Hiren M. Shah, MD, MBA

Setting and Study Design

The study was conducted at Northwestern Memorial Hospital (NMH), an 897‐bed tertiary care teaching hospital in Chicago, IL, and was approved by the Institutional Review Board of Northwestern University. The study was a controlled trial of an intervention, SIDR, on collaboration and teamwork on patient care units. One of 2 similar hospitalist service units was randomly selected for the intervention, while the other served as a control unit. SIDR was implemented in August 2008 and data were collected over a 24 week study period.

Each hospitalist service unit consisted of 30 beds and was equipped with continuous cardiac telemetry monitoring. Units were also identical in structure and staffing of nonphysician personnel. The intervention unit included a heart failure‐hospitalist comanagement service. Patients followed at the Center for Heart Failure in the Bluhm Cardiovascular Institute of Northwestern were preferentially admitted to this service. All other patients were admitted to units based on bed availability in a quasi‐randomized fashion. Hospitalists worked 7 consecutive days while on service and cared for patients primarily on the units involved in this study. Therefore, hospitalists cared for patients on both the intervention and control units during their weeks on service. Hospitalists cared for patients independently without the assistance of resident physicians or mid‐level providers (ie, physician assistants or nurse practitioners).

Intervention

SIDR combined a structured format for communication with a forum for regular interdisciplinary meetings. A working group, consisting of nurses, hospitalists, and the unit pharmacist, social worker, and case manager, met weekly for 12 weeks prior to implementation. The working group determined the optimal timing, frequency, and location for SIDR. Additionally, the working group finalized the content of a structured communication tool (Supporting Information) to be used during SIDR. The structured communication tool was modeled after prior research demonstrating the benefit of daily goals of care forms16, 17 and ensured that important elements of the daily plan of care were discussed. Based on the working group's recommendation, SIDR took place each weekday at 11:00 _AM in the unit conference room and lasted approximately 30 minutes. The nurse manager and a unit medical director co‐led rounds each day. SIDR was attended by all nurses and hospitalists caring for patients on the unit, as well as the pharmacist, social worker, and case manager assigned to the unit.

Provider Survey

Nurses working on the intervention and control units during the study period were administered a survey 16 weeks to 20 weeks after implementation of SIDR to assess ratings of collaboration and teamwork. The first portion of the survey was based on previously published surveys assessing teamwork attitudes among providers.6, 7 We asked nurses to rate the quality of communication and collaboration they had experienced with hospitalists using a 5‐point ordinal scale (1 = very low, 2 = low, 3 = adequate, 4 = high, 5 = very high). The second portion of the survey assessed teamwork and safety climate using the teamwork and safety domains of the Safety Attitudes Questionnaire (SAQ) developed by Sexton et al.18 The SAQ is based on previous research in aviation and medicine and has been validated in clinical settings.19, 20 Because hospitalists worked with nurses on both units, and in light of our prior research demonstrating that hospitalists rate the quality of collaboration with nurses highly,8 we did not assess hospitalists' ratings of collaboration. A final portion of the survey assessed nurses' perceptions of whether SIDR improved efficiency of communication, collaboration among team members, and patient care using a 5‐point Likert scale (1 = strongly disagree; 2 = disagree; 3 = neutral; 4 = agree; 5 = strongly agree). Hospitalists also received this portion of the survey at the completion of each clinical rotation. All surveys were administered in a web‐based format using an internet link (www.formsite.com from Vroman Systems, Inc.) delivered through email. Respondents entered the survey website using a unique login, which allowed for identification of nonresponders. However, survey responses were de‐identified. We sent nonresponders up to 3 reminder emails. The low number of social workers, case managers, and pharmacists on each unit precluded our ability to meaningfully assess their perceptions of collaboration and ratings of teamwork and safety climate.

SIDR Characteristics and Attendance

The unit medical director recorded the duration of SIDR, the number of patients on the unit, and the number of patients discussed each day. Attendance for each discipline was also recorded each day during the study period.

Data Analysis

Provider demographic data were obtained from completed surveys and group comparisons were done using chi‐square and t tests. The percentage of nurses on each unit rating of the quality of communication and collaboration with hospitalist physicians as high or very high was compared using chi‐square. Teamwork and safety climate scores were compared using the Mann Whitney U test.

Patient data were obtained from administrative databases for both the control and intervention unit during the study period as well as for the intervention unit in the 24 weeks preceding the study period. Demographic data were compared using chi‐square and t tests. Primary discharge diagnosis ICD‐9 codes were grouped into diagnosis clusters using the Healthcare Cost and Utilization Project system of the Agency for Healthcare Research and Quality.21 Diagnosis clusters were then analyzed using the chi‐square test. Because of case mix differences between patients on the intervention and control units, we analyzed LOS and cost using a concurrent control as well as an historic control. Unadjusted LOS and costs were compared using the Mann Whitney U test. We then conducted multivariable linear regression analyses to assess the impact of SIDR on LOS and cost. To satisfy normality requirements and distribution of residuals, we explored 2 methods of transforming skewed data on LOS and cost: logarithmic conversion and truncation at the mean LOS + 3 standard deviations (SDs). Since both techniques yielded similar results, we chose to present results by using truncation. Covariates for multivariable analyses included age, gender, race, payor, admission source, case‐mix, discharge disposition, presence of ICU stay during hospitalization, and Medicare Severity‐Diagnosis Related Group (MS‐DRG) weight. We used standard errors robust to the clustering of patients within each physician. All analyses were conducted using Stata version 10.0 (College Station, TX).

Characteristics of Providers, Patients, and SIDR

Forty‐nine of 58 (84%) nurses completed the survey. Eighty‐eight of 96 (92%) surveys were completed by hospitalists at the end of their week on service. Hospitalist surveys represented 33 different hospitalists because individuals may have worked on study units more than once during the study period. Nurses were a mean 35.0 10.4 years of age and had been working at the hospital for a mean 5.0 6.3 years. Hospitalists were a mean 32.8 2.8 years of age and had been working at the hospital for a mean 2.6 1.9 years.

Patient characteristics are shown in Table 1. Intervention unit patients were admitted from the Emergency Department slightly more often in the postSIDR period. Patient case mix differed between the control and intervention unit, but was similar when comparing the intervention unit preSIDR and postSIDR. Intervention unit MS‐DRG weight was lower in the postSIDR period.

SIDR occurred each weekday (with the exception of holidays) on the intervention unit and lasted a mean 27.7 4.6 minutes. The unit had a mean 27 patients per day and 86% of patients on the unit were discussed each day. Attendance exceeded 85% for each discipline (hospitalists, nurses, and the unit pharmacist, social worker, and case manager).

Ratings of Teamwork and Perceptions of SIDR

As shown in Figure 1, a larger percentage of nurses rated the quality of communication and collaboration with hospitalists as high or very high on the intervention unit compared to the control unit (80% vs. 54%; P = 0.05).

