Preventable or ameliorable adverse events have been reported to occur in 12% of patients in the period immediately following hospital discharge.1, 2 A potential contributor to this is the inadequate transfer of clinical information at hospital discharge. The discharge summary comprises a vital component of the information transfer between the inpatient and outpatient settings. Unfortunately, discharge summaries are often unavailable at the time of follow‐up care and often lack important content.37

A growing number of hospitals are implementing electronic medical records (EMR). This creates the opportunity to standardize the content of clinical documentation and creates the potential to assemble, immediately at the time of hospital discharge, major components of a discharge summary. With enhanced communication systems, this information can be delivered in a variety of ways with minimal delay. Previously, we reported the results of a survey of medicine faculty at an urban academic medical center evaluating the timeliness and quality of discharge summaries, the perceived incidence of preventable adverse events related to suboptimal information transfer at discharge, and a needs assessment for an electronically generated discharge summary that we planned to design.8 We now report the results of the follow‐up survey of outpatient physicians and an evaluation of the quality and timeliness of the electronic discharge summary we created.

Materials and Methods

Results

Discussion

Our study found that an electronic discharge summary was well accepted by inpatient physicians and significantly improved the quality and timeliness of discharge summaries. Prior studies have shown that the use of electronically entered discharge summaries improved the timeliness of discharge summaries.1416 However, the discharge summaries used in these studies required manual input of data into a computer system separate from the patient's medical record. To our knowledge, this is the first study to report the impact of discharge summaries generated from an EMR. Leveraging the EMR, we were able to automate the insertion of specific patient data elements, streamline delivery, and create an electronic reminder system to inpatient physicians for summaries not completed 24 hours after discharge.

Prior research has shown that the quality of discharges summaries is improved with the use of standardized content.10, 17 Using a standardized template for the electronic discharge summary, we likewise demonstrated improved quality of discharge summaries. Key discharge summary elements, specifically discussion of follow‐up issues, pending test results, and information provided to the patient and/or family, were present more reliably after the implementation of the electronic discharge summary. The importance of identifying pending test results is underscored by a recent study showing that many patients are discharged from hospitals with test results still pending, and that physicians are often unaware when results are abnormal.18 One discharge summary element, significant laboratory findings, was present less often after the implementation of the electronic discharge summary. Our template did not designate significant laboratory findings under a separate heading. Instead, we used a heading entitled Key Results (labs, imaging, pathology). Physicians completing the discharge summaries may have prioritized the report of imaging and pathology results in this section. A simple revision of our discharge summary template to include a separate heading for significant laboratory findings may result in improvement in this regard.

Timeliness of discharge summaries was improved in our study, but remained less than optimal. Although nearly three‐quarters of electronic discharge summaries were completed within 3 days of discharge, our ultimate goal is to have 100% of discharge summaries completed within 3 days. This is especially important for complicated patients requiring outpatient follow‐up soon after discharge. We are currently in the process of designing further modifications to the electronic discharge summary completion process. One modification that may be beneficial is the automation of additional patient specific data elements into the discharge summary. We also plan to link performance of medication reconciliation, completion of patient discharge instructions, and completion of the discharge summary into an integrated set of activities performed in the EMR prior to patient discharge.

We found that fewer outpatient physicians reported 1 or more of their patients having a preventable adverse event or near miss as a result of suboptimal transfer of information at discharge after the implementation of the electronic discharge summary. Although we did not measure preventable adverse events directly in our study, this is an important finding in light of the large number of patients who sustain preventable adverse events after hospital discharge1, 2 and prior research showing that the absence of discharge summaries at postdischarge follow‐up visits increased the risk for hospital readmission.19

We had wondered what effect the electronic discharge summary would have on the length and clarity of discharge summaries. A published commentary suggested that notes performed in EMRs were inordinately long and often difficult to read.20 We were pleased to discover that electronic discharge summaries were similar in length to previous discharge summaries and were rated higher with regard to clarity.

Our study has several limitations. First, many inpatient physicians began to use electronic discharge summaries prior to our creation of the final electronic discharge summary product. We had explicitly instructed physicians to continue to dictate discharge summaries in the first phase of our study. The fact that physicians quickly adopted the practice of completing discharge summaries electronically suggests that they preferred this method for completion and may help to explain the improvement in timeliness. A second limitation, as previously mentioned, is that our study did not measure adverse events directly. Instead, we asked outpatient physicians to estimate the number of their patients discharged in the last 6 months who had sustained a preventable adverse event or near miss related to suboptimal information transfer at discharge. We had limited space in the survey to define the meaning of a preventable adverse event; therefore, the description in the survey does not exactly match previous definitions.1, 2 Finally, the ordinal scale used to assess clarity of discharge summaries has not been previously validated.

In conclusion, the use of an electronic discharge summary significantly improved the quality and timeliness of discharge summaries. The discharge summary comprises a vital component of the information transfer between the inpatient and outpatient settings during the vulnerable period following hospital discharge. As hospitals expand their use of EMRs, they should take advantage of opportunities to leverage functionality to improve quality and timeliness of discharge summaries.

Preventable or ameliorable adverse events have been reported to occur in 12% of patients in the period immediately following hospital discharge.1, 2 A potential contributor to this is the inadequate transfer of clinical information at hospital discharge. The discharge summary comprises a vital component of the information transfer between the inpatient and outpatient settings. Unfortunately, discharge summaries are often unavailable at the time of follow‐up care and often lack important content.37

A growing number of hospitals are implementing electronic medical records (EMR). This creates the opportunity to standardize the content of clinical documentation and creates the potential to assemble, immediately at the time of hospital discharge, major components of a discharge summary. With enhanced communication systems, this information can be delivered in a variety of ways with minimal delay. Previously, we reported the results of a survey of medicine faculty at an urban academic medical center evaluating the timeliness and quality of discharge summaries, the perceived incidence of preventable adverse events related to suboptimal information transfer at discharge, and a needs assessment for an electronically generated discharge summary that we planned to design.8 We now report the results of the follow‐up survey of outpatient physicians and an evaluation of the quality and timeliness of the electronic discharge summary we created.

Materials and Methods

Results

Discussion

Our study found that an electronic discharge summary was well accepted by inpatient physicians and significantly improved the quality and timeliness of discharge summaries. Prior studies have shown that the use of electronically entered discharge summaries improved the timeliness of discharge summaries.1416 However, the discharge summaries used in these studies required manual input of data into a computer system separate from the patient's medical record. To our knowledge, this is the first study to report the impact of discharge summaries generated from an EMR. Leveraging the EMR, we were able to automate the insertion of specific patient data elements, streamline delivery, and create an electronic reminder system to inpatient physicians for summaries not completed 24 hours after discharge.

Prior research has shown that the quality of discharges summaries is improved with the use of standardized content.10, 17 Using a standardized template for the electronic discharge summary, we likewise demonstrated improved quality of discharge summaries. Key discharge summary elements, specifically discussion of follow‐up issues, pending test results, and information provided to the patient and/or family, were present more reliably after the implementation of the electronic discharge summary. The importance of identifying pending test results is underscored by a recent study showing that many patients are discharged from hospitals with test results still pending, and that physicians are often unaware when results are abnormal.18 One discharge summary element, significant laboratory findings, was present less often after the implementation of the electronic discharge summary. Our template did not designate significant laboratory findings under a separate heading. Instead, we used a heading entitled Key Results (labs, imaging, pathology). Physicians completing the discharge summaries may have prioritized the report of imaging and pathology results in this section. A simple revision of our discharge summary template to include a separate heading for significant laboratory findings may result in improvement in this regard.

Timeliness of discharge summaries was improved in our study, but remained less than optimal. Although nearly three‐quarters of electronic discharge summaries were completed within 3 days of discharge, our ultimate goal is to have 100% of discharge summaries completed within 3 days. This is especially important for complicated patients requiring outpatient follow‐up soon after discharge. We are currently in the process of designing further modifications to the electronic discharge summary completion process. One modification that may be beneficial is the automation of additional patient specific data elements into the discharge summary. We also plan to link performance of medication reconciliation, completion of patient discharge instructions, and completion of the discharge summary into an integrated set of activities performed in the EMR prior to patient discharge.

We found that fewer outpatient physicians reported 1 or more of their patients having a preventable adverse event or near miss as a result of suboptimal transfer of information at discharge after the implementation of the electronic discharge summary. Although we did not measure preventable adverse events directly in our study, this is an important finding in light of the large number of patients who sustain preventable adverse events after hospital discharge1, 2 and prior research showing that the absence of discharge summaries at postdischarge follow‐up visits increased the risk for hospital readmission.19

We had wondered what effect the electronic discharge summary would have on the length and clarity of discharge summaries. A published commentary suggested that notes performed in EMRs were inordinately long and often difficult to read.20 We were pleased to discover that electronic discharge summaries were similar in length to previous discharge summaries and were rated higher with regard to clarity.

Our study has several limitations. First, many inpatient physicians began to use electronic discharge summaries prior to our creation of the final electronic discharge summary product. We had explicitly instructed physicians to continue to dictate discharge summaries in the first phase of our study. The fact that physicians quickly adopted the practice of completing discharge summaries electronically suggests that they preferred this method for completion and may help to explain the improvement in timeliness. A second limitation, as previously mentioned, is that our study did not measure adverse events directly. Instead, we asked outpatient physicians to estimate the number of their patients discharged in the last 6 months who had sustained a preventable adverse event or near miss related to suboptimal information transfer at discharge. We had limited space in the survey to define the meaning of a preventable adverse event; therefore, the description in the survey does not exactly match previous definitions.1, 2 Finally, the ordinal scale used to assess clarity of discharge summaries has not been previously validated.

In conclusion, the use of an electronic discharge summary significantly improved the quality and timeliness of discharge summaries. The discharge summary comprises a vital component of the information transfer between the inpatient and outpatient settings during the vulnerable period following hospital discharge. As hospitals expand their use of EMRs, they should take advantage of opportunities to leverage functionality to improve quality and timeliness of discharge summaries.

Preventable or ameliorable adverse events have been reported to occur in 12% of patients in the period immediately following hospital discharge.1, 2 A potential contributor to this is the inadequate transfer of clinical information at hospital discharge. The discharge summary comprises a vital component of the information transfer between the inpatient and outpatient settings. Unfortunately, discharge summaries are often unavailable at the time of follow‐up care and often lack important content.37

A growing number of hospitals are implementing electronic medical records (EMR). This creates the opportunity to standardize the content of clinical documentation and creates the potential to assemble, immediately at the time of hospital discharge, major components of a discharge summary. With enhanced communication systems, this information can be delivered in a variety of ways with minimal delay. Previously, we reported the results of a survey of medicine faculty at an urban academic medical center evaluating the timeliness and quality of discharge summaries, the perceived incidence of preventable adverse events related to suboptimal information transfer at discharge, and a needs assessment for an electronically generated discharge summary that we planned to design.8 We now report the results of the follow‐up survey of outpatient physicians and an evaluation of the quality and timeliness of the electronic discharge summary we created.

Materials and Methods

Results

Discussion

Our study found that an electronic discharge summary was well accepted by inpatient physicians and significantly improved the quality and timeliness of discharge summaries. Prior studies have shown that the use of electronically entered discharge summaries improved the timeliness of discharge summaries.1416 However, the discharge summaries used in these studies required manual input of data into a computer system separate from the patient's medical record. To our knowledge, this is the first study to report the impact of discharge summaries generated from an EMR. Leveraging the EMR, we were able to automate the insertion of specific patient data elements, streamline delivery, and create an electronic reminder system to inpatient physicians for summaries not completed 24 hours after discharge.

Prior research has shown that the quality of discharges summaries is improved with the use of standardized content.10, 17 Using a standardized template for the electronic discharge summary, we likewise demonstrated improved quality of discharge summaries. Key discharge summary elements, specifically discussion of follow‐up issues, pending test results, and information provided to the patient and/or family, were present more reliably after the implementation of the electronic discharge summary. The importance of identifying pending test results is underscored by a recent study showing that many patients are discharged from hospitals with test results still pending, and that physicians are often unaware when results are abnormal.18 One discharge summary element, significant laboratory findings, was present less often after the implementation of the electronic discharge summary. Our template did not designate significant laboratory findings under a separate heading. Instead, we used a heading entitled Key Results (labs, imaging, pathology). Physicians completing the discharge summaries may have prioritized the report of imaging and pathology results in this section. A simple revision of our discharge summary template to include a separate heading for significant laboratory findings may result in improvement in this regard.

Timeliness of discharge summaries was improved in our study, but remained less than optimal. Although nearly three‐quarters of electronic discharge summaries were completed within 3 days of discharge, our ultimate goal is to have 100% of discharge summaries completed within 3 days. This is especially important for complicated patients requiring outpatient follow‐up soon after discharge. We are currently in the process of designing further modifications to the electronic discharge summary completion process. One modification that may be beneficial is the automation of additional patient specific data elements into the discharge summary. We also plan to link performance of medication reconciliation, completion of patient discharge instructions, and completion of the discharge summary into an integrated set of activities performed in the EMR prior to patient discharge.

We found that fewer outpatient physicians reported 1 or more of their patients having a preventable adverse event or near miss as a result of suboptimal transfer of information at discharge after the implementation of the electronic discharge summary. Although we did not measure preventable adverse events directly in our study, this is an important finding in light of the large number of patients who sustain preventable adverse events after hospital discharge1, 2 and prior research showing that the absence of discharge summaries at postdischarge follow‐up visits increased the risk for hospital readmission.19

We had wondered what effect the electronic discharge summary would have on the length and clarity of discharge summaries. A published commentary suggested that notes performed in EMRs were inordinately long and often difficult to read.20 We were pleased to discover that electronic discharge summaries were similar in length to previous discharge summaries and were rated higher with regard to clarity.

