Christopher L. Roy, MD

Failures in communication among healthcare professionals are known threats to patient safety. These failures account for over 60% of root causes of sentinel events, the most serious events reported to The Joint Commission.[1] As such, identifying both patterns of effective communication as well as barriers to successful communication has been a focus of efforts aimed at improving patient safety. However, to date, the majority of this work has centered on improving communication in settings such as the operating room and intensive care unit,[2, 3, 4] or at times of care transitions.[5, 6, 7, 8]

Unique barriers exist for effective interdisciplinary communication in the hospital setting, particularly physiciannurse communication regarding shared hospitalized patients.[9] Traditionally, care of hospitalized patients is provided by physicians, nurses, and other team members working in varied workflow patterns, leading to dispersed team membership, where each team member cares for different groups of patients in different locations across the hospital. This dispersion is further heightened on teaching services, where residents' rotation schedules lead to frequent changes of care team membership, leaving inpatient care teams particularly vulnerable to ineffective communication. Evidence suggests that communication between nurses and physicians is currently suboptimal, leading to frequent disagreement regarding the patient's plan of care.[9, 10] This divergence between physician and nursing perceptions of patients' care plans may leave patients at greater risk of adverse events (AEs).

Several studies have examined the effects of regionalized inpatient care teams, where multidisciplinary team members care for the same patients on the same hospital unit, on communication and patient outcomes.[4, 11, 12, 13, 14] Results of these studies have been inconsistent, perhaps due to the particular characteristics of the care teams or to the study methodology. Thus, further rigorously done studies are required to better understand the impact of team regionalization on patient care. The goal of this study was to examine whether the implementation of regionalized inpatient care teams was associated with improvements in care team communication and preventable AEs.

METHODS

Intervention

On June 20, 2012, regionalized care was implemented on the GMS such that each of 3 GMS teams was localized to 1 of 3, 15‐bed nursing units. Prior to regionalization, the GMS physician care teams, each consisting of 1 hospitalist attending, 1 medical resident, and 2 medical interns, would care for patients on an average of 7 and up to 13 different nursing units on a given day.

Regionalized care consisted of a multifaceted intervention codeveloped by hospitalist, residency, nursing, emergency department, and hospital leadership and included: (1) regionalizing GMS teams as much as possible; (2) change in resident call structure from a traditional 4‐day call cycle to daily admitting; (3) collaborative efforts to enhance GMS patient discharges before noon to promote regionalized placement of patients without prolonging time in the emergency department (ED); (4) daily morning and postround multidisciplinary huddles to prioritize sicker patients and discharges; (5) encouragement of daily rounds at patients' bedsides with presence of physician team, nurse, and team pharmacist if available; (6) creation of unit‐ and team‐level performance reports; and (7) creation of unit‐based physician and nursing co‐leadership (Figure 1).[15]

Figure 1

RESULTS

The fidelity of the intervention in achieving its goal of regionalized care is discussed separately.[15] Briefly, the intervention was successful at achieving 85% regionalization by team (ie, average daily percentage of team's patients assigned to team's unit) and 87% regionalization by unit (ie, average daily percentage of unit's patients with assigned team) following implementation, compared to 20% regionalization by team and unit in the preintervention period. Importantly, the average daily census of physician care teams rose by 32%, from a mean of 10.8 patients/physician care team preregionalization to a mean of 14.3 patients/physician care team postregionalization.

DISCUSSION

In this study of general medicine patients at a large academic medical center, we found that regionalization of care teams on general medicine services was associated with improved recognition of care team members and agreement on estimated date of patient discharge, but was not associated with improvement in overall nurse and physician concordance of the patient care plan, or the odds of preventable AEs.

This intervention importantly addresses the barrier of dispersion of team membership, a well‐recognized barrier to interdisciplinary collaboration,[17, 18] particularly with resident physician teams due to frequently changing team membership. Localization of all team members, in addition to encouragement of daily collaborative bedside rounds as part of the regionalization initiative, likely contributed to our observed improvement in team member identification and discussion of daily care plans. Similarly, regionalization resulted in improved agreement in estimations of date of patient discharge. Focus on early patient discharges was an integral part of the implementation efforts; we therefore hypothesize that mutual focus on discharge planning by both nurses and responding clinicians may have explained this observed result.

On the other hand, regionalization did not appreciably improve the overall concordance of care plan between nurses and interns, despite a significant increase in team members agreeing that the plan had been discussed. Our findings support similar prior research demonstrating that regionalizing hospitalist attendings to single nursing units had limited impact on agreement of care plan between physicians and nurses.[13] Similarly, in settings where physicians and nurses are inherently regionalized, such as the intensive care unit[4] or the operating room,[3] communication between physicians and nurses remains difficult. Collectively, our findings suggest that colocalization of physicians and nurses alone is likely insufficient to improve measured communication between care team members. Existing literature suggests that more standardized approaches to improve communication, such as structured communication tools used during daily inpatient care[19, 20] or formalized team training,[21, 22, 23] lead to improvements in communication and collaboration. Despite these findings, it is important to highlight that this study did not assess other measures of workplace culture, such as teamwork and care team cohesiveness, which may have been positively affected by this intervention, even without measurable effect on concordance of care plan. Additionally, as noted, the average daily census on each team increased by almost a third postintervention, which may have impeded improvements in care team communication.

In addition, we found that our intervention had no significant impact on preventable AEs or severe preventable AEs. Although we cannot exclude the possibility that more subtle AEs were missed with our methodology, our results indicate that regionalized care alone may be inadequate to improve major patient safety outcomes. As discussed, the volume of patients did increase postintervention; thus, another way to state our results is that we were able to increase the daily volume of patients without any significant decreases in patient safety. Nevertheless, the results on patient safety were less than desired. A recent review of interdisciplinary team care interventions on general medical wards similarly demonstrated underwhelming improvements in patient safety outcomes, although the reviewed interventions did not specifically address preventable AEs, a gap in the literature commented on by the authors.[24] Other albeit limited literature has demonstrated improvement in patient safety outcomes via multifaceted efforts aimed at improving care team member communication. Notably, these efforts include colocalization of care team members to single units but also involve additional measures to improve communication and collaboration between care team members, such as structured communication during interdisciplinary rounds, and certification of key interdisciplinary teamwork skills.[11, 14] Although our regionalized care intervention included many similar features to these accountable care units (ACUs) including unit‐based care teams, unit‐level performance reporting, and unit‐based physician and nursing coleadership, significant differences existed. Notably, in addition to the above features, the ACU model also incorporated highly structured communication models for interdisciplinary rounding, and certification processes to ensure an appropriate communication skill base among care team members.[14] Thus, although creation of regionalized care teams is likely a necessary precursor to implementation of these additional measures, alone it may be insufficient to improve patient safety outcomes.

Importantly, in our study we identified that adverse drug events and manifestations of poor glycemic control occurred in high frequency both before and following implementation of regionalized care, supporting other literature that describes the prevalence of these AEs.[11, 25, 26, 27] These results suggest that targeted interventions to address these specific AEs are likely necessary. Notably, the intervention units in our study did not consistently employ clinical pharmacists assigned specifically to that unit's care team to allow for integration within the care team. As prior research has suggested that greater collaboration with clinical pharmacists results in reduction of adverse drug events,[28] next steps may include improved integration of team‐based pharmacists into the activities of the regionalized care teams. Inpatient management of diabetes also requires specific interventions,[29, 30, 31] only some of which may be addressable by having regionalized care and better interdisciplinary communication.

Our findings are subject to several limitations. First, this was a single‐site study and thus our findings may not be generalizable to other institutions. However, regionalized care is increasingly encouraged to optimize communication between care team members.[17, 18] Therefore, our null findings may be pertinent to other institutions looking to improve patient safety outcomes, demonstrating that additional initiatives will likely be required. Second, our modes of outcome measurement possess limitations. In measuring concordance of care plan, although previously used survey techniques were employed,[9] the concordance survey has not been formally validated, and we believe some of the questions may have led to ambiguity on the part of the responders that may have resulted in less accurate responses, thus biasing toward the null. Similarly, in measuring AEs, the screening tool relied on retrospective chart review looking for specific AE types[11] and thus may not have captured more subtle AEs. Additionally, our study may have been underpowered to demonstrate significant reduction in preventable AEs, although other studies of similar methodology demonstrated significant results with similar sample size.[11] This was due in part to our lower‐than‐expected baseline AE rate (6.6% compared with approximately 10.3% in previous studies).[11] Lastly, our study solely examined the association of regionalization with concordance of care plan and preventable AEs, but importantly excluded other clinically important outcomes that may have been positively (or negatively) impacted by these regionalization efforts, such as ED wait times, provider efficiency (eg, fewer pages, less time in transit, more time at the bedside), interdisciplinary teamwork, or patient or provider satisfaction.

CONCLUSION

In summary, our findings suggest that regionalized care teams alone may be insufficient to effectively promote communication between care team members regarding the care plan or to lead to improvements in patient safety, although we recognize that there may have been benefits (or unintended harms) not measured in this study but are nonetheless important for clinical care and workplace culture. This is an important lesson, as many hospitals move toward regionalized care in an effort to improve patient safety outcomes. However, strengthening the infrastructure by colocalizing care team members to maximize opportunity for communication is likely a necessary first step toward facilitating implementation of additional initiatives that may lead to more robust patient safety improvements, such as structured interdisciplinary bedside rounds (eg, facilitating and training all team members to fulfill specific roles), teamwork training, and certification of key interdisciplinary teamwork skills. Additionally, close examination of identified prevalent and preventable AEs can help to determine which additional initiatives are most likely to have greatest impact in improving patient safety.

Disclosures: This research was supported by funds provided by Brigham and Women's Hospital (BWH) and by funds provided by the Department of Medicine at BWH. All authors had full access to all of the data in the study and were integrally involved in the design, implementation, data collection, and analyses. The first author, Dr. Stephanie Mueller, takes responsibility for the integrity for the data and the accuracy of the data analysis. Dr. Schnipper reports grants from Sanofi Aventis, outside the submitted work.

Failures in communication among healthcare professionals are known threats to patient safety. These failures account for over 60% of root causes of sentinel events, the most serious events reported to The Joint Commission.[1] As such, identifying both patterns of effective communication as well as barriers to successful communication has been a focus of efforts aimed at improving patient safety. However, to date, the majority of this work has centered on improving communication in settings such as the operating room and intensive care unit,[2, 3, 4] or at times of care transitions.[5, 6, 7, 8]

Unique barriers exist for effective interdisciplinary communication in the hospital setting, particularly physiciannurse communication regarding shared hospitalized patients.[9] Traditionally, care of hospitalized patients is provided by physicians, nurses, and other team members working in varied workflow patterns, leading to dispersed team membership, where each team member cares for different groups of patients in different locations across the hospital. This dispersion is further heightened on teaching services, where residents' rotation schedules lead to frequent changes of care team membership, leaving inpatient care teams particularly vulnerable to ineffective communication. Evidence suggests that communication between nurses and physicians is currently suboptimal, leading to frequent disagreement regarding the patient's plan of care.[9, 10] This divergence between physician and nursing perceptions of patients' care plans may leave patients at greater risk of adverse events (AEs).

Several studies have examined the effects of regionalized inpatient care teams, where multidisciplinary team members care for the same patients on the same hospital unit, on communication and patient outcomes.[4, 11, 12, 13, 14] Results of these studies have been inconsistent, perhaps due to the particular characteristics of the care teams or to the study methodology. Thus, further rigorously done studies are required to better understand the impact of team regionalization on patient care. The goal of this study was to examine whether the implementation of regionalized inpatient care teams was associated with improvements in care team communication and preventable AEs.

METHODS

Intervention

On June 20, 2012, regionalized care was implemented on the GMS such that each of 3 GMS teams was localized to 1 of 3, 15‐bed nursing units. Prior to regionalization, the GMS physician care teams, each consisting of 1 hospitalist attending, 1 medical resident, and 2 medical interns, would care for patients on an average of 7 and up to 13 different nursing units on a given day.

Regionalized care consisted of a multifaceted intervention codeveloped by hospitalist, residency, nursing, emergency department, and hospital leadership and included: (1) regionalizing GMS teams as much as possible; (2) change in resident call structure from a traditional 4‐day call cycle to daily admitting; (3) collaborative efforts to enhance GMS patient discharges before noon to promote regionalized placement of patients without prolonging time in the emergency department (ED); (4) daily morning and postround multidisciplinary huddles to prioritize sicker patients and discharges; (5) encouragement of daily rounds at patients' bedsides with presence of physician team, nurse, and team pharmacist if available; (6) creation of unit‐ and team‐level performance reports; and (7) creation of unit‐based physician and nursing co‐leadership (Figure 1).[15]

Figure 1

RESULTS

The fidelity of the intervention in achieving its goal of regionalized care is discussed separately.[15] Briefly, the intervention was successful at achieving 85% regionalization by team (ie, average daily percentage of team's patients assigned to team's unit) and 87% regionalization by unit (ie, average daily percentage of unit's patients with assigned team) following implementation, compared to 20% regionalization by team and unit in the preintervention period. Importantly, the average daily census of physician care teams rose by 32%, from a mean of 10.8 patients/physician care team preregionalization to a mean of 14.3 patients/physician care team postregionalization.

DISCUSSION

In this study of general medicine patients at a large academic medical center, we found that regionalization of care teams on general medicine services was associated with improved recognition of care team members and agreement on estimated date of patient discharge, but was not associated with improvement in overall nurse and physician concordance of the patient care plan, or the odds of preventable AEs.

This intervention importantly addresses the barrier of dispersion of team membership, a well‐recognized barrier to interdisciplinary collaboration,[17, 18] particularly with resident physician teams due to frequently changing team membership. Localization of all team members, in addition to encouragement of daily collaborative bedside rounds as part of the regionalization initiative, likely contributed to our observed improvement in team member identification and discussion of daily care plans. Similarly, regionalization resulted in improved agreement in estimations of date of patient discharge. Focus on early patient discharges was an integral part of the implementation efforts; we therefore hypothesize that mutual focus on discharge planning by both nurses and responding clinicians may have explained this observed result.

On the other hand, regionalization did not appreciably improve the overall concordance of care plan between nurses and interns, despite a significant increase in team members agreeing that the plan had been discussed. Our findings support similar prior research demonstrating that regionalizing hospitalist attendings to single nursing units had limited impact on agreement of care plan between physicians and nurses.[13] Similarly, in settings where physicians and nurses are inherently regionalized, such as the intensive care unit[4] or the operating room,[3] communication between physicians and nurses remains difficult. Collectively, our findings suggest that colocalization of physicians and nurses alone is likely insufficient to improve measured communication between care team members. Existing literature suggests that more standardized approaches to improve communication, such as structured communication tools used during daily inpatient care[19, 20] or formalized team training,[21, 22, 23] lead to improvements in communication and collaboration. Despite these findings, it is important to highlight that this study did not assess other measures of workplace culture, such as teamwork and care team cohesiveness, which may have been positively affected by this intervention, even without measurable effect on concordance of care plan. Additionally, as noted, the average daily census on each team increased by almost a third postintervention, which may have impeded improvements in care team communication.

In addition, we found that our intervention had no significant impact on preventable AEs or severe preventable AEs. Although we cannot exclude the possibility that more subtle AEs were missed with our methodology, our results indicate that regionalized care alone may be inadequate to improve major patient safety outcomes. As discussed, the volume of patients did increase postintervention; thus, another way to state our results is that we were able to increase the daily volume of patients without any significant decreases in patient safety. Nevertheless, the results on patient safety were less than desired. A recent review of interdisciplinary team care interventions on general medical wards similarly demonstrated underwhelming improvements in patient safety outcomes, although the reviewed interventions did not specifically address preventable AEs, a gap in the literature commented on by the authors.[24] Other albeit limited literature has demonstrated improvement in patient safety outcomes via multifaceted efforts aimed at improving care team member communication. Notably, these efforts include colocalization of care team members to single units but also involve additional measures to improve communication and collaboration between care team members, such as structured communication during interdisciplinary rounds, and certification of key interdisciplinary teamwork skills.[11, 14] Although our regionalized care intervention included many similar features to these accountable care units (ACUs) including unit‐based care teams, unit‐level performance reporting, and unit‐based physician and nursing coleadership, significant differences existed. Notably, in addition to the above features, the ACU model also incorporated highly structured communication models for interdisciplinary rounding, and certification processes to ensure an appropriate communication skill base among care team members.[14] Thus, although creation of regionalized care teams is likely a necessary precursor to implementation of these additional measures, alone it may be insufficient to improve patient safety outcomes.

