High-dose vitamin B6 supplements may reduce feelings of anxiety and depression, new research suggests.

Investigators compared supplementation with a 1-month course of vitamin B6 or B12 to supplementation with placebo in almost 500 adults. Results showed that vitamin B6 supplementation was associated with reductions in self-reported anxiety and a trend toward decreased depressive symptoms.

In addition, the vitamin B6 group showed increased levels of gamma-aminobutyric acid (GABA), as indicated by results on a visual test that was administered at the end of the trial. The test results demonstrated subtle changes in participants’ visual performance. The researchers considered this to be consistent with controlled levels of GABA-related brain activity.

However, “before practicing clinicians would recommend taking high doses of vitamin B6, a full-scale clinical trial would have to be carried out to verify the findings, assess any side effects, and find out which types of patients do or don’t benefit,” study investigator David Field, PhD, associate professor, School of Psychological and Clinical Language Sciences, University of Reading (England), told this news organization.

“My relatively small study can only be considered as an initial proof of concept,” Dr. Field said.

The findings were published online in the Journal of Human Psychopharmacology: Clinical and Experimental.
 

Eat Marmite?

“Recent research has connected mood disorders and some other neuropsychiatric conditions with disturbance in this balance, often in the direction of raised levels of brain activity,” Dr. Field noted.

Vitamin B6 is a coenzyme in the synthesis of GABA, an inhibitory neurotransmitter, from glutamate. Some previous research has suggested that vitamins B6 and B12 have a role in improving mood-related outcomes.

Dr. Field had reviewed a 2017 study of the effects on visual processing of eating Marmite, a type of food spread rich in vitamin B, every day for a few weeks.

“Remarkably, the results of that study suggested that eating Marmite had increased the level of the inhibitory neurotransmitter GABA in the visual part of the brain, damping down the level of neural activity slightly,” he said.

However, Marmite contains other B vitamins and other ingredients that might potentially account for this result, “plus, a lot of people don’t like the taste of Marmite,” Dr. Field noted.

Therefore, he wanted to “find out which individual ingredients were driving the effect, and B6 and B12 were the most plausible candidates.”

He decided to test these vitamins individually and to compare them to placebo. “I added the measures of anxiety and depression that were not in the Marmite study because I reasoned that if GABA levels were altered, this could improve those disorders, because we know that decreased levels of GABA in the brain occur in both of those conditions,” Dr. Field added.

Over the course of 5 years, investigators recruited 478 participants aged 18-58 years (mean age, 23 years; 381 women). Of these, 265 reported having anxiety, and 146 reported having depression.

The study participants were randomly assigned to receive either vitamin B6 (100 mg pyroxidine hydrochloride), vitamin B12 (1,000 mg methylcobalmin), or placebo tablets once daily for a month.

They also completed the Screen for Adult Anxiety Related Disorders (SCAARED) and the Mood and Feelings Questionnaire (MFQ) long version at baseline and following supplementation (“post test”), and they underwent three sensory tests that acted as assays of inhibitory function at post test.

In addition, 307 participants completed the Visual Contrast Sensitivity and Surround Suppression, which “measures the minimum percentage contrast between the lighter and darker regions of a striped pattern that can be detected (called the contrast threshold),” the investigators note.

The contrast threshold was measured with and without a suppressive surround mask that increases the threshold – an effect mediated by GABAergic connections in the visual cortex.

Participants (n = 172) also completed the Binocular Rivalry test and the Tactile Test Battery (n = 180). Both tests are designed to measure responses requiring GABAergic inhibitory activity.
 

‘Subtle changes’

ANOVA analyses revealed a “highly significant” reduction in anxiety at post test (F[1,173] = 10.03; P = .002; np 2 = .055), driven primarily by reduced anxiety in the B6 group (t[88] = 3.51; P < .001; d = .37). The placebo group also showed some reduction in anxiety, but it was not deemed significant, and the overall interaction itself did not reach significance.

A comparison of the B12 group with the group that received placebo revealed a significant reduction in anxiety at post test (F[1,175] = 4.08; P = .045; np 2 = .023), similarly driven by reduced anxiety in the B12 group (t[89] = 1.84; P = .069; d = .19) – but the interaction was not significant.

Among the B6 group, there was a highly significant reduction in scores on the generalized anxiety disorder and social anxiety subscales of the SCAARED, and there was a trend toward reductions on the other subscales. Among the B12 group, there was a significant reduction only on scores on the separation anxiety subscale. No significant changes were found in the placebo group.

The ANOVA test analysis of the B6 and placebo group data showed “no uniform direction of change” in depression at post test. The researchers found a “tendency” for depression scores to decrease between baseline and post test in the B6 group but to increase in the placebo group – an interaction that “approached” significance (F[1,96] = 3.08; P = .083; np 2 = .031), they report.

The ANOVA analysis of the B12 and placebo group data revealed no significant or trending effects, and the t-test comparing baseline and post-test scores in the B12 group was similarly nonsignificant.

B6 supplementation did change visual contrast thresholds, but only when a suppressive surround was present. There were “no clear effects” of B6 supplementation on other outcome measures, including binocular rivalry reversal rate and the tactile test battery, the investigators note.

“We found that supplementation with B6 produced subtle changes in tests of visual processing in a way that suggested an increase in the level of the inhibitory neurotransmitter GABA,” Dr. Field reported.

Vitamin B6 is a “cofactor for a metabolic pathway in the brain that converts the excitatory neurotransmitter glutamate into the inhibitory/calming GABA,” he said.

“By increasing the quantity of the cofactor, we slightly speed up the rate of this metabolic process, and so you end up with a bit more of the GABA neurotransmitter and a bit less glutamate. The net effect of this is to slightly reduce the amount of activity in the brain,” Dr. Field added.
 

Most common nutrient deficiency

Carol Johnston, PhD, professor and associate dean for faculty success, College of Health Solutions, Arizona State University, Phoenix, said vitamin B6 is “the most common nutrient deficiency in the United States;” 16% of men and 32% of women are reportedly B6 deficient.

“Young women on birth control are at higher risk for B6 deficiency due to effects of oral contraceptives on B6 metabolism,” whereas vitamin B12 deficiency is more common in older adults, said Dr. Johnston, who was not involved with the study.

The current study’s population mainly consisted of young women, and the interpretation of the data is “limited” because the researchers did not measure blood status for B6 and B12, Dr. Johnston noted. It is possible the sample was low in B6 and that the supplements “improved cognitive measures.”

Because the population was young – no one was older than 60 years – B12 status was likely “adequate in the sample, and supplementation did not have an impact,” she said.

Overall, Dr. Johnston cautioned that it is important to “alert clinicians and the general public about the concerns of overdosing B6.” For example, supplementation at high amounts can cause potentially irreversible sensory neuropathy, she noted.

“The safe upper limit defined by experts is 100 mg per day – the dosage used in this trial. Daily supplementation should not exceed this level,” Dr. Johnston said.

The vitamin tablets used in the study were supplied by Innopure. The investigators and Dr. Johnston have reported no relevant financial relationships.

A version of this article first appeared on Medscape.com.

High-dose vitamin B6 supplements may reduce feelings of anxiety and depression, new research suggests.

Investigators compared supplementation with a 1-month course of vitamin B6 or B12 to supplementation with placebo in almost 500 adults. Results showed that vitamin B6 supplementation was associated with reductions in self-reported anxiety and a trend toward decreased depressive symptoms.

In addition, the vitamin B6 group showed increased levels of gamma-aminobutyric acid (GABA), as indicated by results on a visual test that was administered at the end of the trial. The test results demonstrated subtle changes in participants’ visual performance. The researchers considered this to be consistent with controlled levels of GABA-related brain activity.

However, “before practicing clinicians would recommend taking high doses of vitamin B6, a full-scale clinical trial would have to be carried out to verify the findings, assess any side effects, and find out which types of patients do or don’t benefit,” study investigator David Field, PhD, associate professor, School of Psychological and Clinical Language Sciences, University of Reading (England), told this news organization.

“My relatively small study can only be considered as an initial proof of concept,” Dr. Field said.

The findings were published online in the Journal of Human Psychopharmacology: Clinical and Experimental.
 

Eat Marmite?

“Recent research has connected mood disorders and some other neuropsychiatric conditions with disturbance in this balance, often in the direction of raised levels of brain activity,” Dr. Field noted.

Vitamin B6 is a coenzyme in the synthesis of GABA, an inhibitory neurotransmitter, from glutamate. Some previous research has suggested that vitamins B6 and B12 have a role in improving mood-related outcomes.

Dr. Field had reviewed a 2017 study of the effects on visual processing of eating Marmite, a type of food spread rich in vitamin B, every day for a few weeks.

“Remarkably, the results of that study suggested that eating Marmite had increased the level of the inhibitory neurotransmitter GABA in the visual part of the brain, damping down the level of neural activity slightly,” he said.

However, Marmite contains other B vitamins and other ingredients that might potentially account for this result, “plus, a lot of people don’t like the taste of Marmite,” Dr. Field noted.

Therefore, he wanted to “find out which individual ingredients were driving the effect, and B6 and B12 were the most plausible candidates.”

He decided to test these vitamins individually and to compare them to placebo. “I added the measures of anxiety and depression that were not in the Marmite study because I reasoned that if GABA levels were altered, this could improve those disorders, because we know that decreased levels of GABA in the brain occur in both of those conditions,” Dr. Field added.

Over the course of 5 years, investigators recruited 478 participants aged 18-58 years (mean age, 23 years; 381 women). Of these, 265 reported having anxiety, and 146 reported having depression.

The study participants were randomly assigned to receive either vitamin B6 (100 mg pyroxidine hydrochloride), vitamin B12 (1,000 mg methylcobalmin), or placebo tablets once daily for a month.

They also completed the Screen for Adult Anxiety Related Disorders (SCAARED) and the Mood and Feelings Questionnaire (MFQ) long version at baseline and following supplementation (“post test”), and they underwent three sensory tests that acted as assays of inhibitory function at post test.

In addition, 307 participants completed the Visual Contrast Sensitivity and Surround Suppression, which “measures the minimum percentage contrast between the lighter and darker regions of a striped pattern that can be detected (called the contrast threshold),” the investigators note.

The contrast threshold was measured with and without a suppressive surround mask that increases the threshold – an effect mediated by GABAergic connections in the visual cortex.

Participants (n = 172) also completed the Binocular Rivalry test and the Tactile Test Battery (n = 180). Both tests are designed to measure responses requiring GABAergic inhibitory activity.
 

‘Subtle changes’

ANOVA analyses revealed a “highly significant” reduction in anxiety at post test (F[1,173] = 10.03; P = .002; np 2 = .055), driven primarily by reduced anxiety in the B6 group (t[88] = 3.51; P < .001; d = .37). The placebo group also showed some reduction in anxiety, but it was not deemed significant, and the overall interaction itself did not reach significance.

A comparison of the B12 group with the group that received placebo revealed a significant reduction in anxiety at post test (F[1,175] = 4.08; P = .045; np 2 = .023), similarly driven by reduced anxiety in the B12 group (t[89] = 1.84; P = .069; d = .19) – but the interaction was not significant.

Among the B6 group, there was a highly significant reduction in scores on the generalized anxiety disorder and social anxiety subscales of the SCAARED, and there was a trend toward reductions on the other subscales. Among the B12 group, there was a significant reduction only on scores on the separation anxiety subscale. No significant changes were found in the placebo group.

The ANOVA test analysis of the B6 and placebo group data showed “no uniform direction of change” in depression at post test. The researchers found a “tendency” for depression scores to decrease between baseline and post test in the B6 group but to increase in the placebo group – an interaction that “approached” significance (F[1,96] = 3.08; P = .083; np 2 = .031), they report.

The ANOVA analysis of the B12 and placebo group data revealed no significant or trending effects, and the t-test comparing baseline and post-test scores in the B12 group was similarly nonsignificant.

B6 supplementation did change visual contrast thresholds, but only when a suppressive surround was present. There were “no clear effects” of B6 supplementation on other outcome measures, including binocular rivalry reversal rate and the tactile test battery, the investigators note.

“We found that supplementation with B6 produced subtle changes in tests of visual processing in a way that suggested an increase in the level of the inhibitory neurotransmitter GABA,” Dr. Field reported.

Vitamin B6 is a “cofactor for a metabolic pathway in the brain that converts the excitatory neurotransmitter glutamate into the inhibitory/calming GABA,” he said.

“By increasing the quantity of the cofactor, we slightly speed up the rate of this metabolic process, and so you end up with a bit more of the GABA neurotransmitter and a bit less glutamate. The net effect of this is to slightly reduce the amount of activity in the brain,” Dr. Field added.
 

Most common nutrient deficiency

Carol Johnston, PhD, professor and associate dean for faculty success, College of Health Solutions, Arizona State University, Phoenix, said vitamin B6 is “the most common nutrient deficiency in the United States;” 16% of men and 32% of women are reportedly B6 deficient.

“Young women on birth control are at higher risk for B6 deficiency due to effects of oral contraceptives on B6 metabolism,” whereas vitamin B12 deficiency is more common in older adults, said Dr. Johnston, who was not involved with the study.

The current study’s population mainly consisted of young women, and the interpretation of the data is “limited” because the researchers did not measure blood status for B6 and B12, Dr. Johnston noted. It is possible the sample was low in B6 and that the supplements “improved cognitive measures.”

Because the population was young – no one was older than 60 years – B12 status was likely “adequate in the sample, and supplementation did not have an impact,” she said.

Overall, Dr. Johnston cautioned that it is important to “alert clinicians and the general public about the concerns of overdosing B6.” For example, supplementation at high amounts can cause potentially irreversible sensory neuropathy, she noted.

“The safe upper limit defined by experts is 100 mg per day – the dosage used in this trial. Daily supplementation should not exceed this level,” Dr. Johnston said.

The vitamin tablets used in the study were supplied by Innopure. The investigators and Dr. Johnston have reported no relevant financial relationships.

A version of this article first appeared on Medscape.com.

High-dose vitamin B6 supplements may reduce feelings of anxiety and depression, new research suggests.

Investigators compared supplementation with a 1-month course of vitamin B6 or B12 to supplementation with placebo in almost 500 adults. Results showed that vitamin B6 supplementation was associated with reductions in self-reported anxiety and a trend toward decreased depressive symptoms.

In addition, the vitamin B6 group showed increased levels of gamma-aminobutyric acid (GABA), as indicated by results on a visual test that was administered at the end of the trial. The test results demonstrated subtle changes in participants’ visual performance. The researchers considered this to be consistent with controlled levels of GABA-related brain activity.

However, “before practicing clinicians would recommend taking high doses of vitamin B6, a full-scale clinical trial would have to be carried out to verify the findings, assess any side effects, and find out which types of patients do or don’t benefit,” study investigator David Field, PhD, associate professor, School of Psychological and Clinical Language Sciences, University of Reading (England), told this news organization.

“My relatively small study can only be considered as an initial proof of concept,” Dr. Field said.

The findings were published online in the Journal of Human Psychopharmacology: Clinical and Experimental.
 

Eat Marmite?

“Recent research has connected mood disorders and some other neuropsychiatric conditions with disturbance in this balance, often in the direction of raised levels of brain activity,” Dr. Field noted.

Vitamin B6 is a coenzyme in the synthesis of GABA, an inhibitory neurotransmitter, from glutamate. Some previous research has suggested that vitamins B6 and B12 have a role in improving mood-related outcomes.

Dr. Field had reviewed a 2017 study of the effects on visual processing of eating Marmite, a type of food spread rich in vitamin B, every day for a few weeks.

“Remarkably, the results of that study suggested that eating Marmite had increased the level of the inhibitory neurotransmitter GABA in the visual part of the brain, damping down the level of neural activity slightly,” he said.

However, Marmite contains other B vitamins and other ingredients that might potentially account for this result, “plus, a lot of people don’t like the taste of Marmite,” Dr. Field noted.

Therefore, he wanted to “find out which individual ingredients were driving the effect, and B6 and B12 were the most plausible candidates.”

He decided to test these vitamins individually and to compare them to placebo. “I added the measures of anxiety and depression that were not in the Marmite study because I reasoned that if GABA levels were altered, this could improve those disorders, because we know that decreased levels of GABA in the brain occur in both of those conditions,” Dr. Field added.

Over the course of 5 years, investigators recruited 478 participants aged 18-58 years (mean age, 23 years; 381 women). Of these, 265 reported having anxiety, and 146 reported having depression.

The study participants were randomly assigned to receive either vitamin B6 (100 mg pyroxidine hydrochloride), vitamin B12 (1,000 mg methylcobalmin), or placebo tablets once daily for a month.

They also completed the Screen for Adult Anxiety Related Disorders (SCAARED) and the Mood and Feelings Questionnaire (MFQ) long version at baseline and following supplementation (“post test”), and they underwent three sensory tests that acted as assays of inhibitory function at post test.

In addition, 307 participants completed the Visual Contrast Sensitivity and Surround Suppression, which “measures the minimum percentage contrast between the lighter and darker regions of a striped pattern that can be detected (called the contrast threshold),” the investigators note.

The contrast threshold was measured with and without a suppressive surround mask that increases the threshold – an effect mediated by GABAergic connections in the visual cortex.

Participants (n = 172) also completed the Binocular Rivalry test and the Tactile Test Battery (n = 180). Both tests are designed to measure responses requiring GABAergic inhibitory activity.
 

‘Subtle changes’

ANOVA analyses revealed a “highly significant” reduction in anxiety at post test (F[1,173] = 10.03; P = .002; np 2 = .055), driven primarily by reduced anxiety in the B6 group (t[88] = 3.51; P < .001; d = .37). The placebo group also showed some reduction in anxiety, but it was not deemed significant, and the overall interaction itself did not reach significance.

A comparison of the B12 group with the group that received placebo revealed a significant reduction in anxiety at post test (F[1,175] = 4.08; P = .045; np 2 = .023), similarly driven by reduced anxiety in the B12 group (t[89] = 1.84; P = .069; d = .19) – but the interaction was not significant.

Among the B6 group, there was a highly significant reduction in scores on the generalized anxiety disorder and social anxiety subscales of the SCAARED, and there was a trend toward reductions on the other subscales. Among the B12 group, there was a significant reduction only on scores on the separation anxiety subscale. No significant changes were found in the placebo group.

The ANOVA test analysis of the B6 and placebo group data showed “no uniform direction of change” in depression at post test. The researchers found a “tendency” for depression scores to decrease between baseline and post test in the B6 group but to increase in the placebo group – an interaction that “approached” significance (F[1,96] = 3.08; P = .083; np 2 = .031), they report.

The ANOVA analysis of the B12 and placebo group data revealed no significant or trending effects, and the t-test comparing baseline and post-test scores in the B12 group was similarly nonsignificant.

B6 supplementation did change visual contrast thresholds, but only when a suppressive surround was present. There were “no clear effects” of B6 supplementation on other outcome measures, including binocular rivalry reversal rate and the tactile test battery, the investigators note.

“We found that supplementation with B6 produced subtle changes in tests of visual processing in a way that suggested an increase in the level of the inhibitory neurotransmitter GABA,” Dr. Field reported.

Vitamin B6 is a “cofactor for a metabolic pathway in the brain that converts the excitatory neurotransmitter glutamate into the inhibitory/calming GABA,” he said.

“By increasing the quantity of the cofactor, we slightly speed up the rate of this metabolic process, and so you end up with a bit more of the GABA neurotransmitter and a bit less glutamate. The net effect of this is to slightly reduce the amount of activity in the brain,” Dr. Field added.
 

Most common nutrient deficiency

Carol Johnston, PhD, professor and associate dean for faculty success, College of Health Solutions, Arizona State University, Phoenix, said vitamin B6 is “the most common nutrient deficiency in the United States;” 16% of men and 32% of women are reportedly B6 deficient.

“Young women on birth control are at higher risk for B6 deficiency due to effects of oral contraceptives on B6 metabolism,” whereas vitamin B12 deficiency is more common in older adults, said Dr. Johnston, who was not involved with the study.

The current study’s population mainly consisted of young women, and the interpretation of the data is “limited” because the researchers did not measure blood status for B6 and B12, Dr. Johnston noted. It is possible the sample was low in B6 and that the supplements “improved cognitive measures.”

Because the population was young – no one was older than 60 years – B12 status was likely “adequate in the sample, and supplementation did not have an impact,” she said.

Overall, Dr. Johnston cautioned that it is important to “alert clinicians and the general public about the concerns of overdosing B6.” For example, supplementation at high amounts can cause potentially irreversible sensory neuropathy, she noted.

“The safe upper limit defined by experts is 100 mg per day – the dosage used in this trial. Daily supplementation should not exceed this level,” Dr. Johnston said.

The vitamin tablets used in the study were supplied by Innopure. The investigators and Dr. Johnston have reported no relevant financial relationships.

A version of this article first appeared on Medscape.com.

Pharmacogenetic testing, which is used to classify how patients with major depressive disorder (MDD) metabolize medications, reduces adverse drug-gene interactions, new research shows.

In a randomized clinical trial that included almost 2,000 adults with MDD, patients in the pharmacogenomics-guided group were more likely to receive an antidepressant that had no potential drug-gene interaction than the patients who received usual care.

In addition, among the intervention group, the rate of remission over 24 weeks was significantly greater.

“These tests can be helpful in rethinking choices of antidepressants, but clinicians should not expect them to be helpful for every patient,” study investigator David W. Oslin, MD, Corporal Michael J. Crescenz VA Medical Center and professor of psychiatry at Perelman School of Medicine, University of Pennsylvania, Philadelphia, said in an interview.

The findings were published online  in JAMA.

Less trial and error

Pharmacogenomic testing can provide information to inform drug selection or dosing for patients with a genetic variation that alters pharmacokinetics or pharmacodynamics. Such testing may be particularly useful for patients with MDD, as fewer than 40% of these patients achieve clinical remission after an initial treatment with an antidepressant, the investigators note.

“To get to a treatment that works for an individual, it’s not unusual to have to try two or three or four antidepressants,” said Dr. Oslin. “If we could reduce that variance a little bit with a test like this, that would be huge from a public health perspective.”

The study included 676 physicians and 1,944 adults with MDD (mean age, 48 years; 24% women) who were receiving care at 22 Department of Veterans Affairs medical centers. Eligible patients were set to start a new antidepressant monotherapy, and all underwent a pharmacogenomic test using a cheek swab.

Investigators randomly assigned patients to receive test results when available (pharmacogenomic-guided group) or 24 weeks later (usual-care group). For the former group, clinicians were asked to initiate treatment when test results were available, typically within 2-3 days. For the latter group, they were asked to initiate treatment on a day of randomization.

Assessments included the 9-item Patient Health questionnaire (PHQ-9), scores for which range from 0-27 points, with higher scores indicating worse symptoms.

Of the total patient population, 79% completed the 24-week assessment.

Researchers characterized antidepressant medications on the basis of drug-gene interaction categories: no known interactions, moderate interactions, and substantial interactions.

The co-primary outcomes were treatment initiation within 30 days, determined on the basis of drug-gene interaction categories, and remission from depression symptoms, defined as a PHQ-9 score of less than or equal to 5.

Raters who were blinded to clinical care and study randomization assessed outcomes at 4, 8, 12, 18, and 24 weeks.
 

Significant impact?

Results showed that the pharmacogenomic-guided group was more likely to receive an antidepressant that had no potential drug-gene interaction, as opposed to one with a moderate/substantial interaction (odds ratio, 4.32; 95% confidence interval, 3.47-5.39; P < .001).

The usual-care group was more likely to receive a drug with mild potential drug-gene interaction (no/moderate interaction vs. substantial interaction: OR, 2.08; 95% CI, 1.52-2.84; P = .005).

For the intervention group, the estimated rates of receiving an antidepressant with no, moderate, and substantial drug-gene interactions were 59.3%, 30.0%, and 10.7%, respectively. For the usual-care group, the estimates were 25.7%, 54.6%, and 19.7%.

The finding that 1 in 5 patients who received usual care were initially given a medication for which there were significant drug-gene interactions means it is “not a rare event,” said Dr. Oslin. “If we can make an impact on 20% of the people we prescribe to, that’s actually pretty big.”

Rates of remission were greater in the pharmacogenomic-guided group over 24 weeks (OR, 1.28; 95% CI, 1.05-1.57; P = .02; absolute risk difference, 2.8%; 95% CI, 0.6%-5.1%).

The secondary outcomes of response to treatment, defined as at least a 50% decrease in PHQ-9 score, also favored the pharmacogenomic-guided group. This was also the case for the secondary outcome of reduction in symptom severity on the PHQ-9 score.

Some physicians have expressed skepticism about pharmacogenomic testing, but the study provides additional evidence of its usefulness, Dr. Oslin noted.

“While I don’t think testing should be standard of practice, I also don’t think we should put barriers into the testing until we can better understand how to target the testing” to those who will benefit the most, he added.

The tests are available at a commercial cost of about $1,000 – which may not be that expensive if testing has a significant impact on a patient’s life, said Dr. Oslin.
 

Important research, but with several limitations

In an accompanying editorial, Dan V. Iosifescu, MD, associate professor of psychiatry at New York University School of Medicine and director of clinical research at the Nathan Kline Institute for Psychiatric Research, called the study an important addition to the literature on pharmacogenomic testing for patients with MDD.

The study was significantly larger and had broader inclusion criteria and longer follow-up than previous clinical trials and is one of the few investigations not funded by a manufacturer of pharmacogenomic tests, writes Dr. Iosifescu, who was not involved with the research.

However, he notes that an antidepressant was not initiated for 30 days after randomization in 25% of the intervention group and in 31% of the usual-care group, which was “puzzling.” “Because these rates were comparable in the 2 groups, it cannot be explained primarily by the delay of the pharmacogenomic test results in the intervention group,” he writes.

In addition, in the co-primary outcome of symptom remission rate, the difference in clinical improvement in favor of the pharmacogenomic-guided treatment was only “modest” – the gain was of less than 2% in the proportion of patients achieving remission, Dr. Iosifescu adds.

He adds this is “likely not very meaningful clinically despite this difference achieving statistical significance in this large study sample.”

Other potential study limitations he cites include the lack of patient blinding to treatment assignment and the absence of clarity about why rates of MDD response and remission over time were relatively low in both treatment groups.

A possible approach to optimize antidepressant choices could involve integration of pharmacogenomic data into larger predictive models that include clinical and demographic variables, Dr. Iosifescu notes.

“The development of such complex models is challenging, but it is now possible given the recent substantial advances in the proficiency of computational tools,” he writes.