Our study has several limitations. First, many inpatient physicians began to use electronic discharge summaries prior to our creation of the final electronic discharge summary product. We had explicitly instructed physicians to continue to dictate discharge summaries in the first phase of our study. The fact that physicians quickly adopted the practice of completing discharge summaries electronically suggests that they preferred this method for completion and may help to explain the improvement in timeliness. A second limitation, as previously mentioned, is that our study did not measure adverse events directly. Instead, we asked outpatient physicians to estimate the number of their patients discharged in the last 6 months who had sustained a preventable adverse event or near miss related to suboptimal information transfer at discharge. We had limited space in the survey to define the meaning of a preventable adverse event; therefore, the description in the survey does not exactly match previous definitions.1, 2 Finally, the ordinal scale used to assess clarity of discharge summaries has not been previously validated.

In conclusion, the use of an electronic discharge summary significantly improved the quality and timeliness of discharge summaries. The discharge summary comprises a vital component of the information transfer between the inpatient and outpatient settings during the vulnerable period following hospital discharge. As hospitals expand their use of EMRs, they should take advantage of opportunities to leverage functionality to improve quality and timeliness of discharge summaries.

More than a decade ago, we surveyed a group of practicing pulmonologists to determine their attitudes regarding the use of thrombolytic therapy in various settings of acute venous thromboembolism (VTE).1 Since that time, the literature regarding the treatment of acute VTE has grown dramatically.214 However, despite the available evidence, there remains considerable controversy regarding the appropriate setting for thrombolysis in acute pulmonary embolism (PE) or deep‐vein thrombosis (DVT). We therefore sought to better describe the current patterns of thrombolytic use among practicing pulmonologists and to determine if these patterns have changed over the last decade.

Methods

Five‐hundred and ten physicians in the southeastern US were selected from the American Thoracic Society (ATS) membership roster and were e‐mailed a link to an online questionnaire. The roster was searched for physicians who described their subspecialty as pulmonary disease or pulmonary and critical care.

Participants were asked background information and questions regarding hypothetical clinical scenarios. All participants were offered a $50 stipend, and to further improve the response rate, 2 reminder e‐mail messages were sent 30 days and 45 days after the initial request.

Baseline findings of the survey were summarized using descriptive statistics. Differences among participants and their responses were determined by Fisher's exact test. Analyses were performed using SAS E‐Guide Version 3.0 for Windows (SAS Institute, Cary, NC) with 2‐sided P values at the standard 0.05 level used to determine statistical significance.

Results

Use of Thrombolytic Therapy When Contraindications Exist

The vast majority of respondents reported that they would consider giving thrombolytic therapy to a patient with massive PE and hypotension requiring vasopressor therapy despite having a traditional contraindication (relative or absolute) to thrombolysis (Table 3). Most respondents would consider giving thrombolytic therapy to postoperative orthopedic, abdominal, or thoracic surgery patients if they were more than 2 weeks postoperation, and very few would give thrombolytic therapy to patients who were less than 2 days postoperation. Many respondents would also consider giving thrombolytic therapy to a patient with a massive PE and with a history of major gastrointestinal (GI) bleeding (requiring blood transfusion) if the bleed was more than 4 weeks prior to the embolism (Figure 1).

Figure 1

Table 3.Strong Consideration of Thrombolytic Therapy for Hemodynamically Significant PE in the Context of Absolute or Relative Contraindications

Condition	Number of Physicians (%)
Abbreviations: CPR, cardiopulmonary resuscitation; ICH, intracranial hemorrhage.
Age >75 years	58 (72)
Guaiac + stool	54 (67)
CPR in past 10 days	39 (48)
History of ischemic stroke	37 (46)
Recent venipuncture of a noncompressible vessel	33 (41)
History of ICH	6 (7)
Brain tumor	6 (7)
Would never use thrombolytics in these scenarios	7 (9)

Discussion

Given the paucity of data from randomized controlled trials, there remains considerable controversy regarding the indications for thrombolytic therapy. It may be difficult to define those patients in whom the benefit of a rapid reduction in clot burden outweighs the increased hemorrhagic risk. The case for thrombolysis is the strongest in patients with massive PE complicated by hypotension, in whom the mortality rate may be 30%.15 Our survey confirms that the vast majority of practicing pulmonologists would strongly consider systemic thrombolysis in this clinical setting, which is in accordance with current guidelines and with our previous survey results.1, 5, 10, 12

No clinical trial has specifically evaluated thrombolytic therapy in patients with large PE and hypoxemia but without hypotension, and it is interesting that so many physicians would consider thrombolytic therapy in this scenario. As right heart failure is the cause of death in PE, the absence of significant hypotension would imply less cardiovascular risk and thrombolytic use would seemingly be less justifiable from a physiologic point of view. It may be that further study and education is warranted in this area.

Many patients who present with acute, life‐threatening PE have contraindications or relative contraindications to systemic thrombolysis. Our study suggests that most practicing pulmonologists would consider giving thrombolytic therapy in some of these situations, such as if the patient was more than 2 weeks postoperative from major thoracic or abdominal surgery (or even a few days following orthopedic surgery), or in the setting of advanced age or guaiac positive stools. Physicians were appropriately very reluctant to use thrombolytic therapy in the setting of a brain tumor or prior intracranial hemorrhage. These scenarios emphasize the vagaries of the current guidelines and real‐world complexities of considering thrombolytic therapy in clinical practice, in which the risks and benefits must be weighed on a case‐by‐case basis.

One major difference between our current and past findings is the general experience with thrombolytic therapy in acute PE. In our first study, only 54% of physicians queried had employed systemic thrombolysis for acute PE. Our current findings were that 84% of physicians had used thrombolysis for acute PE within the last 2 years, perhaps suggesting a greater comfort with this therapy.

Response bias is a major limitation of our study. We sought to keep questions short and clear, and offered a small stipend to improve the return rate. Despite these measures, only 81 of 510 questionnaires were completed. We selected our list of participants from the ATS roster and by geographic location. As suggested by our findings, the results may have been different had we focused solely on VTE experts or those treating large numbers of VTE patients. One strength of this study is that our sample had approximately even numbers of academic and private practice physicians, and that we could compare current results with our prior findings.

In conclusion, practicing pulmonologists generally agreed that in the absence of contraindications, thrombolytic therapy should be considered in patients with massive PE and hypotension, which is in accordance with current guidelines. Furthermore, a majority would still consider thrombolytic therapy in this scenario even if certain contraindications were present. Although there is less agreement in other scenarios, a majority of physicians would consider using thrombolytics in patients with PE and severe hypoxemia or right ventricular (RV) dysfunction. Despite the evolving data and guidelines, our findings are similar to prior survey results, with the notable exception that more physicians reported thrombolytic therapy use in acute PE in the current study. This emphasizes the need for further physician education and future randomized clinical trials to delineate and unify therapeutic strategies in cases of VTE.

More than a decade ago, we surveyed a group of practicing pulmonologists to determine their attitudes regarding the use of thrombolytic therapy in various settings of acute venous thromboembolism (VTE).1 Since that time, the literature regarding the treatment of acute VTE has grown dramatically.214 However, despite the available evidence, there remains considerable controversy regarding the appropriate setting for thrombolysis in acute pulmonary embolism (PE) or deep‐vein thrombosis (DVT). We therefore sought to better describe the current patterns of thrombolytic use among practicing pulmonologists and to determine if these patterns have changed over the last decade.

Methods

Five‐hundred and ten physicians in the southeastern US were selected from the American Thoracic Society (ATS) membership roster and were e‐mailed a link to an online questionnaire. The roster was searched for physicians who described their subspecialty as pulmonary disease or pulmonary and critical care.

Participants were asked background information and questions regarding hypothetical clinical scenarios. All participants were offered a $50 stipend, and to further improve the response rate, 2 reminder e‐mail messages were sent 30 days and 45 days after the initial request.

Baseline findings of the survey were summarized using descriptive statistics. Differences among participants and their responses were determined by Fisher's exact test. Analyses were performed using SAS E‐Guide Version 3.0 for Windows (SAS Institute, Cary, NC) with 2‐sided P values at the standard 0.05 level used to determine statistical significance.

Results

Use of Thrombolytic Therapy When Contraindications Exist

The vast majority of respondents reported that they would consider giving thrombolytic therapy to a patient with massive PE and hypotension requiring vasopressor therapy despite having a traditional contraindication (relative or absolute) to thrombolysis (Table 3). Most respondents would consider giving thrombolytic therapy to postoperative orthopedic, abdominal, or thoracic surgery patients if they were more than 2 weeks postoperation, and very few would give thrombolytic therapy to patients who were less than 2 days postoperation. Many respondents would also consider giving thrombolytic therapy to a patient with a massive PE and with a history of major gastrointestinal (GI) bleeding (requiring blood transfusion) if the bleed was more than 4 weeks prior to the embolism (Figure 1).

Figure 1

Table 3.Strong Consideration of Thrombolytic Therapy for Hemodynamically Significant PE in the Context of Absolute or Relative Contraindications

Condition	Number of Physicians (%)
Abbreviations: CPR, cardiopulmonary resuscitation; ICH, intracranial hemorrhage.
Age >75 years	58 (72)
Guaiac + stool	54 (67)
CPR in past 10 days	39 (48)
History of ischemic stroke	37 (46)
Recent venipuncture of a noncompressible vessel	33 (41)
History of ICH	6 (7)
Brain tumor	6 (7)
Would never use thrombolytics in these scenarios	7 (9)

Discussion

Given the paucity of data from randomized controlled trials, there remains considerable controversy regarding the indications for thrombolytic therapy. It may be difficult to define those patients in whom the benefit of a rapid reduction in clot burden outweighs the increased hemorrhagic risk. The case for thrombolysis is the strongest in patients with massive PE complicated by hypotension, in whom the mortality rate may be 30%.15 Our survey confirms that the vast majority of practicing pulmonologists would strongly consider systemic thrombolysis in this clinical setting, which is in accordance with current guidelines and with our previous survey results.1, 5, 10, 12

No clinical trial has specifically evaluated thrombolytic therapy in patients with large PE and hypoxemia but without hypotension, and it is interesting that so many physicians would consider thrombolytic therapy in this scenario. As right heart failure is the cause of death in PE, the absence of significant hypotension would imply less cardiovascular risk and thrombolytic use would seemingly be less justifiable from a physiologic point of view. It may be that further study and education is warranted in this area.

Many patients who present with acute, life‐threatening PE have contraindications or relative contraindications to systemic thrombolysis. Our study suggests that most practicing pulmonologists would consider giving thrombolytic therapy in some of these situations, such as if the patient was more than 2 weeks postoperative from major thoracic or abdominal surgery (or even a few days following orthopedic surgery), or in the setting of advanced age or guaiac positive stools. Physicians were appropriately very reluctant to use thrombolytic therapy in the setting of a brain tumor or prior intracranial hemorrhage. These scenarios emphasize the vagaries of the current guidelines and real‐world complexities of considering thrombolytic therapy in clinical practice, in which the risks and benefits must be weighed on a case‐by‐case basis.

One major difference between our current and past findings is the general experience with thrombolytic therapy in acute PE. In our first study, only 54% of physicians queried had employed systemic thrombolysis for acute PE. Our current findings were that 84% of physicians had used thrombolysis for acute PE within the last 2 years, perhaps suggesting a greater comfort with this therapy.

Response bias is a major limitation of our study. We sought to keep questions short and clear, and offered a small stipend to improve the return rate. Despite these measures, only 81 of 510 questionnaires were completed. We selected our list of participants from the ATS roster and by geographic location. As suggested by our findings, the results may have been different had we focused solely on VTE experts or those treating large numbers of VTE patients. One strength of this study is that our sample had approximately even numbers of academic and private practice physicians, and that we could compare current results with our prior findings.

In conclusion, practicing pulmonologists generally agreed that in the absence of contraindications, thrombolytic therapy should be considered in patients with massive PE and hypotension, which is in accordance with current guidelines. Furthermore, a majority would still consider thrombolytic therapy in this scenario even if certain contraindications were present. Although there is less agreement in other scenarios, a majority of physicians would consider using thrombolytics in patients with PE and severe hypoxemia or right ventricular (RV) dysfunction. Despite the evolving data and guidelines, our findings are similar to prior survey results, with the notable exception that more physicians reported thrombolytic therapy use in acute PE in the current study. This emphasizes the need for further physician education and future randomized clinical trials to delineate and unify therapeutic strategies in cases of VTE.

More than a decade ago, we surveyed a group of practicing pulmonologists to determine their attitudes regarding the use of thrombolytic therapy in various settings of acute venous thromboembolism (VTE).1 Since that time, the literature regarding the treatment of acute VTE has grown dramatically.214 However, despite the available evidence, there remains considerable controversy regarding the appropriate setting for thrombolysis in acute pulmonary embolism (PE) or deep‐vein thrombosis (DVT). We therefore sought to better describe the current patterns of thrombolytic use among practicing pulmonologists and to determine if these patterns have changed over the last decade.

Methods

Five‐hundred and ten physicians in the southeastern US were selected from the American Thoracic Society (ATS) membership roster and were e‐mailed a link to an online questionnaire. The roster was searched for physicians who described their subspecialty as pulmonary disease or pulmonary and critical care.

Participants were asked background information and questions regarding hypothetical clinical scenarios. All participants were offered a $50 stipend, and to further improve the response rate, 2 reminder e‐mail messages were sent 30 days and 45 days after the initial request.