Importantly, in our study we identified that adverse drug events and manifestations of poor glycemic control occurred in high frequency both before and following implementation of regionalized care, supporting other literature that describes the prevalence of these AEs.[11, 25, 26, 27] These results suggest that targeted interventions to address these specific AEs are likely necessary. Notably, the intervention units in our study did not consistently employ clinical pharmacists assigned specifically to that unit's care team to allow for integration within the care team. As prior research has suggested that greater collaboration with clinical pharmacists results in reduction of adverse drug events,[28] next steps may include improved integration of team‐based pharmacists into the activities of the regionalized care teams. Inpatient management of diabetes also requires specific interventions,[29, 30, 31] only some of which may be addressable by having regionalized care and better interdisciplinary communication.

Our findings are subject to several limitations. First, this was a single‐site study and thus our findings may not be generalizable to other institutions. However, regionalized care is increasingly encouraged to optimize communication between care team members.[17, 18] Therefore, our null findings may be pertinent to other institutions looking to improve patient safety outcomes, demonstrating that additional initiatives will likely be required. Second, our modes of outcome measurement possess limitations. In measuring concordance of care plan, although previously used survey techniques were employed,[9] the concordance survey has not been formally validated, and we believe some of the questions may have led to ambiguity on the part of the responders that may have resulted in less accurate responses, thus biasing toward the null. Similarly, in measuring AEs, the screening tool relied on retrospective chart review looking for specific AE types[11] and thus may not have captured more subtle AEs. Additionally, our study may have been underpowered to demonstrate significant reduction in preventable AEs, although other studies of similar methodology demonstrated significant results with similar sample size.[11] This was due in part to our lower‐than‐expected baseline AE rate (6.6% compared with approximately 10.3% in previous studies).[11] Lastly, our study solely examined the association of regionalization with concordance of care plan and preventable AEs, but importantly excluded other clinically important outcomes that may have been positively (or negatively) impacted by these regionalization efforts, such as ED wait times, provider efficiency (eg, fewer pages, less time in transit, more time at the bedside), interdisciplinary teamwork, or patient or provider satisfaction.

CONCLUSION

In summary, our findings suggest that regionalized care teams alone may be insufficient to effectively promote communication between care team members regarding the care plan or to lead to improvements in patient safety, although we recognize that there may have been benefits (or unintended harms) not measured in this study but are nonetheless important for clinical care and workplace culture. This is an important lesson, as many hospitals move toward regionalized care in an effort to improve patient safety outcomes. However, strengthening the infrastructure by colocalizing care team members to maximize opportunity for communication is likely a necessary first step toward facilitating implementation of additional initiatives that may lead to more robust patient safety improvements, such as structured interdisciplinary bedside rounds (eg, facilitating and training all team members to fulfill specific roles), teamwork training, and certification of key interdisciplinary teamwork skills. Additionally, close examination of identified prevalent and preventable AEs can help to determine which additional initiatives are most likely to have greatest impact in improving patient safety.

Disclosures: This research was supported by funds provided by Brigham and Women's Hospital (BWH) and by funds provided by the Department of Medicine at BWH. All authors had full access to all of the data in the study and were integrally involved in the design, implementation, data collection, and analyses. The first author, Dr. Stephanie Mueller, takes responsibility for the integrity for the data and the accuracy of the data analysis. Dr. Schnipper reports grants from Sanofi Aventis, outside the submitted work.

Failures in communication among healthcare professionals are known threats to patient safety. These failures account for over 60% of root causes of sentinel events, the most serious events reported to The Joint Commission.[1] As such, identifying both patterns of effective communication as well as barriers to successful communication has been a focus of efforts aimed at improving patient safety. However, to date, the majority of this work has centered on improving communication in settings such as the operating room and intensive care unit,[2, 3, 4] or at times of care transitions.[5, 6, 7, 8]

Unique barriers exist for effective interdisciplinary communication in the hospital setting, particularly physiciannurse communication regarding shared hospitalized patients.[9] Traditionally, care of hospitalized patients is provided by physicians, nurses, and other team members working in varied workflow patterns, leading to dispersed team membership, where each team member cares for different groups of patients in different locations across the hospital. This dispersion is further heightened on teaching services, where residents' rotation schedules lead to frequent changes of care team membership, leaving inpatient care teams particularly vulnerable to ineffective communication. Evidence suggests that communication between nurses and physicians is currently suboptimal, leading to frequent disagreement regarding the patient's plan of care.[9, 10] This divergence between physician and nursing perceptions of patients' care plans may leave patients at greater risk of adverse events (AEs).

Several studies have examined the effects of regionalized inpatient care teams, where multidisciplinary team members care for the same patients on the same hospital unit, on communication and patient outcomes.[4, 11, 12, 13, 14] Results of these studies have been inconsistent, perhaps due to the particular characteristics of the care teams or to the study methodology. Thus, further rigorously done studies are required to better understand the impact of team regionalization on patient care. The goal of this study was to examine whether the implementation of regionalized inpatient care teams was associated with improvements in care team communication and preventable AEs.

METHODS

Intervention

On June 20, 2012, regionalized care was implemented on the GMS such that each of 3 GMS teams was localized to 1 of 3, 15‐bed nursing units. Prior to regionalization, the GMS physician care teams, each consisting of 1 hospitalist attending, 1 medical resident, and 2 medical interns, would care for patients on an average of 7 and up to 13 different nursing units on a given day.

Regionalized care consisted of a multifaceted intervention codeveloped by hospitalist, residency, nursing, emergency department, and hospital leadership and included: (1) regionalizing GMS teams as much as possible; (2) change in resident call structure from a traditional 4‐day call cycle to daily admitting; (3) collaborative efforts to enhance GMS patient discharges before noon to promote regionalized placement of patients without prolonging time in the emergency department (ED); (4) daily morning and postround multidisciplinary huddles to prioritize sicker patients and discharges; (5) encouragement of daily rounds at patients' bedsides with presence of physician team, nurse, and team pharmacist if available; (6) creation of unit‐ and team‐level performance reports; and (7) creation of unit‐based physician and nursing co‐leadership (Figure 1).[15]

Figure 1

RESULTS

The fidelity of the intervention in achieving its goal of regionalized care is discussed separately.[15] Briefly, the intervention was successful at achieving 85% regionalization by team (ie, average daily percentage of team's patients assigned to team's unit) and 87% regionalization by unit (ie, average daily percentage of unit's patients with assigned team) following implementation, compared to 20% regionalization by team and unit in the preintervention period. Importantly, the average daily census of physician care teams rose by 32%, from a mean of 10.8 patients/physician care team preregionalization to a mean of 14.3 patients/physician care team postregionalization.

DISCUSSION

In this study of general medicine patients at a large academic medical center, we found that regionalization of care teams on general medicine services was associated with improved recognition of care team members and agreement on estimated date of patient discharge, but was not associated with improvement in overall nurse and physician concordance of the patient care plan, or the odds of preventable AEs.

This intervention importantly addresses the barrier of dispersion of team membership, a well‐recognized barrier to interdisciplinary collaboration,[17, 18] particularly with resident physician teams due to frequently changing team membership. Localization of all team members, in addition to encouragement of daily collaborative bedside rounds as part of the regionalization initiative, likely contributed to our observed improvement in team member identification and discussion of daily care plans. Similarly, regionalization resulted in improved agreement in estimations of date of patient discharge. Focus on early patient discharges was an integral part of the implementation efforts; we therefore hypothesize that mutual focus on discharge planning by both nurses and responding clinicians may have explained this observed result.

On the other hand, regionalization did not appreciably improve the overall concordance of care plan between nurses and interns, despite a significant increase in team members agreeing that the plan had been discussed. Our findings support similar prior research demonstrating that regionalizing hospitalist attendings to single nursing units had limited impact on agreement of care plan between physicians and nurses.[13] Similarly, in settings where physicians and nurses are inherently regionalized, such as the intensive care unit[4] or the operating room,[3] communication between physicians and nurses remains difficult. Collectively, our findings suggest that colocalization of physicians and nurses alone is likely insufficient to improve measured communication between care team members. Existing literature suggests that more standardized approaches to improve communication, such as structured communication tools used during daily inpatient care[19, 20] or formalized team training,[21, 22, 23] lead to improvements in communication and collaboration. Despite these findings, it is important to highlight that this study did not assess other measures of workplace culture, such as teamwork and care team cohesiveness, which may have been positively affected by this intervention, even without measurable effect on concordance of care plan. Additionally, as noted, the average daily census on each team increased by almost a third postintervention, which may have impeded improvements in care team communication.

In addition, we found that our intervention had no significant impact on preventable AEs or severe preventable AEs. Although we cannot exclude the possibility that more subtle AEs were missed with our methodology, our results indicate that regionalized care alone may be inadequate to improve major patient safety outcomes. As discussed, the volume of patients did increase postintervention; thus, another way to state our results is that we were able to increase the daily volume of patients without any significant decreases in patient safety. Nevertheless, the results on patient safety were less than desired. A recent review of interdisciplinary team care interventions on general medical wards similarly demonstrated underwhelming improvements in patient safety outcomes, although the reviewed interventions did not specifically address preventable AEs, a gap in the literature commented on by the authors.[24] Other albeit limited literature has demonstrated improvement in patient safety outcomes via multifaceted efforts aimed at improving care team member communication. Notably, these efforts include colocalization of care team members to single units but also involve additional measures to improve communication and collaboration between care team members, such as structured communication during interdisciplinary rounds, and certification of key interdisciplinary teamwork skills.[11, 14] Although our regionalized care intervention included many similar features to these accountable care units (ACUs) including unit‐based care teams, unit‐level performance reporting, and unit‐based physician and nursing coleadership, significant differences existed. Notably, in addition to the above features, the ACU model also incorporated highly structured communication models for interdisciplinary rounding, and certification processes to ensure an appropriate communication skill base among care team members.[14] Thus, although creation of regionalized care teams is likely a necessary precursor to implementation of these additional measures, alone it may be insufficient to improve patient safety outcomes.

Importantly, in our study we identified that adverse drug events and manifestations of poor glycemic control occurred in high frequency both before and following implementation of regionalized care, supporting other literature that describes the prevalence of these AEs.[11, 25, 26, 27] These results suggest that targeted interventions to address these specific AEs are likely necessary. Notably, the intervention units in our study did not consistently employ clinical pharmacists assigned specifically to that unit's care team to allow for integration within the care team. As prior research has suggested that greater collaboration with clinical pharmacists results in reduction of adverse drug events,[28] next steps may include improved integration of team‐based pharmacists into the activities of the regionalized care teams. Inpatient management of diabetes also requires specific interventions,[29, 30, 31] only some of which may be addressable by having regionalized care and better interdisciplinary communication.

Our findings are subject to several limitations. First, this was a single‐site study and thus our findings may not be generalizable to other institutions. However, regionalized care is increasingly encouraged to optimize communication between care team members.[17, 18] Therefore, our null findings may be pertinent to other institutions looking to improve patient safety outcomes, demonstrating that additional initiatives will likely be required. Second, our modes of outcome measurement possess limitations. In measuring concordance of care plan, although previously used survey techniques were employed,[9] the concordance survey has not been formally validated, and we believe some of the questions may have led to ambiguity on the part of the responders that may have resulted in less accurate responses, thus biasing toward the null. Similarly, in measuring AEs, the screening tool relied on retrospective chart review looking for specific AE types[11] and thus may not have captured more subtle AEs. Additionally, our study may have been underpowered to demonstrate significant reduction in preventable AEs, although other studies of similar methodology demonstrated significant results with similar sample size.[11] This was due in part to our lower‐than‐expected baseline AE rate (6.6% compared with approximately 10.3% in previous studies).[11] Lastly, our study solely examined the association of regionalization with concordance of care plan and preventable AEs, but importantly excluded other clinically important outcomes that may have been positively (or negatively) impacted by these regionalization efforts, such as ED wait times, provider efficiency (eg, fewer pages, less time in transit, more time at the bedside), interdisciplinary teamwork, or patient or provider satisfaction.

CONCLUSION

In summary, our findings suggest that regionalized care teams alone may be insufficient to effectively promote communication between care team members regarding the care plan or to lead to improvements in patient safety, although we recognize that there may have been benefits (or unintended harms) not measured in this study but are nonetheless important for clinical care and workplace culture. This is an important lesson, as many hospitals move toward regionalized care in an effort to improve patient safety outcomes. However, strengthening the infrastructure by colocalizing care team members to maximize opportunity for communication is likely a necessary first step toward facilitating implementation of additional initiatives that may lead to more robust patient safety improvements, such as structured interdisciplinary bedside rounds (eg, facilitating and training all team members to fulfill specific roles), teamwork training, and certification of key interdisciplinary teamwork skills. Additionally, close examination of identified prevalent and preventable AEs can help to determine which additional initiatives are most likely to have greatest impact in improving patient safety.

Disclosures: This research was supported by funds provided by Brigham and Women's Hospital (BWH) and by funds provided by the Department of Medicine at BWH. All authors had full access to all of the data in the study and were integrally involved in the design, implementation, data collection, and analyses. The first author, Dr. Stephanie Mueller, takes responsibility for the integrity for the data and the accuracy of the data analysis. Dr. Schnipper reports grants from Sanofi Aventis, outside the submitted work.

Effective communication between inpatient and primary care physicians (PCPs) is essential for safe, high‐quality transitions. Unfortunately, PCPs are often not meaningfully engaged in this process; communication is frequently challenging or nonexistent.[1, 2] Instead, information is suboptimally conveyed via lengthy, disorganized discharge summaries.[3] Consequently, timely knowledge is not transferred to PCPs, who instead must seek out and identify actionable information themselves. These deficiencies can lead to misinterpretation of information and patient harm.[4]

An important component of ideal transitions[5] is timely communication of results of tests pending at discharge (TPADs). TPADs are variably documented in discharge summaries, and physician awareness about them is strikingly poor.[3, 6, 7] Communication about TPADs should convey rationales for ordering tests and necessary actions to take in response to finalized results. Most often, this knowledge resides with the inpatient team.

Health information technology (HIT) is an effective strategy for improving test‐result management. We implemented an automated system that notifies inpatient attendings and PCPs of TPAD results via email and demonstrated increased awareness by these physicians at the time of required action.[8, 9] Nevertheless, without timely knowledge transfer, attendings and PCPs may have differing opinions regarding which TPAD results require action. We conducted a secondary analysis of survey respondents from our original clustered randomized controlled trial to measure the degree of agreement between inpatient and ambulatory physicians regarding actionability of TPAD results.

METHODS

The methods of our original study are described elsewhere.[9] In that study, the attending and PCP of each patient were independently surveyed (via email and then by fax if the electronic survey was not completed) to determine their awareness of finalized TPAD results, and to identify actionable results and the types of actions taken (or that would need to be taken). Discharge summaries were available in our electronic medical record (EMR) within 24 hours of discharge. Network physicians (affiliated with Partners HealthCare, Inc.) had access to all components of the EMR, including the discharge summary and test results. Non‐network PCPs were faxed discharge summaries within 48 hours of discharge per institutional policies. For this study, we identified all patients for whom the attending and PCP completed the survey and answered questions about TPAD actionability. We then compared the identified TPADs listed by the attending and PCP in that survey.