The study was funded by the U.S. Department of Veterans Affairs (VA), Health Services Research and Development Service, and the Mental Illness Research, Education, and Clinical Center at the Corporal Michael J. Crescenz VA Medical Center. Dr. Oslin reports having received grants from the VA Office of Research and Development and Janssen Pharmaceuticals and nonfinancial support from Myriad Genetics during the conduct of the study. Dr. Iosifescu report having received personal fees from Alkermes, Allergan, Axsome, Biogen, the Centers for Psychiatric Excellence, Jazz, Lundbeck, Precision Neuroscience, Sage, and Sunovion and grants from Alkermes, AstraZeneca, Brainsway, Litecure, Neosync, Otsuka, Roche, and Shire.

A version of this article first appeared on Medscape.com.

Pharmacogenetic testing, which is used to classify how patients with major depressive disorder (MDD) metabolize medications, reduces adverse drug-gene interactions, new research shows.

In a randomized clinical trial that included almost 2,000 adults with MDD, patients in the pharmacogenomics-guided group were more likely to receive an antidepressant that had no potential drug-gene interaction than the patients who received usual care.

In addition, among the intervention group, the rate of remission over 24 weeks was significantly greater.

“These tests can be helpful in rethinking choices of antidepressants, but clinicians should not expect them to be helpful for every patient,” study investigator David W. Oslin, MD, Corporal Michael J. Crescenz VA Medical Center and professor of psychiatry at Perelman School of Medicine, University of Pennsylvania, Philadelphia, said in an interview.

The findings were published online  in JAMA.

Less trial and error

Pharmacogenomic testing can provide information to inform drug selection or dosing for patients with a genetic variation that alters pharmacokinetics or pharmacodynamics. Such testing may be particularly useful for patients with MDD, as fewer than 40% of these patients achieve clinical remission after an initial treatment with an antidepressant, the investigators note.

“To get to a treatment that works for an individual, it’s not unusual to have to try two or three or four antidepressants,” said Dr. Oslin. “If we could reduce that variance a little bit with a test like this, that would be huge from a public health perspective.”

The study included 676 physicians and 1,944 adults with MDD (mean age, 48 years; 24% women) who were receiving care at 22 Department of Veterans Affairs medical centers. Eligible patients were set to start a new antidepressant monotherapy, and all underwent a pharmacogenomic test using a cheek swab.

Investigators randomly assigned patients to receive test results when available (pharmacogenomic-guided group) or 24 weeks later (usual-care group). For the former group, clinicians were asked to initiate treatment when test results were available, typically within 2-3 days. For the latter group, they were asked to initiate treatment on a day of randomization.

Assessments included the 9-item Patient Health questionnaire (PHQ-9), scores for which range from 0-27 points, with higher scores indicating worse symptoms.

Of the total patient population, 79% completed the 24-week assessment.

Researchers characterized antidepressant medications on the basis of drug-gene interaction categories: no known interactions, moderate interactions, and substantial interactions.

The co-primary outcomes were treatment initiation within 30 days, determined on the basis of drug-gene interaction categories, and remission from depression symptoms, defined as a PHQ-9 score of less than or equal to 5.

Raters who were blinded to clinical care and study randomization assessed outcomes at 4, 8, 12, 18, and 24 weeks.
 

Significant impact?

Results showed that the pharmacogenomic-guided group was more likely to receive an antidepressant that had no potential drug-gene interaction, as opposed to one with a moderate/substantial interaction (odds ratio, 4.32; 95% confidence interval, 3.47-5.39; P < .001).

The usual-care group was more likely to receive a drug with mild potential drug-gene interaction (no/moderate interaction vs. substantial interaction: OR, 2.08; 95% CI, 1.52-2.84; P = .005).

For the intervention group, the estimated rates of receiving an antidepressant with no, moderate, and substantial drug-gene interactions were 59.3%, 30.0%, and 10.7%, respectively. For the usual-care group, the estimates were 25.7%, 54.6%, and 19.7%.

The finding that 1 in 5 patients who received usual care were initially given a medication for which there were significant drug-gene interactions means it is “not a rare event,” said Dr. Oslin. “If we can make an impact on 20% of the people we prescribe to, that’s actually pretty big.”

Rates of remission were greater in the pharmacogenomic-guided group over 24 weeks (OR, 1.28; 95% CI, 1.05-1.57; P = .02; absolute risk difference, 2.8%; 95% CI, 0.6%-5.1%).

The secondary outcomes of response to treatment, defined as at least a 50% decrease in PHQ-9 score, also favored the pharmacogenomic-guided group. This was also the case for the secondary outcome of reduction in symptom severity on the PHQ-9 score.

Some physicians have expressed skepticism about pharmacogenomic testing, but the study provides additional evidence of its usefulness, Dr. Oslin noted.

“While I don’t think testing should be standard of practice, I also don’t think we should put barriers into the testing until we can better understand how to target the testing” to those who will benefit the most, he added.

The tests are available at a commercial cost of about $1,000 – which may not be that expensive if testing has a significant impact on a patient’s life, said Dr. Oslin.
 

Important research, but with several limitations

In an accompanying editorial, Dan V. Iosifescu, MD, associate professor of psychiatry at New York University School of Medicine and director of clinical research at the Nathan Kline Institute for Psychiatric Research, called the study an important addition to the literature on pharmacogenomic testing for patients with MDD.

The study was significantly larger and had broader inclusion criteria and longer follow-up than previous clinical trials and is one of the few investigations not funded by a manufacturer of pharmacogenomic tests, writes Dr. Iosifescu, who was not involved with the research.

However, he notes that an antidepressant was not initiated for 30 days after randomization in 25% of the intervention group and in 31% of the usual-care group, which was “puzzling.” “Because these rates were comparable in the 2 groups, it cannot be explained primarily by the delay of the pharmacogenomic test results in the intervention group,” he writes.

In addition, in the co-primary outcome of symptom remission rate, the difference in clinical improvement in favor of the pharmacogenomic-guided treatment was only “modest” – the gain was of less than 2% in the proportion of patients achieving remission, Dr. Iosifescu adds.

He adds this is “likely not very meaningful clinically despite this difference achieving statistical significance in this large study sample.”

Other potential study limitations he cites include the lack of patient blinding to treatment assignment and the absence of clarity about why rates of MDD response and remission over time were relatively low in both treatment groups.

A possible approach to optimize antidepressant choices could involve integration of pharmacogenomic data into larger predictive models that include clinical and demographic variables, Dr. Iosifescu notes.

“The development of such complex models is challenging, but it is now possible given the recent substantial advances in the proficiency of computational tools,” he writes.

The study was funded by the U.S. Department of Veterans Affairs (VA), Health Services Research and Development Service, and the Mental Illness Research, Education, and Clinical Center at the Corporal Michael J. Crescenz VA Medical Center. Dr. Oslin reports having received grants from the VA Office of Research and Development and Janssen Pharmaceuticals and nonfinancial support from Myriad Genetics during the conduct of the study. Dr. Iosifescu report having received personal fees from Alkermes, Allergan, Axsome, Biogen, the Centers for Psychiatric Excellence, Jazz, Lundbeck, Precision Neuroscience, Sage, and Sunovion and grants from Alkermes, AstraZeneca, Brainsway, Litecure, Neosync, Otsuka, Roche, and Shire.

A version of this article first appeared on Medscape.com.

Pharmacogenetic testing, which is used to classify how patients with major depressive disorder (MDD) metabolize medications, reduces adverse drug-gene interactions, new research shows.

In a randomized clinical trial that included almost 2,000 adults with MDD, patients in the pharmacogenomics-guided group were more likely to receive an antidepressant that had no potential drug-gene interaction than the patients who received usual care.

In addition, among the intervention group, the rate of remission over 24 weeks was significantly greater.

“These tests can be helpful in rethinking choices of antidepressants, but clinicians should not expect them to be helpful for every patient,” study investigator David W. Oslin, MD, Corporal Michael J. Crescenz VA Medical Center and professor of psychiatry at Perelman School of Medicine, University of Pennsylvania, Philadelphia, said in an interview.

The findings were published online  in JAMA.

Less trial and error

Pharmacogenomic testing can provide information to inform drug selection or dosing for patients with a genetic variation that alters pharmacokinetics or pharmacodynamics. Such testing may be particularly useful for patients with MDD, as fewer than 40% of these patients achieve clinical remission after an initial treatment with an antidepressant, the investigators note.

“To get to a treatment that works for an individual, it’s not unusual to have to try two or three or four antidepressants,” said Dr. Oslin. “If we could reduce that variance a little bit with a test like this, that would be huge from a public health perspective.”

The study included 676 physicians and 1,944 adults with MDD (mean age, 48 years; 24% women) who were receiving care at 22 Department of Veterans Affairs medical centers. Eligible patients were set to start a new antidepressant monotherapy, and all underwent a pharmacogenomic test using a cheek swab.

Investigators randomly assigned patients to receive test results when available (pharmacogenomic-guided group) or 24 weeks later (usual-care group). For the former group, clinicians were asked to initiate treatment when test results were available, typically within 2-3 days. For the latter group, they were asked to initiate treatment on a day of randomization.

Assessments included the 9-item Patient Health questionnaire (PHQ-9), scores for which range from 0-27 points, with higher scores indicating worse symptoms.

Of the total patient population, 79% completed the 24-week assessment.

Researchers characterized antidepressant medications on the basis of drug-gene interaction categories: no known interactions, moderate interactions, and substantial interactions.

The co-primary outcomes were treatment initiation within 30 days, determined on the basis of drug-gene interaction categories, and remission from depression symptoms, defined as a PHQ-9 score of less than or equal to 5.

Raters who were blinded to clinical care and study randomization assessed outcomes at 4, 8, 12, 18, and 24 weeks.
 

Significant impact?

Results showed that the pharmacogenomic-guided group was more likely to receive an antidepressant that had no potential drug-gene interaction, as opposed to one with a moderate/substantial interaction (odds ratio, 4.32; 95% confidence interval, 3.47-5.39; P < .001).

The usual-care group was more likely to receive a drug with mild potential drug-gene interaction (no/moderate interaction vs. substantial interaction: OR, 2.08; 95% CI, 1.52-2.84; P = .005).

For the intervention group, the estimated rates of receiving an antidepressant with no, moderate, and substantial drug-gene interactions were 59.3%, 30.0%, and 10.7%, respectively. For the usual-care group, the estimates were 25.7%, 54.6%, and 19.7%.

The finding that 1 in 5 patients who received usual care were initially given a medication for which there were significant drug-gene interactions means it is “not a rare event,” said Dr. Oslin. “If we can make an impact on 20% of the people we prescribe to, that’s actually pretty big.”

Rates of remission were greater in the pharmacogenomic-guided group over 24 weeks (OR, 1.28; 95% CI, 1.05-1.57; P = .02; absolute risk difference, 2.8%; 95% CI, 0.6%-5.1%).

The secondary outcomes of response to treatment, defined as at least a 50% decrease in PHQ-9 score, also favored the pharmacogenomic-guided group. This was also the case for the secondary outcome of reduction in symptom severity on the PHQ-9 score.

Some physicians have expressed skepticism about pharmacogenomic testing, but the study provides additional evidence of its usefulness, Dr. Oslin noted.

“While I don’t think testing should be standard of practice, I also don’t think we should put barriers into the testing until we can better understand how to target the testing” to those who will benefit the most, he added.

The tests are available at a commercial cost of about $1,000 – which may not be that expensive if testing has a significant impact on a patient’s life, said Dr. Oslin.
 

Important research, but with several limitations

In an accompanying editorial, Dan V. Iosifescu, MD, associate professor of psychiatry at New York University School of Medicine and director of clinical research at the Nathan Kline Institute for Psychiatric Research, called the study an important addition to the literature on pharmacogenomic testing for patients with MDD.

The study was significantly larger and had broader inclusion criteria and longer follow-up than previous clinical trials and is one of the few investigations not funded by a manufacturer of pharmacogenomic tests, writes Dr. Iosifescu, who was not involved with the research.

However, he notes that an antidepressant was not initiated for 30 days after randomization in 25% of the intervention group and in 31% of the usual-care group, which was “puzzling.” “Because these rates were comparable in the 2 groups, it cannot be explained primarily by the delay of the pharmacogenomic test results in the intervention group,” he writes.

In addition, in the co-primary outcome of symptom remission rate, the difference in clinical improvement in favor of the pharmacogenomic-guided treatment was only “modest” – the gain was of less than 2% in the proportion of patients achieving remission, Dr. Iosifescu adds.

He adds this is “likely not very meaningful clinically despite this difference achieving statistical significance in this large study sample.”

Other potential study limitations he cites include the lack of patient blinding to treatment assignment and the absence of clarity about why rates of MDD response and remission over time were relatively low in both treatment groups.

A possible approach to optimize antidepressant choices could involve integration of pharmacogenomic data into larger predictive models that include clinical and demographic variables, Dr. Iosifescu notes.

“The development of such complex models is challenging, but it is now possible given the recent substantial advances in the proficiency of computational tools,” he writes.

The study was funded by the U.S. Department of Veterans Affairs (VA), Health Services Research and Development Service, and the Mental Illness Research, Education, and Clinical Center at the Corporal Michael J. Crescenz VA Medical Center. Dr. Oslin reports having received grants from the VA Office of Research and Development and Janssen Pharmaceuticals and nonfinancial support from Myriad Genetics during the conduct of the study. Dr. Iosifescu report having received personal fees from Alkermes, Allergan, Axsome, Biogen, the Centers for Psychiatric Excellence, Jazz, Lundbeck, Precision Neuroscience, Sage, and Sunovion and grants from Alkermes, AstraZeneca, Brainsway, Litecure, Neosync, Otsuka, Roche, and Shire.

A version of this article first appeared on Medscape.com.

Young women with a new diagnosis of node-negative triple-negative breast cancer (TNBC) who have high levels of stromal tumor-infiltrating lymphocytes (sTILs) have a very good long-term prognosis, and may be suitable candidates for reduced intensity pre- or postoperative chemotherapy, according to a team of European investigators.

Among 441 women in a Dutch cancer registry who were younger than 40 when they were diagnosed with node-negative TNBC and had not undergone systemic therapy, those who had 75% or more TILs in the intratumoral stromal area had a 15-year cumulative incidence of distant metastases or death of just 2.1%, and every 10% increase in sTILs was associated with a 19% decrease in the risk of death.

In contrast, the 15-year cumulative incidence of distant metastases was 38.4% for women with stromal TIL scores of less than 30%, according to researchers writing in the Journal of Clinical Oncology.

“These data could be used as a starting point for designing a randomized controlled chemotherapy de-escalation trial. The current study confirms the importance of sTILs as a valuable addition to the set of standard prognostic factors in patients with TNBC,” wrote the researchers, who were led by Sabine C. Linn, MD, of the Netherlands Cancer Institute, Amsterdam.
 

Markers for immune response

Stromal TILs, a mixture of mononuclear immune cells, have been shown in previous studies to be prognostic for outcomes in patients with early-stage TNBC treated either with or without neoadjuvant or adjuvant chemotherapy.

For example, investigators cited a study published in JCO in 2014, that showed among women with TNBC enrolled in the phase 3 ECOG 2197 clinical trial and the related ECOG 119 clinical trial, after a nearly 11-year follow-up, higher sTIL scores were associated with significantly better prognosis with every 10% increase translating into a 14% reduction in the risk of recurrence or death (P = .02).

“The prognostic importance of sTILs is, however, unexplored in patients diagnosed under age 40 years, let alone in the subgroup of systemic therapy–naive patients,” Dr. Linn and colleagues wrote.
 

Retrospective study

To see whether the prognostic value of sTILs was as strong among young, systemic therapy–naive women, the investigators conducted a retrospective study of women enrolled in the Netherlands Cancer Registry who were diagnosed with node-negative TNBC from 1989 to 2000. The patients selected had undergone only locoregional treatment, including axillary node dissection, but had not received any systemic therapy.

Pathologists reviewed samples, with TILs reported for the stromal compartment. The samples were grouped by sTIL score categories of high (75% or greater), intermediate (30% to less than 75%), or low (less than 30%). The investigators looked at overall survival (OS) and distant metastasis-free survival (DMFS) stratified by sTIL scores,

During a median follow-up of 15 years, 107 women died or developed distant metastases, and 78 experienced a second primary cancer.

The results were as noted, with patients in the highest category of sTILs having very low rates of either death or distant metastases during follow-up.

“We confirm the prognostic value of sTILs in young patients with early-stage N0 TNBC who are systemic therapy naive by taking advantage of a prospectively collected population-based cohort. Increasing sTILs are significantly associated with improved OS and DMFS. Patients with high sTILs (> 75%) had an excellent 10-year overall survival and a very low 10-year incidence of distant metastasis or death.

The study was supported by grants from The Netherlands Organization for Health Research and Development, A Sister’s Hope, De Vrienden van UMC Utrecht, Agilent Technologies, the Dutch Cancer Society, and Breast Cancer Research Foundation. Dr. Linn reported consulting with and receiving compensation from Daiichi Sankyo, as well as receiving research funding from Genentech/Roche, AstraZeneca, Bristol-Myers Squibb, Tesaro, Merck, Immunomedics, Eurocept Pharmaceuticals, Agendia, and Novartis.

Young women with a new diagnosis of node-negative triple-negative breast cancer (TNBC) who have high levels of stromal tumor-infiltrating lymphocytes (sTILs) have a very good long-term prognosis, and may be suitable candidates for reduced intensity pre- or postoperative chemotherapy, according to a team of European investigators.

Among 441 women in a Dutch cancer registry who were younger than 40 when they were diagnosed with node-negative TNBC and had not undergone systemic therapy, those who had 75% or more TILs in the intratumoral stromal area had a 15-year cumulative incidence of distant metastases or death of just 2.1%, and every 10% increase in sTILs was associated with a 19% decrease in the risk of death.

In contrast, the 15-year cumulative incidence of distant metastases was 38.4% for women with stromal TIL scores of less than 30%, according to researchers writing in the Journal of Clinical Oncology.

“These data could be used as a starting point for designing a randomized controlled chemotherapy de-escalation trial. The current study confirms the importance of sTILs as a valuable addition to the set of standard prognostic factors in patients with TNBC,” wrote the researchers, who were led by Sabine C. Linn, MD, of the Netherlands Cancer Institute, Amsterdam.
 

Markers for immune response

Stromal TILs, a mixture of mononuclear immune cells, have been shown in previous studies to be prognostic for outcomes in patients with early-stage TNBC treated either with or without neoadjuvant or adjuvant chemotherapy.

For example, investigators cited a study published in JCO in 2014, that showed among women with TNBC enrolled in the phase 3 ECOG 2197 clinical trial and the related ECOG 119 clinical trial, after a nearly 11-year follow-up, higher sTIL scores were associated with significantly better prognosis with every 10% increase translating into a 14% reduction in the risk of recurrence or death (P = .02).

“The prognostic importance of sTILs is, however, unexplored in patients diagnosed under age 40 years, let alone in the subgroup of systemic therapy–naive patients,” Dr. Linn and colleagues wrote.
 

Retrospective study

To see whether the prognostic value of sTILs was as strong among young, systemic therapy–naive women, the investigators conducted a retrospective study of women enrolled in the Netherlands Cancer Registry who were diagnosed with node-negative TNBC from 1989 to 2000. The patients selected had undergone only locoregional treatment, including axillary node dissection, but had not received any systemic therapy.

Pathologists reviewed samples, with TILs reported for the stromal compartment. The samples were grouped by sTIL score categories of high (75% or greater), intermediate (30% to less than 75%), or low (less than 30%). The investigators looked at overall survival (OS) and distant metastasis-free survival (DMFS) stratified by sTIL scores,

During a median follow-up of 15 years, 107 women died or developed distant metastases, and 78 experienced a second primary cancer.

The results were as noted, with patients in the highest category of sTILs having very low rates of either death or distant metastases during follow-up.

“We confirm the prognostic value of sTILs in young patients with early-stage N0 TNBC who are systemic therapy naive by taking advantage of a prospectively collected population-based cohort. Increasing sTILs are significantly associated with improved OS and DMFS. Patients with high sTILs (> 75%) had an excellent 10-year overall survival and a very low 10-year incidence of distant metastasis or death.

The study was supported by grants from The Netherlands Organization for Health Research and Development, A Sister’s Hope, De Vrienden van UMC Utrecht, Agilent Technologies, the Dutch Cancer Society, and Breast Cancer Research Foundation. Dr. Linn reported consulting with and receiving compensation from Daiichi Sankyo, as well as receiving research funding from Genentech/Roche, AstraZeneca, Bristol-Myers Squibb, Tesaro, Merck, Immunomedics, Eurocept Pharmaceuticals, Agendia, and Novartis.

Young women with a new diagnosis of node-negative triple-negative breast cancer (TNBC) who have high levels of stromal tumor-infiltrating lymphocytes (sTILs) have a very good long-term prognosis, and may be suitable candidates for reduced intensity pre- or postoperative chemotherapy, according to a team of European investigators.

Among 441 women in a Dutch cancer registry who were younger than 40 when they were diagnosed with node-negative TNBC and had not undergone systemic therapy, those who had 75% or more TILs in the intratumoral stromal area had a 15-year cumulative incidence of distant metastases or death of just 2.1%, and every 10% increase in sTILs was associated with a 19% decrease in the risk of death.

In contrast, the 15-year cumulative incidence of distant metastases was 38.4% for women with stromal TIL scores of less than 30%, according to researchers writing in the Journal of Clinical Oncology.

“These data could be used as a starting point for designing a randomized controlled chemotherapy de-escalation trial. The current study confirms the importance of sTILs as a valuable addition to the set of standard prognostic factors in patients with TNBC,” wrote the researchers, who were led by Sabine C. Linn, MD, of the Netherlands Cancer Institute, Amsterdam.
 

Markers for immune response

Stromal TILs, a mixture of mononuclear immune cells, have been shown in previous studies to be prognostic for outcomes in patients with early-stage TNBC treated either with or without neoadjuvant or adjuvant chemotherapy.

For example, investigators cited a study published in JCO in 2014, that showed among women with TNBC enrolled in the phase 3 ECOG 2197 clinical trial and the related ECOG 119 clinical trial, after a nearly 11-year follow-up, higher sTIL scores were associated with significantly better prognosis with every 10% increase translating into a 14% reduction in the risk of recurrence or death (P = .02).

“The prognostic importance of sTILs is, however, unexplored in patients diagnosed under age 40 years, let alone in the subgroup of systemic therapy–naive patients,” Dr. Linn and colleagues wrote.
 

Retrospective study

To see whether the prognostic value of sTILs was as strong among young, systemic therapy–naive women, the investigators conducted a retrospective study of women enrolled in the Netherlands Cancer Registry who were diagnosed with node-negative TNBC from 1989 to 2000. The patients selected had undergone only locoregional treatment, including axillary node dissection, but had not received any systemic therapy.

Pathologists reviewed samples, with TILs reported for the stromal compartment. The samples were grouped by sTIL score categories of high (75% or greater), intermediate (30% to less than 75%), or low (less than 30%). The investigators looked at overall survival (OS) and distant metastasis-free survival (DMFS) stratified by sTIL scores,

During a median follow-up of 15 years, 107 women died or developed distant metastases, and 78 experienced a second primary cancer.

The results were as noted, with patients in the highest category of sTILs having very low rates of either death or distant metastases during follow-up.

“We confirm the prognostic value of sTILs in young patients with early-stage N0 TNBC who are systemic therapy naive by taking advantage of a prospectively collected population-based cohort. Increasing sTILs are significantly associated with improved OS and DMFS. Patients with high sTILs (> 75%) had an excellent 10-year overall survival and a very low 10-year incidence of distant metastasis or death.

The study was supported by grants from The Netherlands Organization for Health Research and Development, A Sister’s Hope, De Vrienden van UMC Utrecht, Agilent Technologies, the Dutch Cancer Society, and Breast Cancer Research Foundation. Dr. Linn reported consulting with and receiving compensation from Daiichi Sankyo, as well as receiving research funding from Genentech/Roche, AstraZeneca, Bristol-Myers Squibb, Tesaro, Merck, Immunomedics, Eurocept Pharmaceuticals, Agendia, and Novartis.

Dr. George Sakoulas is an infectious diseases clinician at Sharp Memorial Hospital in San Diego and professor of pediatrics at the University of California, San Diego School of Medicine. He was the lead investigator in a study published in the May/June 2022 issue of JCOM that found that, when allocated to the appropriate patient type, intravenous immunoglobulin can reduce hospital costs for COVID-19 care. ¹ He joined JCOM’s Editor-in-Chief, Dr. Ebrahim Barkoudah, to discuss the study’s background and highlight its main findings.

The following has been edited for length and clarity.

Dr. Barkoudah Dr. Sakoulas is an investigator and a clinician, bridging both worlds to bring the best evidence to our patients. We’re discussing his new article regarding intravenous immunoglobulin in treating nonventilated COVID-19 patients with moderate-to-severe hypoxia. Dr. Sakoulas, could you please share with our readers the clinical question your study addressed and what your work around COVID-19 management means for clinical practice?

Dr. Sakoulas Thank you. I’m an infectious disease physician. I’ve been treating patients with viral acute respiratory distress syndrome for almost 20 years as an ID doctor. Most of these cases are due to influenza or other viruses. And from time to time, anecdotally and supported by some literature, we’ve been using IVIG, or intravenous immunoglobulin, in some of these cases. And again, I can report anecdotal success with that over the years.

So when COVID emerged in March of 2020, we deployed IVIG in a couple of patients early who were heading downhill. Remember, in March of 2020, we didn’t have the knowledge of steroids helping, patients being ventilated very promptly, and we saw some patients who made a turnaround after treatment with IVIG. We were able to get some support from an industry sponsor and perform and publish a pilot study, enrolling patients early in the pandemic. That study actually showed benefits, which then led the sponsor to fund a phase 3 multicenter clinical trial. Unfortunately, a couple of things happened. First, the trial was designed with the knowledge we had in April of 2020, and again, this is before steroids, before we incorporated proning patients in the ICU, or started ventilating people early. So there were some management changes and evolutions and improvements that happened. And second, the trial was enrolling a very broad repertoire of patients. There were no age limitations, and the trial, ultimately a phase 3 multicenter trial, failed to meet its endpoint.

There were some trends for benefit in younger patients, and as the trial was ongoing, we continued to evolve our knowledge, and we really honed it down to seeing a benefit of using IVIG in patients with COVID with specific criteria in mind. They had to be relatively younger patients, under 65, and not have any major comorbidities. In other words, they weren’t dialysis patients or end-stage disease patients, heart failure patients, cancer or malignancy patients. So, you know, we’re looking at the patients under 65 with obesity, diabetes, and hypertension, who are rapidly declining, going from room air to BiPAP or high-flow oxygen in a short amount of time. And we learned that when using IVIG early, we actually saw patients improve and turn around.

