Eugene S. Chu, MD

Emergency department (ED) crowding leads to ambulance diversion,1 which can delay care and worsen outcomes, including mortality.2 A national survey showed that 90% of EDs were overcrowded, and 70% reported time on diversion.3 One of the causes of ED crowding is boarding of admitted patients.4 Boarding admitted patients decreases quality of care and satisfaction.57

Improved ED triage, bedside registration, physical expansion of hospitals, and regional ambulance programs have been implemented to decrease ED diversion.812 Despite these attempts, ED diversion continues to be prevalent.

Interventions involving hospitalists have been tested to improve throughput and quality of care for admitted medicine patients boarded in the ED. Howell and colleagues decreased ED diversion through active bed management by hospitalists.13 Briones and colleagues dedicated a hospitalist team to patients boarded in the ED and improved their quality of care.14

Denver Health Medical Center (DHMC) is an urban, academic safety net hospital. In 2009, the ED saw an average of 133 patients daily and an average of 25 were admitted to the medical service. DHMC's ED diversion rate was a mean of 12.4% in 2009. Boarded medicine patients occupied 16% of ED medicine bed capacity. Teaching and nonteaching medical floor teams cared for patients in the ED awaiting inpatient beds, who were the last to be seen. Nursing supervisors transferred boarded patients from the ED to hospital units. Patients with the greatest duration of time in the ED had priority for open beds.

ED diversion is costly.15, 16 DHMC implemented codified diversion criteria, calling the administrator on‐call prior to diversion, and increasing frequency of rounding in the ED, with no sustained effect seen in the rate of ED diversion.

In 2009, the DHMC Hospital Medicine Service addressed the issue of ED crowding, ED diversion, and care of boarded ED patients by creating a hospital medicine ED (HMED) team with 2 functions: (1) to provide ongoing care for medicine patients in the ED awaiting inpatient beds; and (2) to work with nursing supervisors to improve patient flow by adding physician clinical expertise to bed management.

METHODS

Intervention

In 2009, DHMC, which uses Toyota Lean for quality improvement, performed a Rapid Improvement Event (RIE) to address ED diversion and care of admitted patients boarded in the ED. The RIE team consisted of hospital medicine physicians, ED physicians, social workers, and nurses. Over a 4‐day period, the team examined the present state, created an ideal future state, devised a solution, and tested this solution.

Based upon the results of the RIE, DHMC implemented an HMED team to care for admitted patients boarded in the ED and assist in active bed management. The HMED team is a 24/7 service. During the day shift, the HMED team is composed of 1 dedicated attending and 1 allied health provider (AHP). Since the medicine services were already staffing existing patients in the ED, the 2.0 full‐time equivalent (FTE) needed to staff the HMED team attending and the AHP was reallocated from existing FTE within the hospitalist division. During the evening and night shifts, the HMED team's responsibilities were rolled into existing hospitalist duties.

The HMED team provides clinical care for 2 groups of patients in the ED. The first group represents admitted patients who are still awaiting a medicine ward bed as of 7:00 AM. The HMED team provides ongoing care until discharge from the ED or transfer to a medicine floor. The second group of patients includes new admissions that need to stay in the ED due to a lack of available medicine floor beds. For these patients, the HMED team initiates and continues care until discharge from the ED or transfer to a medical floor (Figure 1).

Figure 1

The physician on the HMED team assists nursing supervisors with bed management by providing detailed clinical knowledge, including proximity to discharge as well as updated information on telemetry and intensive care unit (ICU) appropriateness. The HMED team's physician maintains constant knowledge of hospital census via an electronic bed board, and communicates regularly with medical floors about anticipated discharges and transfers to understand the hospital's patient flow status (Figure 2).

Figure 2

The RIE that resulted in the HMED team was part of the Inpatient Medicine Value Stream, which had the overall goal of saving DHMC $300,000 for 2009. Ten RIEs were planned for this value stream in 2009, with an average of $30,000 of savings expected from each RIE.

RESULTS

The ED saw 48,595 patients during the intervention period (August 1, 2009June 30,2010) which did not differ statistically from the 50,469 patients seen in the control period (August 1, 2008June 30, 2009). The number of admissions to the medicine service during the control period (9727) and intervention period (10,013), and the number of total medical/surgical admissions during the control (20,716) and intervention (20,574) periods did not statistically differ. ED staffing during the intervention did not change. The overall number of licensed beds did not increase during the study period. During the control period, staffed medical/surgical beds increased from 395 to 400 beds, while the number of staffed medical/surgical beds decreased from 400 to 397 beds during the intervention period. Patient characteristics were similar during the 2 time periods, with the exception of race (Table 1).

DISCUSSION

This study suggests that an HMED team can decrease ED diversion, due to hospital bed capacity, by improving patient flow and timeliness of care for boarded medicine patients in the ED.

After participating in bed management, ED diversion due to a lack of medicine beds decreased. This is consistent with findings by Howell and colleagues who were able to improve throughput and decrease ED diversion with active bed management.13 Howell and colleagues decreased diversion hours due to temporary ED overload, and diversion hours due to a lack of telemetry or critical care beds. At DHMC, diversion is attributed to either a lack of ED capacity or lack of hospital beds. The primary outcome was the diversion rate due to lack of hospital beds, but it is possible that increased discharges directly from the ED contributed to the decrease in diversion due to ED capacity, underestimating the effect our intervention had on total ED diversion. There were no other initiatives to decrease diversion due to ED capacity during the study periods, and ED capacity and volume did not change during the intervention period.

While there were no statistically significant changes in staffed medical/surgical beds or medicine admissions, staffed medical/surgical beds during the intervention period decreased while there were more admissions to medicine. Both of these variables would increase diversion, resulting in an underestimation of the effect of the intervention.

Howell and colleagues improved throughput in the ED by implementing a service which provided active bed management without clinical responsibilities,13 while Briones and colleagues improved clinical care of patients boarded in the ED without affecting throughput.14 The HMED team improved throughput and decreased ED diversion while improving timeliness of care and perception of care quality for patients boarding in the ED.

By decreasing unnecessary transfers to medicine units and increasing discharges from the ED, patient flow was improved. While there was no difference in ED LOS, there was a trend towards decreased total LOS. A larger sample size or a longer period of observation would be necessary to determine if the trend toward decreased total LOS is statistically significant. ED LOS may not have been decreased because patients who would have been sent to the floor only to be discharged within 8 hours were kept in the ED to expedite testing and discharge, while sicker patients were sent to the medical floor. This decreased the turnover time of inpatient beds and allowed more boarded patients to be moved to floor units.

There was concern that an HMED team would fragment care, which would lead to an increased LOS for those patients who were transferred to a medical floor and cared for by an additional medicine team before discharge.17 As noted, there was a trend towards a decreased LOS for patients initially cared for by the HMED team.