RESULTS

We enrolled 441 patients in our original study. We sent 441 surveys to 117 attendings and 353 surveys to 273 PCPs. Eighty‐eight patients did not have an identified PCP. We received 275 responses from 83 attendings (62% response rate), and 152 responses from 112 PCPs (43% response rate). Patient and physician characteristics are reported elsewhere.[9]

For this analysis, we identified the 98 patients (aged 6018 years, 44 male, 52 Caucasian, 46 non‐Caucasian, 85 network, 13 non‐network) cared for by 46 attendings (aged 4411 years, 33 male, 22 hospitalists, 24 nonhospitalists) and 79 PCPs (aged 4512.5, 33 male, 66 network, 13 non‐network) for whom we received completed surveys from both physicians. For 59 patients, both thought none of the TPAD results were actionable. For 12 patients, both thought at least 1 was actionable, and they identified the same actionable TPAD result for all 12. Overall, attendings and PCPs agreed on actionability in 72.5% (71/98) (Kappa 0.29, 95% confidence interval: 0.09‐0.50). Table 1 shows the type of action taken by responsible providers. There were 9 patients (9%) for whom the attending alone thought at least 1 TPAD result was actionable; of these, subsequent attending‐initiated communication occurred in 77.8% (7/9). There were 18 patients (18%) for whom the PCP alone thought at least 1 TPAD result was actionable; of these, subsequent PCP‐initiated communication occurred in 77.8% (14/18). Table 2 shows concordance of actionable TPAD by type. In instances of disagreement, the attending frequently reported microbiology TPADs (eg, culture data, viral serologies) as actionable, whereas the PCP reported all TPAD types (eg, culture data, colon biopsy, vitamin D, magnetic resonance imaging) as actionable.

DISCUSSION

We found fair agreement between attendings and PCPs regarding actionability of TPAD results. In 27 patients (27.5%), either the attending or PCP considered TPAD results actionable when the other did not. Possible explanations for this include different thresholds for taking action (eg, inpatient physicians may view vitamin D levels as acceptable within broader ranges than PCPs, and PCPs may view negative results as actionable if they need to contact the patient whereas attendings may not), varying clinical context (eg, rationale for why microbiology culture data is actionable), and varying practices for escalating care (eg, referring patients back to the hospital).

Our study was limited by small sample size and low PCP response rate. Nonetheless, the findings suggest that poor concordance between inpatient and ambulatory physicians will persist without tools that promote more effective communication. Greater awareness alone may be insufficient to mitigate consequences of missed TPAD results if physicians are not on the same page regarding which results require action.

To better engage PCPs, healthcare systems require HIT infrastructure that facilitates seamless care team communication across care settings.[2] When optimally configured, HIT can facilitate greater PCP involvement in postdischarge communication. For example, our system promoted subsequent postdischarge communication in 78% of initial discordance in TPAD actionability; however, most of it was not between the attending and the PCP. Thus, improvements could be made to facilitate more effective communication among key inpatient and ambulatory providers. Furthermore, when configured to facilitate conversation among these providers regarding the discharge care plan throughout a patient's entire hospital course, HIT can promote effective knowledge transfer by virtue of adding clinical context to test ordering and follow‐up. Additional work is needed to understand whether such communication clarifies contingencies and facilitates appropriate postdischarge action. Nevertheless, current electronic solutions (eg, passive placement into results in‐baskets) will likely be ineffective because they do not reliably improve awareness and active communication about context, rationale, interpretation, suggested action, or transfer of responsibility.

In summary, discrepancies in TPAD actionability by inpatient and ambulatory providers still exist, even when awareness of TPAD results is improved by HIT. By fostering more effective communication among key care‐team members across care settings, HIT could mitigate the consequences of suboptimal care transitions. With regard to TPAD results, this may favorably impact unnecessary testing, diagnostic and therapeutic delays, and medical errors.

Disclosures: This article is based on research funded through AHRQ grant #R21HS018229; the authors have no other disclosures or conflicts or interest.

Effective communication between inpatient and primary care physicians (PCPs) is essential for safe, high‐quality transitions. Unfortunately, PCPs are often not meaningfully engaged in this process; communication is frequently challenging or nonexistent.[1, 2] Instead, information is suboptimally conveyed via lengthy, disorganized discharge summaries.[3] Consequently, timely knowledge is not transferred to PCPs, who instead must seek out and identify actionable information themselves. These deficiencies can lead to misinterpretation of information and patient harm.[4]

An important component of ideal transitions[5] is timely communication of results of tests pending at discharge (TPADs). TPADs are variably documented in discharge summaries, and physician awareness about them is strikingly poor.[3, 6, 7] Communication about TPADs should convey rationales for ordering tests and necessary actions to take in response to finalized results. Most often, this knowledge resides with the inpatient team.

Health information technology (HIT) is an effective strategy for improving test‐result management. We implemented an automated system that notifies inpatient attendings and PCPs of TPAD results via email and demonstrated increased awareness by these physicians at the time of required action.[8, 9] Nevertheless, without timely knowledge transfer, attendings and PCPs may have differing opinions regarding which TPAD results require action. We conducted a secondary analysis of survey respondents from our original clustered randomized controlled trial to measure the degree of agreement between inpatient and ambulatory physicians regarding actionability of TPAD results.

METHODS

The methods of our original study are described elsewhere.[9] In that study, the attending and PCP of each patient were independently surveyed (via email and then by fax if the electronic survey was not completed) to determine their awareness of finalized TPAD results, and to identify actionable results and the types of actions taken (or that would need to be taken). Discharge summaries were available in our electronic medical record (EMR) within 24 hours of discharge. Network physicians (affiliated with Partners HealthCare, Inc.) had access to all components of the EMR, including the discharge summary and test results. Non‐network PCPs were faxed discharge summaries within 48 hours of discharge per institutional policies. For this study, we identified all patients for whom the attending and PCP completed the survey and answered questions about TPAD actionability. We then compared the identified TPADs listed by the attending and PCP in that survey.

RESULTS

We enrolled 441 patients in our original study. We sent 441 surveys to 117 attendings and 353 surveys to 273 PCPs. Eighty‐eight patients did not have an identified PCP. We received 275 responses from 83 attendings (62% response rate), and 152 responses from 112 PCPs (43% response rate). Patient and physician characteristics are reported elsewhere.[9]

For this analysis, we identified the 98 patients (aged 6018 years, 44 male, 52 Caucasian, 46 non‐Caucasian, 85 network, 13 non‐network) cared for by 46 attendings (aged 4411 years, 33 male, 22 hospitalists, 24 nonhospitalists) and 79 PCPs (aged 4512.5, 33 male, 66 network, 13 non‐network) for whom we received completed surveys from both physicians. For 59 patients, both thought none of the TPAD results were actionable. For 12 patients, both thought at least 1 was actionable, and they identified the same actionable TPAD result for all 12. Overall, attendings and PCPs agreed on actionability in 72.5% (71/98) (Kappa 0.29, 95% confidence interval: 0.09‐0.50). Table 1 shows the type of action taken by responsible providers. There were 9 patients (9%) for whom the attending alone thought at least 1 TPAD result was actionable; of these, subsequent attending‐initiated communication occurred in 77.8% (7/9). There were 18 patients (18%) for whom the PCP alone thought at least 1 TPAD result was actionable; of these, subsequent PCP‐initiated communication occurred in 77.8% (14/18). Table 2 shows concordance of actionable TPAD by type. In instances of disagreement, the attending frequently reported microbiology TPADs (eg, culture data, viral serologies) as actionable, whereas the PCP reported all TPAD types (eg, culture data, colon biopsy, vitamin D, magnetic resonance imaging) as actionable.

DISCUSSION

We found fair agreement between attendings and PCPs regarding actionability of TPAD results. In 27 patients (27.5%), either the attending or PCP considered TPAD results actionable when the other did not. Possible explanations for this include different thresholds for taking action (eg, inpatient physicians may view vitamin D levels as acceptable within broader ranges than PCPs, and PCPs may view negative results as actionable if they need to contact the patient whereas attendings may not), varying clinical context (eg, rationale for why microbiology culture data is actionable), and varying practices for escalating care (eg, referring patients back to the hospital).

Our study was limited by small sample size and low PCP response rate. Nonetheless, the findings suggest that poor concordance between inpatient and ambulatory physicians will persist without tools that promote more effective communication. Greater awareness alone may be insufficient to mitigate consequences of missed TPAD results if physicians are not on the same page regarding which results require action.

To better engage PCPs, healthcare systems require HIT infrastructure that facilitates seamless care team communication across care settings.[2] When optimally configured, HIT can facilitate greater PCP involvement in postdischarge communication. For example, our system promoted subsequent postdischarge communication in 78% of initial discordance in TPAD actionability; however, most of it was not between the attending and the PCP. Thus, improvements could be made to facilitate more effective communication among key inpatient and ambulatory providers. Furthermore, when configured to facilitate conversation among these providers regarding the discharge care plan throughout a patient's entire hospital course, HIT can promote effective knowledge transfer by virtue of adding clinical context to test ordering and follow‐up. Additional work is needed to understand whether such communication clarifies contingencies and facilitates appropriate postdischarge action. Nevertheless, current electronic solutions (eg, passive placement into results in‐baskets) will likely be ineffective because they do not reliably improve awareness and active communication about context, rationale, interpretation, suggested action, or transfer of responsibility.

In summary, discrepancies in TPAD actionability by inpatient and ambulatory providers still exist, even when awareness of TPAD results is improved by HIT. By fostering more effective communication among key care‐team members across care settings, HIT could mitigate the consequences of suboptimal care transitions. With regard to TPAD results, this may favorably impact unnecessary testing, diagnostic and therapeutic delays, and medical errors.

Disclosures: This article is based on research funded through AHRQ grant #R21HS018229; the authors have no other disclosures or conflicts or interest.

Effective communication between inpatient and primary care physicians (PCPs) is essential for safe, high‐quality transitions. Unfortunately, PCPs are often not meaningfully engaged in this process; communication is frequently challenging or nonexistent.[1, 2] Instead, information is suboptimally conveyed via lengthy, disorganized discharge summaries.[3] Consequently, timely knowledge is not transferred to PCPs, who instead must seek out and identify actionable information themselves. These deficiencies can lead to misinterpretation of information and patient harm.[4]

An important component of ideal transitions[5] is timely communication of results of tests pending at discharge (TPADs). TPADs are variably documented in discharge summaries, and physician awareness about them is strikingly poor.[3, 6, 7] Communication about TPADs should convey rationales for ordering tests and necessary actions to take in response to finalized results. Most often, this knowledge resides with the inpatient team.

Health information technology (HIT) is an effective strategy for improving test‐result management. We implemented an automated system that notifies inpatient attendings and PCPs of TPAD results via email and demonstrated increased awareness by these physicians at the time of required action.[8, 9] Nevertheless, without timely knowledge transfer, attendings and PCPs may have differing opinions regarding which TPAD results require action. We conducted a secondary analysis of survey respondents from our original clustered randomized controlled trial to measure the degree of agreement between inpatient and ambulatory physicians regarding actionability of TPAD results.

METHODS

The methods of our original study are described elsewhere.[9] In that study, the attending and PCP of each patient were independently surveyed (via email and then by fax if the electronic survey was not completed) to determine their awareness of finalized TPAD results, and to identify actionable results and the types of actions taken (or that would need to be taken). Discharge summaries were available in our electronic medical record (EMR) within 24 hours of discharge. Network physicians (affiliated with Partners HealthCare, Inc.) had access to all components of the EMR, including the discharge summary and test results. Non‐network PCPs were faxed discharge summaries within 48 hours of discharge per institutional policies. For this study, we identified all patients for whom the attending and PCP completed the survey and answered questions about TPAD actionability. We then compared the identified TPADs listed by the attending and PCP in that survey.

RESULTS

We enrolled 441 patients in our original study. We sent 441 surveys to 117 attendings and 353 surveys to 273 PCPs. Eighty‐eight patients did not have an identified PCP. We received 275 responses from 83 attendings (62% response rate), and 152 responses from 112 PCPs (43% response rate). Patient and physician characteristics are reported elsewhere.[9]

For this analysis, we identified the 98 patients (aged 6018 years, 44 male, 52 Caucasian, 46 non‐Caucasian, 85 network, 13 non‐network) cared for by 46 attendings (aged 4411 years, 33 male, 22 hospitalists, 24 nonhospitalists) and 79 PCPs (aged 4512.5, 33 male, 66 network, 13 non‐network) for whom we received completed surveys from both physicians. For 59 patients, both thought none of the TPAD results were actionable. For 12 patients, both thought at least 1 was actionable, and they identified the same actionable TPAD result for all 12. Overall, attendings and PCPs agreed on actionability in 72.5% (71/98) (Kappa 0.29, 95% confidence interval: 0.09‐0.50). Table 1 shows the type of action taken by responsible providers. There were 9 patients (9%) for whom the attending alone thought at least 1 TPAD result was actionable; of these, subsequent attending‐initiated communication occurred in 77.8% (7/9). There were 18 patients (18%) for whom the PCP alone thought at least 1 TPAD result was actionable; of these, subsequent PCP‐initiated communication occurred in 77.8% (14/18). Table 2 shows concordance of actionable TPAD by type. In instances of disagreement, the attending frequently reported microbiology TPADs (eg, culture data, viral serologies) as actionable, whereas the PCP reported all TPAD types (eg, culture data, colon biopsy, vitamin D, magnetic resonance imaging) as actionable.

DISCUSSION

We found fair agreement between attendings and PCPs regarding actionability of TPAD results. In 27 patients (27.5%), either the attending or PCP considered TPAD results actionable when the other did not. Possible explanations for this include different thresholds for taking action (eg, inpatient physicians may view vitamin D levels as acceptable within broader ranges than PCPs, and PCPs may view negative results as actionable if they need to contact the patient whereas attendings may not), varying clinical context (eg, rationale for why microbiology culture data is actionable), and varying practices for escalating care (eg, referring patients back to the hospital).

Our study was limited by small sample size and low PCP response rate. Nonetheless, the findings suggest that poor concordance between inpatient and ambulatory physicians will persist without tools that promote more effective communication. Greater awareness alone may be insufficient to mitigate consequences of missed TPAD results if physicians are not on the same page regarding which results require action.

To better engage PCPs, healthcare systems require HIT infrastructure that facilitates seamless care team communication across care settings.[2] When optimally configured, HIT can facilitate greater PCP involvement in postdischarge communication. For example, our system promoted subsequent postdischarge communication in 78% of initial discordance in TPAD actionability; however, most of it was not between the attending and the PCP. Thus, improvements could be made to facilitate more effective communication among key inpatient and ambulatory providers. Furthermore, when configured to facilitate conversation among these providers regarding the discharge care plan throughout a patient's entire hospital course, HIT can promote effective knowledge transfer by virtue of adding clinical context to test ordering and follow‐up. Additional work is needed to understand whether such communication clarifies contingencies and facilitates appropriate postdischarge action. Nevertheless, current electronic solutions (eg, passive placement into results in‐baskets) will likely be ineffective because they do not reliably improve awareness and active communication about context, rationale, interpretation, suggested action, or transfer of responsibility.

In summary, discrepancies in TPAD actionability by inpatient and ambulatory providers still exist, even when awareness of TPAD results is improved by HIT. By fostering more effective communication among key care‐team members across care settings, HIT could mitigate the consequences of suboptimal care transitions. With regard to TPAD results, this may favorably impact unnecessary testing, diagnostic and therapeutic delays, and medical errors.

Disclosures: This article is based on research funded through AHRQ grant #R21HS018229; the authors have no other disclosures or conflicts or interest.

If you wish to receive credit for this activity, which beginson the next page, please refer to the website: www.blackwellpublishing.com/cme.

Accreditation and Designation Statement

Blackwell Futura Media Services designates this educational activity for a 1 AMA PRA Category 1 Credit. Physicians should only claim credit commensurate with the extent of their participation in the activity.

Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians.

Educational Objectives

Continuous participation in the Journal of Hospital Medicine CME program will enable learners to be better able to:

• Interpret clinical guidelines and their applications for higher quality and more efficient care for all hospitalized patients.

• Describe the standard of care for common illnesses and conditions treated in the hospital; such as pneumonia, COPD exacerbation, acute coronary syndrome, HF exacerbation, glycemic control, venous thromboembolic disease, stroke, etc.

• Discuss evidence‐based recommendations involving transitions of care, including the hospital discharge process.