What this article in JCOM highlighted was, number one, incorporating that outcome or that patient type and then looking at the cost of hospitalization of patients who received IVIG versus those that did not. There were 2 groups that were studied. One was the group of patients in that original pilot trial that I discussed who were randomized to receive 1 or the other prospectively; it was an unblinded randomized study. And the second group was a matched case-control study where we had patients treated with IVIG matched by age and comorbidity status and level of hypoxia to patients that did not receive IVIG. We saw a financial benefit in shortening or reducing hospitalizations, really coming down to getting rid of that 20% tail of patients that wound up going to the ICU, getting intubated, and using a high amount of hospital resources that would ramp up the cost of hospitalization. We saw great mitigation of that with IVIG, and even with a small subset of patients, we were able to show a benefit.

Dr. Barkoudah Any thoughts on where we can implement the new findings from your article in our practice at the moment, knowing we now have practice guidelines and protocols to treat COVID-19? There was a tangible benefit in treating the patients the way you approached it in your important work. Could you share with us what would be implementable at the moment?

Dr. Sakoulas I think, fortunately, with the increasing host immunity in the population and decreased virulence of the virus, perhaps we won’t see as many patients of the type that were in these trials going forward, but I suspect we will perhaps in the unvaccinated patients that remain. I believe one-third of the United States is not vaccinated. So there is certainly a vulnerable group of people out there. Potentially, an unvaccinated patient who winds up getting very sick, the patient who is relatively young—what I’m looking at is the 30- to 65-year-old obese, hypertensive, or diabetic patient who comes in and, despite the steroids and the antivirals, rapidly deteriorates into requiring high-flow oxygen. I think implementing IVIG in that patient type would be helpful. I don’t think it’s going to be as helpful in patients who are very elderly, because I think the mechanism of the disease is different in an 80-year-old versus a 50-year-old patient. So again, hopefully, it will not amount to a lot of patients, but I still suspect hospitals are going to see, perhaps in the fall, when they’re expecting a greater number of cases, a trickling of patients that do meet the criteria that I described.

Dr. Barkoudah JCOM’s audience are the QI implementers and hospital leadership. And what caught my eye in your article is your perspective on the pharmacoeconomics of treating COVID-19, and I really appreciate your looking at the cost aspect. Would you talk about the economics of inpatient care, the total care that we provide now that we’re in the age of tocilizumab, and the current state of multiple layers of therapy?

Dr. Sakoulas The reason to look at the economics of it is because IVIG—which is actually not a drug, it’s a blood product—is very expensive. So, we received a considerable amount of administrative pushback implementing this treatment at the beginning outside of the clinical trial setting because it hadn’t been studied on a large scale and because the cost was so high, even though, as a clinician at the bedside, I was seeing a benefit in patients. This study came out of my trying to demonstrate to the folks that are keeping the economics of medicine in mind that, in fact, investing several thousand dollars of treatment in IVIG will save you cost of care, the cost of an ICU bed, the cost of a ventilator, and the cost even of ECMO, which is hugely expensive.

If you look at the numbers in the study, for two-thirds or three-quarters of the patients, your cost of care is actually greater than the controls because you’re giving them IVIG, and it’s increasing the cost of their care, even though three-quarters of the patients are going to do just as well without it. It’s that 20% to 25% of patients that really are going to benefit from it, where you’re reducing your cost of care so much, and you’re getting rid of that very, very expensive 20%, that there’s a cost savings across the board per patient. So, it’s hard to understand when you say you’re losing money on three-quarters of the patients, you’re only saving money on a quarter of the patients, but that cost of saving on that small subset is so substantial it’s really impacting all numbers.

Also, abandoning the outlier principle is sort of an underlying theme in how we think of things. We tend to ignore outliers, not consider them, but I think we really have to pay attention to the more extreme cases because those patients are the ones that drive not just the financial cost of care. Remember, if you’re down to 1 ventilator and you can cut down the use of scarce ICU resources, the cost is sort of even beyond the cost of money. It’s the cost of resources that may become scarce in some settings. So, I think it speaks to that as well.

A lot of the drugs that we use, for example, tocilizumab, were able to be studied in thousands of patients. If you look at the absolute numbers, the benefit of tocilizumab from a magnitude standpoint—low to mid twenties to high twenties—you know, reducing mortality from 29% to 24%. I mean, just take a step back and think about that. Even though it’s statistically significant, try telling a patient, “Well, I’m going to give you this treatment that’s going to reduce mortality from 29% to 24%.” You know, that doesn’t really change anything from a clinical significance standpoint. But they have a P value less than .05, which is our standard, and they were able to do a study with thousands of patients. We didn’t have that luxury with IVIG. No one studied thousands of patients, only retrospectively, and those retrospective studies don’t get the attention because they’re considered biased with all their limitations. But I think one of the difficulties we have here is the balance between statistical and clinical significance. For example, in our pilot study, our ventilation rate was 58% with the non-IVIG patients versus 14% for IVIG patients. So you might say, magnitude-wise, that’s a big number, but the statistical significance of it is borderline because of small numbers.

Anyway, that’s a challenge that we have as clinicians trying to incorporate what’s published—the balancing of statistics, absolute numbers, and practicalities of delivering care. And I think this study highlights some of the nuances that go into that incorporation and those clinical decisions.

Dr. Barkoudah Would you mind sharing with our audience how we can make the connection between the medical outcomes and pharmacoeconomics findings from your article and link it to the bedside and treatment of our patients?

Dr. Sakoulas One of the points this article brings out is the importance of bringing together not just level 1A data, but also small studies with data such as this, where the magnitude of the effect is pretty big but you lose the statistics because of the small numbers. And then also the patients’ aspects of things. I think, as a bedside clinician, you appreciate things, the nuances, much sooner than what percolates out from a level 1A study. Case in point, in the sponsored phase 3 study that we did, and in some other studies that were prospectively done as well, these studies of IVIG simply had an enrollment of patients that was very broad, and not every patient benefits from the same therapy. A great example of this is the sepsis trials with Xigris and those types of agents that failed. You know, there are clinicians to this day who believe that there is a subset of patients that benefit from agents like this. The IVIG story falls a little bit into that category. It comes down to trying to identify the subset of patients that might benefit. And I think we’ve outlined this subset pretty well in our study: the younger, obese diabetic or hypertensive patient who’s rapidly declining.

It really brings together the need to not necessarily toss out these smaller studies, but kind of summarize everything together, and clinicians who are bedside, who are more in tune with the nuances of individual decisions at the individual patient level, might better appreciate these kinds of data. But I think we all have to put it together. IVIG does not make treatment guidelines at national levels and so forth. It’s not even listed in many of them. But there are patients out there who, if you ask them specifically how they felt, including a friend of mine who received the medication, there’s no question from their end, how they felt about this treatment option. Now, some people will get it and will not benefit. We just have to be really tuned into the fact that the same drug does not have the same result for every patient. And just to consider this in the high-risk patients that we talked about in our study.

Dr. Barkoudah While we were prepping for this interview, you made an analogy regarding clinical evidence along the lines of, “Do we need randomized clinical trials to do a parachute-type of experiment,” and we chatted about clinical wisdom. Would you mind sharing with our readers your thoughts on that?

Dr. Sakoulas Sometimes, we try a treatment and it’s very obvious for that particular patient that it helped them. Then you study the treatment in a large trial setting and it doesn’t work. For us bedside clinicians, there are some interventions sometimes that do appear as beneficial as a parachute would be, but yet, there has never been a randomized clinical trial proving that parachutes work. Again, a part of the challenge we have is patients are so different, their immunology is different, the pathogen infecting them is different, the time they present is different. Some present early, some present late. There are just so many moving parts to treating an infection that only a subset of people are going to benefit. And sometimes as clinicians, we’re so nuanced, that we identify a specific subset of patients where we know we can help them. And it’s so obvious for us, like a parachute would be, but to people who are looking at the world from 30,000 feet, they don’t necessarily grasp that because, when you look at all comers, it doesn’t show a benefit.

So the problem is that now those treatments that might help a subset of patients are being denied, and the subset of patients that are going to benefit never get the treatment. Now we have to balance that with a lot of stuff that went on during the pandemic with, you know, ivermectin, hydroxychloroquine, and people pushing those things. Someone asked me once what I thought about hydroxychloroquine, and I said, “Well, somebody in the lab probably showed that it was beneficial, analogous to lighting tissue paper on fire on a plate and taking a cup of water and putting the fire out. Well, now, if you take that cup of water to the Caldor fire that’s burning in California on thousands of acres, you’re not going to be able to put the fire out with that cup of water.” So while it might work in the lab, it’s truly not going to work in a clinical setting. We have to balance individualizing care for patients with some information people are pushing out there that may not be necessarily translatable to the clinical setting.

I think there’s nothing better than being at the bedside, though, and being able to implement something and seeing what works. And really, experience goes a long way in being able to individually treat a patient optimally.

Dr. Barkoudah Thank you for everything you do at the bedside and your work on improving the treatment we have and how we can leverage knowledge to treat our patients. Thank you very much for your time and your scholarly contribution. We appreciate it and I hope the work will continue. We will keep working on treating COVID-19 patients with the best knowledge we have.

Q&A participants: George Sakoulas, MD, Sharp Rees-Stealy Medical Group, La Jolla, CA, and University of California San Diego School of Medicine, San Diego, CA; and Ebrahim Barkoudah, MD, MPH, Department of Medicine, Brigham and Women’s Hospital, Boston, MA.

Disclosures: None reported.

Dr. George Sakoulas is an infectious diseases clinician at Sharp Memorial Hospital in San Diego and professor of pediatrics at the University of California, San Diego School of Medicine. He was the lead investigator in a study published in the May/June 2022 issue of JCOM that found that, when allocated to the appropriate patient type, intravenous immunoglobulin can reduce hospital costs for COVID-19 care. ¹ He joined JCOM’s Editor-in-Chief, Dr. Ebrahim Barkoudah, to discuss the study’s background and highlight its main findings.

The following has been edited for length and clarity.

Dr. Barkoudah Dr. Sakoulas is an investigator and a clinician, bridging both worlds to bring the best evidence to our patients. We’re discussing his new article regarding intravenous immunoglobulin in treating nonventilated COVID-19 patients with moderate-to-severe hypoxia. Dr. Sakoulas, could you please share with our readers the clinical question your study addressed and what your work around COVID-19 management means for clinical practice?

Dr. Sakoulas Thank you. I’m an infectious disease physician. I’ve been treating patients with viral acute respiratory distress syndrome for almost 20 years as an ID doctor. Most of these cases are due to influenza or other viruses. And from time to time, anecdotally and supported by some literature, we’ve been using IVIG, or intravenous immunoglobulin, in some of these cases. And again, I can report anecdotal success with that over the years.

So when COVID emerged in March of 2020, we deployed IVIG in a couple of patients early who were heading downhill. Remember, in March of 2020, we didn’t have the knowledge of steroids helping, patients being ventilated very promptly, and we saw some patients who made a turnaround after treatment with IVIG. We were able to get some support from an industry sponsor and perform and publish a pilot study, enrolling patients early in the pandemic. That study actually showed benefits, which then led the sponsor to fund a phase 3 multicenter clinical trial. Unfortunately, a couple of things happened. First, the trial was designed with the knowledge we had in April of 2020, and again, this is before steroids, before we incorporated proning patients in the ICU, or started ventilating people early. So there were some management changes and evolutions and improvements that happened. And second, the trial was enrolling a very broad repertoire of patients. There were no age limitations, and the trial, ultimately a phase 3 multicenter trial, failed to meet its endpoint.

There were some trends for benefit in younger patients, and as the trial was ongoing, we continued to evolve our knowledge, and we really honed it down to seeing a benefit of using IVIG in patients with COVID with specific criteria in mind. They had to be relatively younger patients, under 65, and not have any major comorbidities. In other words, they weren’t dialysis patients or end-stage disease patients, heart failure patients, cancer or malignancy patients. So, you know, we’re looking at the patients under 65 with obesity, diabetes, and hypertension, who are rapidly declining, going from room air to BiPAP or high-flow oxygen in a short amount of time. And we learned that when using IVIG early, we actually saw patients improve and turn around.

What this article in JCOM highlighted was, number one, incorporating that outcome or that patient type and then looking at the cost of hospitalization of patients who received IVIG versus those that did not. There were 2 groups that were studied. One was the group of patients in that original pilot trial that I discussed who were randomized to receive 1 or the other prospectively; it was an unblinded randomized study. And the second group was a matched case-control study where we had patients treated with IVIG matched by age and comorbidity status and level of hypoxia to patients that did not receive IVIG. We saw a financial benefit in shortening or reducing hospitalizations, really coming down to getting rid of that 20% tail of patients that wound up going to the ICU, getting intubated, and using a high amount of hospital resources that would ramp up the cost of hospitalization. We saw great mitigation of that with IVIG, and even with a small subset of patients, we were able to show a benefit.

Dr. Barkoudah Any thoughts on where we can implement the new findings from your article in our practice at the moment, knowing we now have practice guidelines and protocols to treat COVID-19? There was a tangible benefit in treating the patients the way you approached it in your important work. Could you share with us what would be implementable at the moment?

Dr. Sakoulas I think, fortunately, with the increasing host immunity in the population and decreased virulence of the virus, perhaps we won’t see as many patients of the type that were in these trials going forward, but I suspect we will perhaps in the unvaccinated patients that remain. I believe one-third of the United States is not vaccinated. So there is certainly a vulnerable group of people out there. Potentially, an unvaccinated patient who winds up getting very sick, the patient who is relatively young—what I’m looking at is the 30- to 65-year-old obese, hypertensive, or diabetic patient who comes in and, despite the steroids and the antivirals, rapidly deteriorates into requiring high-flow oxygen. I think implementing IVIG in that patient type would be helpful. I don’t think it’s going to be as helpful in patients who are very elderly, because I think the mechanism of the disease is different in an 80-year-old versus a 50-year-old patient. So again, hopefully, it will not amount to a lot of patients, but I still suspect hospitals are going to see, perhaps in the fall, when they’re expecting a greater number of cases, a trickling of patients that do meet the criteria that I described.

Dr. Barkoudah JCOM’s audience are the QI implementers and hospital leadership. And what caught my eye in your article is your perspective on the pharmacoeconomics of treating COVID-19, and I really appreciate your looking at the cost aspect. Would you talk about the economics of inpatient care, the total care that we provide now that we’re in the age of tocilizumab, and the current state of multiple layers of therapy?

Dr. Sakoulas The reason to look at the economics of it is because IVIG—which is actually not a drug, it’s a blood product—is very expensive. So, we received a considerable amount of administrative pushback implementing this treatment at the beginning outside of the clinical trial setting because it hadn’t been studied on a large scale and because the cost was so high, even though, as a clinician at the bedside, I was seeing a benefit in patients. This study came out of my trying to demonstrate to the folks that are keeping the economics of medicine in mind that, in fact, investing several thousand dollars of treatment in IVIG will save you cost of care, the cost of an ICU bed, the cost of a ventilator, and the cost even of ECMO, which is hugely expensive.

If you look at the numbers in the study, for two-thirds or three-quarters of the patients, your cost of care is actually greater than the controls because you’re giving them IVIG, and it’s increasing the cost of their care, even though three-quarters of the patients are going to do just as well without it. It’s that 20% to 25% of patients that really are going to benefit from it, where you’re reducing your cost of care so much, and you’re getting rid of that very, very expensive 20%, that there’s a cost savings across the board per patient. So, it’s hard to understand when you say you’re losing money on three-quarters of the patients, you’re only saving money on a quarter of the patients, but that cost of saving on that small subset is so substantial it’s really impacting all numbers.

Also, abandoning the outlier principle is sort of an underlying theme in how we think of things. We tend to ignore outliers, not consider them, but I think we really have to pay attention to the more extreme cases because those patients are the ones that drive not just the financial cost of care. Remember, if you’re down to 1 ventilator and you can cut down the use of scarce ICU resources, the cost is sort of even beyond the cost of money. It’s the cost of resources that may become scarce in some settings. So, I think it speaks to that as well.

A lot of the drugs that we use, for example, tocilizumab, were able to be studied in thousands of patients. If you look at the absolute numbers, the benefit of tocilizumab from a magnitude standpoint—low to mid twenties to high twenties—you know, reducing mortality from 29% to 24%. I mean, just take a step back and think about that. Even though it’s statistically significant, try telling a patient, “Well, I’m going to give you this treatment that’s going to reduce mortality from 29% to 24%.” You know, that doesn’t really change anything from a clinical significance standpoint. But they have a P value less than .05, which is our standard, and they were able to do a study with thousands of patients. We didn’t have that luxury with IVIG. No one studied thousands of patients, only retrospectively, and those retrospective studies don’t get the attention because they’re considered biased with all their limitations. But I think one of the difficulties we have here is the balance between statistical and clinical significance. For example, in our pilot study, our ventilation rate was 58% with the non-IVIG patients versus 14% for IVIG patients. So you might say, magnitude-wise, that’s a big number, but the statistical significance of it is borderline because of small numbers.

Anyway, that’s a challenge that we have as clinicians trying to incorporate what’s published—the balancing of statistics, absolute numbers, and practicalities of delivering care. And I think this study highlights some of the nuances that go into that incorporation and those clinical decisions.

Dr. Barkoudah Would you mind sharing with our audience how we can make the connection between the medical outcomes and pharmacoeconomics findings from your article and link it to the bedside and treatment of our patients?

Dr. Sakoulas One of the points this article brings out is the importance of bringing together not just level 1A data, but also small studies with data such as this, where the magnitude of the effect is pretty big but you lose the statistics because of the small numbers. And then also the patients’ aspects of things. I think, as a bedside clinician, you appreciate things, the nuances, much sooner than what percolates out from a level 1A study. Case in point, in the sponsored phase 3 study that we did, and in some other studies that were prospectively done as well, these studies of IVIG simply had an enrollment of patients that was very broad, and not every patient benefits from the same therapy. A great example of this is the sepsis trials with Xigris and those types of agents that failed. You know, there are clinicians to this day who believe that there is a subset of patients that benefit from agents like this. The IVIG story falls a little bit into that category. It comes down to trying to identify the subset of patients that might benefit. And I think we’ve outlined this subset pretty well in our study: the younger, obese diabetic or hypertensive patient who’s rapidly declining.

It really brings together the need to not necessarily toss out these smaller studies, but kind of summarize everything together, and clinicians who are bedside, who are more in tune with the nuances of individual decisions at the individual patient level, might better appreciate these kinds of data. But I think we all have to put it together. IVIG does not make treatment guidelines at national levels and so forth. It’s not even listed in many of them. But there are patients out there who, if you ask them specifically how they felt, including a friend of mine who received the medication, there’s no question from their end, how they felt about this treatment option. Now, some people will get it and will not benefit. We just have to be really tuned into the fact that the same drug does not have the same result for every patient. And just to consider this in the high-risk patients that we talked about in our study.

Dr. Barkoudah While we were prepping for this interview, you made an analogy regarding clinical evidence along the lines of, “Do we need randomized clinical trials to do a parachute-type of experiment,” and we chatted about clinical wisdom. Would you mind sharing with our readers your thoughts on that?

Dr. Sakoulas Sometimes, we try a treatment and it’s very obvious for that particular patient that it helped them. Then you study the treatment in a large trial setting and it doesn’t work. For us bedside clinicians, there are some interventions sometimes that do appear as beneficial as a parachute would be, but yet, there has never been a randomized clinical trial proving that parachutes work. Again, a part of the challenge we have is patients are so different, their immunology is different, the pathogen infecting them is different, the time they present is different. Some present early, some present late. There are just so many moving parts to treating an infection that only a subset of people are going to benefit. And sometimes as clinicians, we’re so nuanced, that we identify a specific subset of patients where we know we can help them. And it’s so obvious for us, like a parachute would be, but to people who are looking at the world from 30,000 feet, they don’t necessarily grasp that because, when you look at all comers, it doesn’t show a benefit.

So the problem is that now those treatments that might help a subset of patients are being denied, and the subset of patients that are going to benefit never get the treatment. Now we have to balance that with a lot of stuff that went on during the pandemic with, you know, ivermectin, hydroxychloroquine, and people pushing those things. Someone asked me once what I thought about hydroxychloroquine, and I said, “Well, somebody in the lab probably showed that it was beneficial, analogous to lighting tissue paper on fire on a plate and taking a cup of water and putting the fire out. Well, now, if you take that cup of water to the Caldor fire that’s burning in California on thousands of acres, you’re not going to be able to put the fire out with that cup of water.” So while it might work in the lab, it’s truly not going to work in a clinical setting. We have to balance individualizing care for patients with some information people are pushing out there that may not be necessarily translatable to the clinical setting.

I think there’s nothing better than being at the bedside, though, and being able to implement something and seeing what works. And really, experience goes a long way in being able to individually treat a patient optimally.

Dr. Barkoudah Thank you for everything you do at the bedside and your work on improving the treatment we have and how we can leverage knowledge to treat our patients. Thank you very much for your time and your scholarly contribution. We appreciate it and I hope the work will continue. We will keep working on treating COVID-19 patients with the best knowledge we have.

Q&A participants: George Sakoulas, MD, Sharp Rees-Stealy Medical Group, La Jolla, CA, and University of California San Diego School of Medicine, San Diego, CA; and Ebrahim Barkoudah, MD, MPH, Department of Medicine, Brigham and Women’s Hospital, Boston, MA.

Disclosures: None reported.

Dr. George Sakoulas is an infectious diseases clinician at Sharp Memorial Hospital in San Diego and professor of pediatrics at the University of California, San Diego School of Medicine. He was the lead investigator in a study published in the May/June 2022 issue of JCOM that found that, when allocated to the appropriate patient type, intravenous immunoglobulin can reduce hospital costs for COVID-19 care. ¹ He joined JCOM’s Editor-in-Chief, Dr. Ebrahim Barkoudah, to discuss the study’s background and highlight its main findings.

The following has been edited for length and clarity.

Dr. Barkoudah Dr. Sakoulas is an investigator and a clinician, bridging both worlds to bring the best evidence to our patients. We’re discussing his new article regarding intravenous immunoglobulin in treating nonventilated COVID-19 patients with moderate-to-severe hypoxia. Dr. Sakoulas, could you please share with our readers the clinical question your study addressed and what your work around COVID-19 management means for clinical practice?

Dr. Sakoulas Thank you. I’m an infectious disease physician. I’ve been treating patients with viral acute respiratory distress syndrome for almost 20 years as an ID doctor. Most of these cases are due to influenza or other viruses. And from time to time, anecdotally and supported by some literature, we’ve been using IVIG, or intravenous immunoglobulin, in some of these cases. And again, I can report anecdotal success with that over the years.

So when COVID emerged in March of 2020, we deployed IVIG in a couple of patients early who were heading downhill. Remember, in March of 2020, we didn’t have the knowledge of steroids helping, patients being ventilated very promptly, and we saw some patients who made a turnaround after treatment with IVIG. We were able to get some support from an industry sponsor and perform and publish a pilot study, enrolling patients early in the pandemic. That study actually showed benefits, which then led the sponsor to fund a phase 3 multicenter clinical trial. Unfortunately, a couple of things happened. First, the trial was designed with the knowledge we had in April of 2020, and again, this is before steroids, before we incorporated proning patients in the ICU, or started ventilating people early. So there were some management changes and evolutions and improvements that happened. And second, the trial was enrolling a very broad repertoire of patients. There were no age limitations, and the trial, ultimately a phase 3 multicenter trial, failed to meet its endpoint.

There were some trends for benefit in younger patients, and as the trial was ongoing, we continued to evolve our knowledge, and we really honed it down to seeing a benefit of using IVIG in patients with COVID with specific criteria in mind. They had to be relatively younger patients, under 65, and not have any major comorbidities. In other words, they weren’t dialysis patients or end-stage disease patients, heart failure patients, cancer or malignancy patients. So, you know, we’re looking at the patients under 65 with obesity, diabetes, and hypertension, who are rapidly declining, going from room air to BiPAP or high-flow oxygen in a short amount of time. And we learned that when using IVIG early, we actually saw patients improve and turn around.

What this article in JCOM highlighted was, number one, incorporating that outcome or that patient type and then looking at the cost of hospitalization of patients who received IVIG versus those that did not. There were 2 groups that were studied. One was the group of patients in that original pilot trial that I discussed who were randomized to receive 1 or the other prospectively; it was an unblinded randomized study. And the second group was a matched case-control study where we had patients treated with IVIG matched by age and comorbidity status and level of hypoxia to patients that did not receive IVIG. We saw a financial benefit in shortening or reducing hospitalizations, really coming down to getting rid of that 20% tail of patients that wound up going to the ICU, getting intubated, and using a high amount of hospital resources that would ramp up the cost of hospitalization. We saw great mitigation of that with IVIG, and even with a small subset of patients, we were able to show a benefit.

Dr. Barkoudah Any thoughts on where we can implement the new findings from your article in our practice at the moment, knowing we now have practice guidelines and protocols to treat COVID-19? There was a tangible benefit in treating the patients the way you approached it in your important work. Could you share with us what would be implementable at the moment?

Dr. Sakoulas I think, fortunately, with the increasing host immunity in the population and decreased virulence of the virus, perhaps we won’t see as many patients of the type that were in these trials going forward, but I suspect we will perhaps in the unvaccinated patients that remain. I believe one-third of the United States is not vaccinated. So there is certainly a vulnerable group of people out there. Potentially, an unvaccinated patient who winds up getting very sick, the patient who is relatively young—what I’m looking at is the 30- to 65-year-old obese, hypertensive, or diabetic patient who comes in and, despite the steroids and the antivirals, rapidly deteriorates into requiring high-flow oxygen. I think implementing IVIG in that patient type would be helpful. I don’t think it’s going to be as helpful in patients who are very elderly, because I think the mechanism of the disease is different in an 80-year-old versus a 50-year-old patient. So again, hopefully, it will not amount to a lot of patients, but I still suspect hospitals are going to see, perhaps in the fall, when they’re expecting a greater number of cases, a trickling of patients that do meet the criteria that I described.

Dr. Barkoudah JCOM’s audience are the QI implementers and hospital leadership. And what caught my eye in your article is your perspective on the pharmacoeconomics of treating COVID-19, and I really appreciate your looking at the cost aspect. Would you talk about the economics of inpatient care, the total care that we provide now that we’re in the age of tocilizumab, and the current state of multiple layers of therapy?

Dr. Sakoulas The reason to look at the economics of it is because IVIG—which is actually not a drug, it’s a blood product—is very expensive. So, we received a considerable amount of administrative pushback implementing this treatment at the beginning outside of the clinical trial setting because it hadn’t been studied on a large scale and because the cost was so high, even though, as a clinician at the bedside, I was seeing a benefit in patients. This study came out of my trying to demonstrate to the folks that are keeping the economics of medicine in mind that, in fact, investing several thousand dollars of treatment in IVIG will save you cost of care, the cost of an ICU bed, the cost of a ventilator, and the cost even of ECMO, which is hugely expensive.