In this intervention, hospital medicine physicians provided information regarding ongoing care of patients boarded in the ED to nursing supervisors. Prior to the intervention, nursing supervisors relied upon information from the ED staff and the boarded patient's time in the ED to assign a medical floor. However, ED staff was not providing care to boarded patients and did not know the most up‐to‐date status of the patient. This queuing process and lack of communication resulted in patients ready for discharge being transferred to floor beds and discharged within a few hours of transfer. The HMED team allowed nursing supervisors to have direct knowledge regarding clinical status, including telemetry and ICU criteria (similar to Howell and colleagues13), and readiness for discharge from the physician taking care of the patient.

By managing boarded patients, an HMED team can improve timeliness and coordination of care. Prior to the intervention, boarded ED patients were the last to be seen on rounds. The HMED team rounds only in the ED, expediting care and discharges. The increased proportion of boarded patients discharged from the ED by the HMED team is consistent with Briones and colleagues' clinically oriented team managing boarding patients in the ED.14

Potential adverse effects of our intervention included increased returns to the ED, increased ICU transfer rate, and decreased housestaff satisfaction. There was no increase in the 48‐hour return rate and no increase in the ICU transfer rate for patients cared for by the HMED team. Housestaff at DHMC are satisfied with the HMED team, since the presence of the HMED team allows them to concentrate on patients on the medical floors.

This intervention provides DHMC with an additional $525,600 in revenue annually. Since existing FTE were reallocated to create the HMED team, no additional FTE were required. In our facility, AHPs take on duties of housestaff. However, only 1 physician may be needed to staff an HMED team. This physician's clinical productivity is about 75% of other physicians; therefore, 25% of time is spent in bed management. At DHMC, other medicine teams picked up for the decreased clinical productivity of the HMED team, so the budget was neutral. However, using 2 FTE to staff 1 physician daily for 365 days a year, one would need to allocate 0.5 physician FTE (0.25 decrease in clinical productivity 2 FTE) for an HMED team.

Our study has several limitations. As a single center study, our findings may not extrapolate to other settings. The study used historical controls, therefore, undetected confounders may exist. We could not control for simultaneous changes in the hospital, however, we did not know of any other concurrent interventions aimed at decreasing ED diversion. Also, the decision to admit or not is partially based on individual ED attendings, which causes variability in practice. Finally, while we were able to measure rounding times as a process measure to reflect timeliness of care and staff perceptions of quality of care, due to our data infrastructure and the way our housestaff and attendings rotate, we were not able to assess more downstream measures of quality of care.

CONCLUSION

ED crowding decreases throughput and worsens clinical care; there are few proven solutions. This study demonstrates an intervention that reduced the percentage of patients transferred to a medicine floor and discharged within 8 hours, increased the number of discharges from the ED of admitted medicine patients, and decreased ED diversion while improving the timeliness of clinical care for patients boarded in the ED.

Emergency department (ED) crowding leads to ambulance diversion,1 which can delay care and worsen outcomes, including mortality.2 A national survey showed that 90% of EDs were overcrowded, and 70% reported time on diversion.3 One of the causes of ED crowding is boarding of admitted patients.4 Boarding admitted patients decreases quality of care and satisfaction.57

Improved ED triage, bedside registration, physical expansion of hospitals, and regional ambulance programs have been implemented to decrease ED diversion.812 Despite these attempts, ED diversion continues to be prevalent.

Interventions involving hospitalists have been tested to improve throughput and quality of care for admitted medicine patients boarded in the ED. Howell and colleagues decreased ED diversion through active bed management by hospitalists.13 Briones and colleagues dedicated a hospitalist team to patients boarded in the ED and improved their quality of care.14

Denver Health Medical Center (DHMC) is an urban, academic safety net hospital. In 2009, the ED saw an average of 133 patients daily and an average of 25 were admitted to the medical service. DHMC's ED diversion rate was a mean of 12.4% in 2009. Boarded medicine patients occupied 16% of ED medicine bed capacity. Teaching and nonteaching medical floor teams cared for patients in the ED awaiting inpatient beds, who were the last to be seen. Nursing supervisors transferred boarded patients from the ED to hospital units. Patients with the greatest duration of time in the ED had priority for open beds.

ED diversion is costly.15, 16 DHMC implemented codified diversion criteria, calling the administrator on‐call prior to diversion, and increasing frequency of rounding in the ED, with no sustained effect seen in the rate of ED diversion.

In 2009, the DHMC Hospital Medicine Service addressed the issue of ED crowding, ED diversion, and care of boarded ED patients by creating a hospital medicine ED (HMED) team with 2 functions: (1) to provide ongoing care for medicine patients in the ED awaiting inpatient beds; and (2) to work with nursing supervisors to improve patient flow by adding physician clinical expertise to bed management.

METHODS

Intervention

In 2009, DHMC, which uses Toyota Lean for quality improvement, performed a Rapid Improvement Event (RIE) to address ED diversion and care of admitted patients boarded in the ED. The RIE team consisted of hospital medicine physicians, ED physicians, social workers, and nurses. Over a 4‐day period, the team examined the present state, created an ideal future state, devised a solution, and tested this solution.

Based upon the results of the RIE, DHMC implemented an HMED team to care for admitted patients boarded in the ED and assist in active bed management. The HMED team is a 24/7 service. During the day shift, the HMED team is composed of 1 dedicated attending and 1 allied health provider (AHP). Since the medicine services were already staffing existing patients in the ED, the 2.0 full‐time equivalent (FTE) needed to staff the HMED team attending and the AHP was reallocated from existing FTE within the hospitalist division. During the evening and night shifts, the HMED team's responsibilities were rolled into existing hospitalist duties.

The HMED team provides clinical care for 2 groups of patients in the ED. The first group represents admitted patients who are still awaiting a medicine ward bed as of 7:00 AM. The HMED team provides ongoing care until discharge from the ED or transfer to a medicine floor. The second group of patients includes new admissions that need to stay in the ED due to a lack of available medicine floor beds. For these patients, the HMED team initiates and continues care until discharge from the ED or transfer to a medical floor (Figure 1).

Figure 1

The physician on the HMED team assists nursing supervisors with bed management by providing detailed clinical knowledge, including proximity to discharge as well as updated information on telemetry and intensive care unit (ICU) appropriateness. The HMED team's physician maintains constant knowledge of hospital census via an electronic bed board, and communicates regularly with medical floors about anticipated discharges and transfers to understand the hospital's patient flow status (Figure 2).

Figure 2

The RIE that resulted in the HMED team was part of the Inpatient Medicine Value Stream, which had the overall goal of saving DHMC $300,000 for 2009. Ten RIEs were planned for this value stream in 2009, with an average of $30,000 of savings expected from each RIE.