• Gain insights into the roles of hospitalists as medical educators, researchers, medical ethicists, palliative care providers, and hospital‐based geriatricians.

• Incorporate best practices for hospitalist administration, including quality improvement, patient safety, practice management, leadership, and demonstrating hospitalist value.

• Identify evidence‐based best practices and trends for both adult and pediatric hospital medicine.

Instructions on Receiving Credit

For information on applicability and acceptance of continuing medical education credit for this activity, please consult your professional licensing board.

This activity is designed to be completed within the time designated on the title page; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period that is noted on the title page.

Follow these steps to earn credit:

• Log on to www.blackwellpublishing.com/cme.

• Read the target audience, learning objectives, and author disclosures.

• Read the article in print or online format.

• Reflect on the article.

• Access the CME Exam, and choose the best answer to each question.

• Complete the required evaluation component of the activity.

If you wish to receive credit for this activity, which beginson the next page, please refer to the website: www.blackwellpublishing.com/cme.

Accreditation and Designation Statement

Blackwell Futura Media Services designates this educational activity for a 1 AMA PRA Category 1 Credit. Physicians should only claim credit commensurate with the extent of their participation in the activity.

Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians.

Educational Objectives

Continuous participation in the Journal of Hospital Medicine CME program will enable learners to be better able to:

• Interpret clinical guidelines and their applications for higher quality and more efficient care for all hospitalized patients.

• Describe the standard of care for common illnesses and conditions treated in the hospital; such as pneumonia, COPD exacerbation, acute coronary syndrome, HF exacerbation, glycemic control, venous thromboembolic disease, stroke, etc.

• Discuss evidence‐based recommendations involving transitions of care, including the hospital discharge process.

• Gain insights into the roles of hospitalists as medical educators, researchers, medical ethicists, palliative care providers, and hospital‐based geriatricians.

• Incorporate best practices for hospitalist administration, including quality improvement, patient safety, practice management, leadership, and demonstrating hospitalist value.

• Identify evidence‐based best practices and trends for both adult and pediatric hospital medicine.

Instructions on Receiving Credit

For information on applicability and acceptance of continuing medical education credit for this activity, please consult your professional licensing board.

This activity is designed to be completed within the time designated on the title page; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period that is noted on the title page.

Follow these steps to earn credit:

• Log on to www.blackwellpublishing.com/cme.

• Read the target audience, learning objectives, and author disclosures.

• Read the article in print or online format.

• Reflect on the article.

• Access the CME Exam, and choose the best answer to each question.

• Complete the required evaluation component of the activity.

If you wish to receive credit for this activity, which beginson the next page, please refer to the website: www.blackwellpublishing.com/cme.

Accreditation and Designation Statement

Blackwell Futura Media Services designates this educational activity for a 1 AMA PRA Category 1 Credit. Physicians should only claim credit commensurate with the extent of their participation in the activity.

Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians.

Educational Objectives

Continuous participation in the Journal of Hospital Medicine CME program will enable learners to be better able to:

• Interpret clinical guidelines and their applications for higher quality and more efficient care for all hospitalized patients.

• Describe the standard of care for common illnesses and conditions treated in the hospital; such as pneumonia, COPD exacerbation, acute coronary syndrome, HF exacerbation, glycemic control, venous thromboembolic disease, stroke, etc.

• Discuss evidence‐based recommendations involving transitions of care, including the hospital discharge process.

• Gain insights into the roles of hospitalists as medical educators, researchers, medical ethicists, palliative care providers, and hospital‐based geriatricians.

• Incorporate best practices for hospitalist administration, including quality improvement, patient safety, practice management, leadership, and demonstrating hospitalist value.

• Identify evidence‐based best practices and trends for both adult and pediatric hospital medicine.

Instructions on Receiving Credit

For information on applicability and acceptance of continuing medical education credit for this activity, please consult your professional licensing board.

This activity is designed to be completed within the time designated on the title page; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period that is noted on the title page.

Follow these steps to earn credit:

• Log on to www.blackwellpublishing.com/cme.

• Read the target audience, learning objectives, and author disclosures.

• Read the article in print or online format.

• Reflect on the article.

• Access the CME Exam, and choose the best answer to each question.

• Complete the required evaluation component of the activity.

The period following discharge is a vulnerable time for patientsthe prevalence of medical errors related to this transition is high and has important patient safety and medico‐legal ramifications.13 Factors contributing to this vulnerability include complexity of hospitalized patients, shorter lengths of stay, and increased discontinuity of care. Hospitalists have recognized this threat to patient safety and have worked toward improving information exchange between inpatient and outpatient providers at hospital discharge.46 Nonetheless, the evidence suggests that more work is necessary. A recent study found that discharge summaries are often incomplete, and do not contain important information requiring follow‐up, such as pending tests.7 Additionally, a review by Kripalani et al. characterizing information deficits at hospital discharge found few interventions which specifically improve communication of pending tests at hospital discharge.8

In a prior study we determined that 41% of patients left the hospital before all laboratory and radiology test results were finalized. Of these results, 9.4% were potentially actionable and could have altered management. Physicians were aware of only 38% of post‐discharge test results.9 This awareness gap is a consequence of several factors including the lack of systems to track and alert providers of test results finalized post discharge. Also, it is unclear who is responsible for pending tests at discharge, since these tests are ordered by the inpatient physicians but often reported in the time period between hospital discharge and the patient's first follow‐up appointment with the primary care physician (PCP). Because responsibility is not explicitly made in the final communication between physicians at discharge, such test results may not be reviewed in a timely manner, potentially resulting in delays in treatment, a need for readmission, or other unfavorable outcomes.

Even in integrated health systems with advanced electronic health records, missed test results which result in treatment delays remain prevalent.10, 11 Test result management applications aid clinicians in reviewing and acting upon results as they become available and such systems may provide solutions to this problem. At Partners Healthcare in Boston, the Results Manager (RM) application was developed to help clinicians in the ambulatory setting safely, reliably, and efficiently review and act upon test results. The application enables clinicians to prioritize test results, utilize guidelines, and generate letters to patients. This system also prompts physicians to set reminders for future testing.12 In a 2.5‐year study evaluating the impact of this intervention, PCPs at 26 adult primary care practices were able to expedite communication of outpatient laboratory and imaging test results to patients with the help of RM. Patients of physicians who participated in the project reported greater satisfaction with test result communication and with information provided about their condition than did a control group of similar patients.13 RM has not yet been studied in the inpatient setting or at care transitions. We describe an attempt at modifying the Partners RM application to help inpatient physicians manage pending tests at hospital discharge.

Methods

Test Result Management Application

RM was originally developed by Partners Healthcare to improve timely review and appropriate management of test results in the ambulatory setting. RM was developed for and vetted by primarily ambulatory physicians. The application is browser‐based, provider‐centric, and embedded in the LMR to help ambulatory clinicians review and act upon test results in a safe, reliable, and efficient manner. Although RM has access to all inpatient and outpatient data in the Partners Clinical Data Repository (CDR), given the volume of inpatient tests ordered, hospital‐based results are suppressed by default to limit inundating ambulatory clinicians' queues. Therefore, users of RM only receive results of laboratory and radiology tests ordered in the ambulatory setting. They can track these tests for specific patients for a designated period of time by placing the patient on a watch list. Finally, RM incorporates extensive decision support features to classify the degree of abnormality for each result, presents guidelines to help clinicians manage abnormal results, allows clinicians to generate result letters to patients using predefined, context‐sensitive templates, and prompts physicians to set reminders for future testing. Because RM was developed from the ambulatory perspective, there was limited input from hospitalist physicians with regard to inpatient workflow in the original design of the module.12 See Figure 1 for a screen shot of RM and a description of its features.

Figure 1

For purposes of this pilot, we modified RM to allow results of tests ordered in the inpatient setting to be available for viewing (Hospitalist Results Manager, HRM). This feature was turned on only for inpatient physicians as previously defined. Inpatient tests, including pending tests at discharge, continued to be suppressed from PCP's RM queue (however, any physician could access a patient's test result(s) directly from the Partners CDR). Inpatient physicians could track laboratory and radiology results finalized after discharge by keeping discharged patients on their HRM watch list for a designated period of time. The finalized results would become available for review in their HRM queue and abnormal results were displayed prominently at the top of this queue. Inpatient physicians were trained to use HRM in a series of meetings and demonstrations. Although HRM could be accessed from inpatient clinical workstations, it was not part of the inpatient clinical information system.

Results

A total of 35 inpatient physicians participated in the pilot. Among 649 patients discharged during the study period, there were 1075 tests pending of which 555 were subsequently flagged as abnormal in HRM. Study surveys were sent to the 35 inpatient physician participants and 29 were completed, including partial responses (83% survey response rate). The 35 inpatient physician participants had the following characteristics: 22 were male, 13 were female; 21 were trainees and 14 were nontrainees/faculty. All 21 trainees were PGY2s. Nontrainees and faculty varied in experience level (PGY 15: 5, PGY 610: 7, PGY 1120: 1, PGY 21+: 1). Of 29 survey respondents, 7 were from hospital A and 22 were from hospital B; 19 were trainees and 10 were nontrainees/faculty. Of the 6 nonrespondents, 2 were from hospital A and 4 were from hospital B; 2 were trainees and 4 were nontrainees/faculty.

Table 1 shows the results of our survey of inpatient physicians regarding usage of HRM. Of 29 survey respondents, 14 (48%) reported never using HRM. Thirteen (45%) reported using HRM 1 to 2 times per week. None of the respondents used it more than 4 times per week. The frequency of usage was similar for hospitals A and B. Table 2 details barriers to using HRM. Twenty‐three inpatient physicians (79%) reported barriers. Seventeen (59%) thought that results in their HRM queue were not clinically relevant, 16 (55%) felt that HRM did not fit into their daily workflow, 14 (48%) had limited time to use HRM, and 12 (41%) noted that too many results in their HRM queue were on other physician's patients. Seven (24%) reported operational issues and 3 (10%) reported technical issues prohibiting use of HRM. With regard to preferred elements of an ideal results manager system, 21 (72%) inpatient physician respondents wanted to receive notification of abnormal and clinician‐designated pending test results. Four (14%) wanted to receive only abnormal results and 1 (3%) wanted to receive all results. Twenty‐seven (93%) physicians agreed that an ideally designed computerized test result management application would be valuable for managing pending tests at discharge.

Table 3 provides comments from inpatient physician respondents regarding obstacles prohibiting use of HRM and suggestions for future systems.

Discussion

We describe a pilot implementation of a computerized application for the management of pending tests at hospital discharge. From responses to post‐implementation surveys, we were able to identify multiple factors prohibiting successful implementation of the application. These observations may help inform future interventions and evaluations.

Almost half of inpatient physicians reported never using HRM despite training and reminders. The feedback provided by physicians in our study suggested that HRM was not ideally designed from an inpatient physician perspective. We discovered several barriers to its use: (1) HRM overburdened physicians with clinically irrelevant test results, suggesting that more robust filtering of abnormal but low importance test results may be required (eg, a borderline electrolyte abnormality or low but stable hematocrit); (2) HRM did not integrate well into inpatient workflowthe system was not integrated into the inpatient results viewing and computerized physician order entry (CPOE) applications, and therefore required an extra step to access; (3) there was no mechanism of alerting inpatient physicians that finalized test results were available for viewing in their HRM queues (eg, by email or by an alert in the inpatient computer system); (4) because responsibility for these results was unclear, most inpatient physicians had no formal method of managing them, and for many, using HRM represented an additional task; and finally (5) several physicians commented on finding results in their HRM queue that belonged to other physician's patients, implying that the hospital databases were inaccurate in identifying the discharging physician or that rotation schedules, and therefore patient responsibility, had changed in the intervening period. Table 4 summarizes the advantages and respective limitations of features of HRM available to inpatient physicians.

In the literature, there is little information regarding optimal features of a test result management system for transitions from the inpatient to ambulatory care setting. Prior studies outline important functions for results management systems developed for noninpatient sites of care, including the ambulatory and emergency room setting.12, 14, 15 These include a method of prioritizing by degree of abnormality, the ability to reliably and efficiently act upon results, and an automated alerting system for abnormal results. Findings from our study provide insight in defining core functions for result management systems which focus on transitions from the inpatient to ambulatory care setting. These functions include tight integration with applications used by inpatient physicians, clear assignment of responsibility for test results finalized after hospital discharge (as well as a mechanism to reassign responsibility), automated alerts to responsible providers of test results finalized post‐discharge, and ways to automatically filter test results to avoid over‐burdening physicians with clinically irrelevant results.

Almost all surveyed inpatient physicians agreed that an ideally designed electronic post‐discharge results management system would be valuable. For such systems to be successfully adopted, we offer several principles to help guide future work. These include: (1) clarifying responsibility at the time a test is ordered and again at discharge, (2) understanding workflow and communication patterns among inpatient and outpatient clinicians, and (3) integrating technological solutions into existing systems to minimize workflow disruptions. For example, if the primary responsibility for post‐discharge result follow‐up lies with the ordering physician, the system should be integrated within the EMR most often used by inpatient physicians and become part of inpatient physician workflow. If the system depends on administrative databases to identify the responsible providers, these must be accurate. Alternatively, in organizations with computerized provider order entry, responsibility for the result could be assigned when the test is ordered and confirmed at discharge (ie, the results management system would be integrated into the discharge order such that pending tests are reviewed at the time of discharge). The discharging physician should have the ability to assign responsibility for each pending test and select preferred mode(s) of notification once its result is finalized (eg, e‐mail, alphanumeric page, etc.). The system should have the ability to generate an automatic notification to the inpatient and PCP (and perhaps other designated providers involved in the patient's inpatient care), but it should not burden busy clinicians with unnecessary alerts and warnings. Finally, the rules by which results are prioritized must be robust enough to filter out less urgent results, and should be modified to reflect the severity of illness of recently discharged patients. In essence, in consideration of the time constraints of busy clinicians, an ideal results management system should achieve automated notification of test results while minimizing the risk of alert fatigue from the potentially large volume of alerts generated.

Our study has several important limitations. First, although our survey response rate was high, the sample size of actual participants was small. Second, because the study was conducted in 2 similar, tertiary care academic centers, it may not generalize to other settings (we note that hospital B included a nonteaching service similar to those in nonacademic medical centers). This may be particularly true in assessing the importance of specific barriers to use of results management systems, which may vary at different institutions. Third, the representation of survey respondents were skeweda majority of the responses were from trainees (all post‐graduate level [PGY] level 2) and from hospital B. Fourth, we did not actively monitor physician interaction with the test result management application, and therefore, we depended heavily on physician recollection of use of the system when responding to surveys. Finally, we did not convene focus groups of key individuals with regard to the factors facilitating or prohibiting adoption of the system. Use of semi‐structured, key informant interviews (ie, focus groups) before and after implementation of an electronic results management application, have been shown to be effective in evaluating potential barriers and facilitators of adoption.16 Focus groups of and/or interviews with inpatient and PCPs, physician extenders, and housestaff could have been useful to better characterize the potential barriers and facilitators of adoption noted by survey respondents in our study.

In summary, we offer several lessons from our attempt to implement a system to manage pending tests at hospital discharge. The success of implementing future systems to address this patient safety concern will rely on accurately assigning responsibility for these test results, integrating the system within clinical information systems commonly used by the inpatient physician, addressing workflow issues and time constraints, maximizing appropriateness of alerting, and minimizing alert fatigue.