Baseline findings of the survey were summarized using descriptive statistics. Differences among participants and their responses were determined by Fisher's exact test. Analyses were performed using SAS E‐Guide Version 3.0 for Windows (SAS Institute, Cary, NC) with 2‐sided P values at the standard 0.05 level used to determine statistical significance.

Results

Use of Thrombolytic Therapy When Contraindications Exist

The vast majority of respondents reported that they would consider giving thrombolytic therapy to a patient with massive PE and hypotension requiring vasopressor therapy despite having a traditional contraindication (relative or absolute) to thrombolysis (Table 3). Most respondents would consider giving thrombolytic therapy to postoperative orthopedic, abdominal, or thoracic surgery patients if they were more than 2 weeks postoperation, and very few would give thrombolytic therapy to patients who were less than 2 days postoperation. Many respondents would also consider giving thrombolytic therapy to a patient with a massive PE and with a history of major gastrointestinal (GI) bleeding (requiring blood transfusion) if the bleed was more than 4 weeks prior to the embolism (Figure 1).

Figure 1

Table 3.Strong Consideration of Thrombolytic Therapy for Hemodynamically Significant PE in the Context of Absolute or Relative Contraindications

Condition	Number of Physicians (%)
Abbreviations: CPR, cardiopulmonary resuscitation; ICH, intracranial hemorrhage.
Age >75 years	58 (72)
Guaiac + stool	54 (67)
CPR in past 10 days	39 (48)
History of ischemic stroke	37 (46)
Recent venipuncture of a noncompressible vessel	33 (41)
History of ICH	6 (7)
Brain tumor	6 (7)
Would never use thrombolytics in these scenarios	7 (9)

Discussion

Given the paucity of data from randomized controlled trials, there remains considerable controversy regarding the indications for thrombolytic therapy. It may be difficult to define those patients in whom the benefit of a rapid reduction in clot burden outweighs the increased hemorrhagic risk. The case for thrombolysis is the strongest in patients with massive PE complicated by hypotension, in whom the mortality rate may be 30%.15 Our survey confirms that the vast majority of practicing pulmonologists would strongly consider systemic thrombolysis in this clinical setting, which is in accordance with current guidelines and with our previous survey results.1, 5, 10, 12

No clinical trial has specifically evaluated thrombolytic therapy in patients with large PE and hypoxemia but without hypotension, and it is interesting that so many physicians would consider thrombolytic therapy in this scenario. As right heart failure is the cause of death in PE, the absence of significant hypotension would imply less cardiovascular risk and thrombolytic use would seemingly be less justifiable from a physiologic point of view. It may be that further study and education is warranted in this area.

Many patients who present with acute, life‐threatening PE have contraindications or relative contraindications to systemic thrombolysis. Our study suggests that most practicing pulmonologists would consider giving thrombolytic therapy in some of these situations, such as if the patient was more than 2 weeks postoperative from major thoracic or abdominal surgery (or even a few days following orthopedic surgery), or in the setting of advanced age or guaiac positive stools. Physicians were appropriately very reluctant to use thrombolytic therapy in the setting of a brain tumor or prior intracranial hemorrhage. These scenarios emphasize the vagaries of the current guidelines and real‐world complexities of considering thrombolytic therapy in clinical practice, in which the risks and benefits must be weighed on a case‐by‐case basis.

One major difference between our current and past findings is the general experience with thrombolytic therapy in acute PE. In our first study, only 54% of physicians queried had employed systemic thrombolysis for acute PE. Our current findings were that 84% of physicians had used thrombolysis for acute PE within the last 2 years, perhaps suggesting a greater comfort with this therapy.

Response bias is a major limitation of our study. We sought to keep questions short and clear, and offered a small stipend to improve the return rate. Despite these measures, only 81 of 510 questionnaires were completed. We selected our list of participants from the ATS roster and by geographic location. As suggested by our findings, the results may have been different had we focused solely on VTE experts or those treating large numbers of VTE patients. One strength of this study is that our sample had approximately even numbers of academic and private practice physicians, and that we could compare current results with our prior findings.

In conclusion, practicing pulmonologists generally agreed that in the absence of contraindications, thrombolytic therapy should be considered in patients with massive PE and hypotension, which is in accordance with current guidelines. Furthermore, a majority would still consider thrombolytic therapy in this scenario even if certain contraindications were present. Although there is less agreement in other scenarios, a majority of physicians would consider using thrombolytics in patients with PE and severe hypoxemia or right ventricular (RV) dysfunction. Despite the evolving data and guidelines, our findings are similar to prior survey results, with the notable exception that more physicians reported thrombolytic therapy use in acute PE in the current study. This emphasizes the need for further physician education and future randomized clinical trials to delineate and unify therapeutic strategies in cases of VTE.

Patient Presentation

A 52‐year‐old woman with cirrhosis presented with a large‐volume upper gastrointestinal (GI) bleed. After receiving massive volume and blood product resuscitation (18 U packed red blood cells [PRBCs], 17 U fresh frozen plasma [FFP], 2 U cryoprecipitate, 1 U platelets, and 14 L normal saline), a routine admission electrocardiogram (ECG) was obtained (Figure 1).

Figure 1

Discussion

The ECG shows normal sinus rhythm with a prolonged QT interval (specifically due to a long ST segment). The differential diagnosis of this pattern of QT prolongation includes hypocalcemia, long QT syndrome (variant 3), and hypothermia. Massive transfusion can result in the chelation of calcium by citrate resulting in hypocalcemia. This patient's serum calcium was 7 mg/dL (8.9‐10.2) (ionized 0.21 mmol/L; 1.18‐1.38), her pH was 7.03. There were no overt signs or symptoms of hypocalcemia (paresthesias, twitching, tetany, or seizures.) and with aggressive replacement the calcium and the ECG normalized. Transfusions are an uncommon cause of hypocalcemia, with hypomagnesemia or hypermagnesemia, acute pancreatitis, rhabdomyolysis, tumor lysis syndrome, renal failure, vitamin D deficiency, pseudoparathyroidism and hypoparathyroidism being the more common disorders caused.1, 2 This patient's magnesium was normal on admission (2.1 mg/dL) and she did not have any of the other conditions mentioned. It is possible that hemodilution may also have played a role.

Genetic long QT syndrome (variant 3) can manifest with a similar ECG pattern, although the T‐wave is sometimes peaked or biphasic in that condition. Our patient had no personal or family history of syncope or sudden death, and her ECG normalized with calcium replacement.3

Hypothermia can also present with a long QT interval secondary to a long ST segment. Our patient was normothermic, and had no other ECG findings suggestive of hypothermia (Osborn waves: notching of terminal portion of QRS), shivering artifacts, sinus bradycardia, atrial fibrillation, QRS prolongation, or prolongation of the PR interval.4

The differential diagnosis of a long QT interval with a long ST segment is short, and the clinical scenario will provide the etiology in most cases.

Patient Presentation

A 52‐year‐old woman with cirrhosis presented with a large‐volume upper gastrointestinal (GI) bleed. After receiving massive volume and blood product resuscitation (18 U packed red blood cells [PRBCs], 17 U fresh frozen plasma [FFP], 2 U cryoprecipitate, 1 U platelets, and 14 L normal saline), a routine admission electrocardiogram (ECG) was obtained (Figure 1).

Figure 1

Discussion

The ECG shows normal sinus rhythm with a prolonged QT interval (specifically due to a long ST segment). The differential diagnosis of this pattern of QT prolongation includes hypocalcemia, long QT syndrome (variant 3), and hypothermia. Massive transfusion can result in the chelation of calcium by citrate resulting in hypocalcemia. This patient's serum calcium was 7 mg/dL (8.9‐10.2) (ionized 0.21 mmol/L; 1.18‐1.38), her pH was 7.03. There were no overt signs or symptoms of hypocalcemia (paresthesias, twitching, tetany, or seizures.) and with aggressive replacement the calcium and the ECG normalized. Transfusions are an uncommon cause of hypocalcemia, with hypomagnesemia or hypermagnesemia, acute pancreatitis, rhabdomyolysis, tumor lysis syndrome, renal failure, vitamin D deficiency, pseudoparathyroidism and hypoparathyroidism being the more common disorders caused.1, 2 This patient's magnesium was normal on admission (2.1 mg/dL) and she did not have any of the other conditions mentioned. It is possible that hemodilution may also have played a role.

Genetic long QT syndrome (variant 3) can manifest with a similar ECG pattern, although the T‐wave is sometimes peaked or biphasic in that condition. Our patient had no personal or family history of syncope or sudden death, and her ECG normalized with calcium replacement.3

Hypothermia can also present with a long QT interval secondary to a long ST segment. Our patient was normothermic, and had no other ECG findings suggestive of hypothermia (Osborn waves: notching of terminal portion of QRS), shivering artifacts, sinus bradycardia, atrial fibrillation, QRS prolongation, or prolongation of the PR interval.4

The differential diagnosis of a long QT interval with a long ST segment is short, and the clinical scenario will provide the etiology in most cases.

Patient Presentation

A 52‐year‐old woman with cirrhosis presented with a large‐volume upper gastrointestinal (GI) bleed. After receiving massive volume and blood product resuscitation (18 U packed red blood cells [PRBCs], 17 U fresh frozen plasma [FFP], 2 U cryoprecipitate, 1 U platelets, and 14 L normal saline), a routine admission electrocardiogram (ECG) was obtained (Figure 1).

Figure 1

Discussion

The ECG shows normal sinus rhythm with a prolonged QT interval (specifically due to a long ST segment). The differential diagnosis of this pattern of QT prolongation includes hypocalcemia, long QT syndrome (variant 3), and hypothermia. Massive transfusion can result in the chelation of calcium by citrate resulting in hypocalcemia. This patient's serum calcium was 7 mg/dL (8.9‐10.2) (ionized 0.21 mmol/L; 1.18‐1.38), her pH was 7.03. There were no overt signs or symptoms of hypocalcemia (paresthesias, twitching, tetany, or seizures.) and with aggressive replacement the calcium and the ECG normalized. Transfusions are an uncommon cause of hypocalcemia, with hypomagnesemia or hypermagnesemia, acute pancreatitis, rhabdomyolysis, tumor lysis syndrome, renal failure, vitamin D deficiency, pseudoparathyroidism and hypoparathyroidism being the more common disorders caused.1, 2 This patient's magnesium was normal on admission (2.1 mg/dL) and she did not have any of the other conditions mentioned. It is possible that hemodilution may also have played a role.

Genetic long QT syndrome (variant 3) can manifest with a similar ECG pattern, although the T‐wave is sometimes peaked or biphasic in that condition. Our patient had no personal or family history of syncope or sudden death, and her ECG normalized with calcium replacement.3

Hypothermia can also present with a long QT interval secondary to a long ST segment. Our patient was normothermic, and had no other ECG findings suggestive of hypothermia (Osborn waves: notching of terminal portion of QRS), shivering artifacts, sinus bradycardia, atrial fibrillation, QRS prolongation, or prolongation of the PR interval.4

The differential diagnosis of a long QT interval with a long ST segment is short, and the clinical scenario will provide the etiology in most cases.

Thyrotoxicosis is a relatively common endocrine disorder. An epidemiologic study estimated the prevalence of hyperthyroidism in the United States at 1.3%.1 In our experience newly diagnosed thyrotoxicosis is 1 of the more frequent reasons for inpatient endocrinology specialty consultation. Generally, the diagnosis of thyrotoxicosis is made on clinical and laboratory grounds, and consultation is obtained in order to assist with diagnosing the specific cause and to arrange necessary treatment. A 24‐hour radioactive iodine uptake and scan are useful studies for differentiating the cause of thyrotoxicosis. These tests provide an indication of thyroid pathophysiology based on measurement and imaging of thyroidal uptake of a small dose of orally administered radioiodine. In thyrotoxic patients, the uptake is typically high in those with conditions associated with increased hormone synthesis, such as Graves' disease or autonomously functioning thyroid nodules, and is low in those with conditions associated with decreased hormone synthesis, such as thyroiditis or exogenous thyroid hormone ingestion.

Establishing the correct diagnosis has significant impact on the selection of appropriate therapy for these vastly different etiologies. Radioactive iodine in the form of I‐131 is useful in the treatment of certain thyrotoxic conditions. Radioiodine is the treatment most frequently employed in the United States for Graves' disease,2 and is also the most commonly used treatment of solitary toxic nodules and toxic multinodular goiter. However, it is of no use in treatment of subacute thyroiditis or exogenous ingestion of thyroid hormone. While antithyroid drugs such as methimazole or propylthiouracil can be used to treat conditions associated with increased thyroid hormone synthesis, they also are not effective in treating subacute thyroiditis or thyrotoxicosis factitia; therefore, it is important to establish the diagnosis before using these medications.

Exogenous iodine that has been administered prior to administration of radioiodine can interfere with the uptake of radioiodine by the thyroid, thus rendering the 24‐hour uptake and scan inaccurate as a means of diagnosis, and interfering with the utility of I‐131 as a treatment.3 In our experience, many inpatients with newly diagnosed thyrotoxicosis receive exogenous iodine in the form of intravenous (IV) contrast agents near the time of their presentation, leading to an inability to utilize radioactive iodine to diagnose and potentially treat them. One common reason for the use of iodinated contrast is to enhance computed tomography (CT) scan images. Many symptoms that could presumably lead to the ordering of a CT scan may overlap with symptoms of thyrotoxicosis. Prompt diagnosis of thyrotoxicosis, along with awareness of the interference of IV contrast with use of radioactive iodine and the frequency with which this occurs, may prevent unnecessary CT scans in some patients. Therefore, we undertook this study to quantify the frequency of this occurrence, and the frequency that the IV contrast studies were ultimately useful in the management of these patients.