RESULTS

The ED saw 48,595 patients during the intervention period (August 1, 2009June 30,2010) which did not differ statistically from the 50,469 patients seen in the control period (August 1, 2008June 30, 2009). The number of admissions to the medicine service during the control period (9727) and intervention period (10,013), and the number of total medical/surgical admissions during the control (20,716) and intervention (20,574) periods did not statistically differ. ED staffing during the intervention did not change. The overall number of licensed beds did not increase during the study period. During the control period, staffed medical/surgical beds increased from 395 to 400 beds, while the number of staffed medical/surgical beds decreased from 400 to 397 beds during the intervention period. Patient characteristics were similar during the 2 time periods, with the exception of race (Table 1).

DISCUSSION

This study suggests that an HMED team can decrease ED diversion, due to hospital bed capacity, by improving patient flow and timeliness of care for boarded medicine patients in the ED.

After participating in bed management, ED diversion due to a lack of medicine beds decreased. This is consistent with findings by Howell and colleagues who were able to improve throughput and decrease ED diversion with active bed management.13 Howell and colleagues decreased diversion hours due to temporary ED overload, and diversion hours due to a lack of telemetry or critical care beds. At DHMC, diversion is attributed to either a lack of ED capacity or lack of hospital beds. The primary outcome was the diversion rate due to lack of hospital beds, but it is possible that increased discharges directly from the ED contributed to the decrease in diversion due to ED capacity, underestimating the effect our intervention had on total ED diversion. There were no other initiatives to decrease diversion due to ED capacity during the study periods, and ED capacity and volume did not change during the intervention period.

While there were no statistically significant changes in staffed medical/surgical beds or medicine admissions, staffed medical/surgical beds during the intervention period decreased while there were more admissions to medicine. Both of these variables would increase diversion, resulting in an underestimation of the effect of the intervention.

Howell and colleagues improved throughput in the ED by implementing a service which provided active bed management without clinical responsibilities,13 while Briones and colleagues improved clinical care of patients boarded in the ED without affecting throughput.14 The HMED team improved throughput and decreased ED diversion while improving timeliness of care and perception of care quality for patients boarding in the ED.

By decreasing unnecessary transfers to medicine units and increasing discharges from the ED, patient flow was improved. While there was no difference in ED LOS, there was a trend towards decreased total LOS. A larger sample size or a longer period of observation would be necessary to determine if the trend toward decreased total LOS is statistically significant. ED LOS may not have been decreased because patients who would have been sent to the floor only to be discharged within 8 hours were kept in the ED to expedite testing and discharge, while sicker patients were sent to the medical floor. This decreased the turnover time of inpatient beds and allowed more boarded patients to be moved to floor units.

There was concern that an HMED team would fragment care, which would lead to an increased LOS for those patients who were transferred to a medical floor and cared for by an additional medicine team before discharge.17 As noted, there was a trend towards a decreased LOS for patients initially cared for by the HMED team.

In this intervention, hospital medicine physicians provided information regarding ongoing care of patients boarded in the ED to nursing supervisors. Prior to the intervention, nursing supervisors relied upon information from the ED staff and the boarded patient's time in the ED to assign a medical floor. However, ED staff was not providing care to boarded patients and did not know the most up‐to‐date status of the patient. This queuing process and lack of communication resulted in patients ready for discharge being transferred to floor beds and discharged within a few hours of transfer. The HMED team allowed nursing supervisors to have direct knowledge regarding clinical status, including telemetry and ICU criteria (similar to Howell and colleagues13), and readiness for discharge from the physician taking care of the patient.

By managing boarded patients, an HMED team can improve timeliness and coordination of care. Prior to the intervention, boarded ED patients were the last to be seen on rounds. The HMED team rounds only in the ED, expediting care and discharges. The increased proportion of boarded patients discharged from the ED by the HMED team is consistent with Briones and colleagues' clinically oriented team managing boarding patients in the ED.14

Potential adverse effects of our intervention included increased returns to the ED, increased ICU transfer rate, and decreased housestaff satisfaction. There was no increase in the 48‐hour return rate and no increase in the ICU transfer rate for patients cared for by the HMED team. Housestaff at DHMC are satisfied with the HMED team, since the presence of the HMED team allows them to concentrate on patients on the medical floors.

This intervention provides DHMC with an additional $525,600 in revenue annually. Since existing FTE were reallocated to create the HMED team, no additional FTE were required. In our facility, AHPs take on duties of housestaff. However, only 1 physician may be needed to staff an HMED team. This physician's clinical productivity is about 75% of other physicians; therefore, 25% of time is spent in bed management. At DHMC, other medicine teams picked up for the decreased clinical productivity of the HMED team, so the budget was neutral. However, using 2 FTE to staff 1 physician daily for 365 days a year, one would need to allocate 0.5 physician FTE (0.25 decrease in clinical productivity 2 FTE) for an HMED team.

Our study has several limitations. As a single center study, our findings may not extrapolate to other settings. The study used historical controls, therefore, undetected confounders may exist. We could not control for simultaneous changes in the hospital, however, we did not know of any other concurrent interventions aimed at decreasing ED diversion. Also, the decision to admit or not is partially based on individual ED attendings, which causes variability in practice. Finally, while we were able to measure rounding times as a process measure to reflect timeliness of care and staff perceptions of quality of care, due to our data infrastructure and the way our housestaff and attendings rotate, we were not able to assess more downstream measures of quality of care.

CONCLUSION

ED crowding decreases throughput and worsens clinical care; there are few proven solutions. This study demonstrates an intervention that reduced the percentage of patients transferred to a medicine floor and discharged within 8 hours, increased the number of discharges from the ED of admitted medicine patients, and decreased ED diversion while improving the timeliness of clinical care for patients boarded in the ED.

Emergency department (ED) crowding leads to ambulance diversion,1 which can delay care and worsen outcomes, including mortality.2 A national survey showed that 90% of EDs were overcrowded, and 70% reported time on diversion.3 One of the causes of ED crowding is boarding of admitted patients.4 Boarding admitted patients decreases quality of care and satisfaction.57

Improved ED triage, bedside registration, physical expansion of hospitals, and regional ambulance programs have been implemented to decrease ED diversion.812 Despite these attempts, ED diversion continues to be prevalent.

Interventions involving hospitalists have been tested to improve throughput and quality of care for admitted medicine patients boarded in the ED. Howell and colleagues decreased ED diversion through active bed management by hospitalists.13 Briones and colleagues dedicated a hospitalist team to patients boarded in the ED and improved their quality of care.14

Denver Health Medical Center (DHMC) is an urban, academic safety net hospital. In 2009, the ED saw an average of 133 patients daily and an average of 25 were admitted to the medical service. DHMC's ED diversion rate was a mean of 12.4% in 2009. Boarded medicine patients occupied 16% of ED medicine bed capacity. Teaching and nonteaching medical floor teams cared for patients in the ED awaiting inpatient beds, who were the last to be seen. Nursing supervisors transferred boarded patients from the ED to hospital units. Patients with the greatest duration of time in the ED had priority for open beds.

ED diversion is costly.15, 16 DHMC implemented codified diversion criteria, calling the administrator on‐call prior to diversion, and increasing frequency of rounding in the ED, with no sustained effect seen in the rate of ED diversion.