The period following discharge is a vulnerable time for patientsthe prevalence of medical errors related to this transition is high and has important patient safety and medico‐legal ramifications.13 Factors contributing to this vulnerability include complexity of hospitalized patients, shorter lengths of stay, and increased discontinuity of care. Hospitalists have recognized this threat to patient safety and have worked toward improving information exchange between inpatient and outpatient providers at hospital discharge.46 Nonetheless, the evidence suggests that more work is necessary. A recent study found that discharge summaries are often incomplete, and do not contain important information requiring follow‐up, such as pending tests.7 Additionally, a review by Kripalani et al. characterizing information deficits at hospital discharge found few interventions which specifically improve communication of pending tests at hospital discharge.8

In a prior study we determined that 41% of patients left the hospital before all laboratory and radiology test results were finalized. Of these results, 9.4% were potentially actionable and could have altered management. Physicians were aware of only 38% of post‐discharge test results.9 This awareness gap is a consequence of several factors including the lack of systems to track and alert providers of test results finalized post discharge. Also, it is unclear who is responsible for pending tests at discharge, since these tests are ordered by the inpatient physicians but often reported in the time period between hospital discharge and the patient's first follow‐up appointment with the primary care physician (PCP). Because responsibility is not explicitly made in the final communication between physicians at discharge, such test results may not be reviewed in a timely manner, potentially resulting in delays in treatment, a need for readmission, or other unfavorable outcomes.

Even in integrated health systems with advanced electronic health records, missed test results which result in treatment delays remain prevalent.10, 11 Test result management applications aid clinicians in reviewing and acting upon results as they become available and such systems may provide solutions to this problem. At Partners Healthcare in Boston, the Results Manager (RM) application was developed to help clinicians in the ambulatory setting safely, reliably, and efficiently review and act upon test results. The application enables clinicians to prioritize test results, utilize guidelines, and generate letters to patients. This system also prompts physicians to set reminders for future testing.12 In a 2.5‐year study evaluating the impact of this intervention, PCPs at 26 adult primary care practices were able to expedite communication of outpatient laboratory and imaging test results to patients with the help of RM. Patients of physicians who participated in the project reported greater satisfaction with test result communication and with information provided about their condition than did a control group of similar patients.13 RM has not yet been studied in the inpatient setting or at care transitions. We describe an attempt at modifying the Partners RM application to help inpatient physicians manage pending tests at hospital discharge.

Methods

Test Result Management Application

RM was originally developed by Partners Healthcare to improve timely review and appropriate management of test results in the ambulatory setting. RM was developed for and vetted by primarily ambulatory physicians. The application is browser‐based, provider‐centric, and embedded in the LMR to help ambulatory clinicians review and act upon test results in a safe, reliable, and efficient manner. Although RM has access to all inpatient and outpatient data in the Partners Clinical Data Repository (CDR), given the volume of inpatient tests ordered, hospital‐based results are suppressed by default to limit inundating ambulatory clinicians' queues. Therefore, users of RM only receive results of laboratory and radiology tests ordered in the ambulatory setting. They can track these tests for specific patients for a designated period of time by placing the patient on a watch list. Finally, RM incorporates extensive decision support features to classify the degree of abnormality for each result, presents guidelines to help clinicians manage abnormal results, allows clinicians to generate result letters to patients using predefined, context‐sensitive templates, and prompts physicians to set reminders for future testing. Because RM was developed from the ambulatory perspective, there was limited input from hospitalist physicians with regard to inpatient workflow in the original design of the module.12 See Figure 1 for a screen shot of RM and a description of its features.

Figure 1

For purposes of this pilot, we modified RM to allow results of tests ordered in the inpatient setting to be available for viewing (Hospitalist Results Manager, HRM). This feature was turned on only for inpatient physicians as previously defined. Inpatient tests, including pending tests at discharge, continued to be suppressed from PCP's RM queue (however, any physician could access a patient's test result(s) directly from the Partners CDR). Inpatient physicians could track laboratory and radiology results finalized after discharge by keeping discharged patients on their HRM watch list for a designated period of time. The finalized results would become available for review in their HRM queue and abnormal results were displayed prominently at the top of this queue. Inpatient physicians were trained to use HRM in a series of meetings and demonstrations. Although HRM could be accessed from inpatient clinical workstations, it was not part of the inpatient clinical information system.

Results

A total of 35 inpatient physicians participated in the pilot. Among 649 patients discharged during the study period, there were 1075 tests pending of which 555 were subsequently flagged as abnormal in HRM. Study surveys were sent to the 35 inpatient physician participants and 29 were completed, including partial responses (83% survey response rate). The 35 inpatient physician participants had the following characteristics: 22 were male, 13 were female; 21 were trainees and 14 were nontrainees/faculty. All 21 trainees were PGY2s. Nontrainees and faculty varied in experience level (PGY 15: 5, PGY 610: 7, PGY 1120: 1, PGY 21+: 1). Of 29 survey respondents, 7 were from hospital A and 22 were from hospital B; 19 were trainees and 10 were nontrainees/faculty. Of the 6 nonrespondents, 2 were from hospital A and 4 were from hospital B; 2 were trainees and 4 were nontrainees/faculty.

Table 1 shows the results of our survey of inpatient physicians regarding usage of HRM. Of 29 survey respondents, 14 (48%) reported never using HRM. Thirteen (45%) reported using HRM 1 to 2 times per week. None of the respondents used it more than 4 times per week. The frequency of usage was similar for hospitals A and B. Table 2 details barriers to using HRM. Twenty‐three inpatient physicians (79%) reported barriers. Seventeen (59%) thought that results in their HRM queue were not clinically relevant, 16 (55%) felt that HRM did not fit into their daily workflow, 14 (48%) had limited time to use HRM, and 12 (41%) noted that too many results in their HRM queue were on other physician's patients. Seven (24%) reported operational issues and 3 (10%) reported technical issues prohibiting use of HRM. With regard to preferred elements of an ideal results manager system, 21 (72%) inpatient physician respondents wanted to receive notification of abnormal and clinician‐designated pending test results. Four (14%) wanted to receive only abnormal results and 1 (3%) wanted to receive all results. Twenty‐seven (93%) physicians agreed that an ideally designed computerized test result management application would be valuable for managing pending tests at discharge.

Table 3 provides comments from inpatient physician respondents regarding obstacles prohibiting use of HRM and suggestions for future systems.

Discussion

We describe a pilot implementation of a computerized application for the management of pending tests at hospital discharge. From responses to post‐implementation surveys, we were able to identify multiple factors prohibiting successful implementation of the application. These observations may help inform future interventions and evaluations.

Almost half of inpatient physicians reported never using HRM despite training and reminders. The feedback provided by physicians in our study suggested that HRM was not ideally designed from an inpatient physician perspective. We discovered several barriers to its use: (1) HRM overburdened physicians with clinically irrelevant test results, suggesting that more robust filtering of abnormal but low importance test results may be required (eg, a borderline electrolyte abnormality or low but stable hematocrit); (2) HRM did not integrate well into inpatient workflowthe system was not integrated into the inpatient results viewing and computerized physician order entry (CPOE) applications, and therefore required an extra step to access; (3) there was no mechanism of alerting inpatient physicians that finalized test results were available for viewing in their HRM queues (eg, by email or by an alert in the inpatient computer system); (4) because responsibility for these results was unclear, most inpatient physicians had no formal method of managing them, and for many, using HRM represented an additional task; and finally (5) several physicians commented on finding results in their HRM queue that belonged to other physician's patients, implying that the hospital databases were inaccurate in identifying the discharging physician or that rotation schedules, and therefore patient responsibility, had changed in the intervening period. Table 4 summarizes the advantages and respective limitations of features of HRM available to inpatient physicians.

In the literature, there is little information regarding optimal features of a test result management system for transitions from the inpatient to ambulatory care setting. Prior studies outline important functions for results management systems developed for noninpatient sites of care, including the ambulatory and emergency room setting.12, 14, 15 These include a method of prioritizing by degree of abnormality, the ability to reliably and efficiently act upon results, and an automated alerting system for abnormal results. Findings from our study provide insight in defining core functions for result management systems which focus on transitions from the inpatient to ambulatory care setting. These functions include tight integration with applications used by inpatient physicians, clear assignment of responsibility for test results finalized after hospital discharge (as well as a mechanism to reassign responsibility), automated alerts to responsible providers of test results finalized post‐discharge, and ways to automatically filter test results to avoid over‐burdening physicians with clinically irrelevant results.

Almost all surveyed inpatient physicians agreed that an ideally designed electronic post‐discharge results management system would be valuable. For such systems to be successfully adopted, we offer several principles to help guide future work. These include: (1) clarifying responsibility at the time a test is ordered and again at discharge, (2) understanding workflow and communication patterns among inpatient and outpatient clinicians, and (3) integrating technological solutions into existing systems to minimize workflow disruptions. For example, if the primary responsibility for post‐discharge result follow‐up lies with the ordering physician, the system should be integrated within the EMR most often used by inpatient physicians and become part of inpatient physician workflow. If the system depends on administrative databases to identify the responsible providers, these must be accurate. Alternatively, in organizations with computerized provider order entry, responsibility for the result could be assigned when the test is ordered and confirmed at discharge (ie, the results management system would be integrated into the discharge order such that pending tests are reviewed at the time of discharge). The discharging physician should have the ability to assign responsibility for each pending test and select preferred mode(s) of notification once its result is finalized (eg, e‐mail, alphanumeric page, etc.). The system should have the ability to generate an automatic notification to the inpatient and PCP (and perhaps other designated providers involved in the patient's inpatient care), but it should not burden busy clinicians with unnecessary alerts and warnings. Finally, the rules by which results are prioritized must be robust enough to filter out less urgent results, and should be modified to reflect the severity of illness of recently discharged patients. In essence, in consideration of the time constraints of busy clinicians, an ideal results management system should achieve automated notification of test results while minimizing the risk of alert fatigue from the potentially large volume of alerts generated.

Our study has several important limitations. First, although our survey response rate was high, the sample size of actual participants was small. Second, because the study was conducted in 2 similar, tertiary care academic centers, it may not generalize to other settings (we note that hospital B included a nonteaching service similar to those in nonacademic medical centers). This may be particularly true in assessing the importance of specific barriers to use of results management systems, which may vary at different institutions. Third, the representation of survey respondents were skeweda majority of the responses were from trainees (all post‐graduate level [PGY] level 2) and from hospital B. Fourth, we did not actively monitor physician interaction with the test result management application, and therefore, we depended heavily on physician recollection of use of the system when responding to surveys. Finally, we did not convene focus groups of key individuals with regard to the factors facilitating or prohibiting adoption of the system. Use of semi‐structured, key informant interviews (ie, focus groups) before and after implementation of an electronic results management application, have been shown to be effective in evaluating potential barriers and facilitators of adoption.16 Focus groups of and/or interviews with inpatient and PCPs, physician extenders, and housestaff could have been useful to better characterize the potential barriers and facilitators of adoption noted by survey respondents in our study.

In summary, we offer several lessons from our attempt to implement a system to manage pending tests at hospital discharge. The success of implementing future systems to address this patient safety concern will rely on accurately assigning responsibility for these test results, integrating the system within clinical information systems commonly used by the inpatient physician, addressing workflow issues and time constraints, maximizing appropriateness of alerting, and minimizing alert fatigue.

The period following discharge is a vulnerable time for patientsthe prevalence of medical errors related to this transition is high and has important patient safety and medico‐legal ramifications.13 Factors contributing to this vulnerability include complexity of hospitalized patients, shorter lengths of stay, and increased discontinuity of care. Hospitalists have recognized this threat to patient safety and have worked toward improving information exchange between inpatient and outpatient providers at hospital discharge.46 Nonetheless, the evidence suggests that more work is necessary. A recent study found that discharge summaries are often incomplete, and do not contain important information requiring follow‐up, such as pending tests.7 Additionally, a review by Kripalani et al. characterizing information deficits at hospital discharge found few interventions which specifically improve communication of pending tests at hospital discharge.8

In a prior study we determined that 41% of patients left the hospital before all laboratory and radiology test results were finalized. Of these results, 9.4% were potentially actionable and could have altered management. Physicians were aware of only 38% of post‐discharge test results.9 This awareness gap is a consequence of several factors including the lack of systems to track and alert providers of test results finalized post discharge. Also, it is unclear who is responsible for pending tests at discharge, since these tests are ordered by the inpatient physicians but often reported in the time period between hospital discharge and the patient's first follow‐up appointment with the primary care physician (PCP). Because responsibility is not explicitly made in the final communication between physicians at discharge, such test results may not be reviewed in a timely manner, potentially resulting in delays in treatment, a need for readmission, or other unfavorable outcomes.

Even in integrated health systems with advanced electronic health records, missed test results which result in treatment delays remain prevalent.10, 11 Test result management applications aid clinicians in reviewing and acting upon results as they become available and such systems may provide solutions to this problem. At Partners Healthcare in Boston, the Results Manager (RM) application was developed to help clinicians in the ambulatory setting safely, reliably, and efficiently review and act upon test results. The application enables clinicians to prioritize test results, utilize guidelines, and generate letters to patients. This system also prompts physicians to set reminders for future testing.12 In a 2.5‐year study evaluating the impact of this intervention, PCPs at 26 adult primary care practices were able to expedite communication of outpatient laboratory and imaging test results to patients with the help of RM. Patients of physicians who participated in the project reported greater satisfaction with test result communication and with information provided about their condition than did a control group of similar patients.13 RM has not yet been studied in the inpatient setting or at care transitions. We describe an attempt at modifying the Partners RM application to help inpatient physicians manage pending tests at hospital discharge.

Methods

Test Result Management Application

RM was originally developed by Partners Healthcare to improve timely review and appropriate management of test results in the ambulatory setting. RM was developed for and vetted by primarily ambulatory physicians. The application is browser‐based, provider‐centric, and embedded in the LMR to help ambulatory clinicians review and act upon test results in a safe, reliable, and efficient manner. Although RM has access to all inpatient and outpatient data in the Partners Clinical Data Repository (CDR), given the volume of inpatient tests ordered, hospital‐based results are suppressed by default to limit inundating ambulatory clinicians' queues. Therefore, users of RM only receive results of laboratory and radiology tests ordered in the ambulatory setting. They can track these tests for specific patients for a designated period of time by placing the patient on a watch list. Finally, RM incorporates extensive decision support features to classify the degree of abnormality for each result, presents guidelines to help clinicians manage abnormal results, allows clinicians to generate result letters to patients using predefined, context‐sensitive templates, and prompts physicians to set reminders for future testing. Because RM was developed from the ambulatory perspective, there was limited input from hospitalist physicians with regard to inpatient workflow in the original design of the module.12 See Figure 1 for a screen shot of RM and a description of its features.

Figure 1

For purposes of this pilot, we modified RM to allow results of tests ordered in the inpatient setting to be available for viewing (Hospitalist Results Manager, HRM). This feature was turned on only for inpatient physicians as previously defined. Inpatient tests, including pending tests at discharge, continued to be suppressed from PCP's RM queue (however, any physician could access a patient's test result(s) directly from the Partners CDR). Inpatient physicians could track laboratory and radiology results finalized after discharge by keeping discharged patients on their HRM watch list for a designated period of time. The finalized results would become available for review in their HRM queue and abnormal results were displayed prominently at the top of this queue. Inpatient physicians were trained to use HRM in a series of meetings and demonstrations. Although HRM could be accessed from inpatient clinical workstations, it was not part of the inpatient clinical information system.

Results

A total of 35 inpatient physicians participated in the pilot. Among 649 patients discharged during the study period, there were 1075 tests pending of which 555 were subsequently flagged as abnormal in HRM. Study surveys were sent to the 35 inpatient physician participants and 29 were completed, including partial responses (83% survey response rate). The 35 inpatient physician participants had the following characteristics: 22 were male, 13 were female; 21 were trainees and 14 were nontrainees/faculty. All 21 trainees were PGY2s. Nontrainees and faculty varied in experience level (PGY 15: 5, PGY 610: 7, PGY 1120: 1, PGY 21+: 1). Of 29 survey respondents, 7 were from hospital A and 22 were from hospital B; 19 were trainees and 10 were nontrainees/faculty. Of the 6 nonrespondents, 2 were from hospital A and 4 were from hospital B; 2 were trainees and 4 were nontrainees/faculty.

