtag or verbiage is not seen, then it is the reader’s responsibility to contact the posting physician. The AMA’s code of ethics also recommends physicians to “report the matter to appropriate authorities” if the individual posting “does not take appropriate actions.”⁴ While we realize that verification of consent being obtained may be virtually impossible online, we hope that, as physicians, we hold patient privacy to the highest regard and would never use this hashtag inappropriately. Lastly, it’s important to remember that removing/deleting a post may delete it from the platform, but that post and its contents are not deleted from the internet and may be accessed through another site.

Social media has allowed the healthcare community to develop a voice for individuals and communities; it has allowed for collaboration, open discussion, and education. However, it also asks us to reevaluate the professional ethics and rules we have abided for decades with regard to keeping patient health information safe. We must be proactive to develop solutions regarding patient privacy as our social media presence continues to grow.

The authors have no conflicts of interest to report.

“I have a rare dermatologic disorder. In medical school, I read a case report about treatment for my disorder. I was surprised to read my history and shocked to see my childhood face staring back at me in the figures section. The case report was written when I was a child and my parents had signed a consent form that stated my case and images could be used for ‘educational purposes.’ My parents were not notified that my images and case were published. While surprised and shocked to read my history and see images of myself in a medical journal, I trusted my privacy was protected because the journal would only be read by medical professionals. Fast-forward to today, I do not know how comfortable I would feel if my images were shared on social media, with the potential to reach viewers outside of the medical community. If I were a parent, I would feel even more uncomfortable with reading my child’s case on social media, let alone viewing an image of my child.”

—A.K.

Social media has become ingrained in our society, including many facets of our professional life. According to a 2019 report from the Pew Research Center, 73% of Americans use social media.¹ The PricewaterhouseCoopers Health Institute found 90% of physicians use social media personally, and 65% use it professionally.²

As the Pediatric Hospital Medicine Conference Social Media Cochairs (2015-2019), we managed official profiles on Twitter, Facebook, and Instagram. We also crafted and executed the conference’s social media strategy. During that time, we witnessed a substantial increase in the presence of physicians on social media with little available guidance on best practices. Here, we discuss patient privacy challenges with social media as well as solutions to address them.

In 2011, Greyson et al surveyed executive directors of all medical and osteopathic boards in the United States for online professionalism violations.³ Online violations of patient confidentiality were reported by over 55% of the 48 boards that responded. Of those, 10% reported more than three violations of patient confidentiality, and no actions were initially taken in 25% of violations. While these violations were not specific to social media, they highlight online patient confidentiality breaches are occurring, even if they are not being disciplined.

Several organizations, including the American Medical Association (AMA), the American Academy of Pediatrics (AAP), and the American College of Physicians (ACP) have developed social media guidelines.^4-6 However, these guidelines are not always followed. Fanti Silva and Colleoni studied surgeons and surgical trainees at a university hospital and found that social media guidelines were unknown to 100% of medical students, 85% of residents, and 78% of attendings.⁷ They also found that 53% of medical students, 86% of residents, and 32% of attendings were sharing patient information on social media despite hospitals’ privacy policies.

Social media provides forums for physicians to discuss cases and share experiences in hopes of educating others. These posts may include images or videos. Unfortunately, sharing specific clinical information or improperly deidentifying images may lead to the unintentional identification of patients.⁸ Some information may not be protected by the US Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule, and may lead to patient identification when shared.⁹ Despite disguising or omitting demographics, encounter information, or unique characteristics of the presentation, some physicians—not the posting physician—believe patients may still be able to identify their cases.⁸

Physicians who try to be mindful of patient privacy concerns face challenges with social media platforms themselves. For example, Facebook allows users to create Closed Groups (CGs) in which the group’s “administrators” can grant “admission” to users wishing to join the conversation (eg, Physician Moms Group). These groups are left to govern themselves and comply only with Facebook’s safety standards. The Society of Gastrointestinal and Endoscopic Surgeons used Facebook’s CGs to create a forum for education, consultation, and collaboration for society members. Group administrators grant admittance only after group members have agreed to HIPAA compliance. Group members may then share deidentified images and videos when discussing cases.¹⁰ However, Facebook’s Terms of Service states the company has “a non-exclusive, transferable, sub-licensable, royalty-free, worldwide license to host, use, distribute, modify, run, copy, publicly perform or display, translate, and create derivative works” of the content based on the privacy settings of the individual posting the content.¹¹ Therefore, these CGs may create a false sense of security because many members may assume the content of the CGs are private. Twitter’s Terms of Service are similar to Facebook’s, but state that users should have “obtained, all rights, licenses, consents, permissions, power and/or authority necessary to grant the rights . . . for any Content that is posted.”¹² If a patient’s deidentified story is posted on Twitter, the posting physician may be violating Twitter’s Terms of Service by not obtaining the patient’s consent/permission or explicitly stating so in their tweet.