Methods

The records of inpatient endocrinology consultations over a 48‐month period (January 2003 through January 2007) were reviewed. Consultations requested for thyroid disease were identified, and those which were for thyrotoxicosis were reviewed. Patients with thyrotoxicosis with a cause that had been previously determined, or who had already undergone treatment, were not included in the analysis. The remaining patients with new onset thyrotoxicosis, based on clinical and laboratory findings, in whom radioactive iodine would have been useful for diagnostic and potentially treatment purposes, were included.

Of the patients included, our records were reviewed to determine demographic data, whether intravenous contrast had been administered within 2 weeks of endocrinology consultation, and the type of study that was performed. The results of the contrast studies were reviewed, with attention to findings that potentially could have changed acute or inpatient management of the patients.

Results

A total of 1171 consults were reviewed. Of these, 324 (27.7%) were for thyroid disease. One hundred patients (8.5% of total consults, 30.8% of thyroid disease consults) were identified with previously unevaluated thyrotoxicosis as the primary reason for the consult. Of these patients, 74% were women, and 26% were men. The mean age was 54.6 years (range 18‐93 years).

Forty‐five patients (45%) had been given iodinated contrast within 2 weeks prior to endocrinology evaluation, 43 for CT scanning, and 2 for angiography. There were a total of 50 contrast‐enhanced CT scans done (7 patients had CT scans of multiple anatomic regions). There were 26 chest CT scans, 16 abdominal/pelvic CT scans, and 8 neck CT scans. Indications for the CT scans in these patients are shown in Table 1.

Of the 43 patients who underwent CT scanning, 7 (16%) had a finding that potentially changed their inpatient management (2 with tracheal compression by a goiter, 1 with airway compromise from tonsillar edema, 1 with appendicitis, 1 with pulmonary hypertension, 1 with diffuse lymphoma, and 1 with a rib fracture after trauma). Only 1 of these patients (appendicitis) required emergent treatment of their condition before further diagnostic studies could have been reasonably performed. Among the angiography patients, 1 had undergone cardiac catheterization, which was diagnostic of coronary artery disease, and 1 had undergone angiography of the lower extremities, which revealed an acute arterial clot.

Also of note, in April 2005 at our institution, a new emergency department opened which made it logistically easier to obtain a CT scan. Prior to that time, 16 of 49 patients (33%) evaluated for thyrotoxicosis had received iodinated contrast for a CT scan. After that time, 27 of 51 patients (52%) received a contrast CT study.

Discussion

Indications for contrast‐enhanced CT scans in the acute care setting are numerous. The majority of the CT scans ordered in the above patients were of the chest. CT scans of the chest may be used to evaluate suspected pulmonary emboli or aortic dissection. Coronary CT angiography is also being studied as a method of risk stratification of patients presenting with acute chest pain.4 Common signs and symptoms of the above disorders include chest pain, dyspnea, palpitations, and tachycardia. Many of the most common cardiovascular signs and symptoms associated with thyrotoxicosis overlap with those seen in the above disorders. A recent study of thyrotoxic patients found that on presentation, 73% had palpitations, 60% had dyspnea, 25% had chest pain, and 35% had cough.5 Mean heart rate has been shown to be significantly elevated in thyrotoxic patients compared with normal controls.5, 6 Atrial fibrillation or flutter can occur in about 8% of thyrotoxic patients, especially in males, in older ages, and in those with underlying cardiac disease.7

We also noted that 16 out of the 50 total contrast‐enhanced CT scans performed were of the abdomen and pelvis. While abdominal and gastrointestinal signs and symptoms seem to be less common manifestations of thyrotoxicosis than cardiovascular ones, hyperdefecation with or without diarrhea can be seen in approximately in one‐third of cases. Nausea, vomiting, abdominal pain, and jaundice occur less commonly but can be seen occasionally with severe thyrotoxicosis.3 Weight loss is common, and fever may occur occasionally, which also may trigger evaluation by imaging for malignancy or infection.

Indications for the CT scans ordered for the patients in this study can be seen in Table 1. Some of the studies were ordered for symptoms not likely to be thyroid‐related and may have been unavoidable. However, many of the studies, especially those done for complaints such as palpitations, dyspnea, and weight loss were likely obtained due to symptoms that ultimately turned out to be caused by thyrotoxicosis.

A variety of other factors may uncommonly decrease radioactive iodine uptake, including cardiac decompensation, vomiting after ingestion of the orally dosed radioiodine capsule, and medications, including thioureas. However, exogenous iodine is probably the most common interfering factor.8 Exogenous iodine decreases thyroidal radioiodine uptake both by dilution of the total body iodine pool, and by inhibition of thyroid hormone synthesis via the Wolff‐Chaikoff effect.8 Sources of iodine include diet, medications (amiodarone, which contains approximately 6 mg of inorganic iodine per 200 mg tablet, and potassium iodide), and, most prominently in the hospitalized patient, intravenous contrast material. It has been shown that exposure to as little as 100 g of intravenous iodide can suppress uptake in hyperthyroid patients.9 Both ionic and nonionic intravenous contrast media contain sufficient amounts of inorganic iodide (10 g/ml) to induce this suppression when given in typical doses of 50‐200 ml. This iodide may be derived from deiodination of the contrast media molecule, and also from free iodide contaminants.10 The duration of uptake suppression varies among patients and generally ranges from 4 to 12 weeks, although it may be shorter in hyperthyroid patients.8 This long‐term suppression may be related to the slow deiodination of contrast media remaining in the body.10 Urinary iodine normalization can be used as a means of determining the time when uptake can be measured without interference from exogenous iodine.

Iodine administration can also have deleterious effects on thyroid function apart from its effect on radioiodine uptake. In euthyroid patients, exogenous iodine in large doses inhibits organification of iodide and thyroid hormone synthesis (Wolff‐Chaikoff effect). Normally this effect diminishes after several weeks, but in patients with autoimmune thyroid disease, it may persist, leading to hypothyroidism.11 Iodine ingestion may also lead to hyperthyroidism. In areas of endemic iodine deficiency, this is thought to represent unmasking of thyroid autonomy that had been suppressed by lack of iodine, encompassing such patients as those with Graves' disease or autonomous nodules. In areas of iodine sufficiency such as the United States, the incidence of iodine‐induced thyrotoxicosis is low, and typically occurs in patients with autonomous thyroid nodules or multinodular goiter.11, 12 Cases of thyroid storm have been reported following iodinated contrast administration.13 The elderly may be at greater risk for iodine‐induced thyrotoxicosis,14 which is of concern given their inherent higher likelihood of cardiac disease. It is possible that in some of the patients in our study who had CT scans done for nonthyroid‐related symptoms, iodinated contrast could have precipitated subsequent thyrotoxicosis. In general, caution should be used when administering intravenous contrast agents to patients with known thyroid disease.

CT scanning is being used much more frequently in the acute care setting. One study in the emergency department at a single institution revealed that from the years 2000 to 2005, CT scanning of the chest increased by 226%, and of the abdomen by 72%, despite only a 13% increase in patient volume and a stable level of acuity.15 We refer the reader to this recent review for a discussion of the recent increase in the use of CT scans and the associated risks.16 We have found that a substantial proportion of inpatients with thyrotoxicosis receive intravenous contrast prior to endocrinologic evaluation, thus limiting the ability to use radioactive iodine in diagnosis and treatment. The vast majority of these diagnostic studies are contrast‐enhanced CT scans. Acute findings from these studies in this population that change patient management are uncommon.

The increasing usage of CT scans may exacerbate this problem. Education of physicians that are likely to diagnose thyrotoxicosis prior to subspecialty evaluation (internists, family practitioners, and emergency medicine physicians) regarding the similarity of thyrotoxic symptoms to those typically triggering the request of a CT scan is essential. Clinicians should have an appreciation of the interference of iodinated contrast with the ability to obtain a radioiodine uptake and scan. They should also be aware of the potential effects of iodinated contrast on thyroid function when ordering a contrast‐enhanced CT scan in patients with known thyroid disease or symptoms consistent with the presence of thyrotoxicosis. Under these circumstances consideration of thyroid dysfunction with appropriate blood tests may result in more accurate and timely diagnosis and treatment of underlying thyroid disease and enhance patient outcomes.

Thyrotoxicosis is a relatively common endocrine disorder. An epidemiologic study estimated the prevalence of hyperthyroidism in the United States at 1.3%.1 In our experience newly diagnosed thyrotoxicosis is 1 of the more frequent reasons for inpatient endocrinology specialty consultation. Generally, the diagnosis of thyrotoxicosis is made on clinical and laboratory grounds, and consultation is obtained in order to assist with diagnosing the specific cause and to arrange necessary treatment. A 24‐hour radioactive iodine uptake and scan are useful studies for differentiating the cause of thyrotoxicosis. These tests provide an indication of thyroid pathophysiology based on measurement and imaging of thyroidal uptake of a small dose of orally administered radioiodine. In thyrotoxic patients, the uptake is typically high in those with conditions associated with increased hormone synthesis, such as Graves' disease or autonomously functioning thyroid nodules, and is low in those with conditions associated with decreased hormone synthesis, such as thyroiditis or exogenous thyroid hormone ingestion.

Establishing the correct diagnosis has significant impact on the selection of appropriate therapy for these vastly different etiologies. Radioactive iodine in the form of I‐131 is useful in the treatment of certain thyrotoxic conditions. Radioiodine is the treatment most frequently employed in the United States for Graves' disease,2 and is also the most commonly used treatment of solitary toxic nodules and toxic multinodular goiter. However, it is of no use in treatment of subacute thyroiditis or exogenous ingestion of thyroid hormone. While antithyroid drugs such as methimazole or propylthiouracil can be used to treat conditions associated with increased thyroid hormone synthesis, they also are not effective in treating subacute thyroiditis or thyrotoxicosis factitia; therefore, it is important to establish the diagnosis before using these medications.

Exogenous iodine that has been administered prior to administration of radioiodine can interfere with the uptake of radioiodine by the thyroid, thus rendering the 24‐hour uptake and scan inaccurate as a means of diagnosis, and interfering with the utility of I‐131 as a treatment.3 In our experience, many inpatients with newly diagnosed thyrotoxicosis receive exogenous iodine in the form of intravenous (IV) contrast agents near the time of their presentation, leading to an inability to utilize radioactive iodine to diagnose and potentially treat them. One common reason for the use of iodinated contrast is to enhance computed tomography (CT) scan images. Many symptoms that could presumably lead to the ordering of a CT scan may overlap with symptoms of thyrotoxicosis. Prompt diagnosis of thyrotoxicosis, along with awareness of the interference of IV contrast with use of radioactive iodine and the frequency with which this occurs, may prevent unnecessary CT scans in some patients. Therefore, we undertook this study to quantify the frequency of this occurrence, and the frequency that the IV contrast studies were ultimately useful in the management of these patients.

Methods

The records of inpatient endocrinology consultations over a 48‐month period (January 2003 through January 2007) were reviewed. Consultations requested for thyroid disease were identified, and those which were for thyrotoxicosis were reviewed. Patients with thyrotoxicosis with a cause that had been previously determined, or who had already undergone treatment, were not included in the analysis. The remaining patients with new onset thyrotoxicosis, based on clinical and laboratory findings, in whom radioactive iodine would have been useful for diagnostic and potentially treatment purposes, were included.

Of the patients included, our records were reviewed to determine demographic data, whether intravenous contrast had been administered within 2 weeks of endocrinology consultation, and the type of study that was performed. The results of the contrast studies were reviewed, with attention to findings that potentially could have changed acute or inpatient management of the patients.

Results

A total of 1171 consults were reviewed. Of these, 324 (27.7%) were for thyroid disease. One hundred patients (8.5% of total consults, 30.8% of thyroid disease consults) were identified with previously unevaluated thyrotoxicosis as the primary reason for the consult. Of these patients, 74% were women, and 26% were men. The mean age was 54.6 years (range 18‐93 years).

Forty‐five patients (45%) had been given iodinated contrast within 2 weeks prior to endocrinology evaluation, 43 for CT scanning, and 2 for angiography. There were a total of 50 contrast‐enhanced CT scans done (7 patients had CT scans of multiple anatomic regions). There were 26 chest CT scans, 16 abdominal/pelvic CT scans, and 8 neck CT scans. Indications for the CT scans in these patients are shown in Table 1.

Of the 43 patients who underwent CT scanning, 7 (16%) had a finding that potentially changed their inpatient management (2 with tracheal compression by a goiter, 1 with airway compromise from tonsillar edema, 1 with appendicitis, 1 with pulmonary hypertension, 1 with diffuse lymphoma, and 1 with a rib fracture after trauma). Only 1 of these patients (appendicitis) required emergent treatment of their condition before further diagnostic studies could have been reasonably performed. Among the angiography patients, 1 had undergone cardiac catheterization, which was diagnostic of coronary artery disease, and 1 had undergone angiography of the lower extremities, which revealed an acute arterial clot.

Also of note, in April 2005 at our institution, a new emergency department opened which made it logistically easier to obtain a CT scan. Prior to that time, 16 of 49 patients (33%) evaluated for thyrotoxicosis had received iodinated contrast for a CT scan. After that time, 27 of 51 patients (52%) received a contrast CT study.