Table 1 shows the results of our survey of inpatient physicians regarding usage of HRM. Of 29 survey respondents, 14 (48%) reported never using HRM. Thirteen (45%) reported using HRM 1 to 2 times per week. None of the respondents used it more than 4 times per week. The frequency of usage was similar for hospitals A and B. Table 2 details barriers to using HRM. Twenty‐three inpatient physicians (79%) reported barriers. Seventeen (59%) thought that results in their HRM queue were not clinically relevant, 16 (55%) felt that HRM did not fit into their daily workflow, 14 (48%) had limited time to use HRM, and 12 (41%) noted that too many results in their HRM queue were on other physician's patients. Seven (24%) reported operational issues and 3 (10%) reported technical issues prohibiting use of HRM. With regard to preferred elements of an ideal results manager system, 21 (72%) inpatient physician respondents wanted to receive notification of abnormal and clinician‐designated pending test results. Four (14%) wanted to receive only abnormal results and 1 (3%) wanted to receive all results. Twenty‐seven (93%) physicians agreed that an ideally designed computerized test result management application would be valuable for managing pending tests at discharge.

Table 3 provides comments from inpatient physician respondents regarding obstacles prohibiting use of HRM and suggestions for future systems.

Discussion

We describe a pilot implementation of a computerized application for the management of pending tests at hospital discharge. From responses to post‐implementation surveys, we were able to identify multiple factors prohibiting successful implementation of the application. These observations may help inform future interventions and evaluations.

Almost half of inpatient physicians reported never using HRM despite training and reminders. The feedback provided by physicians in our study suggested that HRM was not ideally designed from an inpatient physician perspective. We discovered several barriers to its use: (1) HRM overburdened physicians with clinically irrelevant test results, suggesting that more robust filtering of abnormal but low importance test results may be required (eg, a borderline electrolyte abnormality or low but stable hematocrit); (2) HRM did not integrate well into inpatient workflowthe system was not integrated into the inpatient results viewing and computerized physician order entry (CPOE) applications, and therefore required an extra step to access; (3) there was no mechanism of alerting inpatient physicians that finalized test results were available for viewing in their HRM queues (eg, by email or by an alert in the inpatient computer system); (4) because responsibility for these results was unclear, most inpatient physicians had no formal method of managing them, and for many, using HRM represented an additional task; and finally (5) several physicians commented on finding results in their HRM queue that belonged to other physician's patients, implying that the hospital databases were inaccurate in identifying the discharging physician or that rotation schedules, and therefore patient responsibility, had changed in the intervening period. Table 4 summarizes the advantages and respective limitations of features of HRM available to inpatient physicians.

In the literature, there is little information regarding optimal features of a test result management system for transitions from the inpatient to ambulatory care setting. Prior studies outline important functions for results management systems developed for noninpatient sites of care, including the ambulatory and emergency room setting.12, 14, 15 These include a method of prioritizing by degree of abnormality, the ability to reliably and efficiently act upon results, and an automated alerting system for abnormal results. Findings from our study provide insight in defining core functions for result management systems which focus on transitions from the inpatient to ambulatory care setting. These functions include tight integration with applications used by inpatient physicians, clear assignment of responsibility for test results finalized after hospital discharge (as well as a mechanism to reassign responsibility), automated alerts to responsible providers of test results finalized post‐discharge, and ways to automatically filter test results to avoid over‐burdening physicians with clinically irrelevant results.

Almost all surveyed inpatient physicians agreed that an ideally designed electronic post‐discharge results management system would be valuable. For such systems to be successfully adopted, we offer several principles to help guide future work. These include: (1) clarifying responsibility at the time a test is ordered and again at discharge, (2) understanding workflow and communication patterns among inpatient and outpatient clinicians, and (3) integrating technological solutions into existing systems to minimize workflow disruptions. For example, if the primary responsibility for post‐discharge result follow‐up lies with the ordering physician, the system should be integrated within the EMR most often used by inpatient physicians and become part of inpatient physician workflow. If the system depends on administrative databases to identify the responsible providers, these must be accurate. Alternatively, in organizations with computerized provider order entry, responsibility for the result could be assigned when the test is ordered and confirmed at discharge (ie, the results management system would be integrated into the discharge order such that pending tests are reviewed at the time of discharge). The discharging physician should have the ability to assign responsibility for each pending test and select preferred mode(s) of notification once its result is finalized (eg, e‐mail, alphanumeric page, etc.). The system should have the ability to generate an automatic notification to the inpatient and PCP (and perhaps other designated providers involved in the patient's inpatient care), but it should not burden busy clinicians with unnecessary alerts and warnings. Finally, the rules by which results are prioritized must be robust enough to filter out less urgent results, and should be modified to reflect the severity of illness of recently discharged patients. In essence, in consideration of the time constraints of busy clinicians, an ideal results management system should achieve automated notification of test results while minimizing the risk of alert fatigue from the potentially large volume of alerts generated.

Our study has several important limitations. First, although our survey response rate was high, the sample size of actual participants was small. Second, because the study was conducted in 2 similar, tertiary care academic centers, it may not generalize to other settings (we note that hospital B included a nonteaching service similar to those in nonacademic medical centers). This may be particularly true in assessing the importance of specific barriers to use of results management systems, which may vary at different institutions. Third, the representation of survey respondents were skeweda majority of the responses were from trainees (all post‐graduate level [PGY] level 2) and from hospital B. Fourth, we did not actively monitor physician interaction with the test result management application, and therefore, we depended heavily on physician recollection of use of the system when responding to surveys. Finally, we did not convene focus groups of key individuals with regard to the factors facilitating or prohibiting adoption of the system. Use of semi‐structured, key informant interviews (ie, focus groups) before and after implementation of an electronic results management application, have been shown to be effective in evaluating potential barriers and facilitators of adoption.16 Focus groups of and/or interviews with inpatient and PCPs, physician extenders, and housestaff could have been useful to better characterize the potential barriers and facilitators of adoption noted by survey respondents in our study.

In summary, we offer several lessons from our attempt to implement a system to manage pending tests at hospital discharge. The success of implementing future systems to address this patient safety concern will rely on accurately assigning responsibility for these test results, integrating the system within clinical information systems commonly used by the inpatient physician, addressing workflow issues and time constraints, maximizing appropriateness of alerting, and minimizing alert fatigue.

Ford and Britting's1 editorial in this month's Journal of Hospital Medicine raises important questions concerning the use of nonphysician providers in hospital medicine. They focus primarily on the use of mid‐level providers (MLPs), namely physician‐assistants (PAs) and nurse practitioners (NPs), as a potential solution to the current physician workforce shortages in our field. While we acknowledge the challenges of meeting workforce needs, we also believe that much is unknown about the optimal use of MLPs on inpatient general medicine services and it is premature to tout MLPs as the solution to hospital medicine staffing problems. This is especially true in those hospitals where hospitalists care for complex, general medical patients with a wide variety of medical conditions, a trend that is especially common in academic medical centers.2

This article discusses the current literature, our own experiences with MLPs, and suggests some future initiatives that might help better integrate MLPs into hospital medicine.

The Literature on MLPs in Inpatient Venues

The existing literature on the use of MLPs in inpatient venues is quite limited, and a recent review, while suggesting that the existing literature does describe benefits of MLPs in the inpatient setting, also states that the overall quality of the evidence is quite poor and that many studies suffer from significant limitations, including small populations, limited patient mixes, use of selected settings, and short durations of outcome assessment.3

Ford and Britting,1 in their article, cite several studies46 as evidence that a MLP model of care either improved outcomes or provided cost benefits. Each of these studies has important limitations that are worth examining.

The study by Myers et al.4 described the use of MLPs in a chest pain unit. NPs partnered with hospitalists to care for a low‐acuity chest pain population. In addition, 5 NPs only staffed the unit during daytime weekday hours. Off‐hour and weekend staffing was accomplished through the use of resident physicians. Notably, the work suggests the service only admitted 113 low‐risk patients over 10 months. The service was staffed by 3 full‐time equivalent (FTE) NPs in addition to involving hospitalists during the day. It is not surprising, given the extremely low volume of patients coupled with a daytime‐only focus, that this service showed efficiency gains. In addition, given the service was only staffed by NPs 40 hours a week and by resident physicians on nights and weekends, the true cost of such an intervention needs to take into account the full cost of 24/7 coverage. In addition, the model of using residents to cover nonteaching patients is no longer permitted by the current Accreditation Council for Graduate Medical Education (ACGME) Internal Medicine Residency Requirements7 and thus implementation of a model such as this in 2009 would require alternative means of nighttime coverage.

The study by Nishimura et al.,5 also describing the use of MLPs in cardiovascular care, has important caveats that make full assessment of the model impossible. The model describes the implementation of a care team consisting of an attending, a fellow, and MLPs to replace a traditional teaching team of an attending, senior resident, and 2 interns. The study states that the model resulted in a lower length of stay (LOS) and lower costs per case. Importantly, the new MLP‐based team only admitted during the hours of 7 _AM to 2 _PM. The study does not fully describe the number of MLPs required nor does it fully describe the role of cardiovascular fellows in the model. The study does state that the cost savings offset the cost of the MLPs but it is not clear if this cost analysis took into account the cost of the fellow's daytime involvement or if it measured attending time required before and after the implementation of the new model. In addition, this model presumes the availability of other services to admit patients during afternoon and nighttime hours and so may not be generalizable to other settings.

The final study by Cowan et al.6 describes the addition of a NP, a hospitalist medical director, and daily multidisciplinary rounds to a traditional teaching service model. Importantly, the NP was not involved in the admission process nor were they the primary providers for day‐to‐day medical care but rather they focused on implementation of care protocols, multidisciplinary coordination of care and discharge planning, and postdischarge follow‐up. In addition, the NP worked only weekdays for about 40 hours a week. It is not surprising that adding multiple additional resources to existing care models might provide benefits but this does not address any issues in terms of the workforce since the care in this model required a higher total input of providers than the usual care model being studied. Cost savings from such a model may make it cost‐effective but it does not represent a workforce solution.

There have been other studies examining the use of MLPs in the inpatient setting in internal medicine. Some of these studies have suggested that MLP‐based models result in equivalent outcomes and efficiency810 to traditional teaching or nonteaching physician‐only models. There are 2 important caveats, however, that must be considered. The total resources required for such models may be quite high, especially taking into account the costs of 24/7 coverage and physician backup of the MLPs, and most importantly there is almost no literature that robustly examines ultimate clinical outcomes in these models. We do note that a recent study11 did show a lower inpatient mortality rate over a 2‐year period of time after substituting a PA‐hospitalist model for a traditional academic medicine residency model in a community hospital. Importantly, however, the new model also added 24/7 hospitalist physicians and night and weekend intensivists that were not present in the prior residency‐based model. Thus, the lower mortality rate could be attributed to the addition of hospitalists or the more robust in‐house physician coverage during off‐hours rather than the use of MLPs.

Notably, while the evidence base in internal medicine is not robust, many studies have described successful use of MLPs in non‐internal medicine inpatient settings.1214 The reasons for this success is debatable, but it may be that MLPs are more successful in settings where the care is either more protocol‐driven or where there is less diagnostic and therapeutic complexity.

Recent Experiences with MLPs in Academic Hospital Medicine

Given the paucity of data, it is clear that further research is needed on the role of MLPs in hospital medicine. While waiting for such evidence to appear, it may be worthwhile to reflect on the recent experience of 3 major medical centers. A recent article described 5 hospitalist models at major academic medical centers across the country. Two of the institutions described at the time (University of Michigan Health System, Ann Arbor, MI; and Brigham and Women's Hospital, Boston, MA) utilized MLPs as a major element of their staffing of nonresident hospitalist services while another (University of California, San Francisco [UCSF] Medical Center at Mt. Zion, San Francisco, CA) had previously used MLPs as part of its model but phased them out about 1 year prior to publication of the article.2 The model used by the Brigham and Women's Hospital was later described in more detail in a subsequent publication.8 Recently 1 of these institutions (Michigan) has chosen to phase out MLPs. At Michigan, a 4‐year experience with PAs on a general‐medicine focused hospitalist service eventually led to the conclusion that continued use of PAs was not cost‐effective. Significant barriers to success included a steep learning curve and the significant time required before PAs developed sufficient autonomy and efficiency in caring for a highly complex heterogeneous patient population. In the Michigan experience, PAs took up to 2 years to attain a significant level of autonomy and efficiency and even then some PAs still required a significant amount of physician oversight. Similar concerns at UCSF Mt. Zion led to the elimination of their MLP program as well. At Brigham and Women's, the MLP service continues but has required additional hospitalist staffing due to difficulties recruiting qualified MLPs with appropriate inpatient experience. In all cases, the models were challenged by high costs and the difficulty of developing MLPs to attain the level of autonomy and efficiency needed to justify their continued use. A key point is that in each institution, MLPs continue to play an important role in some specialty inpatient areas such as Hematology/Oncology and Bone Marrow Transplant, which is where MLPs have traditionally found their niche in inpatient Internal Medicine. These focus shops allow MLPs to develop a niche and expertise in a specialized area, where they may become more autonomous and efficient than house staff. Thus these settings may be more appropriate for MLPs than a heterogeneous general medicine inpatient setting.

Reviewing the Financial Case

In their article, Ford and Britting1 cite potential financial advantages for the use of MLPs in hospital medicine by comparing the relative salaries of MLPs to Hospitalists. What was missing in their analysis was the relative productivity of the 2 types of providers. We do have some limited data from the Society of Hospital Medicine (SHM) annual survey that looks at MLPs in hospital medicine but, again, the number of respondents for most data elements is less than 70, making generalizability difficult. Nonetheless, the data suggest that MLPs in hospital medicine average about 60% to 75% of the productivity of a physician when measured by encounters, although there is wide variability depending on the employment model (academic vs. multispecialty group).15 Importantly, the existing data do not provide any measure of how much physician input is provided to these MLPs but we suspect that in most models there is some physician time and input. If we presume that the MLPs bill independently and collect 85% of the physician fee schedule for a Medicare population, then collections would be about 50% to 65% of a typical physician. Given that median total compensation including benefits from the SHM survey was $120,000 for MLPs and $216,000 for physiciansabout a 55% ratiothis would argue for potential financial neutrality when substituting MLPs for physicians in a 2:1 ratio but only if we presume they require no physician supervision, which in our own experience is not likely in a general medicine population. In an alternative model, in which the physician sees every patient with the MLP and the physician bills, one would need to see roughly 50% more patients to achieve a financially neutral situation. In our experience at our own institutions, this level of increased productivity was not achievable. It is important to note that our figures are median compensation and benefit cost figures and local markets vary widely. We know that in major east and west coast cities MLPs may command far higher salaries while early career hospitalist physicians may be paid somewhat less than the reported medians. Recent market changes have significantly pressured MLP salaries,15, 16 further impacting the financial equation and perhaps tilting it farther against a financial benefit for MLPs. Furthermore, night coverage for MLP services should always be considered in a financial analysis and is not captured in this simple analysis.

Next Steps

Given the current shortage of physicians, we imagine that many hospitalist groups will consider the use of MLPs as a solution to the current workforce issues. However, data on how best to utilize MLPs and the true impact on both the cost and quality of such models is lacking. In addition to urging increased publication and dissemination of existing experiences with NP and PAs, we strongly suggest that groups considering starting a MLP model do so in a way which would facilitate robust analysis and comparison of the model with alternatives. We also suggest that SHM consider the following: modifying its biennial survey to better capture the nuances of MLP productivity (such as assessing the amount of physician input and supervision required); targeting MLPs so as to increase the number of respondents; and doing an additional survey to capture demographics and basic data on existing MLP models given the lack of published literature.

In addition to gathering more data on effective models, a critical gap that we have identified is the development of models for the training and development of MLPs interested in hospital medicine. It would be a mistake to believe that MLPs could function in a manner similar to residency‐trained physicians if they do not undergo similar training. NP/PA programs generally do not have a significant inpatient internal medicine focus and so newly minted graduates often lack the skills needed to succeed in hospital medicine.17 Some hospitalist programs train their MLPs on the job, but many programs cannot afford the amount of time and effort required to do this on their own. There are a small number of advanced training options for MLPs in hospital medicine18 but it is not likely such models will proliferate given the inherent opportunity costs that exist for extended training in the current competitive job market for MLPs. Instead we think that very motivated hospital medicine groups may develop training relationships with PA and NP schools in an effort to train their own. In addition, national initiatives such as the Hospital Medicine Boot Camp for NPs and PAs, which is cosponsored by SHM, the American Association of Physician Assistants (AAPA), and the American Academy of Nurse Practitioners (AANP),19 can help fill the educational needs for MLPs who are already in practice.