If you look at the numbers in the study, for two-thirds or three-quarters of the patients, your cost of care is actually greater than the controls because you’re giving them IVIG, and it’s increasing the cost of their care, even though three-quarters of the patients are going to do just as well without it. It’s that 20% to 25% of patients that really are going to benefit from it, where you’re reducing your cost of care so much, and you’re getting rid of that very, very expensive 20%, that there’s a cost savings across the board per patient. So, it’s hard to understand when you say you’re losing money on three-quarters of the patients, you’re only saving money on a quarter of the patients, but that cost of saving on that small subset is so substantial it’s really impacting all numbers.

Also, abandoning the outlier principle is sort of an underlying theme in how we think of things. We tend to ignore outliers, not consider them, but I think we really have to pay attention to the more extreme cases because those patients are the ones that drive not just the financial cost of care. Remember, if you’re down to 1 ventilator and you can cut down the use of scarce ICU resources, the cost is sort of even beyond the cost of money. It’s the cost of resources that may become scarce in some settings. So, I think it speaks to that as well.

A lot of the drugs that we use, for example, tocilizumab, were able to be studied in thousands of patients. If you look at the absolute numbers, the benefit of tocilizumab from a magnitude standpoint—low to mid twenties to high twenties—you know, reducing mortality from 29% to 24%. I mean, just take a step back and think about that. Even though it’s statistically significant, try telling a patient, “Well, I’m going to give you this treatment that’s going to reduce mortality from 29% to 24%.” You know, that doesn’t really change anything from a clinical significance standpoint. But they have a P value less than .05, which is our standard, and they were able to do a study with thousands of patients. We didn’t have that luxury with IVIG. No one studied thousands of patients, only retrospectively, and those retrospective studies don’t get the attention because they’re considered biased with all their limitations. But I think one of the difficulties we have here is the balance between statistical and clinical significance. For example, in our pilot study, our ventilation rate was 58% with the non-IVIG patients versus 14% for IVIG patients. So you might say, magnitude-wise, that’s a big number, but the statistical significance of it is borderline because of small numbers.

Anyway, that’s a challenge that we have as clinicians trying to incorporate what’s published—the balancing of statistics, absolute numbers, and practicalities of delivering care. And I think this study highlights some of the nuances that go into that incorporation and those clinical decisions.

Dr. Barkoudah Would you mind sharing with our audience how we can make the connection between the medical outcomes and pharmacoeconomics findings from your article and link it to the bedside and treatment of our patients?

Dr. Sakoulas One of the points this article brings out is the importance of bringing together not just level 1A data, but also small studies with data such as this, where the magnitude of the effect is pretty big but you lose the statistics because of the small numbers. And then also the patients’ aspects of things. I think, as a bedside clinician, you appreciate things, the nuances, much sooner than what percolates out from a level 1A study. Case in point, in the sponsored phase 3 study that we did, and in some other studies that were prospectively done as well, these studies of IVIG simply had an enrollment of patients that was very broad, and not every patient benefits from the same therapy. A great example of this is the sepsis trials with Xigris and those types of agents that failed. You know, there are clinicians to this day who believe that there is a subset of patients that benefit from agents like this. The IVIG story falls a little bit into that category. It comes down to trying to identify the subset of patients that might benefit. And I think we’ve outlined this subset pretty well in our study: the younger, obese diabetic or hypertensive patient who’s rapidly declining.

It really brings together the need to not necessarily toss out these smaller studies, but kind of summarize everything together, and clinicians who are bedside, who are more in tune with the nuances of individual decisions at the individual patient level, might better appreciate these kinds of data. But I think we all have to put it together. IVIG does not make treatment guidelines at national levels and so forth. It’s not even listed in many of them. But there are patients out there who, if you ask them specifically how they felt, including a friend of mine who received the medication, there’s no question from their end, how they felt about this treatment option. Now, some people will get it and will not benefit. We just have to be really tuned into the fact that the same drug does not have the same result for every patient. And just to consider this in the high-risk patients that we talked about in our study.

Dr. Barkoudah While we were prepping for this interview, you made an analogy regarding clinical evidence along the lines of, “Do we need randomized clinical trials to do a parachute-type of experiment,” and we chatted about clinical wisdom. Would you mind sharing with our readers your thoughts on that?

Dr. Sakoulas Sometimes, we try a treatment and it’s very obvious for that particular patient that it helped them. Then you study the treatment in a large trial setting and it doesn’t work. For us bedside clinicians, there are some interventions sometimes that do appear as beneficial as a parachute would be, but yet, there has never been a randomized clinical trial proving that parachutes work. Again, a part of the challenge we have is patients are so different, their immunology is different, the pathogen infecting them is different, the time they present is different. Some present early, some present late. There are just so many moving parts to treating an infection that only a subset of people are going to benefit. And sometimes as clinicians, we’re so nuanced, that we identify a specific subset of patients where we know we can help them. And it’s so obvious for us, like a parachute would be, but to people who are looking at the world from 30,000 feet, they don’t necessarily grasp that because, when you look at all comers, it doesn’t show a benefit.

So the problem is that now those treatments that might help a subset of patients are being denied, and the subset of patients that are going to benefit never get the treatment. Now we have to balance that with a lot of stuff that went on during the pandemic with, you know, ivermectin, hydroxychloroquine, and people pushing those things. Someone asked me once what I thought about hydroxychloroquine, and I said, “Well, somebody in the lab probably showed that it was beneficial, analogous to lighting tissue paper on fire on a plate and taking a cup of water and putting the fire out. Well, now, if you take that cup of water to the Caldor fire that’s burning in California on thousands of acres, you’re not going to be able to put the fire out with that cup of water.” So while it might work in the lab, it’s truly not going to work in a clinical setting. We have to balance individualizing care for patients with some information people are pushing out there that may not be necessarily translatable to the clinical setting.

I think there’s nothing better than being at the bedside, though, and being able to implement something and seeing what works. And really, experience goes a long way in being able to individually treat a patient optimally.

Dr. Barkoudah Thank you for everything you do at the bedside and your work on improving the treatment we have and how we can leverage knowledge to treat our patients. Thank you very much for your time and your scholarly contribution. We appreciate it and I hope the work will continue. We will keep working on treating COVID-19 patients with the best knowledge we have.

Q&A participants: George Sakoulas, MD, Sharp Rees-Stealy Medical Group, La Jolla, CA, and University of California San Diego School of Medicine, San Diego, CA; and Ebrahim Barkoudah, MD, MPH, Department of Medicine, Brigham and Women’s Hospital, Boston, MA.

Disclosures: None reported.

From the University of Chicago Medical Center, Chicago, IL.

Abstract

Background: Epistaxis is a common chief complaint addressed by otolaryngologists. A review of the literature showed that there is a deficit in epistaxis education within the nursing community. Conversations with our nursing colleagues confirmed this unmet demand.

Objective: This quality improvement project aimed to increase general epistaxis knowledge, perceived comfort level managing nosebleeds, and perceived ability to stop nosebleeds among our nursing staff.

Methods: Data were collected through a survey administered before and after our intervention. The survey tested general epistaxis knowledge and assessed comfort and confidence in stopping epistaxis. Our intervention was an educational session covering pertinent epistaxis etiology and management. Quality improvement principles were used to optimize delivery of the intervention.

Results: A total of 51 nurses participated in the project. After participating in the in-service educational session, nurses answered significantly more epistaxis general knowledge questions correctly (mean [SD] difference, 2.07 [1.10] questions; 95% CI, 1.74-2.39; P < .001). There was no statistically significant difference in additional correct questions when stratified by clinical experience or clinical setting (P = .128 and P = 0.446, respectively). Nurses also reported feeling significantly more comfortable and significantly more confident in managing nosebleeds after the in-service (P = .007 and P < 0.001, respectively); 74.46% of nurses had an improvement in comfort level in managing epistaxis and 43.90% of nurses had an improvement in confidence in stopping epistaxis. After we moved the educational session from mid-shift to shift change, the nursing staff reported more satisfaction while maintaining similar improvements in knowledge and confidence.

Conclusion: We were able to significantly increase epistaxis knowledge, improve comfort levels managing epistaxis, and improve confidence in successful epistaxis management. Nurses of varying clinical experience and different clinical settings benefitted equally from our intervention.

Keywords: nosebleed; in-service; quality improvement.

Epistaxis, or nosebleed, is estimated to be the chief complaint in 1 in 200 emergency department visits in the United States.¹ Additionally, it represents up to one-third of otolaryngology-related emergency room admissions.² There is no existing literature, to our best knowledge, specifically investigating the incidence of epistaxis after a patient is admitted. Anecdotally, inpatients who develop epistaxis account for an appreciable number of consults to otolaryngology (ENT). Epistaxis is a cross-disciplinary issue, occurring in a range of clinical settings. For example, patients with epistaxis can present to the emergency department or to an outpatient primary care clinic before being referred to ENT. Additionally, inpatients on many different services can develop spontaneous epistaxis due to a variety of environmental and iatrogenic factors, such as dry air, use of nasal cannula, and initiation of anticoagulation. Based on the experience of our ENT providers and discussions with our nursing colleagues, we concluded that there was an interest in epistaxis management training among our nursing workforce.

The presence of unmet demand for epistaxis education among our nursing colleagues was supported by our literature review. A study performed in England surveyed emergency department nurses on first aid measures for management of epistaxis, including ideal head positioning, location of pressure application, and duration of pressure application.³ Overall, only 12% to 14% of the nursing staff answered all 3 questions correctly.³ Additionally, 73% to 78% of the nursing staff felt that their training in epistaxis management was inadequate, and 88% desired further training in epistaxis management.³ If generalized, this study confirms the demand for further epistaxis education among nurses.

In-services have previously been shown to be effective educational tools within the nursing community. A study in Ethiopia that evaluated pain management knowledge and attitudes before and after an in-service found a significant improvement in mean rank score of nurses’ knowledge and attitudes regarding pain management after they participated in the in-service.⁴ Scores on the knowledge survey improved from 41.4% before the intervention to 63.0% post intervention.⁴ A study in Connecticut evaluated nurses’ confidence in discussing suicidal ideation with patients and knowledge surrounding suicide precautions.⁵ After participating in an in-service, nurses were significantly more confident in discussing suicidal ideation with patients; application of appropriate suicide precautions also increased after the in-service.⁵

Our aim was for nurses to have an improvement in overall epistaxis knowledge, perceived comfort level managing nosebleeds, and perceived ability to stop nosebleeds after attending our in-service. Additionally, an overarching priority was to provide high-quality epistaxis education based on the literature and best practice guidelines.

Methods

Setting

This study was carried out at an 811-bed quaternary care center located in Chicago, Illinois. In fiscal year 2021, there were 91 643 emergency department visits and 33 805 hospital admissions. At our flagship hospital, 2658 patients were diagnosed with epistaxis during fiscal year 2021. The emergency department saw 533 patients with epistaxis, with 342 requiring admission and 191 being discharged. Separately, 566 inpatients received a diagnosis of epistaxis during their admission. The remainder of the patients with epistaxis were seen on an outpatient basis.

Data Collection

Data were collected from nurses on 5 different inpatient units. An email with information about the in-service was sent to the nurse managers of the inpatient units. These 5 units were included because the nurse managers responded to the email and facilitated delivery of the in-service. Data collection took place from August to December 2020.

Intervention

A quality improvement team composed of a resident physician champion, nurse educators, and nurse managers was formed. The physician champion was a senior otolaryngology resident who was responsible for designing and administering the pre-test, in-service, and post test. The nurse educators and nurse managers helped coordinate times for the in-service and promoted the in-service for their staff.

Our intervention was an educational in-service, a technique that is commonly used at our institution for nurse education. In-services typically involve delivering a lecture on a clinically relevant topic to a group of nurses on a unit. In developing the in-service, a top priority was to present high-quality evidence-based material. There is an abundance of information in the literature surrounding epistaxis management. The clinical practice guideline published by the American Academy of Otolaryngology lists nasal compression, application of vasoconstrictors, nasal packing, and nasal cautery as first-line treatments for the management of epistaxis.⁶ Nasal packing and nasal cautery tend to be perceived as interventions that require a certain level of expertise and specialized supplies. As such, these interventions are not often performed by floor nurses. In contrast, nasal compression and application of vasoconstrictors require only a few easily accessible supplies, and the risks are relatively minimal. When performing nasal compression, the clinical practice guidelines recommend firm, sustained compression to the lower third of the nose for 5 minutes or longer.⁶ Topical vasoconstrictors are generally underutilized in epistaxis management. In a study looking at a random sample of all US emergency department visits from 1992 to 2001, only 18% of visits used an epistaxis-related medication.² Oxymetazoline hydrochloride is a topical vasoconstrictor that is commonly used as a nasal decongestant. However, its vasoconstrictor properties also make it a useful tool for controlling epistaxis. In a study looking at emergency department visits at the University of Texas Health Science Center, 65% of patients had resolution of nosebleed with application of oxymetazoline hydrochloride as the only intervention, with another 18% experiencing resolution of nosebleed with a combination of oxymetazoline hydrochloride and silver nitrate cautery.⁷ Based on review of the literature, nasal compression and application of vasoconstrictors seemed to be low-resource interventions with minimal morbidity. Therefore, management centered around nasal compression and use of topical vasoconstrictors seemed appropriate for our nursing staff.

The in-service included information about the etiology and management of epistaxis. Particular emphasis was placed on addressing and debunking common misconceptions about nosebleed management. With regards to management, our presentation focused on the use of topical vasoconstrictors and firm pressure to the lower third of the nose for at least 5 minutes. Nasal packing and nasal cautery were presented as procedures that ENT would perform. After the in-service, questions from the nurses were answered as time permitted.

Testing and Outcomes

A pre-test was administered before each in-service. The pre-test components comprised a knowledge survey and a descriptive survey. The general epistaxis knowledge questions on the pre-test included the location of blood vessels most commonly responsible for nosebleeds, the ideal positioning of a patient during a nosebleed, the appropriate location to hold pressure during a nosebleed, and the appropriate duration to hold pressure during a nosebleed. The descriptive survey portion asked nurses to rate whether they felt “very comfortable,” “comfortable,” “uncomfortable,” or “very uncomfortable” managing nosebleeds. It also asked whether nurses thought they would be able to “always,” “usually,” “rarely,” or “never” stop nosebleeds on the floor. We collected demographic information, including gender identity, years of clinical experience, and primary clinical environment.

The post test asked the same questions as the pre-test and was administered immediately after the in-service in order to assess its impact. We also established an ongoing dialogue with our nursing colleagues to obtain feedback on the sessions.

Primary outcomes of interest were the difference in general epistaxis knowledge questions answered correctly between the pre-test and the post test; the difference in comfort level in managing epistaxis before and after the in-service; and the difference in confidence to stop nosebleeds before and after the in-service. A secondary outcome was determining the audience for the in-service. Specifically, we wanted to determine whether there were different outcomes based on clinical setting or years of clinical experience. If nurses in a certain clinical environment or beyond a certain experience level did not show significant improvement from pre-test to post test, we would not target them for the in-service. Another secondary outcome was determining optimal timing for delivery of the in-service. We wanted to determine if there was a nursing preference for delivering the in-service at mid-shift vs shift change.

Analysis

Statistical calculations were performed using Stata 15 (StataCorp LLC). A P value < .05 was considered to be statistically significant. Where applicable, 95% confidence intervals (CI) were calculated. T-test was used to determine whether there was a statistically significant difference between pre-test and post-test epistaxis knowledge question scores. T-test was also used to determine whether there was a statistically significant difference in test scores between nurses receiving the in-service at mid-shift vs shift change. Pearson chi-squared tests were used to determine if there was a statistically significant difference between pre-test and post-test perceptions of epistaxis management, and to investigate outcomes between different subsets of nurses.

SQUIRE 2.0 guidelines were utilized to provide a framework for this project and to structure the manuscript.⁸ This study met criteria for exemption from institutional review board approval.

Results

Fifty-one nurses took part in this project (Table). The majority of participants identified as female (88.24%), and just over half worked on medical floors (52.94%), with most of the remainder working in intensive care (25.49%) and surgical (15.69%) settings. There was a wide range of clinical experience, with 1.96% reporting 0 to 1 years of experience, 29.41% reporting 2 to 5 years, 23.53% reporting 5 to 10 years, 25.49% reporting 10 to 20 years, and 17.65% reporting more than 20 years.

There were unanswered questions on both the pre-test and post test. There was no consistently unanswered question. Omitted answers on the epistaxis knowledge questions were recorded as an “incorrect” answer. Omitted answers on the perception questions were considered null values and not considered in final analysis.

Primary Measures

General epistaxis knowledge (Figure, part A) improved from the pre-test, where out of 4 questions, the mean (SD) score was 1.74 (1.02) correct questions, to the post-test, where out of 4 questions, the mean score was 3.80 (0.40) correct questions. After participating in the in-service, nurses answered significantly more questions about epistaxis general knowledge correctly (mean difference, 2.07 [1.10]; 95% CI, 1.74-2.39; P < .001), and 80.43% of them got a perfect score on the epistaxis knowledge questions.

The second primary measure was the difference in comfort level in managing nosebleed. After participating in the in-service, nurses felt significantly more comfortable in managing nosebleeds (Figure, part B; P = .007), with 74.46% of nurses having an improved comfort level managing nosebleeds. Before the in-service, 12.76% of nurses felt “very comfortable” in managing nosebleeds vs more than three-quarters (76.59%) after the in-service. Of those who answered that they felt “comfortable” managing nosebleeds on the pre-test, 82.35% improved to feeling “very comfortable” in managing nosebleeds. Before the in-service, 14.89% of nurses felt “uncomfortable” or “very uncomfortable” in managing nosebleeds, and this decreased to 0 post intervention. After the in-service, 100.00% of nurses felt “comfortable” or “very comfortable” in managing nosebleeds.

After receiving the in-service, nurses felt significantly more confident in stopping nosebleeds (Figure, part C; P < .001), with 43.90% of them having an improvement in confidence in stopping epistaxis. Before the in-service, 7.31% of nurses felt that they would “always” be able to stop a nose-bleed, and this increased to 41.46% after the in-service. Of those who answered that they felt that they would “usually” be able to stop a nosebleed on the pre-test, 36.67% changed their answer to state that they would “always” be able to stop a nosebleed on the post test. Before the in-service, 19.51% of nurses felt that they would “rarely” or “never” be able to stop a nosebleed, and this decreased to 2.44% after the in-service.

Secondary Measures

All of the nurses who participated either “strongly agreed” or “agreed” that they learned something new from the in-service. However, to determine whether there was a population who would benefit most from the in-service, we stratified the data by years of clinical experience. There was no statistically significant difference in whether nurses with varying clinical experience learned something new (P = .148): 100% of nurses with 0-1 years of experience, 80.00% of nurses with 2-5 years of experience, 100% of nurses with 5-10 years of experience, 69.23% of nurses with 10-20 years of experience, and 100% of nurses with >20 years of experience “strongly agreed” that they learned something new from this in-service. There was no statistically significant difference on the post test compared to the pre-test in additional correct questions when stratified by clinical experience (P = .128). Second, when we stratified by clinical setting, we did not find a statistically significant difference in whether nurses in different clinical settings learned something new (P = .929): 88.89% of nurses in the medical setting, 87.50% of nurses in the surgical setting, and 84.62% of nurses in the intensive care setting “strongly agreed” that they learned something new from this presentation. On investigating additional questions correct on the post test compared to the pre-test, there was no statistically significant difference in additional correct questions when stratified by clinical setting (P = .446).

Optimal timing of the in-service was another important outcome. Initially, the in-service was administered at mid-shift, with 9 nurses participating at mid-shift, but our nursing colleagues gave unanimous feedback that this was a suboptimal time for delivery of an in-service. We changed the timing of the in-service to shift change; 42 nurses received the in-service at shift-change. There was no statistically significant difference in scores on the epistaxis knowledge questions between these two groups (P = .123). This indicated to us that changing the timing of the delivery resulted in similarly improved outcomes while having the added benefit of being preferred by our nursing colleagues.

Discussion

In undertaking this project, our primary aims were to improve epistaxis knowledge and perceived management in our nursing staff. Among our nursing staff, we were able to significantly increase epistaxis knowledge, improve comfort levels managing epistaxis, and improve confidence in successful epistaxis management. We also found that nurses of varying clinical experience and different clinical settings benefited equally from our intervention. Using quality improvement principles, we optimized our delivery. Our in-service focused on educating nurses to use epistaxis management techniques that were resource-efficient and low risk.

After participating in the in-service, nurses answered significantly more questions about epistaxis general knowledge correctly (Figure, part A; mean difference, 2.07 questions [1.10]; 95% CI, 1.74-2.39; P < .001), felt significantly more comfortable in managing nosebleeds (Figure, part B; P = .007), and felt significantly more confident in stopping nosebleeds (Figure, part C; P < .001). Based on these results, we successfully achieved our primary aims.

Our secondary aim was to determine the audience that would benefit the most from the in-service. All of the nurses who participated either “strongly agreed” or “agreed” that they learned something new from the in-service. There was no statistically significant difference in whether nurses of varying clinical experience learned something new (P = .148) or in additional correct questions when stratified by clinical experience (P =.128). Also, there was no statistically significant difference in whether nurses in different clinical settings learned something new (P = .929) or in additional correct questions when stratified by clinical setting (P = .446). These results indicated to us that all participants learned something new and that there was no specific target audience, but rather that all participants benefitted from our session.

Our nursing colleagues gave us feedback that the timing of the in-service during mid-shift was not ideal. It was difficult to gather nurses mid-shift due to pressing patient-care duties. Nurses also found it difficult to give their full attention at this time. Nurses, nurse educators, and nurse managers suggested that we conduct the in-service at shift change in order to capture a larger population and take advantage of time relatively free of clinical duties. Giving the in-service at a time with relatively fewer clinical responsibilities allowed for a more robust question-and-answer session. It also allowed our nursing colleagues to pay full attention to the in-service. There was no statistically significant difference in epistaxis general knowledge questions answered correctly; this indicates that the quality of the education session did not vary greatly. However, our nursing colleagues strongly preferred the in-service at shift change. By making this modification to our intervention, we were able to optimize our intervention.

The previously mentioned study in England reported that only 12% to 14% of their nursing staff got a perfect score on epistaxis knowledge questions. Prior to our study, there was no literature investigating the impact of an in-service on epistaxis knowledge. After our intervention, 80.43% of our nurses got a perfect score on the epistaxis knowledge questions. We believe that this is a fair comparison because our post-test questions were identical to the survey questions used in the previously mentioned study in England, with the addition of one question.³ Further, the findings of our study are consistent with other studies regarding the positive effect of in-service education on knowledge and attitudes surrounding clinical topics. Similar to the study in Ethiopia investigating nurses’ knowledge surrounding pain management, our study noted a significant improvement in nurses’ knowledge after participating in the in-service.⁴ Also, when comparing our study to the study performed in Connecticut investigating nurses’ confidence surrounding suicide precautions, we found a similar significant improvement in confidence in management after participating in the in-service.⁵

Given our reliance on a survey as a tool to collect information, our study was subject to nonresponse bias. For each main outcome question, there was a handful of nonresponders. While this likely indicated either overlooking a question or deferring to answer due to clinical inexperience or nonapplicable clinical role, it is possible that this may have represented a respondent who did not benefit from the in-service. Another source of possible bias is sampling bias. Attempts were made to capture a wide range of nurses at the in-service. However, if a nurse was not interested in the topic material, whether due to abundant clinical experience or disinterest, it is possible that they may not have attended. Additionally, the cohort was selected purely based on responses from nursing managers to the initial email. It is possible that nonresponding units may have benefitted differently from this in-service.

There were several limitations within our analysis. We did not collect data assessing the long-term retention of epistaxis knowledge and management techniques. It is possible that epistaxis knowledge, comfort in managing nosebleeds, and perceived confidence in stopping nosebleeds decreased back to baseline several months after the in-service. Ideally, we would have been able to collect this data to assess retention of the in-service information. Unfortunately, a significant number of nurses who initially participated in the project became lost to follow-up, making such data collection impossible. Additionally, there was no assessment of actual ability to stop nosebleeds before vs after this in-service. Perceived management of epistaxis vs actual management of epistaxis are 2 vastly different things. However, this data would have been difficult to collect, and it likely would not have been in the best interest of patients, especially before the in-service was administered. As an improvement to this project, we could have assessed how many nosebleeds nurses had seen and successfully stopped after the in-service. As previously mentioned, this was not possible due to losing a significant number of nurses to follow-up. Finally, we did not collect objective data on preference for administration of in-service at mid-shift vs shift change. We relied on subjective data from conversations with our colleagues. By collecting objective data, we could have supported this change to our intervention with data.

The primary challenge to sustainability for this intervention is nursing turnover. With each wave of departing nurses and new nursing hires, the difficulty of ensuring a consistent knowledge base and management standards within our nursing workforce became clearer. After optimizing our intervention, our solution was to provide a hospital-wide in-service, which was recorded and uploaded to an institution-wide in-service library. In this way, a nurse with the desire to learn about epistaxis management could access the material at any point in time. Another solution would have been to appoint champions for epistaxis management within each major department to deliver the epistaxis in-service to new hires and new rotators within the department. However, given the turnover witnessed in our study cohort, this may not be sustainable long term.

Conclusion

Epistaxis is a chief complaint that can present in many different clinical settings and situations. Therefore, the ability to stop epistaxis in a timely and effective fashion is valuable. Our study demonstrated that in-services can improve epistaxis knowledge and improve perceived epistaxis management. Ideally, this intervention will lead to improved patient care. Given that epistaxis is a ubiquitous issue, this study may benefit other institutions who want to improve care for patients with epistaxis.

Next steps for this intervention include utilizing in-services for epistaxis education at other institutions and collecting long-term data within our own institution. Collecting long-term data would allow us to assess the retention of epistaxis knowledge from our in-service.

Acknowledgments: The author thanks the nurse managers, nurse educators, and staff nurses involved in this project, as well as Dr. Louis Portugal for providing mentorship throughout this process and Dr. Dara Adams for assisting with statistical analysis.

Corresponding author: Avery Nelson, MD, University of Chicago Medical Center, 5841 S Maryland Ave, MC 1035, Chicago, IL 60637; [email protected]

Disclosures: None reported.

From the University of Chicago Medical Center, Chicago, IL.

Abstract

Background: Epistaxis is a common chief complaint addressed by otolaryngologists. A review of the literature showed that there is a deficit in epistaxis education within the nursing community. Conversations with our nursing colleagues confirmed this unmet demand.