Discussion

Indications for contrast‐enhanced CT scans in the acute care setting are numerous. The majority of the CT scans ordered in the above patients were of the chest. CT scans of the chest may be used to evaluate suspected pulmonary emboli or aortic dissection. Coronary CT angiography is also being studied as a method of risk stratification of patients presenting with acute chest pain.4 Common signs and symptoms of the above disorders include chest pain, dyspnea, palpitations, and tachycardia. Many of the most common cardiovascular signs and symptoms associated with thyrotoxicosis overlap with those seen in the above disorders. A recent study of thyrotoxic patients found that on presentation, 73% had palpitations, 60% had dyspnea, 25% had chest pain, and 35% had cough.5 Mean heart rate has been shown to be significantly elevated in thyrotoxic patients compared with normal controls.5, 6 Atrial fibrillation or flutter can occur in about 8% of thyrotoxic patients, especially in males, in older ages, and in those with underlying cardiac disease.7

We also noted that 16 out of the 50 total contrast‐enhanced CT scans performed were of the abdomen and pelvis. While abdominal and gastrointestinal signs and symptoms seem to be less common manifestations of thyrotoxicosis than cardiovascular ones, hyperdefecation with or without diarrhea can be seen in approximately in one‐third of cases. Nausea, vomiting, abdominal pain, and jaundice occur less commonly but can be seen occasionally with severe thyrotoxicosis.3 Weight loss is common, and fever may occur occasionally, which also may trigger evaluation by imaging for malignancy or infection.

Indications for the CT scans ordered for the patients in this study can be seen in Table 1. Some of the studies were ordered for symptoms not likely to be thyroid‐related and may have been unavoidable. However, many of the studies, especially those done for complaints such as palpitations, dyspnea, and weight loss were likely obtained due to symptoms that ultimately turned out to be caused by thyrotoxicosis.

A variety of other factors may uncommonly decrease radioactive iodine uptake, including cardiac decompensation, vomiting after ingestion of the orally dosed radioiodine capsule, and medications, including thioureas. However, exogenous iodine is probably the most common interfering factor.8 Exogenous iodine decreases thyroidal radioiodine uptake both by dilution of the total body iodine pool, and by inhibition of thyroid hormone synthesis via the Wolff‐Chaikoff effect.8 Sources of iodine include diet, medications (amiodarone, which contains approximately 6 mg of inorganic iodine per 200 mg tablet, and potassium iodide), and, most prominently in the hospitalized patient, intravenous contrast material. It has been shown that exposure to as little as 100 g of intravenous iodide can suppress uptake in hyperthyroid patients.9 Both ionic and nonionic intravenous contrast media contain sufficient amounts of inorganic iodide (10 g/ml) to induce this suppression when given in typical doses of 50‐200 ml. This iodide may be derived from deiodination of the contrast media molecule, and also from free iodide contaminants.10 The duration of uptake suppression varies among patients and generally ranges from 4 to 12 weeks, although it may be shorter in hyperthyroid patients.8 This long‐term suppression may be related to the slow deiodination of contrast media remaining in the body.10 Urinary iodine normalization can be used as a means of determining the time when uptake can be measured without interference from exogenous iodine.

Iodine administration can also have deleterious effects on thyroid function apart from its effect on radioiodine uptake. In euthyroid patients, exogenous iodine in large doses inhibits organification of iodide and thyroid hormone synthesis (Wolff‐Chaikoff effect). Normally this effect diminishes after several weeks, but in patients with autoimmune thyroid disease, it may persist, leading to hypothyroidism.11 Iodine ingestion may also lead to hyperthyroidism. In areas of endemic iodine deficiency, this is thought to represent unmasking of thyroid autonomy that had been suppressed by lack of iodine, encompassing such patients as those with Graves' disease or autonomous nodules. In areas of iodine sufficiency such as the United States, the incidence of iodine‐induced thyrotoxicosis is low, and typically occurs in patients with autonomous thyroid nodules or multinodular goiter.11, 12 Cases of thyroid storm have been reported following iodinated contrast administration.13 The elderly may be at greater risk for iodine‐induced thyrotoxicosis,14 which is of concern given their inherent higher likelihood of cardiac disease. It is possible that in some of the patients in our study who had CT scans done for nonthyroid‐related symptoms, iodinated contrast could have precipitated subsequent thyrotoxicosis. In general, caution should be used when administering intravenous contrast agents to patients with known thyroid disease.

CT scanning is being used much more frequently in the acute care setting. One study in the emergency department at a single institution revealed that from the years 2000 to 2005, CT scanning of the chest increased by 226%, and of the abdomen by 72%, despite only a 13% increase in patient volume and a stable level of acuity.15 We refer the reader to this recent review for a discussion of the recent increase in the use of CT scans and the associated risks.16 We have found that a substantial proportion of inpatients with thyrotoxicosis receive intravenous contrast prior to endocrinologic evaluation, thus limiting the ability to use radioactive iodine in diagnosis and treatment. The vast majority of these diagnostic studies are contrast‐enhanced CT scans. Acute findings from these studies in this population that change patient management are uncommon.

The increasing usage of CT scans may exacerbate this problem. Education of physicians that are likely to diagnose thyrotoxicosis prior to subspecialty evaluation (internists, family practitioners, and emergency medicine physicians) regarding the similarity of thyrotoxic symptoms to those typically triggering the request of a CT scan is essential. Clinicians should have an appreciation of the interference of iodinated contrast with the ability to obtain a radioiodine uptake and scan. They should also be aware of the potential effects of iodinated contrast on thyroid function when ordering a contrast‐enhanced CT scan in patients with known thyroid disease or symptoms consistent with the presence of thyrotoxicosis. Under these circumstances consideration of thyroid dysfunction with appropriate blood tests may result in more accurate and timely diagnosis and treatment of underlying thyroid disease and enhance patient outcomes.

Thyrotoxicosis is a relatively common endocrine disorder. An epidemiologic study estimated the prevalence of hyperthyroidism in the United States at 1.3%.1 In our experience newly diagnosed thyrotoxicosis is 1 of the more frequent reasons for inpatient endocrinology specialty consultation. Generally, the diagnosis of thyrotoxicosis is made on clinical and laboratory grounds, and consultation is obtained in order to assist with diagnosing the specific cause and to arrange necessary treatment. A 24‐hour radioactive iodine uptake and scan are useful studies for differentiating the cause of thyrotoxicosis. These tests provide an indication of thyroid pathophysiology based on measurement and imaging of thyroidal uptake of a small dose of orally administered radioiodine. In thyrotoxic patients, the uptake is typically high in those with conditions associated with increased hormone synthesis, such as Graves' disease or autonomously functioning thyroid nodules, and is low in those with conditions associated with decreased hormone synthesis, such as thyroiditis or exogenous thyroid hormone ingestion.

Establishing the correct diagnosis has significant impact on the selection of appropriate therapy for these vastly different etiologies. Radioactive iodine in the form of I‐131 is useful in the treatment of certain thyrotoxic conditions. Radioiodine is the treatment most frequently employed in the United States for Graves' disease,2 and is also the most commonly used treatment of solitary toxic nodules and toxic multinodular goiter. However, it is of no use in treatment of subacute thyroiditis or exogenous ingestion of thyroid hormone. While antithyroid drugs such as methimazole or propylthiouracil can be used to treat conditions associated with increased thyroid hormone synthesis, they also are not effective in treating subacute thyroiditis or thyrotoxicosis factitia; therefore, it is important to establish the diagnosis before using these medications.

Exogenous iodine that has been administered prior to administration of radioiodine can interfere with the uptake of radioiodine by the thyroid, thus rendering the 24‐hour uptake and scan inaccurate as a means of diagnosis, and interfering with the utility of I‐131 as a treatment.3 In our experience, many inpatients with newly diagnosed thyrotoxicosis receive exogenous iodine in the form of intravenous (IV) contrast agents near the time of their presentation, leading to an inability to utilize radioactive iodine to diagnose and potentially treat them. One common reason for the use of iodinated contrast is to enhance computed tomography (CT) scan images. Many symptoms that could presumably lead to the ordering of a CT scan may overlap with symptoms of thyrotoxicosis. Prompt diagnosis of thyrotoxicosis, along with awareness of the interference of IV contrast with use of radioactive iodine and the frequency with which this occurs, may prevent unnecessary CT scans in some patients. Therefore, we undertook this study to quantify the frequency of this occurrence, and the frequency that the IV contrast studies were ultimately useful in the management of these patients.

Methods

The records of inpatient endocrinology consultations over a 48‐month period (January 2003 through January 2007) were reviewed. Consultations requested for thyroid disease were identified, and those which were for thyrotoxicosis were reviewed. Patients with thyrotoxicosis with a cause that had been previously determined, or who had already undergone treatment, were not included in the analysis. The remaining patients with new onset thyrotoxicosis, based on clinical and laboratory findings, in whom radioactive iodine would have been useful for diagnostic and potentially treatment purposes, were included.

Of the patients included, our records were reviewed to determine demographic data, whether intravenous contrast had been administered within 2 weeks of endocrinology consultation, and the type of study that was performed. The results of the contrast studies were reviewed, with attention to findings that potentially could have changed acute or inpatient management of the patients.

Results

A total of 1171 consults were reviewed. Of these, 324 (27.7%) were for thyroid disease. One hundred patients (8.5% of total consults, 30.8% of thyroid disease consults) were identified with previously unevaluated thyrotoxicosis as the primary reason for the consult. Of these patients, 74% were women, and 26% were men. The mean age was 54.6 years (range 18‐93 years).

Forty‐five patients (45%) had been given iodinated contrast within 2 weeks prior to endocrinology evaluation, 43 for CT scanning, and 2 for angiography. There were a total of 50 contrast‐enhanced CT scans done (7 patients had CT scans of multiple anatomic regions). There were 26 chest CT scans, 16 abdominal/pelvic CT scans, and 8 neck CT scans. Indications for the CT scans in these patients are shown in Table 1.

Of the 43 patients who underwent CT scanning, 7 (16%) had a finding that potentially changed their inpatient management (2 with tracheal compression by a goiter, 1 with airway compromise from tonsillar edema, 1 with appendicitis, 1 with pulmonary hypertension, 1 with diffuse lymphoma, and 1 with a rib fracture after trauma). Only 1 of these patients (appendicitis) required emergent treatment of their condition before further diagnostic studies could have been reasonably performed. Among the angiography patients, 1 had undergone cardiac catheterization, which was diagnostic of coronary artery disease, and 1 had undergone angiography of the lower extremities, which revealed an acute arterial clot.

Also of note, in April 2005 at our institution, a new emergency department opened which made it logistically easier to obtain a CT scan. Prior to that time, 16 of 49 patients (33%) evaluated for thyrotoxicosis had received iodinated contrast for a CT scan. After that time, 27 of 51 patients (52%) received a contrast CT study.

Discussion

Indications for contrast‐enhanced CT scans in the acute care setting are numerous. The majority of the CT scans ordered in the above patients were of the chest. CT scans of the chest may be used to evaluate suspected pulmonary emboli or aortic dissection. Coronary CT angiography is also being studied as a method of risk stratification of patients presenting with acute chest pain.4 Common signs and symptoms of the above disorders include chest pain, dyspnea, palpitations, and tachycardia. Many of the most common cardiovascular signs and symptoms associated with thyrotoxicosis overlap with those seen in the above disorders. A recent study of thyrotoxic patients found that on presentation, 73% had palpitations, 60% had dyspnea, 25% had chest pain, and 35% had cough.5 Mean heart rate has been shown to be significantly elevated in thyrotoxic patients compared with normal controls.5, 6 Atrial fibrillation or flutter can occur in about 8% of thyrotoxic patients, especially in males, in older ages, and in those with underlying cardiac disease.7

We also noted that 16 out of the 50 total contrast‐enhanced CT scans performed were of the abdomen and pelvis. While abdominal and gastrointestinal signs and symptoms seem to be less common manifestations of thyrotoxicosis than cardiovascular ones, hyperdefecation with or without diarrhea can be seen in approximately in one‐third of cases. Nausea, vomiting, abdominal pain, and jaundice occur less commonly but can be seen occasionally with severe thyrotoxicosis.3 Weight loss is common, and fever may occur occasionally, which also may trigger evaluation by imaging for malignancy or infection.

Indications for the CT scans ordered for the patients in this study can be seen in Table 1. Some of the studies were ordered for symptoms not likely to be thyroid‐related and may have been unavoidable. However, many of the studies, especially those done for complaints such as palpitations, dyspnea, and weight loss were likely obtained due to symptoms that ultimately turned out to be caused by thyrotoxicosis.

A variety of other factors may uncommonly decrease radioactive iodine uptake, including cardiac decompensation, vomiting after ingestion of the orally dosed radioiodine capsule, and medications, including thioureas. However, exogenous iodine is probably the most common interfering factor.8 Exogenous iodine decreases thyroidal radioiodine uptake both by dilution of the total body iodine pool, and by inhibition of thyroid hormone synthesis via the Wolff‐Chaikoff effect.8 Sources of iodine include diet, medications (amiodarone, which contains approximately 6 mg of inorganic iodine per 200 mg tablet, and potassium iodide), and, most prominently in the hospitalized patient, intravenous contrast material. It has been shown that exposure to as little as 100 g of intravenous iodide can suppress uptake in hyperthyroid patients.9 Both ionic and nonionic intravenous contrast media contain sufficient amounts of inorganic iodide (10 g/ml) to induce this suppression when given in typical doses of 50‐200 ml. This iodide may be derived from deiodination of the contrast media molecule, and also from free iodide contaminants.10 The duration of uptake suppression varies among patients and generally ranges from 4 to 12 weeks, although it may be shorter in hyperthyroid patients.8 This long‐term suppression may be related to the slow deiodination of contrast media remaining in the body.10 Urinary iodine normalization can be used as a means of determining the time when uptake can be measured without interference from exogenous iodine.