In light of the challenges faced when posting medical cases on social media, we propose several solutions that the medical community should adopt to mitigate and limit any potential breaches to patient privacy. These are summarized in the Table.

Medical Education

Many medical students and residents are active on social media. However, not all are formally educated on appropriate engagement online and social media etiquette. A recent article from the Association of American Medical Colleges (AAMC) highlights how this “curriculum” is missing from many medical schools and residency programs.¹³ There are plenty of resources outlining how to maintain professionalism on social media in a general sense, but maintaining patient privacy usually is not concretely explored. Consequently, many programs are left to individually provide this education without firm guidance on best practices. We propose that governing organizations for medical education such as the AAMC and Accreditation Council for Graduate Medical Education have formal requirements, guidelines, and example curriculum on educating trainees on best practices for social media activity.

Health Organization Consent Forms

Healthcare organizations have a responsibility to protect patient privacy. We propose that healthcare organizations should develop independent social media consent forms that address sharing of images, videos, and cases. This separate social media consent form would allow patients/guardians to discuss whether they want their information shared. Some organizations have taken this step and developed consent forms for sharing deidentified posts on HIPAA-compliant CGs.¹⁰ However, it is still far from standard of practice for a healthcare organization to develop a separate consent form addressing the educational uses of sharing cases on social media. The Federation of State Medical Board’s (FSMB) Social Media and Electronic Communications policy endorses obtaining “express written consent” from patients.¹⁴ The policy states that “the physician must adequately explain the risks . . . for consent to be fully informed.” The FSMB policy also reminds readers that any social media post is permanent, even after it has been deleted.

Professional Organizations

Many professional organizations have acknowledged the growing role of social media in the professional lives of medical providers and have adopted policy statements and guidelines to address social media use. However, these guidelines are quite variable. All professional organizations should take the time to clarify and discuss the nuances of patient privacy on social media in their guidelines. For example, the American College of Obstetrics and Gynecology statement warns members that “any public communication about work-­related clinical events may violate . . . privacy” and posting of deidentified general events “may be traced, through public vital statistics data, to a specific patient or hospital” directly violating HIPAA.¹⁵ In comparison, the AAP and ACP’s social media guidelines and toolkits fall short when discussing how to maintain patient privacy specifically. Within these toolkits and guidelines, there is no explicit guidance or discussion about maintaining patient privacy with the use of case examples or best practices.^5,6 As physicians on social media, we should be aware of these variable policy statements and guidelines from our professional organizations. Even further, as active members of our professional organizations, we should call on them to update their guidelines to increase details regarding the nuances of patient privacy.

When a case is posted on social media, it should be the posting physician’s responsibility to clearly state in the initial post that consent was obtained. To simplify the process, we propose the use of the hashtag, #ConsentObtained, to easily identify that assurances were made to protect the patient. Moreover, we encourage our physician colleagues to remind others to explicitly state if consent was obtained if it is not mentioned. The AMA’s code of ethics states that if physicians read posts that they feel are unprofessional, then those physicians “have a responsibility to bring that content to the attention of the individual, so that he or she can remove it and/or take other appropriate actions.”⁴ Therefore, we encourage all readers of social media posts to ensure that posts include #ConsentObtained or otherwise clearly state that patient permission was obtained. If the hashtag or verbiage is not seen, then it is the reader’s responsibility to contact the posting physician. The AMA’s code of ethics also recommends physicians to “report the matter to appropriate authorities” if the individual posting “does not take appropriate actions.”⁴ While we realize that verification of consent being obtained may be virtually impossible online, we hope that, as physicians, we hold patient privacy to the highest regard and would never use this hashtag inappropriately. Lastly, it’s important to remember that removing/deleting a post may delete it from the platform, but that post and its contents are not deleted from the internet and may be accessed through another site.