Objective: This quality improvement project aimed to increase general epistaxis knowledge, perceived comfort level managing nosebleeds, and perceived ability to stop nosebleeds among our nursing staff.

Methods: Data were collected through a survey administered before and after our intervention. The survey tested general epistaxis knowledge and assessed comfort and confidence in stopping epistaxis. Our intervention was an educational session covering pertinent epistaxis etiology and management. Quality improvement principles were used to optimize delivery of the intervention.

Results: A total of 51 nurses participated in the project. After participating in the in-service educational session, nurses answered significantly more epistaxis general knowledge questions correctly (mean [SD] difference, 2.07 [1.10] questions; 95% CI, 1.74-2.39; P < .001). There was no statistically significant difference in additional correct questions when stratified by clinical experience or clinical setting (P = .128 and P = 0.446, respectively). Nurses also reported feeling significantly more comfortable and significantly more confident in managing nosebleeds after the in-service (P = .007 and P < 0.001, respectively); 74.46% of nurses had an improvement in comfort level in managing epistaxis and 43.90% of nurses had an improvement in confidence in stopping epistaxis. After we moved the educational session from mid-shift to shift change, the nursing staff reported more satisfaction while maintaining similar improvements in knowledge and confidence.

Conclusion: We were able to significantly increase epistaxis knowledge, improve comfort levels managing epistaxis, and improve confidence in successful epistaxis management. Nurses of varying clinical experience and different clinical settings benefitted equally from our intervention.

Keywords: nosebleed; in-service; quality improvement.

Epistaxis, or nosebleed, is estimated to be the chief complaint in 1 in 200 emergency department visits in the United States.¹ Additionally, it represents up to one-third of otolaryngology-related emergency room admissions.² There is no existing literature, to our best knowledge, specifically investigating the incidence of epistaxis after a patient is admitted. Anecdotally, inpatients who develop epistaxis account for an appreciable number of consults to otolaryngology (ENT). Epistaxis is a cross-disciplinary issue, occurring in a range of clinical settings. For example, patients with epistaxis can present to the emergency department or to an outpatient primary care clinic before being referred to ENT. Additionally, inpatients on many different services can develop spontaneous epistaxis due to a variety of environmental and iatrogenic factors, such as dry air, use of nasal cannula, and initiation of anticoagulation. Based on the experience of our ENT providers and discussions with our nursing colleagues, we concluded that there was an interest in epistaxis management training among our nursing workforce.

The presence of unmet demand for epistaxis education among our nursing colleagues was supported by our literature review. A study performed in England surveyed emergency department nurses on first aid measures for management of epistaxis, including ideal head positioning, location of pressure application, and duration of pressure application.³ Overall, only 12% to 14% of the nursing staff answered all 3 questions correctly.³ Additionally, 73% to 78% of the nursing staff felt that their training in epistaxis management was inadequate, and 88% desired further training in epistaxis management.³ If generalized, this study confirms the demand for further epistaxis education among nurses.

In-services have previously been shown to be effective educational tools within the nursing community. A study in Ethiopia that evaluated pain management knowledge and attitudes before and after an in-service found a significant improvement in mean rank score of nurses’ knowledge and attitudes regarding pain management after they participated in the in-service.⁴ Scores on the knowledge survey improved from 41.4% before the intervention to 63.0% post intervention.⁴ A study in Connecticut evaluated nurses’ confidence in discussing suicidal ideation with patients and knowledge surrounding suicide precautions.⁵ After participating in an in-service, nurses were significantly more confident in discussing suicidal ideation with patients; application of appropriate suicide precautions also increased after the in-service.⁵

Our aim was for nurses to have an improvement in overall epistaxis knowledge, perceived comfort level managing nosebleeds, and perceived ability to stop nosebleeds after attending our in-service. Additionally, an overarching priority was to provide high-quality epistaxis education based on the literature and best practice guidelines.

Methods

Setting

This study was carried out at an 811-bed quaternary care center located in Chicago, Illinois. In fiscal year 2021, there were 91 643 emergency department visits and 33 805 hospital admissions. At our flagship hospital, 2658 patients were diagnosed with epistaxis during fiscal year 2021. The emergency department saw 533 patients with epistaxis, with 342 requiring admission and 191 being discharged. Separately, 566 inpatients received a diagnosis of epistaxis during their admission. The remainder of the patients with epistaxis were seen on an outpatient basis.

Data Collection

Data were collected from nurses on 5 different inpatient units. An email with information about the in-service was sent to the nurse managers of the inpatient units. These 5 units were included because the nurse managers responded to the email and facilitated delivery of the in-service. Data collection took place from August to December 2020.

Intervention

A quality improvement team composed of a resident physician champion, nurse educators, and nurse managers was formed. The physician champion was a senior otolaryngology resident who was responsible for designing and administering the pre-test, in-service, and post test. The nurse educators and nurse managers helped coordinate times for the in-service and promoted the in-service for their staff.

Our intervention was an educational in-service, a technique that is commonly used at our institution for nurse education. In-services typically involve delivering a lecture on a clinically relevant topic to a group of nurses on a unit. In developing the in-service, a top priority was to present high-quality evidence-based material. There is an abundance of information in the literature surrounding epistaxis management. The clinical practice guideline published by the American Academy of Otolaryngology lists nasal compression, application of vasoconstrictors, nasal packing, and nasal cautery as first-line treatments for the management of epistaxis.⁶ Nasal packing and nasal cautery tend to be perceived as interventions that require a certain level of expertise and specialized supplies. As such, these interventions are not often performed by floor nurses. In contrast, nasal compression and application of vasoconstrictors require only a few easily accessible supplies, and the risks are relatively minimal. When performing nasal compression, the clinical practice guidelines recommend firm, sustained compression to the lower third of the nose for 5 minutes or longer.⁶ Topical vasoconstrictors are generally underutilized in epistaxis management. In a study looking at a random sample of all US emergency department visits from 1992 to 2001, only 18% of visits used an epistaxis-related medication.² Oxymetazoline hydrochloride is a topical vasoconstrictor that is commonly used as a nasal decongestant. However, its vasoconstrictor properties also make it a useful tool for controlling epistaxis. In a study looking at emergency department visits at the University of Texas Health Science Center, 65% of patients had resolution of nosebleed with application of oxymetazoline hydrochloride as the only intervention, with another 18% experiencing resolution of nosebleed with a combination of oxymetazoline hydrochloride and silver nitrate cautery.⁷ Based on review of the literature, nasal compression and application of vasoconstrictors seemed to be low-resource interventions with minimal morbidity. Therefore, management centered around nasal compression and use of topical vasoconstrictors seemed appropriate for our nursing staff.

The in-service included information about the etiology and management of epistaxis. Particular emphasis was placed on addressing and debunking common misconceptions about nosebleed management. With regards to management, our presentation focused on the use of topical vasoconstrictors and firm pressure to the lower third of the nose for at least 5 minutes. Nasal packing and nasal cautery were presented as procedures that ENT would perform. After the in-service, questions from the nurses were answered as time permitted.

Testing and Outcomes

A pre-test was administered before each in-service. The pre-test components comprised a knowledge survey and a descriptive survey. The general epistaxis knowledge questions on the pre-test included the location of blood vessels most commonly responsible for nosebleeds, the ideal positioning of a patient during a nosebleed, the appropriate location to hold pressure during a nosebleed, and the appropriate duration to hold pressure during a nosebleed. The descriptive survey portion asked nurses to rate whether they felt “very comfortable,” “comfortable,” “uncomfortable,” or “very uncomfortable” managing nosebleeds. It also asked whether nurses thought they would be able to “always,” “usually,” “rarely,” or “never” stop nosebleeds on the floor. We collected demographic information, including gender identity, years of clinical experience, and primary clinical environment.

The post test asked the same questions as the pre-test and was administered immediately after the in-service in order to assess its impact. We also established an ongoing dialogue with our nursing colleagues to obtain feedback on the sessions.

Primary outcomes of interest were the difference in general epistaxis knowledge questions answered correctly between the pre-test and the post test; the difference in comfort level in managing epistaxis before and after the in-service; and the difference in confidence to stop nosebleeds before and after the in-service. A secondary outcome was determining the audience for the in-service. Specifically, we wanted to determine whether there were different outcomes based on clinical setting or years of clinical experience. If nurses in a certain clinical environment or beyond a certain experience level did not show significant improvement from pre-test to post test, we would not target them for the in-service. Another secondary outcome was determining optimal timing for delivery of the in-service. We wanted to determine if there was a nursing preference for delivering the in-service at mid-shift vs shift change.

Analysis

Statistical calculations were performed using Stata 15 (StataCorp LLC). A P value < .05 was considered to be statistically significant. Where applicable, 95% confidence intervals (CI) were calculated. T-test was used to determine whether there was a statistically significant difference between pre-test and post-test epistaxis knowledge question scores. T-test was also used to determine whether there was a statistically significant difference in test scores between nurses receiving the in-service at mid-shift vs shift change. Pearson chi-squared tests were used to determine if there was a statistically significant difference between pre-test and post-test perceptions of epistaxis management, and to investigate outcomes between different subsets of nurses.

SQUIRE 2.0 guidelines were utilized to provide a framework for this project and to structure the manuscript.⁸ This study met criteria for exemption from institutional review board approval.

Results

Fifty-one nurses took part in this project (Table). The majority of participants identified as female (88.24%), and just over half worked on medical floors (52.94%), with most of the remainder working in intensive care (25.49%) and surgical (15.69%) settings. There was a wide range of clinical experience, with 1.96% reporting 0 to 1 years of experience, 29.41% reporting 2 to 5 years, 23.53% reporting 5 to 10 years, 25.49% reporting 10 to 20 years, and 17.65% reporting more than 20 years.

There were unanswered questions on both the pre-test and post test. There was no consistently unanswered question. Omitted answers on the epistaxis knowledge questions were recorded as an “incorrect” answer. Omitted answers on the perception questions were considered null values and not considered in final analysis.

Primary Measures

General epistaxis knowledge (Figure, part A) improved from the pre-test, where out of 4 questions, the mean (SD) score was 1.74 (1.02) correct questions, to the post-test, where out of 4 questions, the mean score was 3.80 (0.40) correct questions. After participating in the in-service, nurses answered significantly more questions about epistaxis general knowledge correctly (mean difference, 2.07 [1.10]; 95% CI, 1.74-2.39; P < .001), and 80.43% of them got a perfect score on the epistaxis knowledge questions.

The second primary measure was the difference in comfort level in managing nosebleed. After participating in the in-service, nurses felt significantly more comfortable in managing nosebleeds (Figure, part B; P = .007), with 74.46% of nurses having an improved comfort level managing nosebleeds. Before the in-service, 12.76% of nurses felt “very comfortable” in managing nosebleeds vs more than three-quarters (76.59%) after the in-service. Of those who answered that they felt “comfortable” managing nosebleeds on the pre-test, 82.35% improved to feeling “very comfortable” in managing nosebleeds. Before the in-service, 14.89% of nurses felt “uncomfortable” or “very uncomfortable” in managing nosebleeds, and this decreased to 0 post intervention. After the in-service, 100.00% of nurses felt “comfortable” or “very comfortable” in managing nosebleeds.

After receiving the in-service, nurses felt significantly more confident in stopping nosebleeds (Figure, part C; P < .001), with 43.90% of them having an improvement in confidence in stopping epistaxis. Before the in-service, 7.31% of nurses felt that they would “always” be able to stop a nose-bleed, and this increased to 41.46% after the in-service. Of those who answered that they felt that they would “usually” be able to stop a nosebleed on the pre-test, 36.67% changed their answer to state that they would “always” be able to stop a nosebleed on the post test. Before the in-service, 19.51% of nurses felt that they would “rarely” or “never” be able to stop a nosebleed, and this decreased to 2.44% after the in-service.

Secondary Measures

All of the nurses who participated either “strongly agreed” or “agreed” that they learned something new from the in-service. However, to determine whether there was a population who would benefit most from the in-service, we stratified the data by years of clinical experience. There was no statistically significant difference in whether nurses with varying clinical experience learned something new (P = .148): 100% of nurses with 0-1 years of experience, 80.00% of nurses with 2-5 years of experience, 100% of nurses with 5-10 years of experience, 69.23% of nurses with 10-20 years of experience, and 100% of nurses with >20 years of experience “strongly agreed” that they learned something new from this in-service. There was no statistically significant difference on the post test compared to the pre-test in additional correct questions when stratified by clinical experience (P = .128). Second, when we stratified by clinical setting, we did not find a statistically significant difference in whether nurses in different clinical settings learned something new (P = .929): 88.89% of nurses in the medical setting, 87.50% of nurses in the surgical setting, and 84.62% of nurses in the intensive care setting “strongly agreed” that they learned something new from this presentation. On investigating additional questions correct on the post test compared to the pre-test, there was no statistically significant difference in additional correct questions when stratified by clinical setting (P = .446).

Optimal timing of the in-service was another important outcome. Initially, the in-service was administered at mid-shift, with 9 nurses participating at mid-shift, but our nursing colleagues gave unanimous feedback that this was a suboptimal time for delivery of an in-service. We changed the timing of the in-service to shift change; 42 nurses received the in-service at shift-change. There was no statistically significant difference in scores on the epistaxis knowledge questions between these two groups (P = .123). This indicated to us that changing the timing of the delivery resulted in similarly improved outcomes while having the added benefit of being preferred by our nursing colleagues.

Discussion

In undertaking this project, our primary aims were to improve epistaxis knowledge and perceived management in our nursing staff. Among our nursing staff, we were able to significantly increase epistaxis knowledge, improve comfort levels managing epistaxis, and improve confidence in successful epistaxis management. We also found that nurses of varying clinical experience and different clinical settings benefited equally from our intervention. Using quality improvement principles, we optimized our delivery. Our in-service focused on educating nurses to use epistaxis management techniques that were resource-efficient and low risk.

After participating in the in-service, nurses answered significantly more questions about epistaxis general knowledge correctly (Figure, part A; mean difference, 2.07 questions [1.10]; 95% CI, 1.74-2.39; P < .001), felt significantly more comfortable in managing nosebleeds (Figure, part B; P = .007), and felt significantly more confident in stopping nosebleeds (Figure, part C; P < .001). Based on these results, we successfully achieved our primary aims.

Our secondary aim was to determine the audience that would benefit the most from the in-service. All of the nurses who participated either “strongly agreed” or “agreed” that they learned something new from the in-service. There was no statistically significant difference in whether nurses of varying clinical experience learned something new (P = .148) or in additional correct questions when stratified by clinical experience (P =.128). Also, there was no statistically significant difference in whether nurses in different clinical settings learned something new (P = .929) or in additional correct questions when stratified by clinical setting (P = .446). These results indicated to us that all participants learned something new and that there was no specific target audience, but rather that all participants benefitted from our session.

Our nursing colleagues gave us feedback that the timing of the in-service during mid-shift was not ideal. It was difficult to gather nurses mid-shift due to pressing patient-care duties. Nurses also found it difficult to give their full attention at this time. Nurses, nurse educators, and nurse managers suggested that we conduct the in-service at shift change in order to capture a larger population and take advantage of time relatively free of clinical duties. Giving the in-service at a time with relatively fewer clinical responsibilities allowed for a more robust question-and-answer session. It also allowed our nursing colleagues to pay full attention to the in-service. There was no statistically significant difference in epistaxis general knowledge questions answered correctly; this indicates that the quality of the education session did not vary greatly. However, our nursing colleagues strongly preferred the in-service at shift change. By making this modification to our intervention, we were able to optimize our intervention.

The previously mentioned study in England reported that only 12% to 14% of their nursing staff got a perfect score on epistaxis knowledge questions. Prior to our study, there was no literature investigating the impact of an in-service on epistaxis knowledge. After our intervention, 80.43% of our nurses got a perfect score on the epistaxis knowledge questions. We believe that this is a fair comparison because our post-test questions were identical to the survey questions used in the previously mentioned study in England, with the addition of one question.³ Further, the findings of our study are consistent with other studies regarding the positive effect of in-service education on knowledge and attitudes surrounding clinical topics. Similar to the study in Ethiopia investigating nurses’ knowledge surrounding pain management, our study noted a significant improvement in nurses’ knowledge after participating in the in-service.⁴ Also, when comparing our study to the study performed in Connecticut investigating nurses’ confidence surrounding suicide precautions, we found a similar significant improvement in confidence in management after participating in the in-service.⁵

Given our reliance on a survey as a tool to collect information, our study was subject to nonresponse bias. For each main outcome question, there was a handful of nonresponders. While this likely indicated either overlooking a question or deferring to answer due to clinical inexperience or nonapplicable clinical role, it is possible that this may have represented a respondent who did not benefit from the in-service. Another source of possible bias is sampling bias. Attempts were made to capture a wide range of nurses at the in-service. However, if a nurse was not interested in the topic material, whether due to abundant clinical experience or disinterest, it is possible that they may not have attended. Additionally, the cohort was selected purely based on responses from nursing managers to the initial email. It is possible that nonresponding units may have benefitted differently from this in-service.

There were several limitations within our analysis. We did not collect data assessing the long-term retention of epistaxis knowledge and management techniques. It is possible that epistaxis knowledge, comfort in managing nosebleeds, and perceived confidence in stopping nosebleeds decreased back to baseline several months after the in-service. Ideally, we would have been able to collect this data to assess retention of the in-service information. Unfortunately, a significant number of nurses who initially participated in the project became lost to follow-up, making such data collection impossible. Additionally, there was no assessment of actual ability to stop nosebleeds before vs after this in-service. Perceived management of epistaxis vs actual management of epistaxis are 2 vastly different things. However, this data would have been difficult to collect, and it likely would not have been in the best interest of patients, especially before the in-service was administered. As an improvement to this project, we could have assessed how many nosebleeds nurses had seen and successfully stopped after the in-service. As previously mentioned, this was not possible due to losing a significant number of nurses to follow-up. Finally, we did not collect objective data on preference for administration of in-service at mid-shift vs shift change. We relied on subjective data from conversations with our colleagues. By collecting objective data, we could have supported this change to our intervention with data.

The primary challenge to sustainability for this intervention is nursing turnover. With each wave of departing nurses and new nursing hires, the difficulty of ensuring a consistent knowledge base and management standards within our nursing workforce became clearer. After optimizing our intervention, our solution was to provide a hospital-wide in-service, which was recorded and uploaded to an institution-wide in-service library. In this way, a nurse with the desire to learn about epistaxis management could access the material at any point in time. Another solution would have been to appoint champions for epistaxis management within each major department to deliver the epistaxis in-service to new hires and new rotators within the department. However, given the turnover witnessed in our study cohort, this may not be sustainable long term.

Conclusion

Epistaxis is a chief complaint that can present in many different clinical settings and situations. Therefore, the ability to stop epistaxis in a timely and effective fashion is valuable. Our study demonstrated that in-services can improve epistaxis knowledge and improve perceived epistaxis management. Ideally, this intervention will lead to improved patient care. Given that epistaxis is a ubiquitous issue, this study may benefit other institutions who want to improve care for patients with epistaxis.

Next steps for this intervention include utilizing in-services for epistaxis education at other institutions and collecting long-term data within our own institution. Collecting long-term data would allow us to assess the retention of epistaxis knowledge from our in-service.

Acknowledgments: The author thanks the nurse managers, nurse educators, and staff nurses involved in this project, as well as Dr. Louis Portugal for providing mentorship throughout this process and Dr. Dara Adams for assisting with statistical analysis.

Corresponding author: Avery Nelson, MD, University of Chicago Medical Center, 5841 S Maryland Ave, MC 1035, Chicago, IL 60637; [email protected]

Disclosures: None reported.

From the University of Chicago Medical Center, Chicago, IL.

Abstract

Background: Epistaxis is a common chief complaint addressed by otolaryngologists. A review of the literature showed that there is a deficit in epistaxis education within the nursing community. Conversations with our nursing colleagues confirmed this unmet demand.

Objective: This quality improvement project aimed to increase general epistaxis knowledge, perceived comfort level managing nosebleeds, and perceived ability to stop nosebleeds among our nursing staff.

Methods: Data were collected through a survey administered before and after our intervention. The survey tested general epistaxis knowledge and assessed comfort and confidence in stopping epistaxis. Our intervention was an educational session covering pertinent epistaxis etiology and management. Quality improvement principles were used to optimize delivery of the intervention.

Results: A total of 51 nurses participated in the project. After participating in the in-service educational session, nurses answered significantly more epistaxis general knowledge questions correctly (mean [SD] difference, 2.07 [1.10] questions; 95% CI, 1.74-2.39; P < .001). There was no statistically significant difference in additional correct questions when stratified by clinical experience or clinical setting (P = .128 and P = 0.446, respectively). Nurses also reported feeling significantly more comfortable and significantly more confident in managing nosebleeds after the in-service (P = .007 and P < 0.001, respectively); 74.46% of nurses had an improvement in comfort level in managing epistaxis and 43.90% of nurses had an improvement in confidence in stopping epistaxis. After we moved the educational session from mid-shift to shift change, the nursing staff reported more satisfaction while maintaining similar improvements in knowledge and confidence.

Conclusion: We were able to significantly increase epistaxis knowledge, improve comfort levels managing epistaxis, and improve confidence in successful epistaxis management. Nurses of varying clinical experience and different clinical settings benefitted equally from our intervention.

Keywords: nosebleed; in-service; quality improvement.

Epistaxis, or nosebleed, is estimated to be the chief complaint in 1 in 200 emergency department visits in the United States.¹ Additionally, it represents up to one-third of otolaryngology-related emergency room admissions.² There is no existing literature, to our best knowledge, specifically investigating the incidence of epistaxis after a patient is admitted. Anecdotally, inpatients who develop epistaxis account for an appreciable number of consults to otolaryngology (ENT). Epistaxis is a cross-disciplinary issue, occurring in a range of clinical settings. For example, patients with epistaxis can present to the emergency department or to an outpatient primary care clinic before being referred to ENT. Additionally, inpatients on many different services can develop spontaneous epistaxis due to a variety of environmental and iatrogenic factors, such as dry air, use of nasal cannula, and initiation of anticoagulation. Based on the experience of our ENT providers and discussions with our nursing colleagues, we concluded that there was an interest in epistaxis management training among our nursing workforce.

The presence of unmet demand for epistaxis education among our nursing colleagues was supported by our literature review. A study performed in England surveyed emergency department nurses on first aid measures for management of epistaxis, including ideal head positioning, location of pressure application, and duration of pressure application.³ Overall, only 12% to 14% of the nursing staff answered all 3 questions correctly.³ Additionally, 73% to 78% of the nursing staff felt that their training in epistaxis management was inadequate, and 88% desired further training in epistaxis management.³ If generalized, this study confirms the demand for further epistaxis education among nurses.

In-services have previously been shown to be effective educational tools within the nursing community. A study in Ethiopia that evaluated pain management knowledge and attitudes before and after an in-service found a significant improvement in mean rank score of nurses’ knowledge and attitudes regarding pain management after they participated in the in-service.⁴ Scores on the knowledge survey improved from 41.4% before the intervention to 63.0% post intervention.⁴ A study in Connecticut evaluated nurses’ confidence in discussing suicidal ideation with patients and knowledge surrounding suicide precautions.⁵ After participating in an in-service, nurses were significantly more confident in discussing suicidal ideation with patients; application of appropriate suicide precautions also increased after the in-service.⁵

Our aim was for nurses to have an improvement in overall epistaxis knowledge, perceived comfort level managing nosebleeds, and perceived ability to stop nosebleeds after attending our in-service. Additionally, an overarching priority was to provide high-quality epistaxis education based on the literature and best practice guidelines.

Methods

Setting

This study was carried out at an 811-bed quaternary care center located in Chicago, Illinois. In fiscal year 2021, there were 91 643 emergency department visits and 33 805 hospital admissions. At our flagship hospital, 2658 patients were diagnosed with epistaxis during fiscal year 2021. The emergency department saw 533 patients with epistaxis, with 342 requiring admission and 191 being discharged. Separately, 566 inpatients received a diagnosis of epistaxis during their admission. The remainder of the patients with epistaxis were seen on an outpatient basis.

Data Collection

Data were collected from nurses on 5 different inpatient units. An email with information about the in-service was sent to the nurse managers of the inpatient units. These 5 units were included because the nurse managers responded to the email and facilitated delivery of the in-service. Data collection took place from August to December 2020.

Intervention

A quality improvement team composed of a resident physician champion, nurse educators, and nurse managers was formed. The physician champion was a senior otolaryngology resident who was responsible for designing and administering the pre-test, in-service, and post test. The nurse educators and nurse managers helped coordinate times for the in-service and promoted the in-service for their staff.

Our intervention was an educational in-service, a technique that is commonly used at our institution for nurse education. In-services typically involve delivering a lecture on a clinically relevant topic to a group of nurses on a unit. In developing the in-service, a top priority was to present high-quality evidence-based material. There is an abundance of information in the literature surrounding epistaxis management. The clinical practice guideline published by the American Academy of Otolaryngology lists nasal compression, application of vasoconstrictors, nasal packing, and nasal cautery as first-line treatments for the management of epistaxis.⁶ Nasal packing and nasal cautery tend to be perceived as interventions that require a certain level of expertise and specialized supplies. As such, these interventions are not often performed by floor nurses. In contrast, nasal compression and application of vasoconstrictors require only a few easily accessible supplies, and the risks are relatively minimal. When performing nasal compression, the clinical practice guidelines recommend firm, sustained compression to the lower third of the nose for 5 minutes or longer.⁶ Topical vasoconstrictors are generally underutilized in epistaxis management. In a study looking at a random sample of all US emergency department visits from 1992 to 2001, only 18% of visits used an epistaxis-related medication.² Oxymetazoline hydrochloride is a topical vasoconstrictor that is commonly used as a nasal decongestant. However, its vasoconstrictor properties also make it a useful tool for controlling epistaxis. In a study looking at emergency department visits at the University of Texas Health Science Center, 65% of patients had resolution of nosebleed with application of oxymetazoline hydrochloride as the only intervention, with another 18% experiencing resolution of nosebleed with a combination of oxymetazoline hydrochloride and silver nitrate cautery.⁷ Based on review of the literature, nasal compression and application of vasoconstrictors seemed to be low-resource interventions with minimal morbidity. Therefore, management centered around nasal compression and use of topical vasoconstrictors seemed appropriate for our nursing staff.

The in-service included information about the etiology and management of epistaxis. Particular emphasis was placed on addressing and debunking common misconceptions about nosebleed management. With regards to management, our presentation focused on the use of topical vasoconstrictors and firm pressure to the lower third of the nose for at least 5 minutes. Nasal packing and nasal cautery were presented as procedures that ENT would perform. After the in-service, questions from the nurses were answered as time permitted.