Figure 1

Nurses' ratings of the teamwork and safety climate are summarized in Table 2. The median teamwork climate score was 85.7 (interquartile range [IQR], 75.092.9) for the intervention unit as compared to 61.6 (IQR, 48.283.9) for the control unit (P = 0.008). The median safety climate score was 75.0 (IQR, 70.581.3) for the intervention unit as compared to 61.1 (IQR, 30.281.3) for the control unit (P = 0.03).

Sixty‐five of 88 (74%) hospitalists and 18 of 24 (75%) nurses agreed that SIDR improved the efficiency of their work day. Eighty of 88 (91%) hospitalists and 18 of 24 (75%) nurses agreed that SIDR improved team collaboration. Seventy‐six of 88 (86%) hospitalists and 18 of 24 (75%) nurses agreed that SIDR improved patient care. Sixty‐seven of 88 (76%) hospitalists and 22 of 25 (88%) nurses indicated that they wanted SIDR to continue indefinitely.

SIDR Impact on LOS and Cost

The unadjusted mean LOS was significantly higher for the intervention unit postSIDR as compared to the control unit (4.0 3.4 vs. 3.7 3.3 days; P = 0.03). However, the unadjusted mean LOS was not significantly different for the intervention unit postSIDR as compared to the intervention unit preSIDR (4.0 3.4 vs. 4.26 3.5 days; P = 0.10). The unadjusted cost was lower for the intervention unit postSIDR as compared to the control unit ($7,513.23 7,085.10 vs. $8,588.66 7,381.03; P < 0.001). The unadjusted mean cost was not significantly different for the invention unit postSIDR as compared to the intervention unit preSIDR ($7,513.23 7,085.10 vs. $7,937.00 7,512.23; P = 0.19).

Multivariable analyses of LOS and cost are summarized in Table 3. The adjusted LOS was not significantly different when comparing the intervention unit postSIDR to either the control unit or the intervention unit preSIDR. The adjusted cost for the intervention unit postSIDR was $739.55 less than the control unit (P = 0.02). The adjusted cost was not significantly different when comparing the intervention unit postSIDR to the intervention unit preSIDR.

Setting and Study Design

The study was conducted at Northwestern Memorial Hospital (NMH), an 897‐bed tertiary care teaching hospital in Chicago, IL, and was approved by the Institutional Review Board of Northwestern University. The study was a controlled trial of an intervention, SIDR, on collaboration and teamwork on patient care units. One of 2 similar hospitalist service units was randomly selected for the intervention, while the other served as a control unit. SIDR was implemented in August 2008 and data were collected over a 24 week study period.

Each hospitalist service unit consisted of 30 beds and was equipped with continuous cardiac telemetry monitoring. Units were also identical in structure and staffing of nonphysician personnel. The intervention unit included a heart failure‐hospitalist comanagement service. Patients followed at the Center for Heart Failure in the Bluhm Cardiovascular Institute of Northwestern were preferentially admitted to this service. All other patients were admitted to units based on bed availability in a quasi‐randomized fashion. Hospitalists worked 7 consecutive days while on service and cared for patients primarily on the units involved in this study. Therefore, hospitalists cared for patients on both the intervention and control units during their weeks on service. Hospitalists cared for patients independently without the assistance of resident physicians or mid‐level providers (ie, physician assistants or nurse practitioners).

Intervention

SIDR combined a structured format for communication with a forum for regular interdisciplinary meetings. A working group, consisting of nurses, hospitalists, and the unit pharmacist, social worker, and case manager, met weekly for 12 weeks prior to implementation. The working group determined the optimal timing, frequency, and location for SIDR. Additionally, the working group finalized the content of a structured communication tool (Supporting Information) to be used during SIDR. The structured communication tool was modeled after prior research demonstrating the benefit of daily goals of care forms16, 17 and ensured that important elements of the daily plan of care were discussed. Based on the working group's recommendation, SIDR took place each weekday at 11:00 _AM in the unit conference room and lasted approximately 30 minutes. The nurse manager and a unit medical director co‐led rounds each day. SIDR was attended by all nurses and hospitalists caring for patients on the unit, as well as the pharmacist, social worker, and case manager assigned to the unit.

Provider Survey

Nurses working on the intervention and control units during the study period were administered a survey 16 weeks to 20 weeks after implementation of SIDR to assess ratings of collaboration and teamwork. The first portion of the survey was based on previously published surveys assessing teamwork attitudes among providers.6, 7 We asked nurses to rate the quality of communication and collaboration they had experienced with hospitalists using a 5‐point ordinal scale (1 = very low, 2 = low, 3 = adequate, 4 = high, 5 = very high). The second portion of the survey assessed teamwork and safety climate using the teamwork and safety domains of the Safety Attitudes Questionnaire (SAQ) developed by Sexton et al.18 The SAQ is based on previous research in aviation and medicine and has been validated in clinical settings.19, 20 Because hospitalists worked with nurses on both units, and in light of our prior research demonstrating that hospitalists rate the quality of collaboration with nurses highly,8 we did not assess hospitalists' ratings of collaboration. A final portion of the survey assessed nurses' perceptions of whether SIDR improved efficiency of communication, collaboration among team members, and patient care using a 5‐point Likert scale (1 = strongly disagree; 2 = disagree; 3 = neutral; 4 = agree; 5 = strongly agree). Hospitalists also received this portion of the survey at the completion of each clinical rotation. All surveys were administered in a web‐based format using an internet link (www.formsite.com from Vroman Systems, Inc.) delivered through email. Respondents entered the survey website using a unique login, which allowed for identification of nonresponders. However, survey responses were de‐identified. We sent nonresponders up to 3 reminder emails. The low number of social workers, case managers, and pharmacists on each unit precluded our ability to meaningfully assess their perceptions of collaboration and ratings of teamwork and safety climate.

SIDR Characteristics and Attendance

The unit medical director recorded the duration of SIDR, the number of patients on the unit, and the number of patients discussed each day. Attendance for each discipline was also recorded each day during the study period.

Data Analysis

Provider demographic data were obtained from completed surveys and group comparisons were done using chi‐square and t tests. The percentage of nurses on each unit rating of the quality of communication and collaboration with hospitalist physicians as high or very high was compared using chi‐square. Teamwork and safety climate scores were compared using the Mann Whitney U test.