Social media has allowed the healthcare community to develop a voice for individuals and communities; it has allowed for collaboration, open discussion, and education. However, it also asks us to reevaluate the professional ethics and rules we have abided for decades with regard to keeping patient health information safe. We must be proactive to develop solutions regarding patient privacy as our social media presence continues to grow.

The authors have no conflicts of interest to report.

“I have a rare dermatologic disorder. In medical school, I read a case report about treatment for my disorder. I was surprised to read my history and shocked to see my childhood face staring back at me in the figures section. The case report was written when I was a child and my parents had signed a consent form that stated my case and images could be used for ‘educational purposes.’ My parents were not notified that my images and case were published. While surprised and shocked to read my history and see images of myself in a medical journal, I trusted my privacy was protected because the journal would only be read by medical professionals. Fast-forward to today, I do not know how comfortable I would feel if my images were shared on social media, with the potential to reach viewers outside of the medical community. If I were a parent, I would feel even more uncomfortable with reading my child’s case on social media, let alone viewing an image of my child.”

—A.K.

Social media has become ingrained in our society, including many facets of our professional life. According to a 2019 report from the Pew Research Center, 73% of Americans use social media.¹ The PricewaterhouseCoopers Health Institute found 90% of physicians use social media personally, and 65% use it professionally.²

As the Pediatric Hospital Medicine Conference Social Media Cochairs (2015-2019), we managed official profiles on Twitter, Facebook, and Instagram. We also crafted and executed the conference’s social media strategy. During that time, we witnessed a substantial increase in the presence of physicians on social media with little available guidance on best practices. Here, we discuss patient privacy challenges with social media as well as solutions to address them.

In 2011, Greyson et al surveyed executive directors of all medical and osteopathic boards in the United States for online professionalism violations.³ Online violations of patient confidentiality were reported by over 55% of the 48 boards that responded. Of those, 10% reported more than three violations of patient confidentiality, and no actions were initially taken in 25% of violations. While these violations were not specific to social media, they highlight online patient confidentiality breaches are occurring, even if they are not being disciplined.

Several organizations, including the American Medical Association (AMA), the American Academy of Pediatrics (AAP), and the American College of Physicians (ACP) have developed social media guidelines.^4-6 However, these guidelines are not always followed. Fanti Silva and Colleoni studied surgeons and surgical trainees at a university hospital and found that social media guidelines were unknown to 100% of medical students, 85% of residents, and 78% of attendings.⁷ They also found that 53% of medical students, 86% of residents, and 32% of attendings were sharing patient information on social media despite hospitals’ privacy policies.

Social media provides forums for physicians to discuss cases and share experiences in hopes of educating others. These posts may include images or videos. Unfortunately, sharing specific clinical information or improperly deidentifying images may lead to the unintentional identification of patients.⁸ Some information may not be protected by the US Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule, and may lead to patient identification when shared.⁹ Despite disguising or omitting demographics, encounter information, or unique characteristics of the presentation, some physicians—not the posting physician—believe patients may still be able to identify their cases.⁸

Physicians who try to be mindful of patient privacy concerns face challenges with social media platforms themselves. For example, Facebook allows users to create Closed Groups (CGs) in which the group’s “administrators” can grant “admission” to users wishing to join the conversation (eg, Physician Moms Group). These groups are left to govern themselves and comply only with Facebook’s safety standards. The Society of Gastrointestinal and Endoscopic Surgeons used Facebook’s CGs to create a forum for education, consultation, and collaboration for society members. Group administrators grant admittance only after group members have agreed to HIPAA compliance. Group members may then share deidentified images and videos when discussing cases.¹⁰ However, Facebook’s Terms of Service states the company has “a non-exclusive, transferable, sub-licensable, royalty-free, worldwide license to host, use, distribute, modify, run, copy, publicly perform or display, translate, and create derivative works” of the content based on the privacy settings of the individual posting the content.¹¹ Therefore, these CGs may create a false sense of security because many members may assume the content of the CGs are private. Twitter’s Terms of Service are similar to Facebook’s, but state that users should have “obtained, all rights, licenses, consents, permissions, power and/or authority necessary to grant the rights . . . for any Content that is posted.”¹² If a patient’s deidentified story is posted on Twitter, the posting physician may be violating Twitter’s Terms of Service by not obtaining the patient’s consent/permission or explicitly stating so in their tweet.