Testing and Outcomes

A pre-test was administered before each in-service. The pre-test components comprised a knowledge survey and a descriptive survey. The general epistaxis knowledge questions on the pre-test included the location of blood vessels most commonly responsible for nosebleeds, the ideal positioning of a patient during a nosebleed, the appropriate location to hold pressure during a nosebleed, and the appropriate duration to hold pressure during a nosebleed. The descriptive survey portion asked nurses to rate whether they felt “very comfortable,” “comfortable,” “uncomfortable,” or “very uncomfortable” managing nosebleeds. It also asked whether nurses thought they would be able to “always,” “usually,” “rarely,” or “never” stop nosebleeds on the floor. We collected demographic information, including gender identity, years of clinical experience, and primary clinical environment.

The post test asked the same questions as the pre-test and was administered immediately after the in-service in order to assess its impact. We also established an ongoing dialogue with our nursing colleagues to obtain feedback on the sessions.

Primary outcomes of interest were the difference in general epistaxis knowledge questions answered correctly between the pre-test and the post test; the difference in comfort level in managing epistaxis before and after the in-service; and the difference in confidence to stop nosebleeds before and after the in-service. A secondary outcome was determining the audience for the in-service. Specifically, we wanted to determine whether there were different outcomes based on clinical setting or years of clinical experience. If nurses in a certain clinical environment or beyond a certain experience level did not show significant improvement from pre-test to post test, we would not target them for the in-service. Another secondary outcome was determining optimal timing for delivery of the in-service. We wanted to determine if there was a nursing preference for delivering the in-service at mid-shift vs shift change.

Analysis

Statistical calculations were performed using Stata 15 (StataCorp LLC). A P value < .05 was considered to be statistically significant. Where applicable, 95% confidence intervals (CI) were calculated. T-test was used to determine whether there was a statistically significant difference between pre-test and post-test epistaxis knowledge question scores. T-test was also used to determine whether there was a statistically significant difference in test scores between nurses receiving the in-service at mid-shift vs shift change. Pearson chi-squared tests were used to determine if there was a statistically significant difference between pre-test and post-test perceptions of epistaxis management, and to investigate outcomes between different subsets of nurses.

SQUIRE 2.0 guidelines were utilized to provide a framework for this project and to structure the manuscript.⁸ This study met criteria for exemption from institutional review board approval.

Results

Fifty-one nurses took part in this project (Table). The majority of participants identified as female (88.24%), and just over half worked on medical floors (52.94%), with most of the remainder working in intensive care (25.49%) and surgical (15.69%) settings. There was a wide range of clinical experience, with 1.96% reporting 0 to 1 years of experience, 29.41% reporting 2 to 5 years, 23.53% reporting 5 to 10 years, 25.49% reporting 10 to 20 years, and 17.65% reporting more than 20 years.

There were unanswered questions on both the pre-test and post test. There was no consistently unanswered question. Omitted answers on the epistaxis knowledge questions were recorded as an “incorrect” answer. Omitted answers on the perception questions were considered null values and not considered in final analysis.

Primary Measures

General epistaxis knowledge (Figure, part A) improved from the pre-test, where out of 4 questions, the mean (SD) score was 1.74 (1.02) correct questions, to the post-test, where out of 4 questions, the mean score was 3.80 (0.40) correct questions. After participating in the in-service, nurses answered significantly more questions about epistaxis general knowledge correctly (mean difference, 2.07 [1.10]; 95% CI, 1.74-2.39; P < .001), and 80.43% of them got a perfect score on the epistaxis knowledge questions.

The second primary measure was the difference in comfort level in managing nosebleed. After participating in the in-service, nurses felt significantly more comfortable in managing nosebleeds (Figure, part B; P = .007), with 74.46% of nurses having an improved comfort level managing nosebleeds. Before the in-service, 12.76% of nurses felt “very comfortable” in managing nosebleeds vs more than three-quarters (76.59%) after the in-service. Of those who answered that they felt “comfortable” managing nosebleeds on the pre-test, 82.35% improved to feeling “very comfortable” in managing nosebleeds. Before the in-service, 14.89% of nurses felt “uncomfortable” or “very uncomfortable” in managing nosebleeds, and this decreased to 0 post intervention. After the in-service, 100.00% of nurses felt “comfortable” or “very comfortable” in managing nosebleeds.

After receiving the in-service, nurses felt significantly more confident in stopping nosebleeds (Figure, part C; P < .001), with 43.90% of them having an improvement in confidence in stopping epistaxis. Before the in-service, 7.31% of nurses felt that they would “always” be able to stop a nose-bleed, and this increased to 41.46% after the in-service. Of those who answered that they felt that they would “usually” be able to stop a nosebleed on the pre-test, 36.67% changed their answer to state that they would “always” be able to stop a nosebleed on the post test. Before the in-service, 19.51% of nurses felt that they would “rarely” or “never” be able to stop a nosebleed, and this decreased to 2.44% after the in-service.

Secondary Measures

All of the nurses who participated either “strongly agreed” or “agreed” that they learned something new from the in-service. However, to determine whether there was a population who would benefit most from the in-service, we stratified the data by years of clinical experience. There was no statistically significant difference in whether nurses with varying clinical experience learned something new (P = .148): 100% of nurses with 0-1 years of experience, 80.00% of nurses with 2-5 years of experience, 100% of nurses with 5-10 years of experience, 69.23% of nurses with 10-20 years of experience, and 100% of nurses with >20 years of experience “strongly agreed” that they learned something new from this in-service. There was no statistically significant difference on the post test compared to the pre-test in additional correct questions when stratified by clinical experience (P = .128). Second, when we stratified by clinical setting, we did not find a statistically significant difference in whether nurses in different clinical settings learned something new (P = .929): 88.89% of nurses in the medical setting, 87.50% of nurses in the surgical setting, and 84.62% of nurses in the intensive care setting “strongly agreed” that they learned something new from this presentation. On investigating additional questions correct on the post test compared to the pre-test, there was no statistically significant difference in additional correct questions when stratified by clinical setting (P = .446).

Optimal timing of the in-service was another important outcome. Initially, the in-service was administered at mid-shift, with 9 nurses participating at mid-shift, but our nursing colleagues gave unanimous feedback that this was a suboptimal time for delivery of an in-service. We changed the timing of the in-service to shift change; 42 nurses received the in-service at shift-change. There was no statistically significant difference in scores on the epistaxis knowledge questions between these two groups (P = .123). This indicated to us that changing the timing of the delivery resulted in similarly improved outcomes while having the added benefit of being preferred by our nursing colleagues.

Discussion

In undertaking this project, our primary aims were to improve epistaxis knowledge and perceived management in our nursing staff. Among our nursing staff, we were able to significantly increase epistaxis knowledge, improve comfort levels managing epistaxis, and improve confidence in successful epistaxis management. We also found that nurses of varying clinical experience and different clinical settings benefited equally from our intervention. Using quality improvement principles, we optimized our delivery. Our in-service focused on educating nurses to use epistaxis management techniques that were resource-efficient and low risk.

After participating in the in-service, nurses answered significantly more questions about epistaxis general knowledge correctly (Figure, part A; mean difference, 2.07 questions [1.10]; 95% CI, 1.74-2.39; P < .001), felt significantly more comfortable in managing nosebleeds (Figure, part B; P = .007), and felt significantly more confident in stopping nosebleeds (Figure, part C; P < .001). Based on these results, we successfully achieved our primary aims.

Our secondary aim was to determine the audience that would benefit the most from the in-service. All of the nurses who participated either “strongly agreed” or “agreed” that they learned something new from the in-service. There was no statistically significant difference in whether nurses of varying clinical experience learned something new (P = .148) or in additional correct questions when stratified by clinical experience (P =.128). Also, there was no statistically significant difference in whether nurses in different clinical settings learned something new (P = .929) or in additional correct questions when stratified by clinical setting (P = .446). These results indicated to us that all participants learned something new and that there was no specific target audience, but rather that all participants benefitted from our session.

Our nursing colleagues gave us feedback that the timing of the in-service during mid-shift was not ideal. It was difficult to gather nurses mid-shift due to pressing patient-care duties. Nurses also found it difficult to give their full attention at this time. Nurses, nurse educators, and nurse managers suggested that we conduct the in-service at shift change in order to capture a larger population and take advantage of time relatively free of clinical duties. Giving the in-service at a time with relatively fewer clinical responsibilities allowed for a more robust question-and-answer session. It also allowed our nursing colleagues to pay full attention to the in-service. There was no statistically significant difference in epistaxis general knowledge questions answered correctly; this indicates that the quality of the education session did not vary greatly. However, our nursing colleagues strongly preferred the in-service at shift change. By making this modification to our intervention, we were able to optimize our intervention.

The previously mentioned study in England reported that only 12% to 14% of their nursing staff got a perfect score on epistaxis knowledge questions. Prior to our study, there was no literature investigating the impact of an in-service on epistaxis knowledge. After our intervention, 80.43% of our nurses got a perfect score on the epistaxis knowledge questions. We believe that this is a fair comparison because our post-test questions were identical to the survey questions used in the previously mentioned study in England, with the addition of one question.³ Further, the findings of our study are consistent with other studies regarding the positive effect of in-service education on knowledge and attitudes surrounding clinical topics. Similar to the study in Ethiopia investigating nurses’ knowledge surrounding pain management, our study noted a significant improvement in nurses’ knowledge after participating in the in-service.⁴ Also, when comparing our study to the study performed in Connecticut investigating nurses’ confidence surrounding suicide precautions, we found a similar significant improvement in confidence in management after participating in the in-service.⁵

Given our reliance on a survey as a tool to collect information, our study was subject to nonresponse bias. For each main outcome question, there was a handful of nonresponders. While this likely indicated either overlooking a question or deferring to answer due to clinical inexperience or nonapplicable clinical role, it is possible that this may have represented a respondent who did not benefit from the in-service. Another source of possible bias is sampling bias. Attempts were made to capture a wide range of nurses at the in-service. However, if a nurse was not interested in the topic material, whether due to abundant clinical experience or disinterest, it is possible that they may not have attended. Additionally, the cohort was selected purely based on responses from nursing managers to the initial email. It is possible that nonresponding units may have benefitted differently from this in-service.

There were several limitations within our analysis. We did not collect data assessing the long-term retention of epistaxis knowledge and management techniques. It is possible that epistaxis knowledge, comfort in managing nosebleeds, and perceived confidence in stopping nosebleeds decreased back to baseline several months after the in-service. Ideally, we would have been able to collect this data to assess retention of the in-service information. Unfortunately, a significant number of nurses who initially participated in the project became lost to follow-up, making such data collection impossible. Additionally, there was no assessment of actual ability to stop nosebleeds before vs after this in-service. Perceived management of epistaxis vs actual management of epistaxis are 2 vastly different things. However, this data would have been difficult to collect, and it likely would not have been in the best interest of patients, especially before the in-service was administered. As an improvement to this project, we could have assessed how many nosebleeds nurses had seen and successfully stopped after the in-service. As previously mentioned, this was not possible due to losing a significant number of nurses to follow-up. Finally, we did not collect objective data on preference for administration of in-service at mid-shift vs shift change. We relied on subjective data from conversations with our colleagues. By collecting objective data, we could have supported this change to our intervention with data.

The primary challenge to sustainability for this intervention is nursing turnover. With each wave of departing nurses and new nursing hires, the difficulty of ensuring a consistent knowledge base and management standards within our nursing workforce became clearer. After optimizing our intervention, our solution was to provide a hospital-wide in-service, which was recorded and uploaded to an institution-wide in-service library. In this way, a nurse with the desire to learn about epistaxis management could access the material at any point in time. Another solution would have been to appoint champions for epistaxis management within each major department to deliver the epistaxis in-service to new hires and new rotators within the department. However, given the turnover witnessed in our study cohort, this may not be sustainable long term.

Conclusion

Epistaxis is a chief complaint that can present in many different clinical settings and situations. Therefore, the ability to stop epistaxis in a timely and effective fashion is valuable. Our study demonstrated that in-services can improve epistaxis knowledge and improve perceived epistaxis management. Ideally, this intervention will lead to improved patient care. Given that epistaxis is a ubiquitous issue, this study may benefit other institutions who want to improve care for patients with epistaxis.

Next steps for this intervention include utilizing in-services for epistaxis education at other institutions and collecting long-term data within our own institution. Collecting long-term data would allow us to assess the retention of epistaxis knowledge from our in-service.

Acknowledgments: The author thanks the nurse managers, nurse educators, and staff nurses involved in this project, as well as Dr. Louis Portugal for providing mentorship throughout this process and Dr. Dara Adams for assisting with statistical analysis.

Corresponding author: Avery Nelson, MD, University of Chicago Medical Center, 5841 S Maryland Ave, MC 1035, Chicago, IL 60637; [email protected]

Disclosures: None reported.

Authors' Response

We agree with the valid comments made by Dr. Kerguelen and will respond to each set of questions in order.

Regarding the first set of questions on how we knew that our CMI was low and our patient acuity was under- represented, the University of Miami Health System is a designated cancer center with a Prospective Payment System exempt model (PPS exempt), and is one of 11 hospitals in the United States excluded for payment under the Inpatient Prospective Payment System. We know, therefore, that we care for a very complex patient population. Additionally, we benchmark ourselves against other academic medical centers (AMCs) with similarly complex patients and had noted that our patients appeared “less complex.” Specifically, our baseline CMI was 1.77 in early 2018 compared with an overall higher CMI for the AMC cohort; also, the total number of diagnoses we captured was lower than that in other AMCs. These 2 facts together alerted us that we likely had coding and clinical documentation improvement (CDI) opportunities. We recognized that our complexity was not being captured both because the clinical information was not documented in a manner readily translatable to ICD-10 codes and codes were missed when the documentation did exist. To remedy these problems, we implemented multiple immediate “fixes,” which included revamping our CDI efforts, re-education, and enhancements to our electronic health record for providers, CDIs, and coders. Since publication of our article, our CMI has continued to increase month over month, up to 2.57 most recently in May 2022, as we have continued to focus on several additional initiatives to impact both better documentation and coding.

The second set of questions asked whether the perceived low CMI was causing problems with payers and about the risk of artificially increasing the CMI through overdiagnosis as well as audit mechanisms to avoid this, and changes in expected mortality and observed mortality. To our knowledge, the lower CMI did not cause any problems with payers, but this is something we are currently tracking. Coding and documentation are constantly audited both internally (by our quality department) and externally (using Inter-Rater Reliability audits and validation), with no noted trend or targeted opportunities. We only include comorbidities that are current, actively monitored/managed, and pertinent to the care of our patients. We have not noted a change in denials, which gives us confidence we are not now overdiagnosing.

Our observed mortality has also increased. We, like all institutions, experienced the confounding factor of the COVID-19 pandemic, which coincided with the higher observed mortality over the course of the past 2 years. While the observed mortality (indicating sicker patients assuming no worsening of care processes) may partly explain our increased coding complexity, our decreasing mortality index (observed:expected mortality) suggests that our efforts to improve documentation and coding likely reflect improved capture of missed complexity (Figure).

We understand the concerns raised by Dr. Kerguelen about potential mis(over)coding. As part of this quality initiative, therefore, we plan long-term evaluations of our processes and metrics to better determine and guide our understanding of the impact of what we have already implemented and future interventions. In fact, we are in the process of analyzing additional interventions and hope to share results from these evaluations soon.

Marie Anne Sosa, MD
Tanira Ferreira, MD
Hayley Gershengorn, MD
Melissa Soto
Estin Kelly
Ameena Shrestha
Julianne Burgos
Sandeep Devabhaktuni
Dipen Parekh, MD
Maritza Suarez, MD
University of Miami Hospital and Clinics, Miami, FL
[email protected]

Disclosures: None reported.

Authors' Response

We agree with the valid comments made by Dr. Kerguelen and will respond to each set of questions in order.

Regarding the first set of questions on how we knew that our CMI was low and our patient acuity was under- represented, the University of Miami Health System is a designated cancer center with a Prospective Payment System exempt model (PPS exempt), and is one of 11 hospitals in the United States excluded for payment under the Inpatient Prospective Payment System. We know, therefore, that we care for a very complex patient population. Additionally, we benchmark ourselves against other academic medical centers (AMCs) with similarly complex patients and had noted that our patients appeared “less complex.” Specifically, our baseline CMI was 1.77 in early 2018 compared with an overall higher CMI for the AMC cohort; also, the total number of diagnoses we captured was lower than that in other AMCs. These 2 facts together alerted us that we likely had coding and clinical documentation improvement (CDI) opportunities. We recognized that our complexity was not being captured both because the clinical information was not documented in a manner readily translatable to ICD-10 codes and codes were missed when the documentation did exist. To remedy these problems, we implemented multiple immediate “fixes,” which included revamping our CDI efforts, re-education, and enhancements to our electronic health record for providers, CDIs, and coders. Since publication of our article, our CMI has continued to increase month over month, up to 2.57 most recently in May 2022, as we have continued to focus on several additional initiatives to impact both better documentation and coding.

The second set of questions asked whether the perceived low CMI was causing problems with payers and about the risk of artificially increasing the CMI through overdiagnosis as well as audit mechanisms to avoid this, and changes in expected mortality and observed mortality. To our knowledge, the lower CMI did not cause any problems with payers, but this is something we are currently tracking. Coding and documentation are constantly audited both internally (by our quality department) and externally (using Inter-Rater Reliability audits and validation), with no noted trend or targeted opportunities. We only include comorbidities that are current, actively monitored/managed, and pertinent to the care of our patients. We have not noted a change in denials, which gives us confidence we are not now overdiagnosing.

Our observed mortality has also increased. We, like all institutions, experienced the confounding factor of the COVID-19 pandemic, which coincided with the higher observed mortality over the course of the past 2 years. While the observed mortality (indicating sicker patients assuming no worsening of care processes) may partly explain our increased coding complexity, our decreasing mortality index (observed:expected mortality) suggests that our efforts to improve documentation and coding likely reflect improved capture of missed complexity (Figure).

We understand the concerns raised by Dr. Kerguelen about potential mis(over)coding. As part of this quality initiative, therefore, we plan long-term evaluations of our processes and metrics to better determine and guide our understanding of the impact of what we have already implemented and future interventions. In fact, we are in the process of analyzing additional interventions and hope to share results from these evaluations soon.

Marie Anne Sosa, MD
Tanira Ferreira, MD
Hayley Gershengorn, MD
Melissa Soto
Estin Kelly
Ameena Shrestha
Julianne Burgos
Sandeep Devabhaktuni
Dipen Parekh, MD
Maritza Suarez, MD
University of Miami Hospital and Clinics, Miami, FL
[email protected]

Disclosures: None reported.

Authors' Response

We agree with the valid comments made by Dr. Kerguelen and will respond to each set of questions in order.

Regarding the first set of questions on how we knew that our CMI was low and our patient acuity was under- represented, the University of Miami Health System is a designated cancer center with a Prospective Payment System exempt model (PPS exempt), and is one of 11 hospitals in the United States excluded for payment under the Inpatient Prospective Payment System. We know, therefore, that we care for a very complex patient population. Additionally, we benchmark ourselves against other academic medical centers (AMCs) with similarly complex patients and had noted that our patients appeared “less complex.” Specifically, our baseline CMI was 1.77 in early 2018 compared with an overall higher CMI for the AMC cohort; also, the total number of diagnoses we captured was lower than that in other AMCs. These 2 facts together alerted us that we likely had coding and clinical documentation improvement (CDI) opportunities. We recognized that our complexity was not being captured both because the clinical information was not documented in a manner readily translatable to ICD-10 codes and codes were missed when the documentation did exist. To remedy these problems, we implemented multiple immediate “fixes,” which included revamping our CDI efforts, re-education, and enhancements to our electronic health record for providers, CDIs, and coders. Since publication of our article, our CMI has continued to increase month over month, up to 2.57 most recently in May 2022, as we have continued to focus on several additional initiatives to impact both better documentation and coding.

The second set of questions asked whether the perceived low CMI was causing problems with payers and about the risk of artificially increasing the CMI through overdiagnosis as well as audit mechanisms to avoid this, and changes in expected mortality and observed mortality. To our knowledge, the lower CMI did not cause any problems with payers, but this is something we are currently tracking. Coding and documentation are constantly audited both internally (by our quality department) and externally (using Inter-Rater Reliability audits and validation), with no noted trend or targeted opportunities. We only include comorbidities that are current, actively monitored/managed, and pertinent to the care of our patients. We have not noted a change in denials, which gives us confidence we are not now overdiagnosing.

Our observed mortality has also increased. We, like all institutions, experienced the confounding factor of the COVID-19 pandemic, which coincided with the higher observed mortality over the course of the past 2 years. While the observed mortality (indicating sicker patients assuming no worsening of care processes) may partly explain our increased coding complexity, our decreasing mortality index (observed:expected mortality) suggests that our efforts to improve documentation and coding likely reflect improved capture of missed complexity (Figure).

We understand the concerns raised by Dr. Kerguelen about potential mis(over)coding. As part of this quality initiative, therefore, we plan long-term evaluations of our processes and metrics to better determine and guide our understanding of the impact of what we have already implemented and future interventions. In fact, we are in the process of analyzing additional interventions and hope to share results from these evaluations soon.

Marie Anne Sosa, MD
Tanira Ferreira, MD
Hayley Gershengorn, MD
Melissa Soto
Estin Kelly
Ameena Shrestha
Julianne Burgos
Sandeep Devabhaktuni
Dipen Parekh, MD
Maritza Suarez, MD
University of Miami Hospital and Clinics, Miami, FL
[email protected]

Disclosures: None reported.

To the Editor:

I read with interest the article by Sosa and colleagues¹ in which they present some stimulating analyses pertaining to a topic that we have been discussing at my institution for several years. Part of this discussion deals with the complexity of our hospital and how complexity is affected by comorbidity coding.

In 2013, we implemented the International Refined-DRGs (IR-DRGs) system to measure complexity at our hospital in Bogotá, Colombia. Our perception at that time was that the case mix index (CMI) was very low (0.7566), even for a general hospital with a high volume of pathologies with low relative weight (RW). Two medical auditors were assigned to review the medical records in order to improve the quality, quantity, and order of diagnoses. Emphasis was placed on patients with stays longer than 5 days and with only 1 diagnosis coded at admission. Additionally, International Classification of Diseases 10th Revision (World Health Organization version) diagnoses from chapters R (Symptoms and Signs Not Elsewhere Classified) and V through Y (External Causes) were blocked in the electronic health record. With these measures, our CMI increased 74%, reaching 1.3151 by the end of 2021, with a maximum peak of 1.6743 in May 2021, which coincided with the third peak of COVID-19 in Colombia.

However, the article by Sosa and colleagues draws my attention to the following: why do the authors state that their CMI is low and the patient acuity was under-represented? Is this due to a comparison with similar hospitals, or to a recommendation from a regulatory agency? We have found our CMI remains low because of a high volume of nonsurgical care (60%), deliveries, and digestive, respiratory, and urinary pathologies of low RW.

Also, was the perceived low CMI causing problems with payers? And further, how did the authors avoid the risk of artificially increasing the CMI through overdiagnosis of patients, and were there audit mechanisms to avoid this? While there was a clear change in expected mortality, did the observed mortality also change with the strategies implemented? This last question is relevant because, if the observed mortality were maintained, this would provide evidence that a coding problem was the cause of their hospital’s low CMI.

I reiterate my congratulations to the authors for presenting analyses that are very useful to other providers and researchers worldwide interested in addressing management issues related to the correct identification and classification of patients.

Carlos Kerguelen, MD, MA
Fundacion Santa Fe de Bogotá, Bogotá, Colombia
[email protected]

Disclosures: None reported.

To the Editor:

I read with interest the article by Sosa and colleagues¹ in which they present some stimulating analyses pertaining to a topic that we have been discussing at my institution for several years. Part of this discussion deals with the complexity of our hospital and how complexity is affected by comorbidity coding.

In 2013, we implemented the International Refined-DRGs (IR-DRGs) system to measure complexity at our hospital in Bogotá, Colombia. Our perception at that time was that the case mix index (CMI) was very low (0.7566), even for a general hospital with a high volume of pathologies with low relative weight (RW). Two medical auditors were assigned to review the medical records in order to improve the quality, quantity, and order of diagnoses. Emphasis was placed on patients with stays longer than 5 days and with only 1 diagnosis coded at admission. Additionally, International Classification of Diseases 10th Revision (World Health Organization version) diagnoses from chapters R (Symptoms and Signs Not Elsewhere Classified) and V through Y (External Causes) were blocked in the electronic health record. With these measures, our CMI increased 74%, reaching 1.3151 by the end of 2021, with a maximum peak of 1.6743 in May 2021, which coincided with the third peak of COVID-19 in Colombia.

However, the article by Sosa and colleagues draws my attention to the following: why do the authors state that their CMI is low and the patient acuity was under-represented? Is this due to a comparison with similar hospitals, or to a recommendation from a regulatory agency? We have found our CMI remains low because of a high volume of nonsurgical care (60%), deliveries, and digestive, respiratory, and urinary pathologies of low RW.

Also, was the perceived low CMI causing problems with payers? And further, how did the authors avoid the risk of artificially increasing the CMI through overdiagnosis of patients, and were there audit mechanisms to avoid this? While there was a clear change in expected mortality, did the observed mortality also change with the strategies implemented? This last question is relevant because, if the observed mortality were maintained, this would provide evidence that a coding problem was the cause of their hospital’s low CMI.

I reiterate my congratulations to the authors for presenting analyses that are very useful to other providers and researchers worldwide interested in addressing management issues related to the correct identification and classification of patients.

Carlos Kerguelen, MD, MA
Fundacion Santa Fe de Bogotá, Bogotá, Colombia
[email protected]

Disclosures: None reported.

To the Editor:

I read with interest the article by Sosa and colleagues¹ in which they present some stimulating analyses pertaining to a topic that we have been discussing at my institution for several years. Part of this discussion deals with the complexity of our hospital and how complexity is affected by comorbidity coding.

In 2013, we implemented the International Refined-DRGs (IR-DRGs) system to measure complexity at our hospital in Bogotá, Colombia. Our perception at that time was that the case mix index (CMI) was very low (0.7566), even for a general hospital with a high volume of pathologies with low relative weight (RW). Two medical auditors were assigned to review the medical records in order to improve the quality, quantity, and order of diagnoses. Emphasis was placed on patients with stays longer than 5 days and with only 1 diagnosis coded at admission. Additionally, International Classification of Diseases 10th Revision (World Health Organization version) diagnoses from chapters R (Symptoms and Signs Not Elsewhere Classified) and V through Y (External Causes) were blocked in the electronic health record. With these measures, our CMI increased 74%, reaching 1.3151 by the end of 2021, with a maximum peak of 1.6743 in May 2021, which coincided with the third peak of COVID-19 in Colombia.