Patient data were obtained from administrative databases for both the control and intervention unit during the study period as well as for the intervention unit in the 24 weeks preceding the study period. Demographic data were compared using chi‐square and t tests. Primary discharge diagnosis ICD‐9 codes were grouped into diagnosis clusters using the Healthcare Cost and Utilization Project system of the Agency for Healthcare Research and Quality.21 Diagnosis clusters were then analyzed using the chi‐square test. Because of case mix differences between patients on the intervention and control units, we analyzed LOS and cost using a concurrent control as well as an historic control. Unadjusted LOS and costs were compared using the Mann Whitney U test. We then conducted multivariable linear regression analyses to assess the impact of SIDR on LOS and cost. To satisfy normality requirements and distribution of residuals, we explored 2 methods of transforming skewed data on LOS and cost: logarithmic conversion and truncation at the mean LOS + 3 standard deviations (SDs). Since both techniques yielded similar results, we chose to present results by using truncation. Covariates for multivariable analyses included age, gender, race, payor, admission source, case‐mix, discharge disposition, presence of ICU stay during hospitalization, and Medicare Severity‐Diagnosis Related Group (MS‐DRG) weight. We used standard errors robust to the clustering of patients within each physician. All analyses were conducted using Stata version 10.0 (College Station, TX).

Characteristics of Providers, Patients, and SIDR

Forty‐nine of 58 (84%) nurses completed the survey. Eighty‐eight of 96 (92%) surveys were completed by hospitalists at the end of their week on service. Hospitalist surveys represented 33 different hospitalists because individuals may have worked on study units more than once during the study period. Nurses were a mean 35.0 10.4 years of age and had been working at the hospital for a mean 5.0 6.3 years. Hospitalists were a mean 32.8 2.8 years of age and had been working at the hospital for a mean 2.6 1.9 years.

Patient characteristics are shown in Table 1. Intervention unit patients were admitted from the Emergency Department slightly more often in the postSIDR period. Patient case mix differed between the control and intervention unit, but was similar when comparing the intervention unit preSIDR and postSIDR. Intervention unit MS‐DRG weight was lower in the postSIDR period.

SIDR occurred each weekday (with the exception of holidays) on the intervention unit and lasted a mean 27.7 4.6 minutes. The unit had a mean 27 patients per day and 86% of patients on the unit were discussed each day. Attendance exceeded 85% for each discipline (hospitalists, nurses, and the unit pharmacist, social worker, and case manager).

Ratings of Teamwork and Perceptions of SIDR

As shown in Figure 1, a larger percentage of nurses rated the quality of communication and collaboration with hospitalists as high or very high on the intervention unit compared to the control unit (80% vs. 54%; P = 0.05).

Figure 1

Nurses' ratings of the teamwork and safety climate are summarized in Table 2. The median teamwork climate score was 85.7 (interquartile range [IQR], 75.092.9) for the intervention unit as compared to 61.6 (IQR, 48.283.9) for the control unit (P = 0.008). The median safety climate score was 75.0 (IQR, 70.581.3) for the intervention unit as compared to 61.1 (IQR, 30.281.3) for the control unit (P = 0.03).

Sixty‐five of 88 (74%) hospitalists and 18 of 24 (75%) nurses agreed that SIDR improved the efficiency of their work day. Eighty of 88 (91%) hospitalists and 18 of 24 (75%) nurses agreed that SIDR improved team collaboration. Seventy‐six of 88 (86%) hospitalists and 18 of 24 (75%) nurses agreed that SIDR improved patient care. Sixty‐seven of 88 (76%) hospitalists and 22 of 25 (88%) nurses indicated that they wanted SIDR to continue indefinitely.

SIDR Impact on LOS and Cost

The unadjusted mean LOS was significantly higher for the intervention unit postSIDR as compared to the control unit (4.0 3.4 vs. 3.7 3.3 days; P = 0.03). However, the unadjusted mean LOS was not significantly different for the intervention unit postSIDR as compared to the intervention unit preSIDR (4.0 3.4 vs. 4.26 3.5 days; P = 0.10). The unadjusted cost was lower for the intervention unit postSIDR as compared to the control unit ($7,513.23 7,085.10 vs. $8,588.66 7,381.03; P < 0.001). The unadjusted mean cost was not significantly different for the invention unit postSIDR as compared to the intervention unit preSIDR ($7,513.23 7,085.10 vs. $7,937.00 7,512.23; P = 0.19).

Multivariable analyses of LOS and cost are summarized in Table 3. The adjusted LOS was not significantly different when comparing the intervention unit postSIDR to either the control unit or the intervention unit preSIDR. The adjusted cost for the intervention unit postSIDR was $739.55 less than the control unit (P = 0.02). The adjusted cost was not significantly different when comparing the intervention unit postSIDR to the intervention unit preSIDR.

Setting and Study Design

The study was conducted at Northwestern Memorial Hospital (NMH), an 897‐bed tertiary care teaching hospital in Chicago, IL, and was approved by the Institutional Review Board of Northwestern University. The study was a controlled trial of an intervention, SIDR, on collaboration and teamwork on patient care units. One of 2 similar hospitalist service units was randomly selected for the intervention, while the other served as a control unit. SIDR was implemented in August 2008 and data were collected over a 24 week study period.

Each hospitalist service unit consisted of 30 beds and was equipped with continuous cardiac telemetry monitoring. Units were also identical in structure and staffing of nonphysician personnel. The intervention unit included a heart failure‐hospitalist comanagement service. Patients followed at the Center for Heart Failure in the Bluhm Cardiovascular Institute of Northwestern were preferentially admitted to this service. All other patients were admitted to units based on bed availability in a quasi‐randomized fashion. Hospitalists worked 7 consecutive days while on service and cared for patients primarily on the units involved in this study. Therefore, hospitalists cared for patients on both the intervention and control units during their weeks on service. Hospitalists cared for patients independently without the assistance of resident physicians or mid‐level providers (ie, physician assistants or nurse practitioners).

Intervention

SIDR combined a structured format for communication with a forum for regular interdisciplinary meetings. A working group, consisting of nurses, hospitalists, and the unit pharmacist, social worker, and case manager, met weekly for 12 weeks prior to implementation. The working group determined the optimal timing, frequency, and location for SIDR. Additionally, the working group finalized the content of a structured communication tool (Supporting Information) to be used during SIDR. The structured communication tool was modeled after prior research demonstrating the benefit of daily goals of care forms16, 17 and ensured that important elements of the daily plan of care were discussed. Based on the working group's recommendation, SIDR took place each weekday at 11:00 _AM in the unit conference room and lasted approximately 30 minutes. The nurse manager and a unit medical director co‐led rounds each day. SIDR was attended by all nurses and hospitalists caring for patients on the unit, as well as the pharmacist, social worker, and case manager assigned to the unit.