In light of the challenges faced when posting medical cases on social media, we propose several solutions that the medical community should adopt to mitigate and limit any potential breaches to patient privacy. These are summarized in the Table.

Medical Education

Many medical students and residents are active on social media. However, not all are formally educated on appropriate engagement online and social media etiquette. A recent article from the Association of American Medical Colleges (AAMC) highlights how this “curriculum” is missing from many medical schools and residency programs.¹³ There are plenty of resources outlining how to maintain professionalism on social media in a general sense, but maintaining patient privacy usually is not concretely explored. Consequently, many programs are left to individually provide this education without firm guidance on best practices. We propose that governing organizations for medical education such as the AAMC and Accreditation Council for Graduate Medical Education have formal requirements, guidelines, and example curriculum on educating trainees on best practices for social media activity.

Health Organization Consent Forms

Healthcare organizations have a responsibility to protect patient privacy. We propose that healthcare organizations should develop independent social media consent forms that address sharing of images, videos, and cases. This separate social media consent form would allow patients/guardians to discuss whether they want their information shared. Some organizations have taken this step and developed consent forms for sharing deidentified posts on HIPAA-compliant CGs.¹⁰ However, it is still far from standard of practice for a healthcare organization to develop a separate consent form addressing the educational uses of sharing cases on social media. The Federation of State Medical Board’s (FSMB) Social Media and Electronic Communications policy endorses obtaining “express written consent” from patients.¹⁴ The policy states that “the physician must adequately explain the risks . . . for consent to be fully informed.” The FSMB policy also reminds readers that any social media post is permanent, even after it has been deleted.

Professional Organizations

Many professional organizations have acknowledged the growing role of social media in the professional lives of medical providers and have adopted policy statements and guidelines to address social media use. However, these guidelines are quite variable. All professional organizations should take the time to clarify and discuss the nuances of patient privacy on social media in their guidelines. For example, the American College of Obstetrics and Gynecology statement warns members that “any public communication about work-­related clinical events may violate . . . privacy” and posting of deidentified general events “may be traced, through public vital statistics data, to a specific patient or hospital” directly violating HIPAA.¹⁵ In comparison, the AAP and ACP’s social media guidelines and toolkits fall short when discussing how to maintain patient privacy specifically. Within these toolkits and guidelines, there is no explicit guidance or discussion about maintaining patient privacy with the use of case examples or best practices.^5,6 As physicians on social media, we should be aware of these variable policy statements and guidelines from our professional organizations. Even further, as active members of our professional organizations, we should call on them to update their guidelines to increase details regarding the nuances of patient privacy.

When a case is posted on social media, it should be the posting physician’s responsibility to clearly state in the initial post that consent was obtained. To simplify the process, we propose the use of the hashtag, #ConsentObtained, to easily identify that assurances were made to protect the patient. Moreover, we encourage our physician colleagues to remind others to explicitly state if consent was obtained if it is not mentioned. The AMA’s code of ethics states that if physicians read posts that they feel are unprofessional, then those physicians “have a responsibility to bring that content to the attention of the individual, so that he or she can remove it and/or take other appropriate actions.”⁴ Therefore, we encourage all readers of social media posts to ensure that posts include #ConsentObtained or otherwise clearly state that patient permission was obtained. If the hashtag or verbiage is not seen, then it is the reader’s responsibility to contact the posting physician. The AMA’s code of ethics also recommends physicians to “report the matter to appropriate authorities” if the individual posting “does not take appropriate actions.”⁴ While we realize that verification of consent being obtained may be virtually impossible online, we hope that, as physicians, we hold patient privacy to the highest regard and would never use this hashtag inappropriately. Lastly, it’s important to remember that removing/deleting a post may delete it from the platform, but that post and its contents are not deleted from the internet and may be accessed through another site.

Social media has allowed the healthcare community to develop a voice for individuals and communities; it has allowed for collaboration, open discussion, and education. However, it also asks us to reevaluate the professional ethics and rules we have abided for decades with regard to keeping patient health information safe. We must be proactive to develop solutions regarding patient privacy as our social media presence continues to grow.

The authors have no conflicts of interest to report.