Conclusions

While some literature exists that suggests that MLPs can successfully be used in the inpatient internal medicine setting, it is important to note that the evidence is quite limited and cannot be generalized across all care settings and patient populations. There is an urgent need to gather more data and share our collective experiences to better inform our decision‐making before we state that MLPs are the solution to workforce shortages in hospital medicine. In addition, existing data and experience suggest that MLPs may not be a cost‐effective workforce solution for complex general medical patients who require significant physician input. We believe that redesigning the clinical training of MLPs to focus on inpatient skills may hold promise and encourage interested parties to consider developing partnerships with MLP training programs and hospital medicine groups, as a way to build a more robust and successful hospital medicine MLP workforce.

Ford and Britting's1 editorial in this month's Journal of Hospital Medicine raises important questions concerning the use of nonphysician providers in hospital medicine. They focus primarily on the use of mid‐level providers (MLPs), namely physician‐assistants (PAs) and nurse practitioners (NPs), as a potential solution to the current physician workforce shortages in our field. While we acknowledge the challenges of meeting workforce needs, we also believe that much is unknown about the optimal use of MLPs on inpatient general medicine services and it is premature to tout MLPs as the solution to hospital medicine staffing problems. This is especially true in those hospitals where hospitalists care for complex, general medical patients with a wide variety of medical conditions, a trend that is especially common in academic medical centers.2

This article discusses the current literature, our own experiences with MLPs, and suggests some future initiatives that might help better integrate MLPs into hospital medicine.

The Literature on MLPs in Inpatient Venues

The existing literature on the use of MLPs in inpatient venues is quite limited, and a recent review, while suggesting that the existing literature does describe benefits of MLPs in the inpatient setting, also states that the overall quality of the evidence is quite poor and that many studies suffer from significant limitations, including small populations, limited patient mixes, use of selected settings, and short durations of outcome assessment.3

Ford and Britting,1 in their article, cite several studies46 as evidence that a MLP model of care either improved outcomes or provided cost benefits. Each of these studies has important limitations that are worth examining.

The study by Myers et al.4 described the use of MLPs in a chest pain unit. NPs partnered with hospitalists to care for a low‐acuity chest pain population. In addition, 5 NPs only staffed the unit during daytime weekday hours. Off‐hour and weekend staffing was accomplished through the use of resident physicians. Notably, the work suggests the service only admitted 113 low‐risk patients over 10 months. The service was staffed by 3 full‐time equivalent (FTE) NPs in addition to involving hospitalists during the day. It is not surprising, given the extremely low volume of patients coupled with a daytime‐only focus, that this service showed efficiency gains. In addition, given the service was only staffed by NPs 40 hours a week and by resident physicians on nights and weekends, the true cost of such an intervention needs to take into account the full cost of 24/7 coverage. In addition, the model of using residents to cover nonteaching patients is no longer permitted by the current Accreditation Council for Graduate Medical Education (ACGME) Internal Medicine Residency Requirements7 and thus implementation of a model such as this in 2009 would require alternative means of nighttime coverage.

The study by Nishimura et al.,5 also describing the use of MLPs in cardiovascular care, has important caveats that make full assessment of the model impossible. The model describes the implementation of a care team consisting of an attending, a fellow, and MLPs to replace a traditional teaching team of an attending, senior resident, and 2 interns. The study states that the model resulted in a lower length of stay (LOS) and lower costs per case. Importantly, the new MLP‐based team only admitted during the hours of 7 _AM to 2 _PM. The study does not fully describe the number of MLPs required nor does it fully describe the role of cardiovascular fellows in the model. The study does state that the cost savings offset the cost of the MLPs but it is not clear if this cost analysis took into account the cost of the fellow's daytime involvement or if it measured attending time required before and after the implementation of the new model. In addition, this model presumes the availability of other services to admit patients during afternoon and nighttime hours and so may not be generalizable to other settings.

The final study by Cowan et al.6 describes the addition of a NP, a hospitalist medical director, and daily multidisciplinary rounds to a traditional teaching service model. Importantly, the NP was not involved in the admission process nor were they the primary providers for day‐to‐day medical care but rather they focused on implementation of care protocols, multidisciplinary coordination of care and discharge planning, and postdischarge follow‐up. In addition, the NP worked only weekdays for about 40 hours a week. It is not surprising that adding multiple additional resources to existing care models might provide benefits but this does not address any issues in terms of the workforce since the care in this model required a higher total input of providers than the usual care model being studied. Cost savings from such a model may make it cost‐effective but it does not represent a workforce solution.

There have been other studies examining the use of MLPs in the inpatient setting in internal medicine. Some of these studies have suggested that MLP‐based models result in equivalent outcomes and efficiency810 to traditional teaching or nonteaching physician‐only models. There are 2 important caveats, however, that must be considered. The total resources required for such models may be quite high, especially taking into account the costs of 24/7 coverage and physician backup of the MLPs, and most importantly there is almost no literature that robustly examines ultimate clinical outcomes in these models. We do note that a recent study11 did show a lower inpatient mortality rate over a 2‐year period of time after substituting a PA‐hospitalist model for a traditional academic medicine residency model in a community hospital. Importantly, however, the new model also added 24/7 hospitalist physicians and night and weekend intensivists that were not present in the prior residency‐based model. Thus, the lower mortality rate could be attributed to the addition of hospitalists or the more robust in‐house physician coverage during off‐hours rather than the use of MLPs.

Notably, while the evidence base in internal medicine is not robust, many studies have described successful use of MLPs in non‐internal medicine inpatient settings.1214 The reasons for this success is debatable, but it may be that MLPs are more successful in settings where the care is either more protocol‐driven or where there is less diagnostic and therapeutic complexity.

Recent Experiences with MLPs in Academic Hospital Medicine

Given the paucity of data, it is clear that further research is needed on the role of MLPs in hospital medicine. While waiting for such evidence to appear, it may be worthwhile to reflect on the recent experience of 3 major medical centers. A recent article described 5 hospitalist models at major academic medical centers across the country. Two of the institutions described at the time (University of Michigan Health System, Ann Arbor, MI; and Brigham and Women's Hospital, Boston, MA) utilized MLPs as a major element of their staffing of nonresident hospitalist services while another (University of California, San Francisco [UCSF] Medical Center at Mt. Zion, San Francisco, CA) had previously used MLPs as part of its model but phased them out about 1 year prior to publication of the article.2 The model used by the Brigham and Women's Hospital was later described in more detail in a subsequent publication.8 Recently 1 of these institutions (Michigan) has chosen to phase out MLPs. At Michigan, a 4‐year experience with PAs on a general‐medicine focused hospitalist service eventually led to the conclusion that continued use of PAs was not cost‐effective. Significant barriers to success included a steep learning curve and the significant time required before PAs developed sufficient autonomy and efficiency in caring for a highly complex heterogeneous patient population. In the Michigan experience, PAs took up to 2 years to attain a significant level of autonomy and efficiency and even then some PAs still required a significant amount of physician oversight. Similar concerns at UCSF Mt. Zion led to the elimination of their MLP program as well. At Brigham and Women's, the MLP service continues but has required additional hospitalist staffing due to difficulties recruiting qualified MLPs with appropriate inpatient experience. In all cases, the models were challenged by high costs and the difficulty of developing MLPs to attain the level of autonomy and efficiency needed to justify their continued use. A key point is that in each institution, MLPs continue to play an important role in some specialty inpatient areas such as Hematology/Oncology and Bone Marrow Transplant, which is where MLPs have traditionally found their niche in inpatient Internal Medicine. These focus shops allow MLPs to develop a niche and expertise in a specialized area, where they may become more autonomous and efficient than house staff. Thus these settings may be more appropriate for MLPs than a heterogeneous general medicine inpatient setting.

Reviewing the Financial Case

In their article, Ford and Britting1 cite potential financial advantages for the use of MLPs in hospital medicine by comparing the relative salaries of MLPs to Hospitalists. What was missing in their analysis was the relative productivity of the 2 types of providers. We do have some limited data from the Society of Hospital Medicine (SHM) annual survey that looks at MLPs in hospital medicine but, again, the number of respondents for most data elements is less than 70, making generalizability difficult. Nonetheless, the data suggest that MLPs in hospital medicine average about 60% to 75% of the productivity of a physician when measured by encounters, although there is wide variability depending on the employment model (academic vs. multispecialty group).15 Importantly, the existing data do not provide any measure of how much physician input is provided to these MLPs but we suspect that in most models there is some physician time and input. If we presume that the MLPs bill independently and collect 85% of the physician fee schedule for a Medicare population, then collections would be about 50% to 65% of a typical physician. Given that median total compensation including benefits from the SHM survey was $120,000 for MLPs and $216,000 for physiciansabout a 55% ratiothis would argue for potential financial neutrality when substituting MLPs for physicians in a 2:1 ratio but only if we presume they require no physician supervision, which in our own experience is not likely in a general medicine population. In an alternative model, in which the physician sees every patient with the MLP and the physician bills, one would need to see roughly 50% more patients to achieve a financially neutral situation. In our experience at our own institutions, this level of increased productivity was not achievable. It is important to note that our figures are median compensation and benefit cost figures and local markets vary widely. We know that in major east and west coast cities MLPs may command far higher salaries while early career hospitalist physicians may be paid somewhat less than the reported medians. Recent market changes have significantly pressured MLP salaries,15, 16 further impacting the financial equation and perhaps tilting it farther against a financial benefit for MLPs. Furthermore, night coverage for MLP services should always be considered in a financial analysis and is not captured in this simple analysis.

Next Steps

Given the current shortage of physicians, we imagine that many hospitalist groups will consider the use of MLPs as a solution to the current workforce issues. However, data on how best to utilize MLPs and the true impact on both the cost and quality of such models is lacking. In addition to urging increased publication and dissemination of existing experiences with NP and PAs, we strongly suggest that groups considering starting a MLP model do so in a way which would facilitate robust analysis and comparison of the model with alternatives. We also suggest that SHM consider the following: modifying its biennial survey to better capture the nuances of MLP productivity (such as assessing the amount of physician input and supervision required); targeting MLPs so as to increase the number of respondents; and doing an additional survey to capture demographics and basic data on existing MLP models given the lack of published literature.

In addition to gathering more data on effective models, a critical gap that we have identified is the development of models for the training and development of MLPs interested in hospital medicine. It would be a mistake to believe that MLPs could function in a manner similar to residency‐trained physicians if they do not undergo similar training. NP/PA programs generally do not have a significant inpatient internal medicine focus and so newly minted graduates often lack the skills needed to succeed in hospital medicine.17 Some hospitalist programs train their MLPs on the job, but many programs cannot afford the amount of time and effort required to do this on their own. There are a small number of advanced training options for MLPs in hospital medicine18 but it is not likely such models will proliferate given the inherent opportunity costs that exist for extended training in the current competitive job market for MLPs. Instead we think that very motivated hospital medicine groups may develop training relationships with PA and NP schools in an effort to train their own. In addition, national initiatives such as the Hospital Medicine Boot Camp for NPs and PAs, which is cosponsored by SHM, the American Association of Physician Assistants (AAPA), and the American Academy of Nurse Practitioners (AANP),19 can help fill the educational needs for MLPs who are already in practice.

Conclusions

While some literature exists that suggests that MLPs can successfully be used in the inpatient internal medicine setting, it is important to note that the evidence is quite limited and cannot be generalized across all care settings and patient populations. There is an urgent need to gather more data and share our collective experiences to better inform our decision‐making before we state that MLPs are the solution to workforce shortages in hospital medicine. In addition, existing data and experience suggest that MLPs may not be a cost‐effective workforce solution for complex general medical patients who require significant physician input. We believe that redesigning the clinical training of MLPs to focus on inpatient skills may hold promise and encourage interested parties to consider developing partnerships with MLP training programs and hospital medicine groups, as a way to build a more robust and successful hospital medicine MLP workforce.

Ford and Britting's1 editorial in this month's Journal of Hospital Medicine raises important questions concerning the use of nonphysician providers in hospital medicine. They focus primarily on the use of mid‐level providers (MLPs), namely physician‐assistants (PAs) and nurse practitioners (NPs), as a potential solution to the current physician workforce shortages in our field. While we acknowledge the challenges of meeting workforce needs, we also believe that much is unknown about the optimal use of MLPs on inpatient general medicine services and it is premature to tout MLPs as the solution to hospital medicine staffing problems. This is especially true in those hospitals where hospitalists care for complex, general medical patients with a wide variety of medical conditions, a trend that is especially common in academic medical centers.2

This article discusses the current literature, our own experiences with MLPs, and suggests some future initiatives that might help better integrate MLPs into hospital medicine.

The Literature on MLPs in Inpatient Venues

The existing literature on the use of MLPs in inpatient venues is quite limited, and a recent review, while suggesting that the existing literature does describe benefits of MLPs in the inpatient setting, also states that the overall quality of the evidence is quite poor and that many studies suffer from significant limitations, including small populations, limited patient mixes, use of selected settings, and short durations of outcome assessment.3

Ford and Britting,1 in their article, cite several studies46 as evidence that a MLP model of care either improved outcomes or provided cost benefits. Each of these studies has important limitations that are worth examining.

The study by Myers et al.4 described the use of MLPs in a chest pain unit. NPs partnered with hospitalists to care for a low‐acuity chest pain population. In addition, 5 NPs only staffed the unit during daytime weekday hours. Off‐hour and weekend staffing was accomplished through the use of resident physicians. Notably, the work suggests the service only admitted 113 low‐risk patients over 10 months. The service was staffed by 3 full‐time equivalent (FTE) NPs in addition to involving hospitalists during the day. It is not surprising, given the extremely low volume of patients coupled with a daytime‐only focus, that this service showed efficiency gains. In addition, given the service was only staffed by NPs 40 hours a week and by resident physicians on nights and weekends, the true cost of such an intervention needs to take into account the full cost of 24/7 coverage. In addition, the model of using residents to cover nonteaching patients is no longer permitted by the current Accreditation Council for Graduate Medical Education (ACGME) Internal Medicine Residency Requirements7 and thus implementation of a model such as this in 2009 would require alternative means of nighttime coverage.

The study by Nishimura et al.,5 also describing the use of MLPs in cardiovascular care, has important caveats that make full assessment of the model impossible. The model describes the implementation of a care team consisting of an attending, a fellow, and MLPs to replace a traditional teaching team of an attending, senior resident, and 2 interns. The study states that the model resulted in a lower length of stay (LOS) and lower costs per case. Importantly, the new MLP‐based team only admitted during the hours of 7 _AM to 2 _PM. The study does not fully describe the number of MLPs required nor does it fully describe the role of cardiovascular fellows in the model. The study does state that the cost savings offset the cost of the MLPs but it is not clear if this cost analysis took into account the cost of the fellow's daytime involvement or if it measured attending time required before and after the implementation of the new model. In addition, this model presumes the availability of other services to admit patients during afternoon and nighttime hours and so may not be generalizable to other settings.

The final study by Cowan et al.6 describes the addition of a NP, a hospitalist medical director, and daily multidisciplinary rounds to a traditional teaching service model. Importantly, the NP was not involved in the admission process nor were they the primary providers for day‐to‐day medical care but rather they focused on implementation of care protocols, multidisciplinary coordination of care and discharge planning, and postdischarge follow‐up. In addition, the NP worked only weekdays for about 40 hours a week. It is not surprising that adding multiple additional resources to existing care models might provide benefits but this does not address any issues in terms of the workforce since the care in this model required a higher total input of providers than the usual care model being studied. Cost savings from such a model may make it cost‐effective but it does not represent a workforce solution.