Provider Survey

Nurses working on the intervention and control units during the study period were administered a survey 16 weeks to 20 weeks after implementation of SIDR to assess ratings of collaboration and teamwork. The first portion of the survey was based on previously published surveys assessing teamwork attitudes among providers.6, 7 We asked nurses to rate the quality of communication and collaboration they had experienced with hospitalists using a 5‐point ordinal scale (1 = very low, 2 = low, 3 = adequate, 4 = high, 5 = very high). The second portion of the survey assessed teamwork and safety climate using the teamwork and safety domains of the Safety Attitudes Questionnaire (SAQ) developed by Sexton et al.18 The SAQ is based on previous research in aviation and medicine and has been validated in clinical settings.19, 20 Because hospitalists worked with nurses on both units, and in light of our prior research demonstrating that hospitalists rate the quality of collaboration with nurses highly,8 we did not assess hospitalists' ratings of collaboration. A final portion of the survey assessed nurses' perceptions of whether SIDR improved efficiency of communication, collaboration among team members, and patient care using a 5‐point Likert scale (1 = strongly disagree; 2 = disagree; 3 = neutral; 4 = agree; 5 = strongly agree). Hospitalists also received this portion of the survey at the completion of each clinical rotation. All surveys were administered in a web‐based format using an internet link (www.formsite.com from Vroman Systems, Inc.) delivered through email. Respondents entered the survey website using a unique login, which allowed for identification of nonresponders. However, survey responses were de‐identified. We sent nonresponders up to 3 reminder emails. The low number of social workers, case managers, and pharmacists on each unit precluded our ability to meaningfully assess their perceptions of collaboration and ratings of teamwork and safety climate.

SIDR Characteristics and Attendance

The unit medical director recorded the duration of SIDR, the number of patients on the unit, and the number of patients discussed each day. Attendance for each discipline was also recorded each day during the study period.

Data Analysis

Provider demographic data were obtained from completed surveys and group comparisons were done using chi‐square and t tests. The percentage of nurses on each unit rating of the quality of communication and collaboration with hospitalist physicians as high or very high was compared using chi‐square. Teamwork and safety climate scores were compared using the Mann Whitney U test.

Patient data were obtained from administrative databases for both the control and intervention unit during the study period as well as for the intervention unit in the 24 weeks preceding the study period. Demographic data were compared using chi‐square and t tests. Primary discharge diagnosis ICD‐9 codes were grouped into diagnosis clusters using the Healthcare Cost and Utilization Project system of the Agency for Healthcare Research and Quality.21 Diagnosis clusters were then analyzed using the chi‐square test. Because of case mix differences between patients on the intervention and control units, we analyzed LOS and cost using a concurrent control as well as an historic control. Unadjusted LOS and costs were compared using the Mann Whitney U test. We then conducted multivariable linear regression analyses to assess the impact of SIDR on LOS and cost. To satisfy normality requirements and distribution of residuals, we explored 2 methods of transforming skewed data on LOS and cost: logarithmic conversion and truncation at the mean LOS + 3 standard deviations (SDs). Since both techniques yielded similar results, we chose to present results by using truncation. Covariates for multivariable analyses included age, gender, race, payor, admission source, case‐mix, discharge disposition, presence of ICU stay during hospitalization, and Medicare Severity‐Diagnosis Related Group (MS‐DRG) weight. We used standard errors robust to the clustering of patients within each physician. All analyses were conducted using Stata version 10.0 (College Station, TX).

Characteristics of Providers, Patients, and SIDR

Forty‐nine of 58 (84%) nurses completed the survey. Eighty‐eight of 96 (92%) surveys were completed by hospitalists at the end of their week on service. Hospitalist surveys represented 33 different hospitalists because individuals may have worked on study units more than once during the study period. Nurses were a mean 35.0 10.4 years of age and had been working at the hospital for a mean 5.0 6.3 years. Hospitalists were a mean 32.8 2.8 years of age and had been working at the hospital for a mean 2.6 1.9 years.

Patient characteristics are shown in Table 1. Intervention unit patients were admitted from the Emergency Department slightly more often in the postSIDR period. Patient case mix differed between the control and intervention unit, but was similar when comparing the intervention unit preSIDR and postSIDR. Intervention unit MS‐DRG weight was lower in the postSIDR period.

SIDR occurred each weekday (with the exception of holidays) on the intervention unit and lasted a mean 27.7 4.6 minutes. The unit had a mean 27 patients per day and 86% of patients on the unit were discussed each day. Attendance exceeded 85% for each discipline (hospitalists, nurses, and the unit pharmacist, social worker, and case manager).

Ratings of Teamwork and Perceptions of SIDR

As shown in Figure 1, a larger percentage of nurses rated the quality of communication and collaboration with hospitalists as high or very high on the intervention unit compared to the control unit (80% vs. 54%; P = 0.05).

Figure 1

Nurses' ratings of the teamwork and safety climate are summarized in Table 2. The median teamwork climate score was 85.7 (interquartile range [IQR], 75.092.9) for the intervention unit as compared to 61.6 (IQR, 48.283.9) for the control unit (P = 0.008). The median safety climate score was 75.0 (IQR, 70.581.3) for the intervention unit as compared to 61.1 (IQR, 30.281.3) for the control unit (P = 0.03).

Sixty‐five of 88 (74%) hospitalists and 18 of 24 (75%) nurses agreed that SIDR improved the efficiency of their work day. Eighty of 88 (91%) hospitalists and 18 of 24 (75%) nurses agreed that SIDR improved team collaboration. Seventy‐six of 88 (86%) hospitalists and 18 of 24 (75%) nurses agreed that SIDR improved patient care. Sixty‐seven of 88 (76%) hospitalists and 22 of 25 (88%) nurses indicated that they wanted SIDR to continue indefinitely.

SIDR Impact on LOS and Cost

The unadjusted mean LOS was significantly higher for the intervention unit postSIDR as compared to the control unit (4.0 3.4 vs. 3.7 3.3 days; P = 0.03). However, the unadjusted mean LOS was not significantly different for the intervention unit postSIDR as compared to the intervention unit preSIDR (4.0 3.4 vs. 4.26 3.5 days; P = 0.10). The unadjusted cost was lower for the intervention unit postSIDR as compared to the control unit ($7,513.23 7,085.10 vs. $8,588.66 7,381.03; P < 0.001). The unadjusted mean cost was not significantly different for the invention unit postSIDR as compared to the intervention unit preSIDR ($7,513.23 7,085.10 vs. $7,937.00 7,512.23; P = 0.19).

Multivariable analyses of LOS and cost are summarized in Table 3. The adjusted LOS was not significantly different when comparing the intervention unit postSIDR to either the control unit or the intervention unit preSIDR. The adjusted cost for the intervention unit postSIDR was $739.55 less than the control unit (P = 0.02). The adjusted cost was not significantly different when comparing the intervention unit postSIDR to the intervention unit preSIDR.

The Equities of the System

Prior to developing nonhousestaff services, our academic hospitalist programs scheduled teaching service time in month or half‐month blocks, balancing holidays and weekends. Equity in scheduling became a function of required clinical time, sources of non‐clinical funding (eg, grants, educational or administrative roles), and expectations for scholarship, attributes typical of most subspecialty academic divisions. Given the differing clinical missions that have stimulated academic hospital medicine programs to form, concerns of scheduling equity have grown, posing challenges not experienced in other divisions.