Iodine administration can also have deleterious effects on thyroid function apart from its effect on radioiodine uptake. In euthyroid patients, exogenous iodine in large doses inhibits organification of iodide and thyroid hormone synthesis (Wolff‐Chaikoff effect). Normally this effect diminishes after several weeks, but in patients with autoimmune thyroid disease, it may persist, leading to hypothyroidism.11 Iodine ingestion may also lead to hyperthyroidism. In areas of endemic iodine deficiency, this is thought to represent unmasking of thyroid autonomy that had been suppressed by lack of iodine, encompassing such patients as those with Graves' disease or autonomous nodules. In areas of iodine sufficiency such as the United States, the incidence of iodine‐induced thyrotoxicosis is low, and typically occurs in patients with autonomous thyroid nodules or multinodular goiter.11, 12 Cases of thyroid storm have been reported following iodinated contrast administration.13 The elderly may be at greater risk for iodine‐induced thyrotoxicosis,14 which is of concern given their inherent higher likelihood of cardiac disease. It is possible that in some of the patients in our study who had CT scans done for nonthyroid‐related symptoms, iodinated contrast could have precipitated subsequent thyrotoxicosis. In general, caution should be used when administering intravenous contrast agents to patients with known thyroid disease.

CT scanning is being used much more frequently in the acute care setting. One study in the emergency department at a single institution revealed that from the years 2000 to 2005, CT scanning of the chest increased by 226%, and of the abdomen by 72%, despite only a 13% increase in patient volume and a stable level of acuity.15 We refer the reader to this recent review for a discussion of the recent increase in the use of CT scans and the associated risks.16 We have found that a substantial proportion of inpatients with thyrotoxicosis receive intravenous contrast prior to endocrinologic evaluation, thus limiting the ability to use radioactive iodine in diagnosis and treatment. The vast majority of these diagnostic studies are contrast‐enhanced CT scans. Acute findings from these studies in this population that change patient management are uncommon.

The increasing usage of CT scans may exacerbate this problem. Education of physicians that are likely to diagnose thyrotoxicosis prior to subspecialty evaluation (internists, family practitioners, and emergency medicine physicians) regarding the similarity of thyrotoxic symptoms to those typically triggering the request of a CT scan is essential. Clinicians should have an appreciation of the interference of iodinated contrast with the ability to obtain a radioiodine uptake and scan. They should also be aware of the potential effects of iodinated contrast on thyroid function when ordering a contrast‐enhanced CT scan in patients with known thyroid disease or symptoms consistent with the presence of thyrotoxicosis. Under these circumstances consideration of thyroid dysfunction with appropriate blood tests may result in more accurate and timely diagnosis and treatment of underlying thyroid disease and enhance patient outcomes.

Hospitalist systems focus on providing acute treatments to patients and expediting hospital discharge, sometimes without regard for the need to work in concert with community providers, leading to fragmentation of care.1 This fragmentation, particularly at the transitions of care, such as when patients move from the outpatient setting to a hospitalist system and then back to their primary care providers (PCPs), can lead to communication breakdowns and delays in care, and may compromise patient outcomes.2 Suboptimal treatments, such as medication errors and the ordering of redundant tests can occur in either setting if prior treatment information is not relayed in a timely and accurate fashion. Landrigan et al.3 described a conceptual model in 2001 that recognized the complexity of the hospitalist‐PCP communication system. Specifically, optimal care of hospitalized children includes PCPs, family members, hospitalists, and support staff while meeting the communication needs of families and PCPs. Additionally, mediating a smooth transition into and out of the hospital needs to be measured carefully.3

Previous adult studies have reported that hospitalist systems sometimes create discontinuity of patient care, which can have a negative impact on the quality of care provided to patients if there is poor communication between hospitalists and PCPs.2, 4, 5 Existing research on hospitalist‐PCP communication focuses mainly on adult hospitalist models with little known about the quality of current pediatric hospitalist‐PCP communication.

The objective of this study was to qualitatively explore issues around communication between pediatric hospitalists and PCPs. Specifically, we sought to explore the quality of communication practices and barriers to optimal communication within the hospitalist‐PCP model at a tertiary care children's hospital. The results are serving as a needs assessment to guide the design of a quality improvement project with the aim of improving pediatric hospitalist‐PCP communication.

METHODS

RESULTS

Only 1 physician per practice was interviewed. No PCP who was able to be contacted declined an interview, although some did require multiple phone attempts to schedule the interview. PCPs were located in Salt Lake County (n = 6), in other Utah counties (n = 3), and in surrounding Intermountain West states (n = 3). From January 1, 2004 to December 31, 2004, we estimate that the hospitalist division cared for patients from approximately 35 practices (50% in Salt Lake County, 30% in other Utah counties, and 20% in surrounding Intermountain West states).

Hospitalists and PCPs agreed that overall quality of communication ranged from poor to very good (Table 2). Both parties acknowledged that significant barriers to optimal communication exist, yet the barriers differed for each group. Hospitalists and PCPs also agreed that optimal communication could improve many aspects of patient care and should take place upon discharge and admission of patients and with major clinical changes. Both hospitalists and PCPs also wanted accurate and timely information. One priority that the participants emphasized is the timely transfer of admission notification and the receipt of accurate and timely discharge summaries by PCPs.

DISCUSSION

Both pediatric hospitalists and PCPs agree on what information is important to transmit (diagnoses, medications, follow‐up needs, and pending laboratory test results) and critical times for communication during the hospitalization (at discharge, admission, and during major clinical changes). However, there was discrepancy in the barriers to optimal communication for each group. Identifying and addressing these barriers can help both hospitalists and PCPs implement targeted interventions aimed at improving communication. As the number of pediatric hospitalist programs increases, the risk for hospitalist‐PCP communication breakdowns, which can have a negative impact on patient care, also increases.

Previous adult studies describe the scope of the problem around poor communication between hospitalists and PCPs.1, 912 Kripalani et al.10 reported recently that delays and omissions in communication are common at hospital discharge among adult hospitalists and that computer‐generated summaries, educational interventions, and standardized formats may facilitate more timely transfer of pertinent information. However, there is limited data on pediatric hospitalist‐PCP communication. Srivastava et al.5 found that 60% of community physicians thought hospitalist systems may impair communication with PCPs when evaluating community and hospital‐based physicians' attitudes regarding pediatric hospitalist systems.

PCPs can feel left out when their patients are cared for by hospitalists.13 One PCP in our study commented: Include the referral doc as part of the team. We're the ones who will take care of them after discharge. It seems like an autonomous thing down there and we're excluded from the patient care team. Additionally, patients want their PCPs to remain involved in their care as they transition into and out of the hospital setting.1, 13

The continuity visit model has been proposed by Wachter and Pantilat14 to describe a clinical encounter between the primary physician and hospitalized patient, when the patient has a different physician of record. In this model, the PCP can endorse the hospitalist model and the individual hospitalist, notice subtle findings that differ from the patient's baseline, and help clarify patient preferences regarding difficult situations by drawing on their previous relationship with the patient. This visit may also benefit the PCP by providing insights into the patient's illness, personality, or social support that he or she was unaware of previously. However, in order for the continuity visit to exist, the PCP has to be informed of their patient's admission in the first place. Ethical dilemmas also have been raised regarding who bears primary responsibility for maintaining open lines of communication when patients are hospitalized.15 Lo15 advocates that PCPs can and should be involved in meaningful ways in the inpatient care of their patients even when they are not acting as the treating physicians. Specifically, he suggests that PCPs personally visit particularly ill patients or those with difficult diagnoses and use frequent phone calls to all admitted patients.

Beyond telephone calls and continuity visits, hospitalists and PCPs rely on discharge summaries as a key part of the information transfer about a patient's hospitalization.1, 16, 17 These documents are rendered useless if they are inaccurate, illegible, or not delivered in a timely manner.18 In a study of California family physicians, discharge summaries were thought to be too detailed by 84% of PCPs, and reportedly arrived before the patient's first follow‐up appointment only 33% of the time.1 O'Leary et al.19 found that 41% of the Department of Medicine physicians surveyed believed that at least 1 of their patients hospitalized in the previous 6 months had experienced a preventable adverse event related to poor transfer of information at discharge. In our study, PCPs noted that discharge summaries often arrived in their offices well after the patient had been seen for their follow‐up appointment.

Both hospitalists and PCPs agree that a concise and precise discharge summary should include an overview of the hospitalization with important details highlighted. Similar to the findings of Pantilat et al.,1 in our study PCPs specifically want detailed information with regard to diagnoses, discharge medications, and what to expect when they see the patient in their clinic. Follow‐up phone calls to PCPs to see that they received written information and if they require further details is 1 solution to ensuring good follow‐up, yet this adds to the burden of communication and could be an additional barrier.

The teaching institutions in which physicians train also pose unique obstacles to optimal communication. In academic medical centers, medical students and residents perform a majority of the discharge duties (eg, writing prescriptions, dictating discharge summaries, making follow‐up appointments, and calling PCPs), and teaching these trainees the importance of timely and accurate communication becomes an added challenge. Educators have to find novel ways of providing incentives to residents and medical students to get them to effectively participate in this process. Plauth et al.20 reports that hospitalists feel they needed better training in residency around communicating, noting a meaningful underemphasis during residency training in regard to communication with referring physicians. These skills should be taught in medical school and supported by both hospitalists and PCPs throughout residency training.

Both hospitalists and PCPs also want easy and reliable ways to access their colleagues, which ideally would be automatic. One PCP commented: a weekly or semiweekly phone call would be nice. Another suggested to: fax a short note. One hospitalist acknowledged: a systematic approach would be betterwhether a fax or telephone call and make sure there is a way of checking to make sure the communication has happened. Another hospitalist simply remarked: it needs to be done on every patient.

Thus, it seems an improved communication system should be flexible enough to accommodate unique provider preferences, such as communication via phone, fax, or e‐mail. This is demonstrated by 1 PCP who preferred the phone, but most convenient is the periodic fax updates. I don't have to be taken away from seeing patients.

Lo15 calls for a standard to be established for delivering care within the patient‐PCP‐hospitalist triad. Phone calls and faxes are 2 readily available methods of communication. However, the frequent back‐and‐forth of missed calls, unreturned calls, and days‐off is certainly a factor in determining efficiency and effectiveness of phone calls.

E‐mail, if it is widely used by all participants, may be an effective option for delivery that could provide confirmation of receipt. However, the lack of universal e‐mail usage by all providers remains a barrier. Questions as to which method is more time consuming and for whom, need to be studied further. Patient confidentiality also requires that this protected health information arrive in the proper hands. Personal relationships can also contribute to successful communication. One provider may be more likely to contact another if they know each other through some personal connection, such as medical school, residency, or a social group.

Our study has several limitations. The sample size was small. We obtained responses from a sample of key stakeholders in the hospitalist‐PCP communication process. We were limited by the number of hospitalists at our institution as well as the interest and availability of PCPs to respond. We are unable to determine the total number of patients by respondent PCP practice cared for by the hospitalist division. This could influence the results depending on whether the respondent PCP was a frequent or infrequent utilizer of the hospitalist system. However, we feel reassured that we are not missing important information, because in our methods, a priori, we had intended to stop interviewing PCPs once theoretical saturation had been reached (ie, respondents did not identify any new information). In our study, that occurred with 12 PCPs.

We attempted to interview a single physician in a number of different practice settings in order to gain insight into the perceptions of that individual as well as those of their partners. The views expressed by these individuals may not represent the views of hospitalists and PCPs outside of our practice area. Furthermore, PCMC serves as both a community pediatric hospital and a tertiary‐referral center for a large area, yet the current experience of 1 hospitalist division and 1 cohort of referring PCPs may contain regional variation that contributed bias to the responses.

Selection bias may have been introduced in our study by the inherent nature of phone interviews. We interviewed only providers with previous communication experience with our hospitalist division. These providers may have had a vested interest in the communication process. We did not interview those PCPs who did not have any communication with our hospitalist division or those who may have used the hospitalist division previously and decided to no longer use the division. Interviewing these groups may have provided additional insight into the communication issues mentioned here. Additionally, useful information could have been gleaned from trying to find out more from the 33% of PCPs who felt communication was poor. We anticipate further studies exploring this issue in more depth.

CONCLUSIONS

Hospitalists and PCPs agree that overall quality of communication ranges from poor to very good. Both PCPs and pediatric hospitalists acknowledge that significant barriers to optimal communication exist, yet the barriers differ for each group. They also agree that optimal communication would improve many aspects of patient care and should take place upon discharge and admission of patients and with major clinical changes.

Pediatric hospitalists and PCPs identified issues around optimal communication similar to those noted in the adult hospital medicine literature. Interventions to improve pediatric hospitalist‐PCP communication should at least address these 6 issues: (1) quality of communication; (2) barriers to communication; (3) methods of information sharing; (4) key data element requirements; (5) critical timing; and (6) perceived benefits. Such interventions will likely improve hospitalist‐PCP communication and potentially improve the quality of patient care. However, future studies will need to demonstrate the link between improved hospitalist‐PCP communication and improved patient care and outcomes.

Hospitalist systems focus on providing acute treatments to patients and expediting hospital discharge, sometimes without regard for the need to work in concert with community providers, leading to fragmentation of care.1 This fragmentation, particularly at the transitions of care, such as when patients move from the outpatient setting to a hospitalist system and then back to their primary care providers (PCPs), can lead to communication breakdowns and delays in care, and may compromise patient outcomes.2 Suboptimal treatments, such as medication errors and the ordering of redundant tests can occur in either setting if prior treatment information is not relayed in a timely and accurate fashion. Landrigan et al.3 described a conceptual model in 2001 that recognized the complexity of the hospitalist‐PCP communication system. Specifically, optimal care of hospitalized children includes PCPs, family members, hospitalists, and support staff while meeting the communication needs of families and PCPs. Additionally, mediating a smooth transition into and out of the hospital needs to be measured carefully.3

Previous adult studies have reported that hospitalist systems sometimes create discontinuity of patient care, which can have a negative impact on the quality of care provided to patients if there is poor communication between hospitalists and PCPs.2, 4, 5 Existing research on hospitalist‐PCP communication focuses mainly on adult hospitalist models with little known about the quality of current pediatric hospitalist‐PCP communication.