In 2009, the DHMC Hospital Medicine Service addressed the issue of ED crowding, ED diversion, and care of boarded ED patients by creating a hospital medicine ED (HMED) team with 2 functions: (1) to provide ongoing care for medicine patients in the ED awaiting inpatient beds; and (2) to work with nursing supervisors to improve patient flow by adding physician clinical expertise to bed management.

METHODS

Intervention

In 2009, DHMC, which uses Toyota Lean for quality improvement, performed a Rapid Improvement Event (RIE) to address ED diversion and care of admitted patients boarded in the ED. The RIE team consisted of hospital medicine physicians, ED physicians, social workers, and nurses. Over a 4‐day period, the team examined the present state, created an ideal future state, devised a solution, and tested this solution.

Based upon the results of the RIE, DHMC implemented an HMED team to care for admitted patients boarded in the ED and assist in active bed management. The HMED team is a 24/7 service. During the day shift, the HMED team is composed of 1 dedicated attending and 1 allied health provider (AHP). Since the medicine services were already staffing existing patients in the ED, the 2.0 full‐time equivalent (FTE) needed to staff the HMED team attending and the AHP was reallocated from existing FTE within the hospitalist division. During the evening and night shifts, the HMED team's responsibilities were rolled into existing hospitalist duties.

The HMED team provides clinical care for 2 groups of patients in the ED. The first group represents admitted patients who are still awaiting a medicine ward bed as of 7:00 AM. The HMED team provides ongoing care until discharge from the ED or transfer to a medicine floor. The second group of patients includes new admissions that need to stay in the ED due to a lack of available medicine floor beds. For these patients, the HMED team initiates and continues care until discharge from the ED or transfer to a medical floor (Figure 1).

Figure 1

The physician on the HMED team assists nursing supervisors with bed management by providing detailed clinical knowledge, including proximity to discharge as well as updated information on telemetry and intensive care unit (ICU) appropriateness. The HMED team's physician maintains constant knowledge of hospital census via an electronic bed board, and communicates regularly with medical floors about anticipated discharges and transfers to understand the hospital's patient flow status (Figure 2).

Figure 2

The RIE that resulted in the HMED team was part of the Inpatient Medicine Value Stream, which had the overall goal of saving DHMC $300,000 for 2009. Ten RIEs were planned for this value stream in 2009, with an average of $30,000 of savings expected from each RIE.

RESULTS

The ED saw 48,595 patients during the intervention period (August 1, 2009June 30,2010) which did not differ statistically from the 50,469 patients seen in the control period (August 1, 2008June 30, 2009). The number of admissions to the medicine service during the control period (9727) and intervention period (10,013), and the number of total medical/surgical admissions during the control (20,716) and intervention (20,574) periods did not statistically differ. ED staffing during the intervention did not change. The overall number of licensed beds did not increase during the study period. During the control period, staffed medical/surgical beds increased from 395 to 400 beds, while the number of staffed medical/surgical beds decreased from 400 to 397 beds during the intervention period. Patient characteristics were similar during the 2 time periods, with the exception of race (Table 1).

DISCUSSION

This study suggests that an HMED team can decrease ED diversion, due to hospital bed capacity, by improving patient flow and timeliness of care for boarded medicine patients in the ED.

After participating in bed management, ED diversion due to a lack of medicine beds decreased. This is consistent with findings by Howell and colleagues who were able to improve throughput and decrease ED diversion with active bed management.13 Howell and colleagues decreased diversion hours due to temporary ED overload, and diversion hours due to a lack of telemetry or critical care beds. At DHMC, diversion is attributed to either a lack of ED capacity or lack of hospital beds. The primary outcome was the diversion rate due to lack of hospital beds, but it is possible that increased discharges directly from the ED contributed to the decrease in diversion due to ED capacity, underestimating the effect our intervention had on total ED diversion. There were no other initiatives to decrease diversion due to ED capacity during the study periods, and ED capacity and volume did not change during the intervention period.

While there were no statistically significant changes in staffed medical/surgical beds or medicine admissions, staffed medical/surgical beds during the intervention period decreased while there were more admissions to medicine. Both of these variables would increase diversion, resulting in an underestimation of the effect of the intervention.

Howell and colleagues improved throughput in the ED by implementing a service which provided active bed management without clinical responsibilities,13 while Briones and colleagues improved clinical care of patients boarded in the ED without affecting throughput.14 The HMED team improved throughput and decreased ED diversion while improving timeliness of care and perception of care quality for patients boarding in the ED.

By decreasing unnecessary transfers to medicine units and increasing discharges from the ED, patient flow was improved. While there was no difference in ED LOS, there was a trend towards decreased total LOS. A larger sample size or a longer period of observation would be necessary to determine if the trend toward decreased total LOS is statistically significant. ED LOS may not have been decreased because patients who would have been sent to the floor only to be discharged within 8 hours were kept in the ED to expedite testing and discharge, while sicker patients were sent to the medical floor. This decreased the turnover time of inpatient beds and allowed more boarded patients to be moved to floor units.

There was concern that an HMED team would fragment care, which would lead to an increased LOS for those patients who were transferred to a medical floor and cared for by an additional medicine team before discharge.17 As noted, there was a trend towards a decreased LOS for patients initially cared for by the HMED team.

In this intervention, hospital medicine physicians provided information regarding ongoing care of patients boarded in the ED to nursing supervisors. Prior to the intervention, nursing supervisors relied upon information from the ED staff and the boarded patient's time in the ED to assign a medical floor. However, ED staff was not providing care to boarded patients and did not know the most up‐to‐date status of the patient. This queuing process and lack of communication resulted in patients ready for discharge being transferred to floor beds and discharged within a few hours of transfer. The HMED team allowed nursing supervisors to have direct knowledge regarding clinical status, including telemetry and ICU criteria (similar to Howell and colleagues13), and readiness for discharge from the physician taking care of the patient.

By managing boarded patients, an HMED team can improve timeliness and coordination of care. Prior to the intervention, boarded ED patients were the last to be seen on rounds. The HMED team rounds only in the ED, expediting care and discharges. The increased proportion of boarded patients discharged from the ED by the HMED team is consistent with Briones and colleagues' clinically oriented team managing boarding patients in the ED.14

Potential adverse effects of our intervention included increased returns to the ED, increased ICU transfer rate, and decreased housestaff satisfaction. There was no increase in the 48‐hour return rate and no increase in the ICU transfer rate for patients cared for by the HMED team. Housestaff at DHMC are satisfied with the HMED team, since the presence of the HMED team allows them to concentrate on patients on the medical floors.

This intervention provides DHMC with an additional $525,600 in revenue annually. Since existing FTE were reallocated to create the HMED team, no additional FTE were required. In our facility, AHPs take on duties of housestaff. However, only 1 physician may be needed to staff an HMED team. This physician's clinical productivity is about 75% of other physicians; therefore, 25% of time is spent in bed management. At DHMC, other medicine teams picked up for the decreased clinical productivity of the HMED team, so the budget was neutral. However, using 2 FTE to staff 1 physician daily for 365 days a year, one would need to allocate 0.5 physician FTE (0.25 decrease in clinical productivity 2 FTE) for an HMED team.