Institutions that choose to divide housestaff and nonhousestaff duties among distinct groups of hospitalists create the potential for a 2‐tiered system, one in which those with housestaff roles are more valued and respected by the institution. Hospitalists working on nonhousestaff services admit patients, write orders, and field direct patient calls, a role rarely undertaken by subspecialty attendings or hospitalists on housestaff services. Our collective experiences provide evidence of the danger of this second‐class‐citizen status, one that requires attention to ensure job satisfaction, retention, and necessary career development.

Institutions have accentuated the second‐class‐citizen concern by staffing nonhousestaff roles with 1‐year hospitalistsPGY‐4s. Most of these hires in our institutions are individuals just out of residency and intent on pursuing a fellowship. We speculate that they enjoy the comforts of the AMC where they often trained and accept purely nonhousestaff positions because of what they view as an appealing work schedule and salary. Although this approach addresses the growing need for hospitalists on nonhousestaff services in the short term, these positions must remain attractive enough (both financially and professionally) to encourage residency graduates to pursue an academic hospitalist career instead of a 1‐year position as a transition to fellowship. Otherwise, the approach conveys a message that relatively inexperienced physicians are good enough to be hospitalists.

Developing a cadre of clinically focused hospitalists who provide outstanding patient care and also garner respect as successful academicians is a difficult task. Although 1 group in our sample (Northwestern) shares nonhousestaff responsibilities equally, others may find this impractical, particularly where faculty members were hired before nonhousestaff services were established. Redefining such clinical positions several years into a career may be challenging, as it forces faculty members into roles they didn't sign up for or grandfathers them out of such roles, adding to the risk of a 2‐tiered system. Alternatively, groups may focus on building academic activities into nonhousestaff services, including medical student teaching, quality improvement, or clinical research activities. In this article, we deliberately classified these services as nonhousestaff rather than non‐teaching because the latter fails to acknowledge that these hospitalists often serve as teachers (eg, housestaff conferences, supervision of midlevel providers, and/or rotating medical students)an important if not symbolic distinction. It is imperative that planning for nonhousestaff services balance the larger academic mission of hospital medicine groups with creating equitable, valued, and sustainable job descriptions.

Defining the Patient Mix

Developing an optimal patient mix on nonhousestaff services also carries important implications. For services that work in parallel with the housestaff service and simply take extra patients above the resident cap, this concern may be less significant. However, other nonhousestaff services have been structured to care for lower‐acuity patients (eg, cellulitis, asthma, pneumonia) or select patient populations (eg, sickle cell or inflammatory bowel disease). This distribution system potentially changes the educational experience on the housestaff servicedecreasing the bread‐and‐butter admissionsbut also may affect the job satisfaction of hospitalists and midlevel providers on nonhousestaff services. Building triage criteria, working with emergency department leadership, and avoiding patients being turfed between different services is critical. We strongly recommend a regular process to review admissions to each service and determine when the triage process requires further calibration.

Recruitment and Retention

Traditionally, graduates of residency or fellowship training programs chose academic positions because of an interest in teaching, a desire for scholarship, or a commitment to research. Those interested in primarily clinical roles typically pursued positions in nonacademic settings. The development of nonhousestaff services challenges this paradigm because the objective for academic hospitalist leadership now becomes recruiting pure clinicians as well as academicians. These might be the same individual, a hospitalist who provides both housestaff and nonhousestaff services, or 2 different individuals if the nonhousestaff service is covered by dedicated hospitalists. In addition, with the current promotion structure in academia, a purely clinical position may be less attractive, as it provides fewer opportunities for advancement.

Therefore, recruitment and retention of academic hospitalists will require job descriptions that provide dedicated teaching opportunities, time for participation in quality and safety improvement projects, or pursuit of a scholarly interest in non‐clinical timethe diastole of an academic hospitalist.41 Hospital medicine leadership will also need to better distinguish off‐time from non‐clinical time, as many young hospitalists struggle to balance professional and personal commitmentsa recipe for burnout.42 Regardless of how clinical responsibilities differ between 2 hospitalists, providing them with similar academic resources is what will distinguish their positions from that in the community. Furthermore, many groups have chosen to pay faculty a premium for their nonhousestaff roles or to use specific recruitment incentives such as educational loan forgiveness programs.

With the expected growth of nonhousestaff services and surgical comanagement, hospital medicine programs will also need to determine if new hires will focus on a specific service (eg, orthopedic hospitalist) or whether job descriptions will include a mix of activities (eg, 3 months' teaching service, 3 months' nonhousestaff medical service, and 3 months' surgical comanagement service). A second and equally important question is where does the hospitalist live? If cardiology wants hospitalists to care for their patients, should they be hired and mentored by cardiologists or by hospitalists in a division of general or hospital medicine? In many cases, a graduating resident with plans to pursue a fellowship (eg, cardiology or hematology/oncology) may be a perfect candidate for a 1‐year position on his or her future specialty service. However, in the long term, maintaining all the academic hospitalists under the same umbrella will provide greater mentorship, professional development, opportunities for collaboration, clinical diversity, and sense of belonging to a group, rather than being a token hospitalist for another division.

Compensation and Financial Relationships with AMCs

Salaries for hospitalists working on nonhousestaff services are typically higher at AMCs, which are competing with community standards given the similar level of clinical hours worked. However, although pay for nonhousestaff activities should reflect the nature of the work, compensation models based on clinical productivity alone may prove inadequate. It appears hospitalists working in academic facilities spend significant time on indirect patient care because of these hospitals' inefficiencies, usually not found in community settings.43 Devising compensation for an academic hospitalist requires careful attention and must balance a number of factors because these hospitalists will not generate their entire salary from clinical services. Financial support must come from either the division or medical center, an annual negotiation at AMCs.

Several methods exist to structure hospitalist compensation. A hospitalist's salary may be fixed, may have a base salary with incentives, or may be derived based on clinical productivity. For example, if a hospital medicine program provides both housestaff and nonhousestaff services and employs a fixed‐salary approach, it may choose a menu‐style method to determine compensation (eg, 6 months on nonhousestaff service at x dollars/month + 3 months on housestaff service at x dollars/month = annual salary). If a hospitalist takes on a funded nonclinical role or secures extramural funding, the salary menu gets adjusted accordingly as the clinical time is bought out. Critics of the fixed‐salary approach argue that paying each hospitalist the same salary regardless of the specific job description yields an inequitable system in which some are rewarded with less clinical time.

Compensation should probably have a guaranteed base salary with incentives, which could be determined by a formula that weighs clinical productivity, quality improvement efforts, scholarly activity, and teaching excellence. This model provides financial incentives to develop both clinically and academically but introduces complexity in determining a fair incentive structure. Finally, compensation can be structured without salary guarantee and putting compensation fully at risk based on clinical productivity, although this is an unlikely strategy for any hospital medicine group. This approach does disproportionately reward high volume providers, potentially at the risk of quality and safety, but also creates significant incentives to improve efficiency.