The objective of this study was to qualitatively explore issues around communication between pediatric hospitalists and PCPs. Specifically, we sought to explore the quality of communication practices and barriers to optimal communication within the hospitalist‐PCP model at a tertiary care children's hospital. The results are serving as a needs assessment to guide the design of a quality improvement project with the aim of improving pediatric hospitalist‐PCP communication.

METHODS

RESULTS

Only 1 physician per practice was interviewed. No PCP who was able to be contacted declined an interview, although some did require multiple phone attempts to schedule the interview. PCPs were located in Salt Lake County (n = 6), in other Utah counties (n = 3), and in surrounding Intermountain West states (n = 3). From January 1, 2004 to December 31, 2004, we estimate that the hospitalist division cared for patients from approximately 35 practices (50% in Salt Lake County, 30% in other Utah counties, and 20% in surrounding Intermountain West states).

Hospitalists and PCPs agreed that overall quality of communication ranged from poor to very good (Table 2). Both parties acknowledged that significant barriers to optimal communication exist, yet the barriers differed for each group. Hospitalists and PCPs also agreed that optimal communication could improve many aspects of patient care and should take place upon discharge and admission of patients and with major clinical changes. Both hospitalists and PCPs also wanted accurate and timely information. One priority that the participants emphasized is the timely transfer of admission notification and the receipt of accurate and timely discharge summaries by PCPs.

DISCUSSION

Both pediatric hospitalists and PCPs agree on what information is important to transmit (diagnoses, medications, follow‐up needs, and pending laboratory test results) and critical times for communication during the hospitalization (at discharge, admission, and during major clinical changes). However, there was discrepancy in the barriers to optimal communication for each group. Identifying and addressing these barriers can help both hospitalists and PCPs implement targeted interventions aimed at improving communication. As the number of pediatric hospitalist programs increases, the risk for hospitalist‐PCP communication breakdowns, which can have a negative impact on patient care, also increases.

Previous adult studies describe the scope of the problem around poor communication between hospitalists and PCPs.1, 912 Kripalani et al.10 reported recently that delays and omissions in communication are common at hospital discharge among adult hospitalists and that computer‐generated summaries, educational interventions, and standardized formats may facilitate more timely transfer of pertinent information. However, there is limited data on pediatric hospitalist‐PCP communication. Srivastava et al.5 found that 60% of community physicians thought hospitalist systems may impair communication with PCPs when evaluating community and hospital‐based physicians' attitudes regarding pediatric hospitalist systems.

PCPs can feel left out when their patients are cared for by hospitalists.13 One PCP in our study commented: Include the referral doc as part of the team. We're the ones who will take care of them after discharge. It seems like an autonomous thing down there and we're excluded from the patient care team. Additionally, patients want their PCPs to remain involved in their care as they transition into and out of the hospital setting.1, 13

The continuity visit model has been proposed by Wachter and Pantilat14 to describe a clinical encounter between the primary physician and hospitalized patient, when the patient has a different physician of record. In this model, the PCP can endorse the hospitalist model and the individual hospitalist, notice subtle findings that differ from the patient's baseline, and help clarify patient preferences regarding difficult situations by drawing on their previous relationship with the patient. This visit may also benefit the PCP by providing insights into the patient's illness, personality, or social support that he or she was unaware of previously. However, in order for the continuity visit to exist, the PCP has to be informed of their patient's admission in the first place. Ethical dilemmas also have been raised regarding who bears primary responsibility for maintaining open lines of communication when patients are hospitalized.15 Lo15 advocates that PCPs can and should be involved in meaningful ways in the inpatient care of their patients even when they are not acting as the treating physicians. Specifically, he suggests that PCPs personally visit particularly ill patients or those with difficult diagnoses and use frequent phone calls to all admitted patients.

Beyond telephone calls and continuity visits, hospitalists and PCPs rely on discharge summaries as a key part of the information transfer about a patient's hospitalization.1, 16, 17 These documents are rendered useless if they are inaccurate, illegible, or not delivered in a timely manner.18 In a study of California family physicians, discharge summaries were thought to be too detailed by 84% of PCPs, and reportedly arrived before the patient's first follow‐up appointment only 33% of the time.1 O'Leary et al.19 found that 41% of the Department of Medicine physicians surveyed believed that at least 1 of their patients hospitalized in the previous 6 months had experienced a preventable adverse event related to poor transfer of information at discharge. In our study, PCPs noted that discharge summaries often arrived in their offices well after the patient had been seen for their follow‐up appointment.

Both hospitalists and PCPs agree that a concise and precise discharge summary should include an overview of the hospitalization with important details highlighted. Similar to the findings of Pantilat et al.,1 in our study PCPs specifically want detailed information with regard to diagnoses, discharge medications, and what to expect when they see the patient in their clinic. Follow‐up phone calls to PCPs to see that they received written information and if they require further details is 1 solution to ensuring good follow‐up, yet this adds to the burden of communication and could be an additional barrier.

The teaching institutions in which physicians train also pose unique obstacles to optimal communication. In academic medical centers, medical students and residents perform a majority of the discharge duties (eg, writing prescriptions, dictating discharge summaries, making follow‐up appointments, and calling PCPs), and teaching these trainees the importance of timely and accurate communication becomes an added challenge. Educators have to find novel ways of providing incentives to residents and medical students to get them to effectively participate in this process. Plauth et al.20 reports that hospitalists feel they needed better training in residency around communicating, noting a meaningful underemphasis during residency training in regard to communication with referring physicians. These skills should be taught in medical school and supported by both hospitalists and PCPs throughout residency training.

Both hospitalists and PCPs also want easy and reliable ways to access their colleagues, which ideally would be automatic. One PCP commented: a weekly or semiweekly phone call would be nice. Another suggested to: fax a short note. One hospitalist acknowledged: a systematic approach would be betterwhether a fax or telephone call and make sure there is a way of checking to make sure the communication has happened. Another hospitalist simply remarked: it needs to be done on every patient.

Thus, it seems an improved communication system should be flexible enough to accommodate unique provider preferences, such as communication via phone, fax, or e‐mail. This is demonstrated by 1 PCP who preferred the phone, but most convenient is the periodic fax updates. I don't have to be taken away from seeing patients.

Lo15 calls for a standard to be established for delivering care within the patient‐PCP‐hospitalist triad. Phone calls and faxes are 2 readily available methods of communication. However, the frequent back‐and‐forth of missed calls, unreturned calls, and days‐off is certainly a factor in determining efficiency and effectiveness of phone calls.

E‐mail, if it is widely used by all participants, may be an effective option for delivery that could provide confirmation of receipt. However, the lack of universal e‐mail usage by all providers remains a barrier. Questions as to which method is more time consuming and for whom, need to be studied further. Patient confidentiality also requires that this protected health information arrive in the proper hands. Personal relationships can also contribute to successful communication. One provider may be more likely to contact another if they know each other through some personal connection, such as medical school, residency, or a social group.

Our study has several limitations. The sample size was small. We obtained responses from a sample of key stakeholders in the hospitalist‐PCP communication process. We were limited by the number of hospitalists at our institution as well as the interest and availability of PCPs to respond. We are unable to determine the total number of patients by respondent PCP practice cared for by the hospitalist division. This could influence the results depending on whether the respondent PCP was a frequent or infrequent utilizer of the hospitalist system. However, we feel reassured that we are not missing important information, because in our methods, a priori, we had intended to stop interviewing PCPs once theoretical saturation had been reached (ie, respondents did not identify any new information). In our study, that occurred with 12 PCPs.

We attempted to interview a single physician in a number of different practice settings in order to gain insight into the perceptions of that individual as well as those of their partners. The views expressed by these individuals may not represent the views of hospitalists and PCPs outside of our practice area. Furthermore, PCMC serves as both a community pediatric hospital and a tertiary‐referral center for a large area, yet the current experience of 1 hospitalist division and 1 cohort of referring PCPs may contain regional variation that contributed bias to the responses.

Selection bias may have been introduced in our study by the inherent nature of phone interviews. We interviewed only providers with previous communication experience with our hospitalist division. These providers may have had a vested interest in the communication process. We did not interview those PCPs who did not have any communication with our hospitalist division or those who may have used the hospitalist division previously and decided to no longer use the division. Interviewing these groups may have provided additional insight into the communication issues mentioned here. Additionally, useful information could have been gleaned from trying to find out more from the 33% of PCPs who felt communication was poor. We anticipate further studies exploring this issue in more depth.

CONCLUSIONS

Hospitalists and PCPs agree that overall quality of communication ranges from poor to very good. Both PCPs and pediatric hospitalists acknowledge that significant barriers to optimal communication exist, yet the barriers differ for each group. They also agree that optimal communication would improve many aspects of patient care and should take place upon discharge and admission of patients and with major clinical changes.

Pediatric hospitalists and PCPs identified issues around optimal communication similar to those noted in the adult hospital medicine literature. Interventions to improve pediatric hospitalist‐PCP communication should at least address these 6 issues: (1) quality of communication; (2) barriers to communication; (3) methods of information sharing; (4) key data element requirements; (5) critical timing; and (6) perceived benefits. Such interventions will likely improve hospitalist‐PCP communication and potentially improve the quality of patient care. However, future studies will need to demonstrate the link between improved hospitalist‐PCP communication and improved patient care and outcomes.

Hospitalist systems focus on providing acute treatments to patients and expediting hospital discharge, sometimes without regard for the need to work in concert with community providers, leading to fragmentation of care.1 This fragmentation, particularly at the transitions of care, such as when patients move from the outpatient setting to a hospitalist system and then back to their primary care providers (PCPs), can lead to communication breakdowns and delays in care, and may compromise patient outcomes.2 Suboptimal treatments, such as medication errors and the ordering of redundant tests can occur in either setting if prior treatment information is not relayed in a timely and accurate fashion. Landrigan et al.3 described a conceptual model in 2001 that recognized the complexity of the hospitalist‐PCP communication system. Specifically, optimal care of hospitalized children includes PCPs, family members, hospitalists, and support staff while meeting the communication needs of families and PCPs. Additionally, mediating a smooth transition into and out of the hospital needs to be measured carefully.3

Previous adult studies have reported that hospitalist systems sometimes create discontinuity of patient care, which can have a negative impact on the quality of care provided to patients if there is poor communication between hospitalists and PCPs.2, 4, 5 Existing research on hospitalist‐PCP communication focuses mainly on adult hospitalist models with little known about the quality of current pediatric hospitalist‐PCP communication.

The objective of this study was to qualitatively explore issues around communication between pediatric hospitalists and PCPs. Specifically, we sought to explore the quality of communication practices and barriers to optimal communication within the hospitalist‐PCP model at a tertiary care children's hospital. The results are serving as a needs assessment to guide the design of a quality improvement project with the aim of improving pediatric hospitalist‐PCP communication.

METHODS

RESULTS

Only 1 physician per practice was interviewed. No PCP who was able to be contacted declined an interview, although some did require multiple phone attempts to schedule the interview. PCPs were located in Salt Lake County (n = 6), in other Utah counties (n = 3), and in surrounding Intermountain West states (n = 3). From January 1, 2004 to December 31, 2004, we estimate that the hospitalist division cared for patients from approximately 35 practices (50% in Salt Lake County, 30% in other Utah counties, and 20% in surrounding Intermountain West states).

Hospitalists and PCPs agreed that overall quality of communication ranged from poor to very good (Table 2). Both parties acknowledged that significant barriers to optimal communication exist, yet the barriers differed for each group. Hospitalists and PCPs also agreed that optimal communication could improve many aspects of patient care and should take place upon discharge and admission of patients and with major clinical changes. Both hospitalists and PCPs also wanted accurate and timely information. One priority that the participants emphasized is the timely transfer of admission notification and the receipt of accurate and timely discharge summaries by PCPs.

DISCUSSION

Both pediatric hospitalists and PCPs agree on what information is important to transmit (diagnoses, medications, follow‐up needs, and pending laboratory test results) and critical times for communication during the hospitalization (at discharge, admission, and during major clinical changes). However, there was discrepancy in the barriers to optimal communication for each group. Identifying and addressing these barriers can help both hospitalists and PCPs implement targeted interventions aimed at improving communication. As the number of pediatric hospitalist programs increases, the risk for hospitalist‐PCP communication breakdowns, which can have a negative impact on patient care, also increases.

Previous adult studies describe the scope of the problem around poor communication between hospitalists and PCPs.1, 912 Kripalani et al.10 reported recently that delays and omissions in communication are common at hospital discharge among adult hospitalists and that computer‐generated summaries, educational interventions, and standardized formats may facilitate more timely transfer of pertinent information. However, there is limited data on pediatric hospitalist‐PCP communication. Srivastava et al.5 found that 60% of community physicians thought hospitalist systems may impair communication with PCPs when evaluating community and hospital‐based physicians' attitudes regarding pediatric hospitalist systems.