Our study has several limitations. As a single center study, our findings may not extrapolate to other settings. The study used historical controls, therefore, undetected confounders may exist. We could not control for simultaneous changes in the hospital, however, we did not know of any other concurrent interventions aimed at decreasing ED diversion. Also, the decision to admit or not is partially based on individual ED attendings, which causes variability in practice. Finally, while we were able to measure rounding times as a process measure to reflect timeliness of care and staff perceptions of quality of care, due to our data infrastructure and the way our housestaff and attendings rotate, we were not able to assess more downstream measures of quality of care.

CONCLUSION

ED crowding decreases throughput and worsens clinical care; there are few proven solutions. This study demonstrates an intervention that reduced the percentage of patients transferred to a medicine floor and discharged within 8 hours, increased the number of discharges from the ED of admitted medicine patients, and decreased ED diversion while improving the timeliness of clinical care for patients boarded in the ED.

Hospitalists are physicians whose primary focus is the general medical care of hospitalized patients. Hospitalists are uniquely positioned to implement strategies to improve patient flow and efficiency.1 With emergency department (ED) diversion reaching rates upward of 70%, lack of access to inpatient beds leads to delayed care with worsened outcomes.25

To improve access to hospital beds, hospitals may increase capacity by either adding beds or by more efficiently using existing beds. Operations management principles have been applied to healthcare to ensure efficient use of beds. These include: reducing variability of scheduled admissions, remeasuring length of stay (LOS) and bed demand after implementing strategies to reduce practice variation, and employing queuing theory to generate predictions of optimal beds needed.6 The Joint Commission implemented a leadership standard (LD 04.03.11) that hospitals develop and implement plans to identify and mitigate impediments to efficient patient flow through the hospital.

To improve access, hospital leaders expect hospitalists to staff in inpatient medicine programs, surgical comanagement, short stay and chest pain units, and active bed management.7 In the following review, we define hospitalists' roles in the aforementioned programs and their effect on patient flow. We also touch on preoperative clinics, palliative care, geographic rounding, and flexible staffing models.

ACUTE INPATIENT CARE

Hospitalists are one of the fastest growing physician groups in the United States.810 Hospitalists improve efficiency and quality of care across a variety of demographic, geographic, and healthcare settings.11, 12 A 2002 retrospective cohort study in a community‐based urban teaching hospital showed that hospitalists decreased LOS by 0.61 days and lowered risk for death in the hospital (adjusted relative hazard, 0.71; 95% confidence interval [CI], 0.540.93).13 A 2004 prospective quasi‐experimental observational study done at an academic teaching hospital showed an adjusted LOS that was 16.2% lower, and adjusted cost 9.7% lower, for patients on the hospitalists' service.14 In 2007, Lindenauer and colleagues found that a national sample of hospitalists decreased LOS by 0.4 days and lowered cost by $286 per patient.15 The findings of these individual studies were supported in a 2009 systematic review of 33 studies by Peterson which showed that hospitalists decrease LOS.16 In a recent study, Kuo and Goodwin showed that while hospitalists decrease LOS and cost, the patients they care for have higher Medicare costs after discharge by $322 per patient, and are more likely to be readmitted (odds ratio, 1.08; CI, 1.041.14).17

The hospitalist model of care continues to grow, and hospitalists will soon number as many as 30,000.18 For acute medical inpatients, the evidence suggests that hospitalists improve patient flow by decreasing LOS while improving other aspects of quality of care. However, Kuo and Goodwin's findings suggest that the transition of care from inpatient to outpatient settings still requires attention.17

SURGICAL COMANAGEMENT

The Society of Hospital Medicine (SHM) core competencies include perioperative medicine.19, 20 In the 2006 SHM national survey, 85% of hospital medicine groups indicated that they participated in surgical comanagement.21

Hospitalists have improved patient flow and outcomes for orthopedic patients. Hospitalist management of hip fracture patients decreases time to surgery and LOS compared to standard care.2224 Phy and colleagues studied 466 patients for 2 years after the inception of hospital medicine comanagement of surgical patients, and found that care by hospitalists decreased LOS by 2.2 days.22 In a retrospective study of 118 patients, Roy and colleagues found that hospitalist‐managed patients had shorter time to consultation and surgery, decreased LOS, and lower costs.23 In a retrospective cohort study, Batsis looked at mortality in 466 patients with hip fracture, and found no difference between hospitalist management and standard care.24 In patients undergoing elective hip and knee arthroplasty, Huddleston and colleagues reported that patients managed by hospitalists had fewer complications and shorter LOS. The nurses and orthopedic surgeons preferred the hospitalistorthopedist comanagement model.25

The benefits of hospitalist comanagement are not limited to adult patients undergoing orthopedic surgery. For high‐risk patients undergoing lower extremity reconstruction surgery, Pinzur and colleagues noted that LOS was shorter for a cohort of patients managed by hospitalists than for a group of historical controls not treated by hospitalists.26 Simon and colleagues studied comanagement for pediatric spinal fusion patients, and found a decrease in LOS from 6.5 to 4.8 days.27

Several factors should be considered in developing and implementing a successful comanagement program. Since comanagement duties may fall upon hospitalists in order to protect surgeons' time,28 hospital medicine groups should ensure adequate staffing prior to taking on additional services. Clear guidelines to delineate roles and responsibilities of the comanaging groups also need to be developed and implemented.29, 30

Comanaging may also involve additional training. Hospitalists who manage neurologic, neurosurgical, trauma, and psychiatric patients report being undertrained for such conditions.31, 32 Hospital medicine groups need to ensure training needs are met and supported. Given the successes of comanagement and the increasing complexity of surgical patients,33 this practice will likely expand to a greater variety of non‐medical patients.

SHORT STAY UNITS

In 2003, short stay units (SSU) were present in approximately 20% of US hospitals, with 11% of hospitals planning on opening one in the next year.34 SSU are designed to manage acute, self‐limited medical conditions that require brief staysusually less than 72 hours. Approximately 80% of SSU patients are discharged home, avoiding hospitalization.35 Historically, SSU have been under the domain of the ED; however, there is an emerging role for hospitalist‐run SSU.36

Despite demand for SSU, little research has been performed on hospitalist‐led SSU. In 2000, Abenhaim and colleagues showed that a hospitalist‐run SSU at a university‐affiliated teaching hospital had a shorter LOS and lower rates of complications and readmissions when compared to medicine teaching services.37 In 2008, Northwestern Memorial Hospital opened a 30‐bed hospitalist‐run SSU; for those patients, LOS decreased by 2 days.38 In 2010, Leykum and colleagues showed that a hospitalist‐run observation unit can decrease LOS from 2.4 days to 2.2 days.39 Careful selection of SSU patients is needed to obtain these results. Lucas and colleagues found that whether or not SSU patients required assistance of specialists was the strongest predictor of unsuccessful stays (>72 hours or inpatient conversion) in SSU.36

Whether SSU are run by hospital medicine or emergency medicine is decided at an institutional level. Location of SSU in a specifically designated area is crucial, as it allows physicians to round efficiently on patients and to work with staff trained in observation services. Development of admission criteria that include specific diagnoses which match hospitalists' scope of practice is also important (Table 1).32

The protocol‐based and diagnosis‐specific nature of SSU may enhance quality of care through standardization. Future research may delineate the utility of SSU.