However, the article by Sosa and colleagues draws my attention to the following: why do the authors state that their CMI is low and the patient acuity was under-represented? Is this due to a comparison with similar hospitals, or to a recommendation from a regulatory agency? We have found our CMI remains low because of a high volume of nonsurgical care (60%), deliveries, and digestive, respiratory, and urinary pathologies of low RW.

Also, was the perceived low CMI causing problems with payers? And further, how did the authors avoid the risk of artificially increasing the CMI through overdiagnosis of patients, and were there audit mechanisms to avoid this? While there was a clear change in expected mortality, did the observed mortality also change with the strategies implemented? This last question is relevant because, if the observed mortality were maintained, this would provide evidence that a coding problem was the cause of their hospital’s low CMI.

I reiterate my congratulations to the authors for presenting analyses that are very useful to other providers and researchers worldwide interested in addressing management issues related to the correct identification and classification of patients.

Carlos Kerguelen, MD, MA
Fundacion Santa Fe de Bogotá, Bogotá, Colombia
[email protected]

Disclosures: None reported.

From Memora Health (Dr. Flyckt and Dr. Colbert), San Francisco, CA; and Harvard Medical School (Dr. Colbert), Boston, MA.

A close relative was recently diagnosed with follicular lymphoma. He was cared for at a high-ranked cancer center by physicians with demonstrated expertise, and even had the support of a care navigator. Still, he was often left feeling overwhelmed and confused, holding an inch-thick stack of papers, instructions, and pamphlets. As he left his treatment planning visit, reeling from the emotional burden of his diagnosis and all the unfamiliar terminology, he didn’t know what to do or what to expect. Later, when he experienced early signs of tumor lysis syndrome, he struggled to reach his care team for triage and guidance. When he went to the emergency room, his oncologist was never informed.

This scenario is unfortunately common, and versions of this scenario play out thousands of times each day across the US health system. Within the clinic and hospital setting, patients receive excellent care from their providers, but a disconnect emerges once the patient leaves these medical settings: patients at home struggle to find guidance and support, while care teams lack the tools to engage patients between visits or monitor their health across care settings, providers, or episodes of care.

Leveraging Technology to Move From Episodes of Care to Complex Care Journeys

The use of automated messaging, artificial intelligence and natural language processing–driven chat experiences, and text-based support is becoming more common. However, health care lags behind other industries in the adoption of these technologies.^1,2 The slow pace can be warranted, given that health care is more complicated and higher risk than inquiring about a lost package, ordering groceries, or applying for a mortgage. At the same time, many of the consumer engagement tools used to guide an applicant through the multiple steps and complexities of their home loan process or to prompt viewers to select new shows to binge have applications in health care.

Over the past few years, technologies have emerged that guide patients through complex care journeys and allow care teams to monitor and engage patients between visits. These solutions come in different formats, but generally patients can receive messages on their phones that contain disease-specific educational content, prompts to fill prescriptions and take medications, and reminders and guidance on how to prepare for appointments and procedures. These programs also collect relevant data from patients through survey and electronic patient-reported outcomes instruments, as well as connected patient monitoring devices, that help track patient progress and identify issues as they arise. Many programs also incorporate symptom triage pathways and use natural language processing to respond automatically to patient questions and concerns.^3,4

These technology solutions can automate many tasks that in the past required a care team member to spend hours on the phone. Newly freed from such repetitive tasks, care teams can now focus on more in-depth interactions with those patients who are most in need—the types of interactions that are more satisfying and rewarding. Such assistance is particularly needed today with the staffing shortages faced by most health systems.⁵

In addition, technology allows teams to see the panel of patients they are caring for and to quickly identify and take action on any specific needs or issues. Care teams can focus on any patient and see where they are in their journey. When appropriate, some solutions also allow care teams to engage directly with patients through text-messaging, creating a seamless experience and unified communication channel. Ideally, these solutions should be linked or embedded within the electronic health record or other primary system of record, so that teams can easily access these tools through their existing workflows and avoid creating yet another interface to navigate.

The Impact of Low-Tech Solutions to Deliver High-Touch Support

There is evidence showing that digital patient navigation tools impact patient care. In the oncology setting, patients with a digital navigator have achieved over 95% adherence rates with complex oral chemotherapy regimens (Memora Health Unpublished Data. 2022.). In the postpartum setting, a text message–based program improved screening rates for postpartum depression and did so with very high patient satisfaction ratings.⁶ Particularly notable is the fact that this depression screening program achieved these results in a population that was predominantly low income, with more than half belonging to underrepresented minority populations.⁶

We believe these digital patient navigation technologies, specifically low-tech solutions that don’t require app downloads, portal log-ins, or high-speed internet, will transform care delivery over the next 5 to 10 years. Successful management of complex conditions like diabetes or cancer requires more than 3 hours of care each day,⁷ yet most patients spend only 1 or 2 hours per month directly interacting with their health care providers. However, most patients carry their phones with them at all times, and artificial intelligence–enabled text support is “always on” to provide support, monitoring, and guidance, wherever a patient happens to be when assistance is needed.

Shifting the Model to Support a Lifetime of Care

While still in the early stages of development, these tools have the potential to radically alter the practice of medicine, shifting the focus from episodic interactions to continuous journey-based care delivery. Outside of an acute event bringing a patient into the clinic or emergency room, many patients go a year or more without seeing their primary care providers.⁸ During that time, an immense amount of information is underreported or completely lost. Capturing this information in real-time and more holistically over a person’s lifetime of care could provide physicians better insight to both better manage and more fully evaluate the success of treatment plans by tracking patient symptoms, pain, and functional status over time. With this more longitudinal view of the patient, we see a pathway towards achieving the Quadruple Aim: patients who are more supported will achieve better outcomes at lower cost, they will have a better experience, and care teams will be empowered to focus their time on more satisfying activities rather than repetitive administrative tasks.

Corresponding author: James A. Colbert, MD, MBA; [email protected]

Disclosures: Dr. Flyckt and Dr. Colbert are employed by Memora Health, an organization that helps health care systems digitize and automate care journeys.

From Memora Health (Dr. Flyckt and Dr. Colbert), San Francisco, CA; and Harvard Medical School (Dr. Colbert), Boston, MA.

A close relative was recently diagnosed with follicular lymphoma. He was cared for at a high-ranked cancer center by physicians with demonstrated expertise, and even had the support of a care navigator. Still, he was often left feeling overwhelmed and confused, holding an inch-thick stack of papers, instructions, and pamphlets. As he left his treatment planning visit, reeling from the emotional burden of his diagnosis and all the unfamiliar terminology, he didn’t know what to do or what to expect. Later, when he experienced early signs of tumor lysis syndrome, he struggled to reach his care team for triage and guidance. When he went to the emergency room, his oncologist was never informed.

This scenario is unfortunately common, and versions of this scenario play out thousands of times each day across the US health system. Within the clinic and hospital setting, patients receive excellent care from their providers, but a disconnect emerges once the patient leaves these medical settings: patients at home struggle to find guidance and support, while care teams lack the tools to engage patients between visits or monitor their health across care settings, providers, or episodes of care.

Leveraging Technology to Move From Episodes of Care to Complex Care Journeys

The use of automated messaging, artificial intelligence and natural language processing–driven chat experiences, and text-based support is becoming more common. However, health care lags behind other industries in the adoption of these technologies.^1,2 The slow pace can be warranted, given that health care is more complicated and higher risk than inquiring about a lost package, ordering groceries, or applying for a mortgage. At the same time, many of the consumer engagement tools used to guide an applicant through the multiple steps and complexities of their home loan process or to prompt viewers to select new shows to binge have applications in health care.

Over the past few years, technologies have emerged that guide patients through complex care journeys and allow care teams to monitor and engage patients between visits. These solutions come in different formats, but generally patients can receive messages on their phones that contain disease-specific educational content, prompts to fill prescriptions and take medications, and reminders and guidance on how to prepare for appointments and procedures. These programs also collect relevant data from patients through survey and electronic patient-reported outcomes instruments, as well as connected patient monitoring devices, that help track patient progress and identify issues as they arise. Many programs also incorporate symptom triage pathways and use natural language processing to respond automatically to patient questions and concerns.^3,4

These technology solutions can automate many tasks that in the past required a care team member to spend hours on the phone. Newly freed from such repetitive tasks, care teams can now focus on more in-depth interactions with those patients who are most in need—the types of interactions that are more satisfying and rewarding. Such assistance is particularly needed today with the staffing shortages faced by most health systems.⁵

In addition, technology allows teams to see the panel of patients they are caring for and to quickly identify and take action on any specific needs or issues. Care teams can focus on any patient and see where they are in their journey. When appropriate, some solutions also allow care teams to engage directly with patients through text-messaging, creating a seamless experience and unified communication channel. Ideally, these solutions should be linked or embedded within the electronic health record or other primary system of record, so that teams can easily access these tools through their existing workflows and avoid creating yet another interface to navigate.

The Impact of Low-Tech Solutions to Deliver High-Touch Support

There is evidence showing that digital patient navigation tools impact patient care. In the oncology setting, patients with a digital navigator have achieved over 95% adherence rates with complex oral chemotherapy regimens (Memora Health Unpublished Data. 2022.). In the postpartum setting, a text message–based program improved screening rates for postpartum depression and did so with very high patient satisfaction ratings.⁶ Particularly notable is the fact that this depression screening program achieved these results in a population that was predominantly low income, with more than half belonging to underrepresented minority populations.⁶

We believe these digital patient navigation technologies, specifically low-tech solutions that don’t require app downloads, portal log-ins, or high-speed internet, will transform care delivery over the next 5 to 10 years. Successful management of complex conditions like diabetes or cancer requires more than 3 hours of care each day,⁷ yet most patients spend only 1 or 2 hours per month directly interacting with their health care providers. However, most patients carry their phones with them at all times, and artificial intelligence–enabled text support is “always on” to provide support, monitoring, and guidance, wherever a patient happens to be when assistance is needed.

Shifting the Model to Support a Lifetime of Care

While still in the early stages of development, these tools have the potential to radically alter the practice of medicine, shifting the focus from episodic interactions to continuous journey-based care delivery. Outside of an acute event bringing a patient into the clinic or emergency room, many patients go a year or more without seeing their primary care providers.⁸ During that time, an immense amount of information is underreported or completely lost. Capturing this information in real-time and more holistically over a person’s lifetime of care could provide physicians better insight to both better manage and more fully evaluate the success of treatment plans by tracking patient symptoms, pain, and functional status over time. With this more longitudinal view of the patient, we see a pathway towards achieving the Quadruple Aim: patients who are more supported will achieve better outcomes at lower cost, they will have a better experience, and care teams will be empowered to focus their time on more satisfying activities rather than repetitive administrative tasks.

Corresponding author: James A. Colbert, MD, MBA; [email protected]

Disclosures: Dr. Flyckt and Dr. Colbert are employed by Memora Health, an organization that helps health care systems digitize and automate care journeys.

From Memora Health (Dr. Flyckt and Dr. Colbert), San Francisco, CA; and Harvard Medical School (Dr. Colbert), Boston, MA.

A close relative was recently diagnosed with follicular lymphoma. He was cared for at a high-ranked cancer center by physicians with demonstrated expertise, and even had the support of a care navigator. Still, he was often left feeling overwhelmed and confused, holding an inch-thick stack of papers, instructions, and pamphlets. As he left his treatment planning visit, reeling from the emotional burden of his diagnosis and all the unfamiliar terminology, he didn’t know what to do or what to expect. Later, when he experienced early signs of tumor lysis syndrome, he struggled to reach his care team for triage and guidance. When he went to the emergency room, his oncologist was never informed.

This scenario is unfortunately common, and versions of this scenario play out thousands of times each day across the US health system. Within the clinic and hospital setting, patients receive excellent care from their providers, but a disconnect emerges once the patient leaves these medical settings: patients at home struggle to find guidance and support, while care teams lack the tools to engage patients between visits or monitor their health across care settings, providers, or episodes of care.

Leveraging Technology to Move From Episodes of Care to Complex Care Journeys

The use of automated messaging, artificial intelligence and natural language processing–driven chat experiences, and text-based support is becoming more common. However, health care lags behind other industries in the adoption of these technologies.^1,2 The slow pace can be warranted, given that health care is more complicated and higher risk than inquiring about a lost package, ordering groceries, or applying for a mortgage. At the same time, many of the consumer engagement tools used to guide an applicant through the multiple steps and complexities of their home loan process or to prompt viewers to select new shows to binge have applications in health care.

Over the past few years, technologies have emerged that guide patients through complex care journeys and allow care teams to monitor and engage patients between visits. These solutions come in different formats, but generally patients can receive messages on their phones that contain disease-specific educational content, prompts to fill prescriptions and take medications, and reminders and guidance on how to prepare for appointments and procedures. These programs also collect relevant data from patients through survey and electronic patient-reported outcomes instruments, as well as connected patient monitoring devices, that help track patient progress and identify issues as they arise. Many programs also incorporate symptom triage pathways and use natural language processing to respond automatically to patient questions and concerns.^3,4

These technology solutions can automate many tasks that in the past required a care team member to spend hours on the phone. Newly freed from such repetitive tasks, care teams can now focus on more in-depth interactions with those patients who are most in need—the types of interactions that are more satisfying and rewarding. Such assistance is particularly needed today with the staffing shortages faced by most health systems.⁵

In addition, technology allows teams to see the panel of patients they are caring for and to quickly identify and take action on any specific needs or issues. Care teams can focus on any patient and see where they are in their journey. When appropriate, some solutions also allow care teams to engage directly with patients through text-messaging, creating a seamless experience and unified communication channel. Ideally, these solutions should be linked or embedded within the electronic health record or other primary system of record, so that teams can easily access these tools through their existing workflows and avoid creating yet another interface to navigate.

The Impact of Low-Tech Solutions to Deliver High-Touch Support

There is evidence showing that digital patient navigation tools impact patient care. In the oncology setting, patients with a digital navigator have achieved over 95% adherence rates with complex oral chemotherapy regimens (Memora Health Unpublished Data. 2022.). In the postpartum setting, a text message–based program improved screening rates for postpartum depression and did so with very high patient satisfaction ratings.⁶ Particularly notable is the fact that this depression screening program achieved these results in a population that was predominantly low income, with more than half belonging to underrepresented minority populations.⁶

We believe these digital patient navigation technologies, specifically low-tech solutions that don’t require app downloads, portal log-ins, or high-speed internet, will transform care delivery over the next 5 to 10 years. Successful management of complex conditions like diabetes or cancer requires more than 3 hours of care each day,⁷ yet most patients spend only 1 or 2 hours per month directly interacting with their health care providers. However, most patients carry their phones with them at all times, and artificial intelligence–enabled text support is “always on” to provide support, monitoring, and guidance, wherever a patient happens to be when assistance is needed.

Shifting the Model to Support a Lifetime of Care

While still in the early stages of development, these tools have the potential to radically alter the practice of medicine, shifting the focus from episodic interactions to continuous journey-based care delivery. Outside of an acute event bringing a patient into the clinic or emergency room, many patients go a year or more without seeing their primary care providers.⁸ During that time, an immense amount of information is underreported or completely lost. Capturing this information in real-time and more holistically over a person’s lifetime of care could provide physicians better insight to both better manage and more fully evaluate the success of treatment plans by tracking patient symptoms, pain, and functional status over time. With this more longitudinal view of the patient, we see a pathway towards achieving the Quadruple Aim: patients who are more supported will achieve better outcomes at lower cost, they will have a better experience, and care teams will be empowered to focus their time on more satisfying activities rather than repetitive administrative tasks.

Corresponding author: James A. Colbert, MD, MBA; [email protected]

Disclosures: Dr. Flyckt and Dr. Colbert are employed by Memora Health, an organization that helps health care systems digitize and automate care journeys.

This issue of the Journal of Clinical Outcomes (JCOM) debuts a new cover design that brings forward the articles and features in each issue. Although the Journal’s cover has a new look, JCOM’s goals remain the same—improving care by disseminating evidence of quality improvement in health care and sharing access to the medical literature with our readers. We continue our mission to promote the best medical practice by providing clinicians with updates and communicating advances that lead to measurable improvement in health care delivery, quality, and outcomes.

As we continue the work of improving health care quality, knowledge gaps and unmet needs in the literature remain. These unmet needs are evident throughout all phases of health care delivery. Moreover, the Institutes of Medicine report that centered on efforts to build a safer health care environment by redesigning health care processes remains salient.¹ The journey to continuous improvement in health care, where we achieve threshold change in the quality of each process and across the entire health care system, requires collective effort. Such efforts include establishing clear metrics and measurements for improvement goals throughout the patient’s journey through diagnosis, treatment, transitions of care, and disease management.^2,3 To address evidence and knowledge gaps in the literature, JCOM publishes reports of original studies and quality improvement projects as well as reviews, providing its 30,000 readers with new evidence to implement in daily practice. We welcome submissions of original research reports, reports of quality improvement projects that follow the SQUIRE 2.0 standards,⁴ and perspectives on developments and innovations in health care delivery.

The next chapter in health care delivery improvement will encompass value-based care.⁵ This new era of clinical outcomes management will dictate the metrics and outcomes reporting⁶ and how to plan future investments. The value-based phase will increase innovation and shape policies that advance population health, transforming every step in the care delivery journey.⁷ The next phase in health care delivery will also create a viable financial structure while implementing effective performance measures for optimal outcomes through patient-centered care and optimization of cost and care strategies. In light of health care’s evolution toward a value-based model, JCOM welcomes submissions of manuscripts that explore themes central to this model, including patient-centered care, implementation of best practices, system design, safety, cost-effectiveness, and the balance between cost optimization and quality. For JCOM’s authors and readers, our editorial team remains commited to the highest standards in timely publishing to support our community through our collective expertise and dedication to quality improvement.

Corresponding author: Ebrahim Barkoudah, MD, MPH, Department of Medicine, Brigham and Women’s Hospital, Boston, MA; [email protected]

This issue of the Journal of Clinical Outcomes (JCOM) debuts a new cover design that brings forward the articles and features in each issue. Although the Journal’s cover has a new look, JCOM’s goals remain the same—improving care by disseminating evidence of quality improvement in health care and sharing access to the medical literature with our readers. We continue our mission to promote the best medical practice by providing clinicians with updates and communicating advances that lead to measurable improvement in health care delivery, quality, and outcomes.

As we continue the work of improving health care quality, knowledge gaps and unmet needs in the literature remain. These unmet needs are evident throughout all phases of health care delivery. Moreover, the Institutes of Medicine report that centered on efforts to build a safer health care environment by redesigning health care processes remains salient.¹ The journey to continuous improvement in health care, where we achieve threshold change in the quality of each process and across the entire health care system, requires collective effort. Such efforts include establishing clear metrics and measurements for improvement goals throughout the patient’s journey through diagnosis, treatment, transitions of care, and disease management.^2,3 To address evidence and knowledge gaps in the literature, JCOM publishes reports of original studies and quality improvement projects as well as reviews, providing its 30,000 readers with new evidence to implement in daily practice. We welcome submissions of original research reports, reports of quality improvement projects that follow the SQUIRE 2.0 standards,⁴ and perspectives on developments and innovations in health care delivery.

The next chapter in health care delivery improvement will encompass value-based care.⁵ This new era of clinical outcomes management will dictate the metrics and outcomes reporting⁶ and how to plan future investments. The value-based phase will increase innovation and shape policies that advance population health, transforming every step in the care delivery journey.⁷ The next phase in health care delivery will also create a viable financial structure while implementing effective performance measures for optimal outcomes through patient-centered care and optimization of cost and care strategies. In light of health care’s evolution toward a value-based model, JCOM welcomes submissions of manuscripts that explore themes central to this model, including patient-centered care, implementation of best practices, system design, safety, cost-effectiveness, and the balance between cost optimization and quality. For JCOM’s authors and readers, our editorial team remains commited to the highest standards in timely publishing to support our community through our collective expertise and dedication to quality improvement.

Corresponding author: Ebrahim Barkoudah, MD, MPH, Department of Medicine, Brigham and Women’s Hospital, Boston, MA; [email protected]

This issue of the Journal of Clinical Outcomes (JCOM) debuts a new cover design that brings forward the articles and features in each issue. Although the Journal’s cover has a new look, JCOM’s goals remain the same—improving care by disseminating evidence of quality improvement in health care and sharing access to the medical literature with our readers. We continue our mission to promote the best medical practice by providing clinicians with updates and communicating advances that lead to measurable improvement in health care delivery, quality, and outcomes.

As we continue the work of improving health care quality, knowledge gaps and unmet needs in the literature remain. These unmet needs are evident throughout all phases of health care delivery. Moreover, the Institutes of Medicine report that centered on efforts to build a safer health care environment by redesigning health care processes remains salient.¹ The journey to continuous improvement in health care, where we achieve threshold change in the quality of each process and across the entire health care system, requires collective effort. Such efforts include establishing clear metrics and measurements for improvement goals throughout the patient’s journey through diagnosis, treatment, transitions of care, and disease management.^2,3 To address evidence and knowledge gaps in the literature, JCOM publishes reports of original studies and quality improvement projects as well as reviews, providing its 30,000 readers with new evidence to implement in daily practice. We welcome submissions of original research reports, reports of quality improvement projects that follow the SQUIRE 2.0 standards,⁴ and perspectives on developments and innovations in health care delivery.

The next chapter in health care delivery improvement will encompass value-based care.⁵ This new era of clinical outcomes management will dictate the metrics and outcomes reporting⁶ and how to plan future investments. The value-based phase will increase innovation and shape policies that advance population health, transforming every step in the care delivery journey.⁷ The next phase in health care delivery will also create a viable financial structure while implementing effective performance measures for optimal outcomes through patient-centered care and optimization of cost and care strategies. In light of health care’s evolution toward a value-based model, JCOM welcomes submissions of manuscripts that explore themes central to this model, including patient-centered care, implementation of best practices, system design, safety, cost-effectiveness, and the balance between cost optimization and quality. For JCOM’s authors and readers, our editorial team remains commited to the highest standards in timely publishing to support our community through our collective expertise and dedication to quality improvement.

Corresponding author: Ebrahim Barkoudah, MD, MPH, Department of Medicine, Brigham and Women’s Hospital, Boston, MA; [email protected]

Study 1 Overview (Park et al)

Objective: To examine whether implementation of a geriatric trauma clinical pathway is associated with reduced rates of delirium in older adults with traumatic injury.

Design: Retrospective case-control study of electronic health records.

Setting and participants: Eligible patients were persons aged 65 years or older who were admitted to the trauma service and did not undergo an operation. A Geriatric Trauma Care Pathway was developed by a multidisciplinary Stanford Quality Pathways team and formally launched on November 1, 2018. The clinical pathway was designed to incorporate geriatric best practices, which included order sets (eg, age-appropriate nonpharmacological interventions and pharmacological dosages), guidelines (eg, Institute for Healthcare Improvement Age-Friendly Health systems 4M framework), automated consultations (comprehensive geriatric assessment), and escalation pathways executed by a multidisciplinary team (eg, pain, bowel, and sleep regulation). The clinical pathway began with admission to the emergency department (ED) (ie, automatic trigger of geriatric trauma care admission order set), daily multidisciplinary team meetings during acute hospitalization, and a transitional care team consultation for postdischarge follow-up or home visit.

Main outcome measures: The primary outcome was delirium as determined by a positive Confusion Assessment Method (CAM) score or a diagnosis of delirium by the clinical team. The secondary outcome was hospital length of stay (LOS). Process measures for pathway compliance (eg, achieving adequate pain control, early mobilization, advance care planning) were assessed. Outcome measures were compared between patients who underwent the Geriatric Trauma Care Pathway intervention (postimplementation group) vs patients who were treated prior to pathway implementation (baseline pre-implementation group).

Main results: Of the 859 eligible patients, 712 were included in the analysis (442 [62.1%] in the baseline pre-implementation group and 270 [37.9%] in the postimplementation group); mean (SD) age was 81.4 (9.1) years, and 394 (55.3%) were women. The injury mechanism was similar between groups, with falls being the most common cause of injury (247 [55.9%] in the baseline group vs 162 [60.0%] in the postimplementation group; P = .43). Injuries as measured by Injury Severity Score (ISS) were minor or moderate in both groups (261 [59.0%] in baseline group vs 168 [62.2%] in postimplementation group; P = .87). The adjusted odds ratio (OR) for delirium in the postimplementation group was lower compared to the baseline pre-implementation group (OR, 0.54; 95% CI, 0.37-0.80; P < .001). Measures of advance care planning in the postimplementation group improved, including more frequent goals-of-care documentation (53.7% in postimplementation group vs 16.7% in baseline group; P < .001) and a shortened time to first goals-of-care discussion upon presenting to the ED (36 hours in postimplementation group vs 50 hours in baseline group; P = .03).

Conclusion: Implementation of a multidisciplinary geriatric trauma clinical pathway for older adults with traumatic injury at a single level I trauma center was associated with reduced rates of delirium.

Study 2 Overview (Bryant et al)

Objective: To determine whether an interdisciplinary care pathway for frail trauma patients can improve in-hospital mortality, complications, and 30-day readmissions.

Design: Retrospective cohort study of frail patients.

Setting and participants: Eligible patients were persons aged 65 years or older who were admitted to the trauma service and survived more than 24 hours; admitted to and discharged from the trauma unit; and determined to be pre-frail or frail by a geriatrician’s assessment. A Frailty Identification and Care Pathway designed to reduce delirium and complications in frail older trauma patients was developed by a multidisciplinary team and implemented in 2016. The standardized evidence-based interdisciplinary care pathway included utilization of order sets and interventions for delirium prevention, early ambulation, bowel and pain regimens, nutrition and physical therapy consults, medication management, care-goal setting, and geriatric assessments.

Main outcome measures: The main outcomes were delirium as determined by a positive CAM score, major complications as defined by the Trauma Quality Improvement Project, in-hospital mortality, and 30-day hospital readmission. Outcome measures were compared between patients who underwent Frailty Identification and Care Pathway intervention (postintervention group) vs patients who were treated prior to pathway implementation (pre-intervention group).

Main results: A total of 269 frail patients were included in the analysis (125 in pre-intervention group vs 144 in postintervention group). Patient demographic and admission characteristics were similar between the 2 groups: mean age was 83.5 (7.1) years, 60.6% were women, and median ISS was 10 (interquartile range [IQR], 9-14). The injury mechanism was similar between groups, with falls accounting for 92.8% and 86.1% of injuries in the pre-intervention and postintervention groups, respectively (P = .07). In univariate analysis, the Frailty Identification and Care Pathway intervention was associated with a significant reduction in delirium (12.5% vs 21.6%, P = .04) and 30-day hospital readmission (2.7% vs 9.6%, P = .01) compared to patients in the pre-intervention group. However, rates of major complications (28.5% vs 28.0%, P = 0.93) and in-hospital mortality (4.2% vs 7.2%, P = .28) were similar between the pre-intervention and postintervention groups. In multivariate logistic regression models adjusted for patient characteristics (age, sex, race, ISS), patients in the postintervention group had lower delirium (OR, 0.44; 95% CI, 0.22-0.88; P = .02) and 30-day hospital readmission (OR, 0.25; 95% CI, 0.07-0.84; P = .02) rates compared to those in the pre-intervention group.