PCPs can feel left out when their patients are cared for by hospitalists.13 One PCP in our study commented: Include the referral doc as part of the team. We're the ones who will take care of them after discharge. It seems like an autonomous thing down there and we're excluded from the patient care team. Additionally, patients want their PCPs to remain involved in their care as they transition into and out of the hospital setting.1, 13

The continuity visit model has been proposed by Wachter and Pantilat14 to describe a clinical encounter between the primary physician and hospitalized patient, when the patient has a different physician of record. In this model, the PCP can endorse the hospitalist model and the individual hospitalist, notice subtle findings that differ from the patient's baseline, and help clarify patient preferences regarding difficult situations by drawing on their previous relationship with the patient. This visit may also benefit the PCP by providing insights into the patient's illness, personality, or social support that he or she was unaware of previously. However, in order for the continuity visit to exist, the PCP has to be informed of their patient's admission in the first place. Ethical dilemmas also have been raised regarding who bears primary responsibility for maintaining open lines of communication when patients are hospitalized.15 Lo15 advocates that PCPs can and should be involved in meaningful ways in the inpatient care of their patients even when they are not acting as the treating physicians. Specifically, he suggests that PCPs personally visit particularly ill patients or those with difficult diagnoses and use frequent phone calls to all admitted patients.

Beyond telephone calls and continuity visits, hospitalists and PCPs rely on discharge summaries as a key part of the information transfer about a patient's hospitalization.1, 16, 17 These documents are rendered useless if they are inaccurate, illegible, or not delivered in a timely manner.18 In a study of California family physicians, discharge summaries were thought to be too detailed by 84% of PCPs, and reportedly arrived before the patient's first follow‐up appointment only 33% of the time.1 O'Leary et al.19 found that 41% of the Department of Medicine physicians surveyed believed that at least 1 of their patients hospitalized in the previous 6 months had experienced a preventable adverse event related to poor transfer of information at discharge. In our study, PCPs noted that discharge summaries often arrived in their offices well after the patient had been seen for their follow‐up appointment.

Both hospitalists and PCPs agree that a concise and precise discharge summary should include an overview of the hospitalization with important details highlighted. Similar to the findings of Pantilat et al.,1 in our study PCPs specifically want detailed information with regard to diagnoses, discharge medications, and what to expect when they see the patient in their clinic. Follow‐up phone calls to PCPs to see that they received written information and if they require further details is 1 solution to ensuring good follow‐up, yet this adds to the burden of communication and could be an additional barrier.

The teaching institutions in which physicians train also pose unique obstacles to optimal communication. In academic medical centers, medical students and residents perform a majority of the discharge duties (eg, writing prescriptions, dictating discharge summaries, making follow‐up appointments, and calling PCPs), and teaching these trainees the importance of timely and accurate communication becomes an added challenge. Educators have to find novel ways of providing incentives to residents and medical students to get them to effectively participate in this process. Plauth et al.20 reports that hospitalists feel they needed better training in residency around communicating, noting a meaningful underemphasis during residency training in regard to communication with referring physicians. These skills should be taught in medical school and supported by both hospitalists and PCPs throughout residency training.

Both hospitalists and PCPs also want easy and reliable ways to access their colleagues, which ideally would be automatic. One PCP commented: a weekly or semiweekly phone call would be nice. Another suggested to: fax a short note. One hospitalist acknowledged: a systematic approach would be betterwhether a fax or telephone call and make sure there is a way of checking to make sure the communication has happened. Another hospitalist simply remarked: it needs to be done on every patient.

Thus, it seems an improved communication system should be flexible enough to accommodate unique provider preferences, such as communication via phone, fax, or e‐mail. This is demonstrated by 1 PCP who preferred the phone, but most convenient is the periodic fax updates. I don't have to be taken away from seeing patients.

Lo15 calls for a standard to be established for delivering care within the patient‐PCP‐hospitalist triad. Phone calls and faxes are 2 readily available methods of communication. However, the frequent back‐and‐forth of missed calls, unreturned calls, and days‐off is certainly a factor in determining efficiency and effectiveness of phone calls.

E‐mail, if it is widely used by all participants, may be an effective option for delivery that could provide confirmation of receipt. However, the lack of universal e‐mail usage by all providers remains a barrier. Questions as to which method is more time consuming and for whom, need to be studied further. Patient confidentiality also requires that this protected health information arrive in the proper hands. Personal relationships can also contribute to successful communication. One provider may be more likely to contact another if they know each other through some personal connection, such as medical school, residency, or a social group.

Our study has several limitations. The sample size was small. We obtained responses from a sample of key stakeholders in the hospitalist‐PCP communication process. We were limited by the number of hospitalists at our institution as well as the interest and availability of PCPs to respond. We are unable to determine the total number of patients by respondent PCP practice cared for by the hospitalist division. This could influence the results depending on whether the respondent PCP was a frequent or infrequent utilizer of the hospitalist system. However, we feel reassured that we are not missing important information, because in our methods, a priori, we had intended to stop interviewing PCPs once theoretical saturation had been reached (ie, respondents did not identify any new information). In our study, that occurred with 12 PCPs.

We attempted to interview a single physician in a number of different practice settings in order to gain insight into the perceptions of that individual as well as those of their partners. The views expressed by these individuals may not represent the views of hospitalists and PCPs outside of our practice area. Furthermore, PCMC serves as both a community pediatric hospital and a tertiary‐referral center for a large area, yet the current experience of 1 hospitalist division and 1 cohort of referring PCPs may contain regional variation that contributed bias to the responses.

Selection bias may have been introduced in our study by the inherent nature of phone interviews. We interviewed only providers with previous communication experience with our hospitalist division. These providers may have had a vested interest in the communication process. We did not interview those PCPs who did not have any communication with our hospitalist division or those who may have used the hospitalist division previously and decided to no longer use the division. Interviewing these groups may have provided additional insight into the communication issues mentioned here. Additionally, useful information could have been gleaned from trying to find out more from the 33% of PCPs who felt communication was poor. We anticipate further studies exploring this issue in more depth.

CONCLUSIONS

Hospitalists and PCPs agree that overall quality of communication ranges from poor to very good. Both PCPs and pediatric hospitalists acknowledge that significant barriers to optimal communication exist, yet the barriers differ for each group. They also agree that optimal communication would improve many aspects of patient care and should take place upon discharge and admission of patients and with major clinical changes.

Pediatric hospitalists and PCPs identified issues around optimal communication similar to those noted in the adult hospital medicine literature. Interventions to improve pediatric hospitalist‐PCP communication should at least address these 6 issues: (1) quality of communication; (2) barriers to communication; (3) methods of information sharing; (4) key data element requirements; (5) critical timing; and (6) perceived benefits. Such interventions will likely improve hospitalist‐PCP communication and potentially improve the quality of patient care. However, future studies will need to demonstrate the link between improved hospitalist‐PCP communication and improved patient care and outcomes.

The past decade has seen hospital medicine grow from fewer than 1000 hospitalists nationwide to more than 20,000.1 In fact, survey data suggest that hospital medicine is the fastest growing field of internal medicine in the history of the US, and the growth of hospital medicine has produced a net increase in the number of generalists in the US.2

Although few direct estimates exist, academic hospital medicine (AHM) is also growing rapidly.1 Fueled by potential efficiency gains, a need for increased educational oversight of teaching services, and new residency work hour limitations, many academic medical centers and teaching hospitals have developed large hospital medicine programs. Internal medicine residency graduates interested in general medicine are finding hospital medicine an increasingly popular career choice. As a result, AHM groups have many recent residency graduates with an average age that is generally younger than 40.3

Over 85% of hospitalists are generalists and should find natural alliances with the nonhospitalist side of general internal medicine by collaborating in the course of clinical care, by teaching residents and students, or by designing quality improvement or research projects. In many academic centers, hospitalists are part of the division of general internal medicine, whereas in a few centers, hospitalists either have a separate division or lie outside the internal medicine department (employed by their hospitals).

Despite sharing a common training background and generalist mindset, many new academic hospitalists face different challenges than those faced by pure outpatient‐based academic generalists. First, at many centers, the financial arrangements between the AHM group and the hospital discourage hospitalists from traditional academic pursuits and draw them into clinical, operational, or administrative duties (such as responsibility for utilization review) that, although locally valuable, may not count as academic products in themselves or may take time away from more academic activities. Close alignment between hospitals and AHM may result in hospital administrators dictating hospitalists' practice in a way that further impedes academic viability. Reductions in resident training hours and an increasing need to provide 24‐hour coverage have facilitated growth in AHM into roles beyond those of the traditional academic generalist, such as medical comanagement of surgical patients and coverage of nonteaching services.4, 5 The youth of the field may exacerbate these problems. Most academic hospitalist groups have few senior leaders, whether they are clinical‐, education‐, or research‐focused. Young faculty need senior leaders as mentors to buffer them from relentless clinical demands that would compromise their hopes for academic success.

In order to better characterize these concerns and develop a shared work plan for future activities in support of AHM, the Society of Hospital Medicine (SHM) and the Society of General Internal Medicine (SGIM) convened an AHM consensus conference, a collaborative meeting developed and attended by representatives from SHM, SGIM, the Association of Chiefs of General Internal Medicine (ACGIM), the Association of Professors of Medicine, the Association of Program Directors in Internal Medicine, and the Association of Administrators in Internal Medicine. Using a structured consensus‐building format, we identified key barriers and challenges to AHM, then developed potential solutions.

Consensus Conference Format

Consensus Findings 1: Current Challenges in AHM (Table 1)

Consensus Findings 2: Overcoming Challenges to the Development of AHM (Table 2)

Summit attendees spent considerable time developing and refining solutions to the challenges described previously. Addressing the challenges resulted in a diverse group of proposed products that included educating key stakeholders, designing meetings, courses, or workshops, and gathering and disseminating data. There was considerable overlap among the solutions (Table 2).

Conclusions

AHM is at a crossroads. Unparalleled growth has created a large cadre of hospitalists who are struggling to meet the clinical demands of practice and the requirements for academic promotion; this situation will likely lead to, at a minimum, worsening problems with faculty turnover, and even greater losses of talented and passionate clinicians from the field of academic General Internal Medicine.

The challenges are numerous but not insurmountable, and our process identified issues and potential solutions which address clinical, educational, and research aspects of academic hospitalists' lives. We acknowledge that our findings are most relevant to hospitalists at academic medical centers or large academically oriented community teaching hospitals rather than hospitalists at community hospitals whose work is predominantly clinical with smaller teaching roles. However, we feel the academic hospitalists we targeted are in greater need of assistance. We believe that the most important issues are unsustainable, nonacademic positions, poor job preparation and training, inadequate prioritization of academic roles, and insufficient leadership and mentoring within the field.

It is the hope of all the consensus conference attendees that efforts focusing on academic hospitalists in the short term are not viewed as effort diverted from general internal medicine; in fact, the group felt that while many of the products of the consensus conference were probably most needed by AHM in the short term, these same solutions would likely be useful to outpatient‐based generalists as well. Despite the concerns and challenges outlined, the consensus conference group was also very hopeful that, in the setting where resources and collaboration are appropriately marshaled, that AHM will flourish quickly. In doing so, academic hospitalists will become better role models for residents and students, attracting the next generation of generalists needed to provide care to an increasingly complex patient population, and further advance the mission of General Internal Medicine.

The past decade has seen hospital medicine grow from fewer than 1000 hospitalists nationwide to more than 20,000.1 In fact, survey data suggest that hospital medicine is the fastest growing field of internal medicine in the history of the US, and the growth of hospital medicine has produced a net increase in the number of generalists in the US.2

Although few direct estimates exist, academic hospital medicine (AHM) is also growing rapidly.1 Fueled by potential efficiency gains, a need for increased educational oversight of teaching services, and new residency work hour limitations, many academic medical centers and teaching hospitals have developed large hospital medicine programs. Internal medicine residency graduates interested in general medicine are finding hospital medicine an increasingly popular career choice. As a result, AHM groups have many recent residency graduates with an average age that is generally younger than 40.3

Over 85% of hospitalists are generalists and should find natural alliances with the nonhospitalist side of general internal medicine by collaborating in the course of clinical care, by teaching residents and students, or by designing quality improvement or research projects. In many academic centers, hospitalists are part of the division of general internal medicine, whereas in a few centers, hospitalists either have a separate division or lie outside the internal medicine department (employed by their hospitals).

Despite sharing a common training background and generalist mindset, many new academic hospitalists face different challenges than those faced by pure outpatient‐based academic generalists. First, at many centers, the financial arrangements between the AHM group and the hospital discourage hospitalists from traditional academic pursuits and draw them into clinical, operational, or administrative duties (such as responsibility for utilization review) that, although locally valuable, may not count as academic products in themselves or may take time away from more academic activities. Close alignment between hospitals and AHM may result in hospital administrators dictating hospitalists' practice in a way that further impedes academic viability. Reductions in resident training hours and an increasing need to provide 24‐hour coverage have facilitated growth in AHM into roles beyond those of the traditional academic generalist, such as medical comanagement of surgical patients and coverage of nonteaching services.4, 5 The youth of the field may exacerbate these problems. Most academic hospitalist groups have few senior leaders, whether they are clinical‐, education‐, or research‐focused. Young faculty need senior leaders as mentors to buffer them from relentless clinical demands that would compromise their hopes for academic success.

In order to better characterize these concerns and develop a shared work plan for future activities in support of AHM, the Society of Hospital Medicine (SHM) and the Society of General Internal Medicine (SGIM) convened an AHM consensus conference, a collaborative meeting developed and attended by representatives from SHM, SGIM, the Association of Chiefs of General Internal Medicine (ACGIM), the Association of Professors of Medicine, the Association of Program Directors in Internal Medicine, and the Association of Administrators in Internal Medicine. Using a structured consensus‐building format, we identified key barriers and challenges to AHM, then developed potential solutions.

Consensus Conference Format

Consensus Findings 1: Current Challenges in AHM (Table 1)

Consensus Findings 2: Overcoming Challenges to the Development of AHM (Table 2)

Summit attendees spent considerable time developing and refining solutions to the challenges described previously. Addressing the challenges resulted in a diverse group of proposed products that included educating key stakeholders, designing meetings, courses, or workshops, and gathering and disseminating data. There was considerable overlap among the solutions (Table 2).