CHEST PAIN UNITS

In the United States, in 2004, approximately 6 million patients present annually to EDs with chest pain.40 Cost of care of patients unnecessarily admitted to coronary care units has been estimated to be nearly $3 billion annually.41 Still, as many as 3% of patients with acute myocardial infarction are discharged home.42 Chest pain units (CPU) were developed to facilitate evaluation of patients with chest pain, at low risk for acute coronary syndrome, without requiring inpatient admission. A number of studies have suggested that admission to a CPU is a safe and cost‐effective alternative to hospital admission.4348

CPU have traditionally been staffed by ED physicians and/or cardiologists. In a prepost study, Krantz and colleagues found that a CPU model, incorporating hospitalists at an academic public safety‐net hospital, decreased ED LOS with no difference in 30‐day cardiac event rate.49 Myers and colleagues created a hospitalist‐directed nonteaching service in an academic medical center to admit low‐risk chest pain patients. Patients admitted to the hospitalist service had a statistically significant lower median LOS (23 hours vs 33 hours) and approximately half the median hospital charges than those admitted to teaching services.50 At the same academic medical center, Bayley and colleagues showed that 91% of patients admitted for chest pain waited more than 3 hours for a bed. This adversely affected ED revenue by tying up beds, resulting in an estimated annual loss of $168,300 of hospital revenue. Creation of a hospitalist‐managed service for low‐acuity chest pain patients reduced hospital LOS by 7 hours.51 Somekh and colleagues demonstrated that a protocol‐driven, cardiologist‐run CPU results in a decreased LOS and readmission rate compared to usual care.52 In a non‐peer reviewed case study, Cox Health opened an 8‐bed, hospitalist‐led CPU in 2003. They decreased LOS from 72 to 18 hours, while increasing revenue by $2.5 million a year.53 These studies suggest that hospitalist‐run CPU can decrease LOS, increase revenue, and relieve ED overcrowding.

Development of a successful CPU depends upon clear inclusion/exclusion criteria; close collaboration among ED physicians, hospitalists, and cardiologists; the development of evidence‐based protocols, and the availability of stress testing.

ACTIVE BED MANAGEMENT

As of 2007, 90% of EDs were crowded beyond their capacity.2 ED crowding leads to ambulance diversion,54 which can delay care and increase mortality rates.55 One of the main causes of ED crowding is the boarding of admitted patients.56 Boarded, admitted patients have been shown to have decreased quality of care and patient satisfaction.35

Active bed management (ABM) by hospitalists can decrease ED diversion. Howell and colleagues instituted ABM where hospitalists, as active bed managers, facilitate placement of patients to their inpatient destinations to assist ED flow.57 This 24‐hour, hospitalist‐led, active bed management service decreased both ED LOS and ambulance diversion. The bed manager collaborated real‐time with medicine and ED attending physicians, nursing supervisors, and charge nurses to change patient care status, and assign and facilitate transfer of patients to appropriate units. These hospitalist bed managers were also empowered to activate additional resources when pre‐diversion rounds identified resource limitations and impending ED divert. They found overall ED LOS for admitted patients decreased by 98 minutes, while LOS for non‐admitted patients stayed the same. AMB decreased diversion due to critically ill and telemetry patients by 28% (786 hours), and diversion due to lower acuity patients by 6% (182 hours). This intervention proved cost‐effective. Three full‐time equivalent (FTE) hospitalists' salaries staff 1 active bed manager working 24/7. Nearly 1000 hours of diversion were avoided at an annual savings of $1086 per hour of diversion decreased.

ABM is a new frontier for hospitals in general, and hospitalists in particular. Chadaga and colleagues found that a hospital medicine‐ED team participating in active bed management, while caring for admitted patients boarded in the ED, can decrease ED diversion and improve patient flow. The percentage of patients transferred to a medicine floor and discharged within 8 hours was reduced by 67% (P < 0.01), while the number of discharges from the ED of admitted medicine patients increased by 61% (P < 0.001).58

To decrease initial investment, components of ABM (ED triage, bed assignment, discharge facilitation) can be instituted in parts. Hospital medicine groups with limited resources may only provide a triage service by phone for difficult ED cases. Bedside evaluations and collaboration with nursing staff to improve bed placement may be a next step, with floor and/or intensive care unit (ICU) rounds to facilitate early discharges as a final component.

OTHER AREAS

CONCLUSION

While many hospital medicine groups were started to provide acute inpatient medical care, most have found that their value to hospitals reaches beyond bedside care. With an epidemic of ED diversion and lack of access to hospital beds and services, optimizing throughput has become imperative for hospital systems. While hospital access can be improved with addition of new beds, improving throughput by decreasing LOS maximizes utilization of existing resources.

We have reviewed how hospitalists improve patient flow in acute inpatient medicine, surgical comanagement, short stay units, chest pain units, and active bed management. In each instance, the literature supports measures for decreasing LOS while maintaining or improving quality of care. Hinami and colleagues showed physician satisfaction with hospitalist‐provided patient care.75 Most studies have been limited by tracking upstream effects of improved efficiency. As there is now some evidence that decreasing LOS may increase readmissions,17 future studies should incorporate this metric into their outcomes. The effect of formal operations management principles on patient flow and bed efficiency is not well known and should be further examined.

In addition, we have touched on other areas (perioperative clinics, palliative care, geographic rounding, and flexible staffing models) where hospitalists may impact patient throughput. These areas represent excellent opportunities for future research.

Hospitalist participation in many of these areas is in its infancy. Hospital medicine programs interested in expanding their services, beyond acute inpatient care, have the opportunity to develop standards and continue research on the effect of hospital medicine‐led services on patient care and flow.

Hospitalists are physicians whose primary focus is the general medical care of hospitalized patients. Hospitalists are uniquely positioned to implement strategies to improve patient flow and efficiency.1 With emergency department (ED) diversion reaching rates upward of 70%, lack of access to inpatient beds leads to delayed care with worsened outcomes.25

To improve access to hospital beds, hospitals may increase capacity by either adding beds or by more efficiently using existing beds. Operations management principles have been applied to healthcare to ensure efficient use of beds. These include: reducing variability of scheduled admissions, remeasuring length of stay (LOS) and bed demand after implementing strategies to reduce practice variation, and employing queuing theory to generate predictions of optimal beds needed.6 The Joint Commission implemented a leadership standard (LD 04.03.11) that hospitals develop and implement plans to identify and mitigate impediments to efficient patient flow through the hospital.