There have been other studies examining the use of MLPs in the inpatient setting in internal medicine. Some of these studies have suggested that MLP‐based models result in equivalent outcomes and efficiency810 to traditional teaching or nonteaching physician‐only models. There are 2 important caveats, however, that must be considered. The total resources required for such models may be quite high, especially taking into account the costs of 24/7 coverage and physician backup of the MLPs, and most importantly there is almost no literature that robustly examines ultimate clinical outcomes in these models. We do note that a recent study11 did show a lower inpatient mortality rate over a 2‐year period of time after substituting a PA‐hospitalist model for a traditional academic medicine residency model in a community hospital. Importantly, however, the new model also added 24/7 hospitalist physicians and night and weekend intensivists that were not present in the prior residency‐based model. Thus, the lower mortality rate could be attributed to the addition of hospitalists or the more robust in‐house physician coverage during off‐hours rather than the use of MLPs.

Notably, while the evidence base in internal medicine is not robust, many studies have described successful use of MLPs in non‐internal medicine inpatient settings.1214 The reasons for this success is debatable, but it may be that MLPs are more successful in settings where the care is either more protocol‐driven or where there is less diagnostic and therapeutic complexity.

Recent Experiences with MLPs in Academic Hospital Medicine

Given the paucity of data, it is clear that further research is needed on the role of MLPs in hospital medicine. While waiting for such evidence to appear, it may be worthwhile to reflect on the recent experience of 3 major medical centers. A recent article described 5 hospitalist models at major academic medical centers across the country. Two of the institutions described at the time (University of Michigan Health System, Ann Arbor, MI; and Brigham and Women's Hospital, Boston, MA) utilized MLPs as a major element of their staffing of nonresident hospitalist services while another (University of California, San Francisco [UCSF] Medical Center at Mt. Zion, San Francisco, CA) had previously used MLPs as part of its model but phased them out about 1 year prior to publication of the article.2 The model used by the Brigham and Women's Hospital was later described in more detail in a subsequent publication.8 Recently 1 of these institutions (Michigan) has chosen to phase out MLPs. At Michigan, a 4‐year experience with PAs on a general‐medicine focused hospitalist service eventually led to the conclusion that continued use of PAs was not cost‐effective. Significant barriers to success included a steep learning curve and the significant time required before PAs developed sufficient autonomy and efficiency in caring for a highly complex heterogeneous patient population. In the Michigan experience, PAs took up to 2 years to attain a significant level of autonomy and efficiency and even then some PAs still required a significant amount of physician oversight. Similar concerns at UCSF Mt. Zion led to the elimination of their MLP program as well. At Brigham and Women's, the MLP service continues but has required additional hospitalist staffing due to difficulties recruiting qualified MLPs with appropriate inpatient experience. In all cases, the models were challenged by high costs and the difficulty of developing MLPs to attain the level of autonomy and efficiency needed to justify their continued use. A key point is that in each institution, MLPs continue to play an important role in some specialty inpatient areas such as Hematology/Oncology and Bone Marrow Transplant, which is where MLPs have traditionally found their niche in inpatient Internal Medicine. These focus shops allow MLPs to develop a niche and expertise in a specialized area, where they may become more autonomous and efficient than house staff. Thus these settings may be more appropriate for MLPs than a heterogeneous general medicine inpatient setting.

Reviewing the Financial Case

In their article, Ford and Britting1 cite potential financial advantages for the use of MLPs in hospital medicine by comparing the relative salaries of MLPs to Hospitalists. What was missing in their analysis was the relative productivity of the 2 types of providers. We do have some limited data from the Society of Hospital Medicine (SHM) annual survey that looks at MLPs in hospital medicine but, again, the number of respondents for most data elements is less than 70, making generalizability difficult. Nonetheless, the data suggest that MLPs in hospital medicine average about 60% to 75% of the productivity of a physician when measured by encounters, although there is wide variability depending on the employment model (academic vs. multispecialty group).15 Importantly, the existing data do not provide any measure of how much physician input is provided to these MLPs but we suspect that in most models there is some physician time and input. If we presume that the MLPs bill independently and collect 85% of the physician fee schedule for a Medicare population, then collections would be about 50% to 65% of a typical physician. Given that median total compensation including benefits from the SHM survey was $120,000 for MLPs and $216,000 for physiciansabout a 55% ratiothis would argue for potential financial neutrality when substituting MLPs for physicians in a 2:1 ratio but only if we presume they require no physician supervision, which in our own experience is not likely in a general medicine population. In an alternative model, in which the physician sees every patient with the MLP and the physician bills, one would need to see roughly 50% more patients to achieve a financially neutral situation. In our experience at our own institutions, this level of increased productivity was not achievable. It is important to note that our figures are median compensation and benefit cost figures and local markets vary widely. We know that in major east and west coast cities MLPs may command far higher salaries while early career hospitalist physicians may be paid somewhat less than the reported medians. Recent market changes have significantly pressured MLP salaries,15, 16 further impacting the financial equation and perhaps tilting it farther against a financial benefit for MLPs. Furthermore, night coverage for MLP services should always be considered in a financial analysis and is not captured in this simple analysis.

Next Steps

Given the current shortage of physicians, we imagine that many hospitalist groups will consider the use of MLPs as a solution to the current workforce issues. However, data on how best to utilize MLPs and the true impact on both the cost and quality of such models is lacking. In addition to urging increased publication and dissemination of existing experiences with NP and PAs, we strongly suggest that groups considering starting a MLP model do so in a way which would facilitate robust analysis and comparison of the model with alternatives. We also suggest that SHM consider the following: modifying its biennial survey to better capture the nuances of MLP productivity (such as assessing the amount of physician input and supervision required); targeting MLPs so as to increase the number of respondents; and doing an additional survey to capture demographics and basic data on existing MLP models given the lack of published literature.

In addition to gathering more data on effective models, a critical gap that we have identified is the development of models for the training and development of MLPs interested in hospital medicine. It would be a mistake to believe that MLPs could function in a manner similar to residency‐trained physicians if they do not undergo similar training. NP/PA programs generally do not have a significant inpatient internal medicine focus and so newly minted graduates often lack the skills needed to succeed in hospital medicine.17 Some hospitalist programs train their MLPs on the job, but many programs cannot afford the amount of time and effort required to do this on their own. There are a small number of advanced training options for MLPs in hospital medicine18 but it is not likely such models will proliferate given the inherent opportunity costs that exist for extended training in the current competitive job market for MLPs. Instead we think that very motivated hospital medicine groups may develop training relationships with PA and NP schools in an effort to train their own. In addition, national initiatives such as the Hospital Medicine Boot Camp for NPs and PAs, which is cosponsored by SHM, the American Association of Physician Assistants (AAPA), and the American Academy of Nurse Practitioners (AANP),19 can help fill the educational needs for MLPs who are already in practice.

Conclusions

While some literature exists that suggests that MLPs can successfully be used in the inpatient internal medicine setting, it is important to note that the evidence is quite limited and cannot be generalized across all care settings and patient populations. There is an urgent need to gather more data and share our collective experiences to better inform our decision‐making before we state that MLPs are the solution to workforce shortages in hospital medicine. In addition, existing data and experience suggest that MLPs may not be a cost‐effective workforce solution for complex general medical patients who require significant physician input. We believe that redesigning the clinical training of MLPs to focus on inpatient skills may hold promise and encourage interested parties to consider developing partnerships with MLP training programs and hospital medicine groups, as a way to build a more robust and successful hospital medicine MLP workforce.

Midlevel providers (physician assistants and nurse practitioners) have long been employed by academic medical centers, predominantly on surgical services, or on medical subspecialty services, where they have typically had a limited scope of practice, focused in a narrowly defined area or set of procedures.17 In contrast, there are relatively few reports of experiences deploying midlevel providers to replace house staff on inpatient general medicine services in academic centers,810 and few studies of the effect of midlevel providers on quality and efficiency of care in the academic setting. Despite this, reductions in house officer duty hours as mandated by the Accreditation Council on Graduate Medical Education (ACGME)11 have resulted in academic centers increasingly using midlevel providers to decrease house staff workload on inpatient services.12, 13 In general, midlevel practitioners on general medicine services have been deployed to: (1) care for a population of patients separate from and in parallel with house staff; this population may be narrowly defined (eg, patients with chest pain) or not; (2) assist with the management of patients cared for by house staff by performing certain tasks (eg, scheduling appointments, discharging patients). Even as midlevel providers become more prevalent on academic general medicine services, the best model of care incorporating them into clinical care remains unclear, and few studies have rigorously examined the care provided on services that use them.

We developed an inpatient general medicine service within a large academic medical center staffed by physician assistants and hospitalists to help our residency program meet ACGME duty hour requirements. We hypothesized that by creating a service that is geographically localized and supervised by full‐time hospitalists, by instituting multidisciplinary rounds, and by investing in the professional development of highly‐skilled physician assistants, we could provide care for medically complex, acutely ill general medicine inpatients with similar quality and efficiency as compared to house staff teams. We report our experience during the first year of implementing the service, and compare quality and efficiency of care on this service with that of our traditional house staff services. We also evaluate the effects of this service on patient satisfaction and self‐reported house staff workload.

PATIENTS AND METHODS

RESULTS

Patient Satisfaction

Patients were similarly satisfied with their care on the PACE service and on the house staff services. In specific areas and globally, percentages of patients satisfied with their physicians and with the discharge process were not different, as measured by the Press‐Ganey survey (Press‐Ganey Associates, South Bend, IN; Figures 1 and 2). The survey distinguishes between attendings and residents, but not physician assistants; therefore, Figure 1 only includes responses to the attending questions. Given the sampling procedure of the Press‐Ganey survey, exact response rates cannot be calculated, but Press‐Ganey reports a response rate of about 40% for the English survey and about 20% for the Spanish survey.

Figure 1

Figure 2

DISCUSSION

As academic centers struggle with developing a workforce to provide patient care no longer provided by residents, questions about the ideal structure of nonhouse staff inpatient services abound. Although solutions to this problem will be determined to some extent by local factors such as institutional culture and resources, some lessons learned in developing such services will be more widely applicable. We found that by implementing a geographically localized, physician assistant‐staffed hospitalist service, we were able to provide care of similar quality and efficiency to that of traditional house staff services, despite inexperienced hospitalists staffing the service and a medical residency program commonly recognized as one of the best in the country. Adjusted total costs were slightly lower on the PACE service, but this difference was small and of borderline statistical significance. Likewise, no significant differences were seen in any of several quality measures or in patient satisfaction.

Our findings add to the available evidence supporting the use of physician assistants on academic general medicine services, and are germane to academic centers facing reductions in house staff availability and seeking alternative models of care for inpatients. Several specific characteristics of the PACE service and the implications of these should be considered:

• 1

  The service accepted all patients, regardless of diagnosis, acuity, or complexity of illness. This was unlike many previously described nonhouse staff services which were more limited in scope, and allowed more flexibility with patient flow. However, in the end, patients on the PACE service did have a modestly lower case mix index and Charlson score, suggesting that, despite a lack of triage guidelines, there was some bias in the triage of admissions, possibly due to a perception that physician assistants should take care of lower complexity patients. If it is desirable to have a similar distribution of higher complexity patients across house staff and nonhouse staff services, extra efforts may be necessary to overcome this perception.

• 2

  The service was geographically regionalized. Geographic regionalization offered many important advantages, especially with regards to communication among staff, nursing, and consultants, and allowed for multidisciplinary rounds. However, it is possible that the modest, but not statistically significant, trend toward an increased LOS seen on the PACE service might be a reflection of geographic admitting (less incentive to discharge since discharging a patient means taking a new admission).

• 3

  The education and professional development of the physician assistants was a priority. Physician assistants had considerable autonomy and responsibility, and rather than being assigned only lower level administrative tasks, performed all aspects of patient care. They also received regular teaching from the hospitalists, attended house staff teaching conferences, and developed nonclinical roles in education and quality improvement. The higher standards expected of the physician assistants were quite possibly a factor in the quality of care delivered, and almost certainly contributed to physician assistant satisfaction and retention.

Our findings contrast with those of Myers et al.,9 who found that a nonteaching service staffed by hospitalists and nurse practitioners had a significantly lower median LOS and hospital charges compared to similar patients on resident‐based services. However, unlike ours, their service cared for a select patient population, and only accepted patients with chest pain at low risk for acute coronary syndrome. Van Rhee et al.10 found that physician assistants on a general medicine service used fewer resources for patients with pneumonia, stroke, and congestive heart failure than resident physicians, and did not exceed the resources used by residents in other diagnoses. The authors did not find a difference in LOS, but did find a significantly higher mortality among patients with pneumonia cared for by physician assistants.

Several limitations should be noted. First, the study was a retrospective analysis of administrative data rather than a randomized trial, and although we employed a standard approach to adjust for a wide range of patient characteristics including severity of illness, there may have been undetected differences in the patient populations studied that may have confounded our results. Second, resident moonlighters admitted patients to the PACE service and, at other times, to the house staff services, and this may have diluted any differences between the groups. However, when we limited our analysis to the subgroup of patients admitted during the day, similar results were obtained, with the exception that the PACE service had a lower rate of 14‐day readmissions, an unexpected finding deserving of further study. Third, the study was conducted in a single academic institution and our findings may not be generalizable to others with different needs and resources; indeed, the costs associated with implementing such a service may be prohibitive for some institutions. Fourth, because of simultaneous changes that were taking place in our residency program, we are unable to accurately assess the impact of the PACE service on resident duty hours. However, resident duty hours did not increase over this time period on the general medicine service, and implementation of the service allowed for redistribution of house staff to other services and positions. Fifth, patient satisfaction data were obtained from responses to the mailed Press‐Ganey survey, to which there is a relatively low response rate. Also, we did not survey providers regarding their satisfaction with the service during the study period. Sixth, the study had limited power to detect clinically important differences in mortality and ICU transfers. Finally, this study is unable to compare this particular model of incorporating midlevel providers into general medical services with other models, only with traditional house staff services.

Future research should focus on determining the most effective and efficient ways to incorporate midlevel providers on academic general medicine services. One important question from the standpoint of house staff training is whether such services should be separate but equal, or should house staff gain experience during residency working with midlevel providers, since they are likely to encounter them in the future whether they stay in academics or not. Different models of care will likely have large implications for the quality and efficiency of patient care, house staff education and satisfaction, and physician assistant job satisfaction and turnover.

In summary, our study demonstrates that a geographically regionalized, multidisciplinary service staffed by hospitalists and physician assistants can be a safe alternative to house staff‐based services for the care of general medicine inpatients in an academic medical center.

Midlevel providers (physician assistants and nurse practitioners) have long been employed by academic medical centers, predominantly on surgical services, or on medical subspecialty services, where they have typically had a limited scope of practice, focused in a narrowly defined area or set of procedures.17 In contrast, there are relatively few reports of experiences deploying midlevel providers to replace house staff on inpatient general medicine services in academic centers,810 and few studies of the effect of midlevel providers on quality and efficiency of care in the academic setting. Despite this, reductions in house officer duty hours as mandated by the Accreditation Council on Graduate Medical Education (ACGME)11 have resulted in academic centers increasingly using midlevel providers to decrease house staff workload on inpatient services.12, 13 In general, midlevel practitioners on general medicine services have been deployed to: (1) care for a population of patients separate from and in parallel with house staff; this population may be narrowly defined (eg, patients with chest pain) or not; (2) assist with the management of patients cared for by house staff by performing certain tasks (eg, scheduling appointments, discharging patients). Even as midlevel providers become more prevalent on academic general medicine services, the best model of care incorporating them into clinical care remains unclear, and few studies have rigorously examined the care provided on services that use them.

We developed an inpatient general medicine service within a large academic medical center staffed by physician assistants and hospitalists to help our residency program meet ACGME duty hour requirements. We hypothesized that by creating a service that is geographically localized and supervised by full‐time hospitalists, by instituting multidisciplinary rounds, and by investing in the professional development of highly‐skilled physician assistants, we could provide care for medically complex, acutely ill general medicine inpatients with similar quality and efficiency as compared to house staff teams. We report our experience during the first year of implementing the service, and compare quality and efficiency of care on this service with that of our traditional house staff services. We also evaluate the effects of this service on patient satisfaction and self‐reported house staff workload.

PATIENTS AND METHODS

RESULTS