With respect to AMC relationships, hospital medicine programs must ensure the positive return on investment that drives financial support at their institutions. This fundamental economic dynamic makes AMCs dependent on their hospital medicine groups and vice versa. We caution programs from solely relying on measures such as reduced hospital costs or length of stay as a basis of funding unless there is a reward for maintaining performance once it inevitably plateaus. Moreover, explicitly tying utilization efficiency (ie, length of stay) to salary violates Stark rules44 and carries potential malpractice implications should patient care errors be attributable to premature hospital discharge. Over time hospitalists will need to maintain clinical benchmarks but also provide additional and valued services to their institutions, including quality and safety improvement activities and compliance with residency work hour restrictions.

Defining the Academic Hospitalist

The question is simple and perhaps philosophical: Are hospitalists who work at an AMC academic hospitalists? And what job description truly defines an academic hospitalist? Currently, there are no standards for the clinical activity of an academic hospitalist position (eg, number of weeks, weekends, and hours) or for assessment of nonclinical productivity. Hospital medicine programs face the challenge of defining positions that fulfill the growing clinical mission at AMCs but have little experience or guidance in ensuring they will lead to advancing the academic mission. Specifically, how do hospitalists who provide mostly clinical care, particularly on nonhousestaff services, achieve promotion? Hospital medicine program leadership must create enough opportunity and time for the development of skills in research, education, and quality or systems improvement if academic hospitalists are to succeed.

The Association of Chiefs of General Internal Medicine (ACGIM), the Society of General Internal Medicine (SGIM), and the Society of Hospital Medicine (SHM) are currently collaborating to develop consensus guidelines in this area. Ultimately, through the efforts of these important governing bodies, the specialty of hospital medicine will be able to demonstrate the unique skills and services they provide and move toward advocating for academic promotion criteria that recognize their value and accomplishments.

The Equities of the System

Prior to developing nonhousestaff services, our academic hospitalist programs scheduled teaching service time in month or half‐month blocks, balancing holidays and weekends. Equity in scheduling became a function of required clinical time, sources of non‐clinical funding (eg, grants, educational or administrative roles), and expectations for scholarship, attributes typical of most subspecialty academic divisions. Given the differing clinical missions that have stimulated academic hospital medicine programs to form, concerns of scheduling equity have grown, posing challenges not experienced in other divisions.

Institutions that choose to divide housestaff and nonhousestaff duties among distinct groups of hospitalists create the potential for a 2‐tiered system, one in which those with housestaff roles are more valued and respected by the institution. Hospitalists working on nonhousestaff services admit patients, write orders, and field direct patient calls, a role rarely undertaken by subspecialty attendings or hospitalists on housestaff services. Our collective experiences provide evidence of the danger of this second‐class‐citizen status, one that requires attention to ensure job satisfaction, retention, and necessary career development.

Institutions have accentuated the second‐class‐citizen concern by staffing nonhousestaff roles with 1‐year hospitalistsPGY‐4s. Most of these hires in our institutions are individuals just out of residency and intent on pursuing a fellowship. We speculate that they enjoy the comforts of the AMC where they often trained and accept purely nonhousestaff positions because of what they view as an appealing work schedule and salary. Although this approach addresses the growing need for hospitalists on nonhousestaff services in the short term, these positions must remain attractive enough (both financially and professionally) to encourage residency graduates to pursue an academic hospitalist career instead of a 1‐year position as a transition to fellowship. Otherwise, the approach conveys a message that relatively inexperienced physicians are good enough to be hospitalists.

Developing a cadre of clinically focused hospitalists who provide outstanding patient care and also garner respect as successful academicians is a difficult task. Although 1 group in our sample (Northwestern) shares nonhousestaff responsibilities equally, others may find this impractical, particularly where faculty members were hired before nonhousestaff services were established. Redefining such clinical positions several years into a career may be challenging, as it forces faculty members into roles they didn't sign up for or grandfathers them out of such roles, adding to the risk of a 2‐tiered system. Alternatively, groups may focus on building academic activities into nonhousestaff services, including medical student teaching, quality improvement, or clinical research activities. In this article, we deliberately classified these services as nonhousestaff rather than non‐teaching because the latter fails to acknowledge that these hospitalists often serve as teachers (eg, housestaff conferences, supervision of midlevel providers, and/or rotating medical students)an important if not symbolic distinction. It is imperative that planning for nonhousestaff services balance the larger academic mission of hospital medicine groups with creating equitable, valued, and sustainable job descriptions.

Defining the Patient Mix

Developing an optimal patient mix on nonhousestaff services also carries important implications. For services that work in parallel with the housestaff service and simply take extra patients above the resident cap, this concern may be less significant. However, other nonhousestaff services have been structured to care for lower‐acuity patients (eg, cellulitis, asthma, pneumonia) or select patient populations (eg, sickle cell or inflammatory bowel disease). This distribution system potentially changes the educational experience on the housestaff servicedecreasing the bread‐and‐butter admissionsbut also may affect the job satisfaction of hospitalists and midlevel providers on nonhousestaff services. Building triage criteria, working with emergency department leadership, and avoiding patients being turfed between different services is critical. We strongly recommend a regular process to review admissions to each service and determine when the triage process requires further calibration.

Recruitment and Retention

Traditionally, graduates of residency or fellowship training programs chose academic positions because of an interest in teaching, a desire for scholarship, or a commitment to research. Those interested in primarily clinical roles typically pursued positions in nonacademic settings. The development of nonhousestaff services challenges this paradigm because the objective for academic hospitalist leadership now becomes recruiting pure clinicians as well as academicians. These might be the same individual, a hospitalist who provides both housestaff and nonhousestaff services, or 2 different individuals if the nonhousestaff service is covered by dedicated hospitalists. In addition, with the current promotion structure in academia, a purely clinical position may be less attractive, as it provides fewer opportunities for advancement.

Therefore, recruitment and retention of academic hospitalists will require job descriptions that provide dedicated teaching opportunities, time for participation in quality and safety improvement projects, or pursuit of a scholarly interest in non‐clinical timethe diastole of an academic hospitalist.41 Hospital medicine leadership will also need to better distinguish off‐time from non‐clinical time, as many young hospitalists struggle to balance professional and personal commitmentsa recipe for burnout.42 Regardless of how clinical responsibilities differ between 2 hospitalists, providing them with similar academic resources is what will distinguish their positions from that in the community. Furthermore, many groups have chosen to pay faculty a premium for their nonhousestaff roles or to use specific recruitment incentives such as educational loan forgiveness programs.