To improve access, hospital leaders expect hospitalists to staff in inpatient medicine programs, surgical comanagement, short stay and chest pain units, and active bed management.7 In the following review, we define hospitalists' roles in the aforementioned programs and their effect on patient flow. We also touch on preoperative clinics, palliative care, geographic rounding, and flexible staffing models.

ACUTE INPATIENT CARE

Hospitalists are one of the fastest growing physician groups in the United States.810 Hospitalists improve efficiency and quality of care across a variety of demographic, geographic, and healthcare settings.11, 12 A 2002 retrospective cohort study in a community‐based urban teaching hospital showed that hospitalists decreased LOS by 0.61 days and lowered risk for death in the hospital (adjusted relative hazard, 0.71; 95% confidence interval [CI], 0.540.93).13 A 2004 prospective quasi‐experimental observational study done at an academic teaching hospital showed an adjusted LOS that was 16.2% lower, and adjusted cost 9.7% lower, for patients on the hospitalists' service.14 In 2007, Lindenauer and colleagues found that a national sample of hospitalists decreased LOS by 0.4 days and lowered cost by $286 per patient.15 The findings of these individual studies were supported in a 2009 systematic review of 33 studies by Peterson which showed that hospitalists decrease LOS.16 In a recent study, Kuo and Goodwin showed that while hospitalists decrease LOS and cost, the patients they care for have higher Medicare costs after discharge by $322 per patient, and are more likely to be readmitted (odds ratio, 1.08; CI, 1.041.14).17

The hospitalist model of care continues to grow, and hospitalists will soon number as many as 30,000.18 For acute medical inpatients, the evidence suggests that hospitalists improve patient flow by decreasing LOS while improving other aspects of quality of care. However, Kuo and Goodwin's findings suggest that the transition of care from inpatient to outpatient settings still requires attention.17

SURGICAL COMANAGEMENT

The Society of Hospital Medicine (SHM) core competencies include perioperative medicine.19, 20 In the 2006 SHM national survey, 85% of hospital medicine groups indicated that they participated in surgical comanagement.21

Hospitalists have improved patient flow and outcomes for orthopedic patients. Hospitalist management of hip fracture patients decreases time to surgery and LOS compared to standard care.2224 Phy and colleagues studied 466 patients for 2 years after the inception of hospital medicine comanagement of surgical patients, and found that care by hospitalists decreased LOS by 2.2 days.22 In a retrospective study of 118 patients, Roy and colleagues found that hospitalist‐managed patients had shorter time to consultation and surgery, decreased LOS, and lower costs.23 In a retrospective cohort study, Batsis looked at mortality in 466 patients with hip fracture, and found no difference between hospitalist management and standard care.24 In patients undergoing elective hip and knee arthroplasty, Huddleston and colleagues reported that patients managed by hospitalists had fewer complications and shorter LOS. The nurses and orthopedic surgeons preferred the hospitalistorthopedist comanagement model.25

The benefits of hospitalist comanagement are not limited to adult patients undergoing orthopedic surgery. For high‐risk patients undergoing lower extremity reconstruction surgery, Pinzur and colleagues noted that LOS was shorter for a cohort of patients managed by hospitalists than for a group of historical controls not treated by hospitalists.26 Simon and colleagues studied comanagement for pediatric spinal fusion patients, and found a decrease in LOS from 6.5 to 4.8 days.27

Several factors should be considered in developing and implementing a successful comanagement program. Since comanagement duties may fall upon hospitalists in order to protect surgeons' time,28 hospital medicine groups should ensure adequate staffing prior to taking on additional services. Clear guidelines to delineate roles and responsibilities of the comanaging groups also need to be developed and implemented.29, 30

Comanaging may also involve additional training. Hospitalists who manage neurologic, neurosurgical, trauma, and psychiatric patients report being undertrained for such conditions.31, 32 Hospital medicine groups need to ensure training needs are met and supported. Given the successes of comanagement and the increasing complexity of surgical patients,33 this practice will likely expand to a greater variety of non‐medical patients.

SHORT STAY UNITS

In 2003, short stay units (SSU) were present in approximately 20% of US hospitals, with 11% of hospitals planning on opening one in the next year.34 SSU are designed to manage acute, self‐limited medical conditions that require brief staysusually less than 72 hours. Approximately 80% of SSU patients are discharged home, avoiding hospitalization.35 Historically, SSU have been under the domain of the ED; however, there is an emerging role for hospitalist‐run SSU.36

Despite demand for SSU, little research has been performed on hospitalist‐led SSU. In 2000, Abenhaim and colleagues showed that a hospitalist‐run SSU at a university‐affiliated teaching hospital had a shorter LOS and lower rates of complications and readmissions when compared to medicine teaching services.37 In 2008, Northwestern Memorial Hospital opened a 30‐bed hospitalist‐run SSU; for those patients, LOS decreased by 2 days.38 In 2010, Leykum and colleagues showed that a hospitalist‐run observation unit can decrease LOS from 2.4 days to 2.2 days.39 Careful selection of SSU patients is needed to obtain these results. Lucas and colleagues found that whether or not SSU patients required assistance of specialists was the strongest predictor of unsuccessful stays (>72 hours or inpatient conversion) in SSU.36

Whether SSU are run by hospital medicine or emergency medicine is decided at an institutional level. Location of SSU in a specifically designated area is crucial, as it allows physicians to round efficiently on patients and to work with staff trained in observation services. Development of admission criteria that include specific diagnoses which match hospitalists' scope of practice is also important (Table 1).32

The protocol‐based and diagnosis‐specific nature of SSU may enhance quality of care through standardization. Future research may delineate the utility of SSU.

CHEST PAIN UNITS

In the United States, in 2004, approximately 6 million patients present annually to EDs with chest pain.40 Cost of care of patients unnecessarily admitted to coronary care units has been estimated to be nearly $3 billion annually.41 Still, as many as 3% of patients with acute myocardial infarction are discharged home.42 Chest pain units (CPU) were developed to facilitate evaluation of patients with chest pain, at low risk for acute coronary syndrome, without requiring inpatient admission. A number of studies have suggested that admission to a CPU is a safe and cost‐effective alternative to hospital admission.4348

CPU have traditionally been staffed by ED physicians and/or cardiologists. In a prepost study, Krantz and colleagues found that a CPU model, incorporating hospitalists at an academic public safety‐net hospital, decreased ED LOS with no difference in 30‐day cardiac event rate.49 Myers and colleagues created a hospitalist‐directed nonteaching service in an academic medical center to admit low‐risk chest pain patients. Patients admitted to the hospitalist service had a statistically significant lower median LOS (23 hours vs 33 hours) and approximately half the median hospital charges than those admitted to teaching services.50 At the same academic medical center, Bayley and colleagues showed that 91% of patients admitted for chest pain waited more than 3 hours for a bed. This adversely affected ED revenue by tying up beds, resulting in an estimated annual loss of $168,300 of hospital revenue. Creation of a hospitalist‐managed service for low‐acuity chest pain patients reduced hospital LOS by 7 hours.51 Somekh and colleagues demonstrated that a protocol‐driven, cardiologist‐run CPU results in a decreased LOS and readmission rate compared to usual care.52 In a non‐peer reviewed case study, Cox Health opened an 8‐bed, hospitalist‐led CPU in 2003. They decreased LOS from 72 to 18 hours, while increasing revenue by $2.5 million a year.53 These studies suggest that hospitalist‐run CPU can decrease LOS, increase revenue, and relieve ED overcrowding.