Conclusion: Implementation of an interdisciplinary care protocol for frail geriatric trauma patients significantly decreased their risks for in-hospital delirium and 30-day hospital readmission.

Commentary

Traumatic injuries in older adults are associated with higher morbidity and mortality compared to younger patients, with falls and motor vehicle accidents accounting for a majority of these injuries. Astoundingly, up to one-third of this vulnerable population presenting to hospitals with an ISS greater than 15 may die during hospitalization.¹ As a result, a large number of studies and clinical trials have focused on interventions that are designed to reduce fall risks, and hence reduce adverse consequences of traumatic injuries that may arise after falls.² However, this emphasis on falls prevention has overshadowed a need to develop effective geriatric-centered clinical interventions that aim to improve outcomes in older adults who present to hospitals with traumatic injuries. Furthermore, frailty—a geriatric syndrome indicative of an increased state of vulnerability and predictive of adverse outcomes such as delirium—is highly prevalent in older patients with traumatic injury.³ Thus, there is an urgent need to develop novel, hospital-based, traumatic injury–targeting strategies that incorporate a thoughtful redesign of the care framework that includes evidence-based interventions for geriatric syndromes such as delirium and frailty.

The study reported by Park et al (Study 1) represents the latest effort to evaluate inpatient management strategies designed to improve outcomes in hospitalized older adults who have sustained traumatic injury. Through the implementation of a novel multidisciplinary Geriatric Trauma Care Pathway that incorporates geriatric best practices, this intervention was found to be associated with a 46% lower risk of in-hospital delirium. Because of the inclusion of all age-eligible patients across all strata of traumatic injuries, rather than preselecting for those at the highest risk for poor clinical outcomes, the benefits of this intervention extend to those with minor or moderate injury severity. Furthermore, the improvement in delirium (ie, the primary outcome) is particularly meaningful given that delirium is one of the most common hospital-associated complications that increase hospital LOS, discharge to an institution, and mortality in older adults. Finally, the study’s observed reduced time to a first goals-of-care discussion and increased frequency of goals-of-care documentation after intervention should not be overlooked. The improvements in these 2 process measures are highly significant given that advanced care planning, an intervention that helps to align patients’ values, goals, and treatments, is completed at substantially lower rates in older adults in the acute hospital setting.⁴

Similarly, in an earlier published study, Bryant and colleagues (Study 2) also show that a geriatric-focused interdisciplinary trauma care pathway is associated with delirium risk reduction in hospitalized older trauma patients. Much like Study 1, the Frailty Identification and Care Pathway utilized in Study 2 is an evidence-based interdisciplinary care pathway that includes the use of geriatric assessments, order sets, and geriatric best practices. Moreover, its exclusive inclusion of pre-frail and frail older patients (ie, those at higher risk for poor outcomes) with moderate injury severity (median ISS of 10 [IQR, 9-14]) suggests that this type of care pathway benefits hospitalized older trauma patients, who are particularly vulnerable to adverse complications such as delirium. Moreover, the successful utilization of the FRAIL questionnaire, a validated frailty screening tool, by surgical residents in the ED to initiate this care pathway demonstrates the feasibility of its use in expediting frailty screening in older patients in trauma care.

Application for Clinical Practice and System Implementation

Findings from the 2 studies discussed in this review indicate that implementation of interdisciplinary clinical care pathways predicated on evidence-based geriatric principles and best practices is a promising approach to reduce delirium in hospitalized older trauma patients. These studies have helped to lay the groundwork in outlining the roadmaps (eg, processes and infrastructures) needed to create such clinical pathways. These key elements include: (1) integration of a multidisciplinary committee (eg, representation from trauma, emergency, and geriatric medicine, nursing, physical and occupational therapy, pharmacy, social work) in pathway design and implementation; (2) adaption of evidence-based geriatric best practices (eg, the Institute for Healthcare Improvement Age-Friendly Health System 4M framework [medication, mentation, mobility, what matters]) to prioritize interventions and to design a pathway that incorporates these features; (3) incorporation of comprehensive geriatric assessment by interdisciplinary providers; (4) utilization of validated clinical instruments to assess physical and cognitive functions, frailty, delirium, and social determinants of health; (5) modification of electronic health record systems to encompass order sets that incorporate evidence-based, nonpharmacological and pharmacological interventions to manage symptoms (eg, delirium, pain, bowel movement, sleep, immobility, polypharmacy) essential to quality geriatric care; and (6) integration of patient and caregiver preferences via goals-of-care discussions and corresponding documentation and communication of these goals.

Additionally, these 2 studies imparted some strategies that may facilitate the implementation of interdisciplinary clinical care pathways in trauma care. Examples of such facilitators include: (1) collaboration with champions within each specialty to reinforce education and buy-in; (2) creation of automatically triggered order sets upon patient presentation to the ED that unites distinct features of clinical pathways; (3) adaption and reorganization of existing hospital infrastructures and resources to meet the needs of clinical pathways implementation (eg, utilizing information technology resources to develop electronic health record order sets; using quality department to develop clinical pathway guidelines and electronic outcome dashboards); and (4) development of individualized patient and caregiver education materials based on care needs (eg, principles of delirium prevention and preservation of mobility during hospitalization) to prepare and engage these stakeholders in patient care and recovery.

Practice Points

• A geriatric interdisciplinary care model can be effectively applied to the management of acute trauma in older patients.
• Interdisciplinary clinical pathways should incorporate geriatric best practices and guidelines and age-appropriate order sets to prioritize and integrate care.

—Fred Ko, MD, MS

Study 1 Overview (Park et al)

Objective: To examine whether implementation of a geriatric trauma clinical pathway is associated with reduced rates of delirium in older adults with traumatic injury.

Design: Retrospective case-control study of electronic health records.

Setting and participants: Eligible patients were persons aged 65 years or older who were admitted to the trauma service and did not undergo an operation. A Geriatric Trauma Care Pathway was developed by a multidisciplinary Stanford Quality Pathways team and formally launched on November 1, 2018. The clinical pathway was designed to incorporate geriatric best practices, which included order sets (eg, age-appropriate nonpharmacological interventions and pharmacological dosages), guidelines (eg, Institute for Healthcare Improvement Age-Friendly Health systems 4M framework), automated consultations (comprehensive geriatric assessment), and escalation pathways executed by a multidisciplinary team (eg, pain, bowel, and sleep regulation). The clinical pathway began with admission to the emergency department (ED) (ie, automatic trigger of geriatric trauma care admission order set), daily multidisciplinary team meetings during acute hospitalization, and a transitional care team consultation for postdischarge follow-up or home visit.

Main outcome measures: The primary outcome was delirium as determined by a positive Confusion Assessment Method (CAM) score or a diagnosis of delirium by the clinical team. The secondary outcome was hospital length of stay (LOS). Process measures for pathway compliance (eg, achieving adequate pain control, early mobilization, advance care planning) were assessed. Outcome measures were compared between patients who underwent the Geriatric Trauma Care Pathway intervention (postimplementation group) vs patients who were treated prior to pathway implementation (baseline pre-implementation group).

Main results: Of the 859 eligible patients, 712 were included in the analysis (442 [62.1%] in the baseline pre-implementation group and 270 [37.9%] in the postimplementation group); mean (SD) age was 81.4 (9.1) years, and 394 (55.3%) were women. The injury mechanism was similar between groups, with falls being the most common cause of injury (247 [55.9%] in the baseline group vs 162 [60.0%] in the postimplementation group; P = .43). Injuries as measured by Injury Severity Score (ISS) were minor or moderate in both groups (261 [59.0%] in baseline group vs 168 [62.2%] in postimplementation group; P = .87). The adjusted odds ratio (OR) for delirium in the postimplementation group was lower compared to the baseline pre-implementation group (OR, 0.54; 95% CI, 0.37-0.80; P < .001). Measures of advance care planning in the postimplementation group improved, including more frequent goals-of-care documentation (53.7% in postimplementation group vs 16.7% in baseline group; P < .001) and a shortened time to first goals-of-care discussion upon presenting to the ED (36 hours in postimplementation group vs 50 hours in baseline group; P = .03).

Conclusion: Implementation of a multidisciplinary geriatric trauma clinical pathway for older adults with traumatic injury at a single level I trauma center was associated with reduced rates of delirium.

Study 2 Overview (Bryant et al)

Objective: To determine whether an interdisciplinary care pathway for frail trauma patients can improve in-hospital mortality, complications, and 30-day readmissions.

Design: Retrospective cohort study of frail patients.

Setting and participants: Eligible patients were persons aged 65 years or older who were admitted to the trauma service and survived more than 24 hours; admitted to and discharged from the trauma unit; and determined to be pre-frail or frail by a geriatrician’s assessment. A Frailty Identification and Care Pathway designed to reduce delirium and complications in frail older trauma patients was developed by a multidisciplinary team and implemented in 2016. The standardized evidence-based interdisciplinary care pathway included utilization of order sets and interventions for delirium prevention, early ambulation, bowel and pain regimens, nutrition and physical therapy consults, medication management, care-goal setting, and geriatric assessments.

Main outcome measures: The main outcomes were delirium as determined by a positive CAM score, major complications as defined by the Trauma Quality Improvement Project, in-hospital mortality, and 30-day hospital readmission. Outcome measures were compared between patients who underwent Frailty Identification and Care Pathway intervention (postintervention group) vs patients who were treated prior to pathway implementation (pre-intervention group).

Main results: A total of 269 frail patients were included in the analysis (125 in pre-intervention group vs 144 in postintervention group). Patient demographic and admission characteristics were similar between the 2 groups: mean age was 83.5 (7.1) years, 60.6% were women, and median ISS was 10 (interquartile range [IQR], 9-14). The injury mechanism was similar between groups, with falls accounting for 92.8% and 86.1% of injuries in the pre-intervention and postintervention groups, respectively (P = .07). In univariate analysis, the Frailty Identification and Care Pathway intervention was associated with a significant reduction in delirium (12.5% vs 21.6%, P = .04) and 30-day hospital readmission (2.7% vs 9.6%, P = .01) compared to patients in the pre-intervention group. However, rates of major complications (28.5% vs 28.0%, P = 0.93) and in-hospital mortality (4.2% vs 7.2%, P = .28) were similar between the pre-intervention and postintervention groups. In multivariate logistic regression models adjusted for patient characteristics (age, sex, race, ISS), patients in the postintervention group had lower delirium (OR, 0.44; 95% CI, 0.22-0.88; P = .02) and 30-day hospital readmission (OR, 0.25; 95% CI, 0.07-0.84; P = .02) rates compared to those in the pre-intervention group.

Conclusion: Implementation of an interdisciplinary care protocol for frail geriatric trauma patients significantly decreased their risks for in-hospital delirium and 30-day hospital readmission.

Commentary

Traumatic injuries in older adults are associated with higher morbidity and mortality compared to younger patients, with falls and motor vehicle accidents accounting for a majority of these injuries. Astoundingly, up to one-third of this vulnerable population presenting to hospitals with an ISS greater than 15 may die during hospitalization.¹ As a result, a large number of studies and clinical trials have focused on interventions that are designed to reduce fall risks, and hence reduce adverse consequences of traumatic injuries that may arise after falls.² However, this emphasis on falls prevention has overshadowed a need to develop effective geriatric-centered clinical interventions that aim to improve outcomes in older adults who present to hospitals with traumatic injuries. Furthermore, frailty—a geriatric syndrome indicative of an increased state of vulnerability and predictive of adverse outcomes such as delirium—is highly prevalent in older patients with traumatic injury.³ Thus, there is an urgent need to develop novel, hospital-based, traumatic injury–targeting strategies that incorporate a thoughtful redesign of the care framework that includes evidence-based interventions for geriatric syndromes such as delirium and frailty.

The study reported by Park et al (Study 1) represents the latest effort to evaluate inpatient management strategies designed to improve outcomes in hospitalized older adults who have sustained traumatic injury. Through the implementation of a novel multidisciplinary Geriatric Trauma Care Pathway that incorporates geriatric best practices, this intervention was found to be associated with a 46% lower risk of in-hospital delirium. Because of the inclusion of all age-eligible patients across all strata of traumatic injuries, rather than preselecting for those at the highest risk for poor clinical outcomes, the benefits of this intervention extend to those with minor or moderate injury severity. Furthermore, the improvement in delirium (ie, the primary outcome) is particularly meaningful given that delirium is one of the most common hospital-associated complications that increase hospital LOS, discharge to an institution, and mortality in older adults. Finally, the study’s observed reduced time to a first goals-of-care discussion and increased frequency of goals-of-care documentation after intervention should not be overlooked. The improvements in these 2 process measures are highly significant given that advanced care planning, an intervention that helps to align patients’ values, goals, and treatments, is completed at substantially lower rates in older adults in the acute hospital setting.⁴

Similarly, in an earlier published study, Bryant and colleagues (Study 2) also show that a geriatric-focused interdisciplinary trauma care pathway is associated with delirium risk reduction in hospitalized older trauma patients. Much like Study 1, the Frailty Identification and Care Pathway utilized in Study 2 is an evidence-based interdisciplinary care pathway that includes the use of geriatric assessments, order sets, and geriatric best practices. Moreover, its exclusive inclusion of pre-frail and frail older patients (ie, those at higher risk for poor outcomes) with moderate injury severity (median ISS of 10 [IQR, 9-14]) suggests that this type of care pathway benefits hospitalized older trauma patients, who are particularly vulnerable to adverse complications such as delirium. Moreover, the successful utilization of the FRAIL questionnaire, a validated frailty screening tool, by surgical residents in the ED to initiate this care pathway demonstrates the feasibility of its use in expediting frailty screening in older patients in trauma care.

Application for Clinical Practice and System Implementation

Findings from the 2 studies discussed in this review indicate that implementation of interdisciplinary clinical care pathways predicated on evidence-based geriatric principles and best practices is a promising approach to reduce delirium in hospitalized older trauma patients. These studies have helped to lay the groundwork in outlining the roadmaps (eg, processes and infrastructures) needed to create such clinical pathways. These key elements include: (1) integration of a multidisciplinary committee (eg, representation from trauma, emergency, and geriatric medicine, nursing, physical and occupational therapy, pharmacy, social work) in pathway design and implementation; (2) adaption of evidence-based geriatric best practices (eg, the Institute for Healthcare Improvement Age-Friendly Health System 4M framework [medication, mentation, mobility, what matters]) to prioritize interventions and to design a pathway that incorporates these features; (3) incorporation of comprehensive geriatric assessment by interdisciplinary providers; (4) utilization of validated clinical instruments to assess physical and cognitive functions, frailty, delirium, and social determinants of health; (5) modification of electronic health record systems to encompass order sets that incorporate evidence-based, nonpharmacological and pharmacological interventions to manage symptoms (eg, delirium, pain, bowel movement, sleep, immobility, polypharmacy) essential to quality geriatric care; and (6) integration of patient and caregiver preferences via goals-of-care discussions and corresponding documentation and communication of these goals.

Additionally, these 2 studies imparted some strategies that may facilitate the implementation of interdisciplinary clinical care pathways in trauma care. Examples of such facilitators include: (1) collaboration with champions within each specialty to reinforce education and buy-in; (2) creation of automatically triggered order sets upon patient presentation to the ED that unites distinct features of clinical pathways; (3) adaption and reorganization of existing hospital infrastructures and resources to meet the needs of clinical pathways implementation (eg, utilizing information technology resources to develop electronic health record order sets; using quality department to develop clinical pathway guidelines and electronic outcome dashboards); and (4) development of individualized patient and caregiver education materials based on care needs (eg, principles of delirium prevention and preservation of mobility during hospitalization) to prepare and engage these stakeholders in patient care and recovery.

Practice Points

• A geriatric interdisciplinary care model can be effectively applied to the management of acute trauma in older patients.
• Interdisciplinary clinical pathways should incorporate geriatric best practices and guidelines and age-appropriate order sets to prioritize and integrate care.

—Fred Ko, MD, MS

Study 1 Overview (Park et al)

Objective: To examine whether implementation of a geriatric trauma clinical pathway is associated with reduced rates of delirium in older adults with traumatic injury.

Design: Retrospective case-control study of electronic health records.

Setting and participants: Eligible patients were persons aged 65 years or older who were admitted to the trauma service and did not undergo an operation. A Geriatric Trauma Care Pathway was developed by a multidisciplinary Stanford Quality Pathways team and formally launched on November 1, 2018. The clinical pathway was designed to incorporate geriatric best practices, which included order sets (eg, age-appropriate nonpharmacological interventions and pharmacological dosages), guidelines (eg, Institute for Healthcare Improvement Age-Friendly Health systems 4M framework), automated consultations (comprehensive geriatric assessment), and escalation pathways executed by a multidisciplinary team (eg, pain, bowel, and sleep regulation). The clinical pathway began with admission to the emergency department (ED) (ie, automatic trigger of geriatric trauma care admission order set), daily multidisciplinary team meetings during acute hospitalization, and a transitional care team consultation for postdischarge follow-up or home visit.

Main outcome measures: The primary outcome was delirium as determined by a positive Confusion Assessment Method (CAM) score or a diagnosis of delirium by the clinical team. The secondary outcome was hospital length of stay (LOS). Process measures for pathway compliance (eg, achieving adequate pain control, early mobilization, advance care planning) were assessed. Outcome measures were compared between patients who underwent the Geriatric Trauma Care Pathway intervention (postimplementation group) vs patients who were treated prior to pathway implementation (baseline pre-implementation group).

Main results: Of the 859 eligible patients, 712 were included in the analysis (442 [62.1%] in the baseline pre-implementation group and 270 [37.9%] in the postimplementation group); mean (SD) age was 81.4 (9.1) years, and 394 (55.3%) were women. The injury mechanism was similar between groups, with falls being the most common cause of injury (247 [55.9%] in the baseline group vs 162 [60.0%] in the postimplementation group; P = .43). Injuries as measured by Injury Severity Score (ISS) were minor or moderate in both groups (261 [59.0%] in baseline group vs 168 [62.2%] in postimplementation group; P = .87). The adjusted odds ratio (OR) for delirium in the postimplementation group was lower compared to the baseline pre-implementation group (OR, 0.54; 95% CI, 0.37-0.80; P < .001). Measures of advance care planning in the postimplementation group improved, including more frequent goals-of-care documentation (53.7% in postimplementation group vs 16.7% in baseline group; P < .001) and a shortened time to first goals-of-care discussion upon presenting to the ED (36 hours in postimplementation group vs 50 hours in baseline group; P = .03).

Conclusion: Implementation of a multidisciplinary geriatric trauma clinical pathway for older adults with traumatic injury at a single level I trauma center was associated with reduced rates of delirium.

Study 2 Overview (Bryant et al)

Objective: To determine whether an interdisciplinary care pathway for frail trauma patients can improve in-hospital mortality, complications, and 30-day readmissions.

Design: Retrospective cohort study of frail patients.

Setting and participants: Eligible patients were persons aged 65 years or older who were admitted to the trauma service and survived more than 24 hours; admitted to and discharged from the trauma unit; and determined to be pre-frail or frail by a geriatrician’s assessment. A Frailty Identification and Care Pathway designed to reduce delirium and complications in frail older trauma patients was developed by a multidisciplinary team and implemented in 2016. The standardized evidence-based interdisciplinary care pathway included utilization of order sets and interventions for delirium prevention, early ambulation, bowel and pain regimens, nutrition and physical therapy consults, medication management, care-goal setting, and geriatric assessments.

Main outcome measures: The main outcomes were delirium as determined by a positive CAM score, major complications as defined by the Trauma Quality Improvement Project, in-hospital mortality, and 30-day hospital readmission. Outcome measures were compared between patients who underwent Frailty Identification and Care Pathway intervention (postintervention group) vs patients who were treated prior to pathway implementation (pre-intervention group).

Main results: A total of 269 frail patients were included in the analysis (125 in pre-intervention group vs 144 in postintervention group). Patient demographic and admission characteristics were similar between the 2 groups: mean age was 83.5 (7.1) years, 60.6% were women, and median ISS was 10 (interquartile range [IQR], 9-14). The injury mechanism was similar between groups, with falls accounting for 92.8% and 86.1% of injuries in the pre-intervention and postintervention groups, respectively (P = .07). In univariate analysis, the Frailty Identification and Care Pathway intervention was associated with a significant reduction in delirium (12.5% vs 21.6%, P = .04) and 30-day hospital readmission (2.7% vs 9.6%, P = .01) compared to patients in the pre-intervention group. However, rates of major complications (28.5% vs 28.0%, P = 0.93) and in-hospital mortality (4.2% vs 7.2%, P = .28) were similar between the pre-intervention and postintervention groups. In multivariate logistic regression models adjusted for patient characteristics (age, sex, race, ISS), patients in the postintervention group had lower delirium (OR, 0.44; 95% CI, 0.22-0.88; P = .02) and 30-day hospital readmission (OR, 0.25; 95% CI, 0.07-0.84; P = .02) rates compared to those in the pre-intervention group.

Conclusion: Implementation of an interdisciplinary care protocol for frail geriatric trauma patients significantly decreased their risks for in-hospital delirium and 30-day hospital readmission.

Commentary

Traumatic injuries in older adults are associated with higher morbidity and mortality compared to younger patients, with falls and motor vehicle accidents accounting for a majority of these injuries. Astoundingly, up to one-third of this vulnerable population presenting to hospitals with an ISS greater than 15 may die during hospitalization.¹ As a result, a large number of studies and clinical trials have focused on interventions that are designed to reduce fall risks, and hence reduce adverse consequences of traumatic injuries that may arise after falls.² However, this emphasis on falls prevention has overshadowed a need to develop effective geriatric-centered clinical interventions that aim to improve outcomes in older adults who present to hospitals with traumatic injuries. Furthermore, frailty—a geriatric syndrome indicative of an increased state of vulnerability and predictive of adverse outcomes such as delirium—is highly prevalent in older patients with traumatic injury.³ Thus, there is an urgent need to develop novel, hospital-based, traumatic injury–targeting strategies that incorporate a thoughtful redesign of the care framework that includes evidence-based interventions for geriatric syndromes such as delirium and frailty.

The study reported by Park et al (Study 1) represents the latest effort to evaluate inpatient management strategies designed to improve outcomes in hospitalized older adults who have sustained traumatic injury. Through the implementation of a novel multidisciplinary Geriatric Trauma Care Pathway that incorporates geriatric best practices, this intervention was found to be associated with a 46% lower risk of in-hospital delirium. Because of the inclusion of all age-eligible patients across all strata of traumatic injuries, rather than preselecting for those at the highest risk for poor clinical outcomes, the benefits of this intervention extend to those with minor or moderate injury severity. Furthermore, the improvement in delirium (ie, the primary outcome) is particularly meaningful given that delirium is one of the most common hospital-associated complications that increase hospital LOS, discharge to an institution, and mortality in older adults. Finally, the study’s observed reduced time to a first goals-of-care discussion and increased frequency of goals-of-care documentation after intervention should not be overlooked. The improvements in these 2 process measures are highly significant given that advanced care planning, an intervention that helps to align patients’ values, goals, and treatments, is completed at substantially lower rates in older adults in the acute hospital setting.⁴

Similarly, in an earlier published study, Bryant and colleagues (Study 2) also show that a geriatric-focused interdisciplinary trauma care pathway is associated with delirium risk reduction in hospitalized older trauma patients. Much like Study 1, the Frailty Identification and Care Pathway utilized in Study 2 is an evidence-based interdisciplinary care pathway that includes the use of geriatric assessments, order sets, and geriatric best practices. Moreover, its exclusive inclusion of pre-frail and frail older patients (ie, those at higher risk for poor outcomes) with moderate injury severity (median ISS of 10 [IQR, 9-14]) suggests that this type of care pathway benefits hospitalized older trauma patients, who are particularly vulnerable to adverse complications such as delirium. Moreover, the successful utilization of the FRAIL questionnaire, a validated frailty screening tool, by surgical residents in the ED to initiate this care pathway demonstrates the feasibility of its use in expediting frailty screening in older patients in trauma care.

Application for Clinical Practice and System Implementation

Findings from the 2 studies discussed in this review indicate that implementation of interdisciplinary clinical care pathways predicated on evidence-based geriatric principles and best practices is a promising approach to reduce delirium in hospitalized older trauma patients. These studies have helped to lay the groundwork in outlining the roadmaps (eg, processes and infrastructures) needed to create such clinical pathways. These key elements include: (1) integration of a multidisciplinary committee (eg, representation from trauma, emergency, and geriatric medicine, nursing, physical and occupational therapy, pharmacy, social work) in pathway design and implementation; (2) adaption of evidence-based geriatric best practices (eg, the Institute for Healthcare Improvement Age-Friendly Health System 4M framework [medication, mentation, mobility, what matters]) to prioritize interventions and to design a pathway that incorporates these features; (3) incorporation of comprehensive geriatric assessment by interdisciplinary providers; (4) utilization of validated clinical instruments to assess physical and cognitive functions, frailty, delirium, and social determinants of health; (5) modification of electronic health record systems to encompass order sets that incorporate evidence-based, nonpharmacological and pharmacological interventions to manage symptoms (eg, delirium, pain, bowel movement, sleep, immobility, polypharmacy) essential to quality geriatric care; and (6) integration of patient and caregiver preferences via goals-of-care discussions and corresponding documentation and communication of these goals.

Additionally, these 2 studies imparted some strategies that may facilitate the implementation of interdisciplinary clinical care pathways in trauma care. Examples of such facilitators include: (1) collaboration with champions within each specialty to reinforce education and buy-in; (2) creation of automatically triggered order sets upon patient presentation to the ED that unites distinct features of clinical pathways; (3) adaption and reorganization of existing hospital infrastructures and resources to meet the needs of clinical pathways implementation (eg, utilizing information technology resources to develop electronic health record order sets; using quality department to develop clinical pathway guidelines and electronic outcome dashboards); and (4) development of individualized patient and caregiver education materials based on care needs (eg, principles of delirium prevention and preservation of mobility during hospitalization) to prepare and engage these stakeholders in patient care and recovery.

Practice Points

• A geriatric interdisciplinary care model can be effectively applied to the management of acute trauma in older patients.
• Interdisciplinary clinical pathways should incorporate geriatric best practices and guidelines and age-appropriate order sets to prioritize and integrate care.

—Fred Ko, MD, MS