With the expected growth of nonhousestaff services and surgical comanagement, hospital medicine programs will also need to determine if new hires will focus on a specific service (eg, orthopedic hospitalist) or whether job descriptions will include a mix of activities (eg, 3 months' teaching service, 3 months' nonhousestaff medical service, and 3 months' surgical comanagement service). A second and equally important question is where does the hospitalist live? If cardiology wants hospitalists to care for their patients, should they be hired and mentored by cardiologists or by hospitalists in a division of general or hospital medicine? In many cases, a graduating resident with plans to pursue a fellowship (eg, cardiology or hematology/oncology) may be a perfect candidate for a 1‐year position on his or her future specialty service. However, in the long term, maintaining all the academic hospitalists under the same umbrella will provide greater mentorship, professional development, opportunities for collaboration, clinical diversity, and sense of belonging to a group, rather than being a token hospitalist for another division.

Compensation and Financial Relationships with AMCs

Salaries for hospitalists working on nonhousestaff services are typically higher at AMCs, which are competing with community standards given the similar level of clinical hours worked. However, although pay for nonhousestaff activities should reflect the nature of the work, compensation models based on clinical productivity alone may prove inadequate. It appears hospitalists working in academic facilities spend significant time on indirect patient care because of these hospitals' inefficiencies, usually not found in community settings.43 Devising compensation for an academic hospitalist requires careful attention and must balance a number of factors because these hospitalists will not generate their entire salary from clinical services. Financial support must come from either the division or medical center, an annual negotiation at AMCs.

Several methods exist to structure hospitalist compensation. A hospitalist's salary may be fixed, may have a base salary with incentives, or may be derived based on clinical productivity. For example, if a hospital medicine program provides both housestaff and nonhousestaff services and employs a fixed‐salary approach, it may choose a menu‐style method to determine compensation (eg, 6 months on nonhousestaff service at x dollars/month + 3 months on housestaff service at x dollars/month = annual salary). If a hospitalist takes on a funded nonclinical role or secures extramural funding, the salary menu gets adjusted accordingly as the clinical time is bought out. Critics of the fixed‐salary approach argue that paying each hospitalist the same salary regardless of the specific job description yields an inequitable system in which some are rewarded with less clinical time.

Compensation should probably have a guaranteed base salary with incentives, which could be determined by a formula that weighs clinical productivity, quality improvement efforts, scholarly activity, and teaching excellence. This model provides financial incentives to develop both clinically and academically but introduces complexity in determining a fair incentive structure. Finally, compensation can be structured without salary guarantee and putting compensation fully at risk based on clinical productivity, although this is an unlikely strategy for any hospital medicine group. This approach does disproportionately reward high volume providers, potentially at the risk of quality and safety, but also creates significant incentives to improve efficiency.

With respect to AMC relationships, hospital medicine programs must ensure the positive return on investment that drives financial support at their institutions. This fundamental economic dynamic makes AMCs dependent on their hospital medicine groups and vice versa. We caution programs from solely relying on measures such as reduced hospital costs or length of stay as a basis of funding unless there is a reward for maintaining performance once it inevitably plateaus. Moreover, explicitly tying utilization efficiency (ie, length of stay) to salary violates Stark rules44 and carries potential malpractice implications should patient care errors be attributable to premature hospital discharge. Over time hospitalists will need to maintain clinical benchmarks but also provide additional and valued services to their institutions, including quality and safety improvement activities and compliance with residency work hour restrictions.

Defining the Academic Hospitalist

The question is simple and perhaps philosophical: Are hospitalists who work at an AMC academic hospitalists? And what job description truly defines an academic hospitalist? Currently, there are no standards for the clinical activity of an academic hospitalist position (eg, number of weeks, weekends, and hours) or for assessment of nonclinical productivity. Hospital medicine programs face the challenge of defining positions that fulfill the growing clinical mission at AMCs but have little experience or guidance in ensuring they will lead to advancing the academic mission. Specifically, how do hospitalists who provide mostly clinical care, particularly on nonhousestaff services, achieve promotion? Hospital medicine program leadership must create enough opportunity and time for the development of skills in research, education, and quality or systems improvement if academic hospitalists are to succeed.

The Association of Chiefs of General Internal Medicine (ACGIM), the Society of General Internal Medicine (SGIM), and the Society of Hospital Medicine (SHM) are currently collaborating to develop consensus guidelines in this area. Ultimately, through the efforts of these important governing bodies, the specialty of hospital medicine will be able to demonstrate the unique skills and services they provide and move toward advocating for academic promotion criteria that recognize their value and accomplishments.

The Equities of the System

Prior to developing nonhousestaff services, our academic hospitalist programs scheduled teaching service time in month or half‐month blocks, balancing holidays and weekends. Equity in scheduling became a function of required clinical time, sources of non‐clinical funding (eg, grants, educational or administrative roles), and expectations for scholarship, attributes typical of most subspecialty academic divisions. Given the differing clinical missions that have stimulated academic hospital medicine programs to form, concerns of scheduling equity have grown, posing challenges not experienced in other divisions.

Institutions that choose to divide housestaff and nonhousestaff duties among distinct groups of hospitalists create the potential for a 2‐tiered system, one in which those with housestaff roles are more valued and respected by the institution. Hospitalists working on nonhousestaff services admit patients, write orders, and field direct patient calls, a role rarely undertaken by subspecialty attendings or hospitalists on housestaff services. Our collective experiences provide evidence of the danger of this second‐class‐citizen status, one that requires attention to ensure job satisfaction, retention, and necessary career development.

Institutions have accentuated the second‐class‐citizen concern by staffing nonhousestaff roles with 1‐year hospitalistsPGY‐4s. Most of these hires in our institutions are individuals just out of residency and intent on pursuing a fellowship. We speculate that they enjoy the comforts of the AMC where they often trained and accept purely nonhousestaff positions because of what they view as an appealing work schedule and salary. Although this approach addresses the growing need for hospitalists on nonhousestaff services in the short term, these positions must remain attractive enough (both financially and professionally) to encourage residency graduates to pursue an academic hospitalist career instead of a 1‐year position as a transition to fellowship. Otherwise, the approach conveys a message that relatively inexperienced physicians are good enough to be hospitalists.

Developing a cadre of clinically focused hospitalists who provide outstanding patient care and also garner respect as successful academicians is a difficult task. Although 1 group in our sample (Northwestern) shares nonhousestaff responsibilities equally, others may find this impractical, particularly where faculty members were hired before nonhousestaff services were established. Redefining such clinical positions several years into a career may be challenging, as it forces faculty members into roles they didn't sign up for or grandfathers them out of such roles, adding to the risk of a 2‐tiered system. Alternatively, groups may focus on building academic activities into nonhousestaff services, including medical student teaching, quality improvement, or clinical research activities. In this article, we deliberately classified these services as nonhousestaff rather than non‐teaching because the latter fails to acknowledge that these hospitalists often serve as teachers (eg, housestaff conferences, supervision of midlevel providers, and/or rotating medical students)an important if not symbolic distinction. It is imperative that planning for nonhousestaff services balance the larger academic mission of hospital medicine groups with creating equitable, valued, and sustainable job descriptions.

Defining the Patient Mix

Developing an optimal patient mix on nonhousestaff services also carries important implications. For services that work in parallel with the housestaff service and simply take extra patients above the resident cap, this concern may be less significant. However, other nonhousestaff services have been structured to care for lower‐acuity patients (eg, cellulitis, asthma, pneumonia) or select patient populations (eg, sickle cell or inflammatory bowel disease). This distribution system potentially changes the educational experience on the housestaff servicedecreasing the bread‐and‐butter admissionsbut also may affect the job satisfaction of hospitalists and midlevel providers on nonhousestaff services. Building triage criteria, working with emergency department leadership, and avoiding patients being turfed between different services is critical. We strongly recommend a regular process to review admissions to each service and determine when the triage process requires further calibration.

Recruitment and Retention

Traditionally, graduates of residency or fellowship training programs chose academic positions because of an interest in teaching, a desire for scholarship, or a commitment to research. Those interested in primarily clinical roles typically pursued positions in nonacademic settings. The development of nonhousestaff services challenges this paradigm because the objective for academic hospitalist leadership now becomes recruiting pure clinicians as well as academicians. These might be the same individual, a hospitalist who provides both housestaff and nonhousestaff services, or 2 different individuals if the nonhousestaff service is covered by dedicated hospitalists. In addition, with the current promotion structure in academia, a purely clinical position may be less attractive, as it provides fewer opportunities for advancement.

Therefore, recruitment and retention of academic hospitalists will require job descriptions that provide dedicated teaching opportunities, time for participation in quality and safety improvement projects, or pursuit of a scholarly interest in non‐clinical timethe diastole of an academic hospitalist.41 Hospital medicine leadership will also need to better distinguish off‐time from non‐clinical time, as many young hospitalists struggle to balance professional and personal commitmentsa recipe for burnout.42 Regardless of how clinical responsibilities differ between 2 hospitalists, providing them with similar academic resources is what will distinguish their positions from that in the community. Furthermore, many groups have chosen to pay faculty a premium for their nonhousestaff roles or to use specific recruitment incentives such as educational loan forgiveness programs.

With the expected growth of nonhousestaff services and surgical comanagement, hospital medicine programs will also need to determine if new hires will focus on a specific service (eg, orthopedic hospitalist) or whether job descriptions will include a mix of activities (eg, 3 months' teaching service, 3 months' nonhousestaff medical service, and 3 months' surgical comanagement service). A second and equally important question is where does the hospitalist live? If cardiology wants hospitalists to care for their patients, should they be hired and mentored by cardiologists or by hospitalists in a division of general or hospital medicine? In many cases, a graduating resident with plans to pursue a fellowship (eg, cardiology or hematology/oncology) may be a perfect candidate for a 1‐year position on his or her future specialty service. However, in the long term, maintaining all the academic hospitalists under the same umbrella will provide greater mentorship, professional development, opportunities for collaboration, clinical diversity, and sense of belonging to a group, rather than being a token hospitalist for another division.

Compensation and Financial Relationships with AMCs

Salaries for hospitalists working on nonhousestaff services are typically higher at AMCs, which are competing with community standards given the similar level of clinical hours worked. However, although pay for nonhousestaff activities should reflect the nature of the work, compensation models based on clinical productivity alone may prove inadequate. It appears hospitalists working in academic facilities spend significant time on indirect patient care because of these hospitals' inefficiencies, usually not found in community settings.43 Devising compensation for an academic hospitalist requires careful attention and must balance a number of factors because these hospitalists will not generate their entire salary from clinical services. Financial support must come from either the division or medical center, an annual negotiation at AMCs.

Several methods exist to structure hospitalist compensation. A hospitalist's salary may be fixed, may have a base salary with incentives, or may be derived based on clinical productivity. For example, if a hospital medicine program provides both housestaff and nonhousestaff services and employs a fixed‐salary approach, it may choose a menu‐style method to determine compensation (eg, 6 months on nonhousestaff service at x dollars/month + 3 months on housestaff service at x dollars/month = annual salary). If a hospitalist takes on a funded nonclinical role or secures extramural funding, the salary menu gets adjusted accordingly as the clinical time is bought out. Critics of the fixed‐salary approach argue that paying each hospitalist the same salary regardless of the specific job description yields an inequitable system in which some are rewarded with less clinical time.

Compensation should probably have a guaranteed base salary with incentives, which could be determined by a formula that weighs clinical productivity, quality improvement efforts, scholarly activity, and teaching excellence. This model provides financial incentives to develop both clinically and academically but introduces complexity in determining a fair incentive structure. Finally, compensation can be structured without salary guarantee and putting compensation fully at risk based on clinical productivity, although this is an unlikely strategy for any hospital medicine group. This approach does disproportionately reward high volume providers, potentially at the risk of quality and safety, but also creates significant incentives to improve efficiency.

With respect to AMC relationships, hospital medicine programs must ensure the positive return on investment that drives financial support at their institutions. This fundamental economic dynamic makes AMCs dependent on their hospital medicine groups and vice versa. We caution programs from solely relying on measures such as reduced hospital costs or length of stay as a basis of funding unless there is a reward for maintaining performance once it inevitably plateaus. Moreover, explicitly tying utilization efficiency (ie, length of stay) to salary violates Stark rules44 and carries potential malpractice implications should patient care errors be attributable to premature hospital discharge. Over time hospitalists will need to maintain clinical benchmarks but also provide additional and valued services to their institutions, including quality and safety improvement activities and compliance with residency work hour restrictions.

Defining the Academic Hospitalist

The question is simple and perhaps philosophical: Are hospitalists who work at an AMC academic hospitalists? And what job description truly defines an academic hospitalist? Currently, there are no standards for the clinical activity of an academic hospitalist position (eg, number of weeks, weekends, and hours) or for assessment of nonclinical productivity. Hospital medicine programs face the challenge of defining positions that fulfill the growing clinical mission at AMCs but have little experience or guidance in ensuring they will lead to advancing the academic mission. Specifically, how do hospitalists who provide mostly clinical care, particularly on nonhousestaff services, achieve promotion? Hospital medicine program leadership must create enough opportunity and time for the development of skills in research, education, and quality or systems improvement if academic hospitalists are to succeed.

The Association of Chiefs of General Internal Medicine (ACGIM), the Society of General Internal Medicine (SGIM), and the Society of Hospital Medicine (SHM) are currently collaborating to develop consensus guidelines in this area. Ultimately, through the efforts of these important governing bodies, the specialty of hospital medicine will be able to demonstrate the unique skills and services they provide and move toward advocating for academic promotion criteria that recognize their value and accomplishments.