Development of a successful CPU depends upon clear inclusion/exclusion criteria; close collaboration among ED physicians, hospitalists, and cardiologists; the development of evidence‐based protocols, and the availability of stress testing.

ACTIVE BED MANAGEMENT

As of 2007, 90% of EDs were crowded beyond their capacity.2 ED crowding leads to ambulance diversion,54 which can delay care and increase mortality rates.55 One of the main causes of ED crowding is the boarding of admitted patients.56 Boarded, admitted patients have been shown to have decreased quality of care and patient satisfaction.35

Active bed management (ABM) by hospitalists can decrease ED diversion. Howell and colleagues instituted ABM where hospitalists, as active bed managers, facilitate placement of patients to their inpatient destinations to assist ED flow.57 This 24‐hour, hospitalist‐led, active bed management service decreased both ED LOS and ambulance diversion. The bed manager collaborated real‐time with medicine and ED attending physicians, nursing supervisors, and charge nurses to change patient care status, and assign and facilitate transfer of patients to appropriate units. These hospitalist bed managers were also empowered to activate additional resources when pre‐diversion rounds identified resource limitations and impending ED divert. They found overall ED LOS for admitted patients decreased by 98 minutes, while LOS for non‐admitted patients stayed the same. AMB decreased diversion due to critically ill and telemetry patients by 28% (786 hours), and diversion due to lower acuity patients by 6% (182 hours). This intervention proved cost‐effective. Three full‐time equivalent (FTE) hospitalists' salaries staff 1 active bed manager working 24/7. Nearly 1000 hours of diversion were avoided at an annual savings of $1086 per hour of diversion decreased.

ABM is a new frontier for hospitals in general, and hospitalists in particular. Chadaga and colleagues found that a hospital medicine‐ED team participating in active bed management, while caring for admitted patients boarded in the ED, can decrease ED diversion and improve patient flow. The percentage of patients transferred to a medicine floor and discharged within 8 hours was reduced by 67% (P < 0.01), while the number of discharges from the ED of admitted medicine patients increased by 61% (P < 0.001).58

To decrease initial investment, components of ABM (ED triage, bed assignment, discharge facilitation) can be instituted in parts. Hospital medicine groups with limited resources may only provide a triage service by phone for difficult ED cases. Bedside evaluations and collaboration with nursing staff to improve bed placement may be a next step, with floor and/or intensive care unit (ICU) rounds to facilitate early discharges as a final component.

OTHER AREAS

CONCLUSION

While many hospital medicine groups were started to provide acute inpatient medical care, most have found that their value to hospitals reaches beyond bedside care. With an epidemic of ED diversion and lack of access to hospital beds and services, optimizing throughput has become imperative for hospital systems. While hospital access can be improved with addition of new beds, improving throughput by decreasing LOS maximizes utilization of existing resources.

We have reviewed how hospitalists improve patient flow in acute inpatient medicine, surgical comanagement, short stay units, chest pain units, and active bed management. In each instance, the literature supports measures for decreasing LOS while maintaining or improving quality of care. Hinami and colleagues showed physician satisfaction with hospitalist‐provided patient care.75 Most studies have been limited by tracking upstream effects of improved efficiency. As there is now some evidence that decreasing LOS may increase readmissions,17 future studies should incorporate this metric into their outcomes. The effect of formal operations management principles on patient flow and bed efficiency is not well known and should be further examined.

In addition, we have touched on other areas (perioperative clinics, palliative care, geographic rounding, and flexible staffing models) where hospitalists may impact patient throughput. These areas represent excellent opportunities for future research.

Hospitalist participation in many of these areas is in its infancy. Hospital medicine programs interested in expanding their services, beyond acute inpatient care, have the opportunity to develop standards and continue research on the effect of hospital medicine‐led services on patient care and flow.

Hospitalists are physicians whose primary focus is the general medical care of hospitalized patients. Hospitalists are uniquely positioned to implement strategies to improve patient flow and efficiency.1 With emergency department (ED) diversion reaching rates upward of 70%, lack of access to inpatient beds leads to delayed care with worsened outcomes.25

To improve access to hospital beds, hospitals may increase capacity by either adding beds or by more efficiently using existing beds. Operations management principles have been applied to healthcare to ensure efficient use of beds. These include: reducing variability of scheduled admissions, remeasuring length of stay (LOS) and bed demand after implementing strategies to reduce practice variation, and employing queuing theory to generate predictions of optimal beds needed.6 The Joint Commission implemented a leadership standard (LD 04.03.11) that hospitals develop and implement plans to identify and mitigate impediments to efficient patient flow through the hospital.

To improve access, hospital leaders expect hospitalists to staff in inpatient medicine programs, surgical comanagement, short stay and chest pain units, and active bed management.7 In the following review, we define hospitalists' roles in the aforementioned programs and their effect on patient flow. We also touch on preoperative clinics, palliative care, geographic rounding, and flexible staffing models.

ACUTE INPATIENT CARE

Hospitalists are one of the fastest growing physician groups in the United States.810 Hospitalists improve efficiency and quality of care across a variety of demographic, geographic, and healthcare settings.11, 12 A 2002 retrospective cohort study in a community‐based urban teaching hospital showed that hospitalists decreased LOS by 0.61 days and lowered risk for death in the hospital (adjusted relative hazard, 0.71; 95% confidence interval [CI], 0.540.93).13 A 2004 prospective quasi‐experimental observational study done at an academic teaching hospital showed an adjusted LOS that was 16.2% lower, and adjusted cost 9.7% lower, for patients on the hospitalists' service.14 In 2007, Lindenauer and colleagues found that a national sample of hospitalists decreased LOS by 0.4 days and lowered cost by $286 per patient.15 The findings of these individual studies were supported in a 2009 systematic review of 33 studies by Peterson which showed that hospitalists decrease LOS.16 In a recent study, Kuo and Goodwin showed that while hospitalists decrease LOS and cost, the patients they care for have higher Medicare costs after discharge by $322 per patient, and are more likely to be readmitted (odds ratio, 1.08; CI, 1.041.14).17

The hospitalist model of care continues to grow, and hospitalists will soon number as many as 30,000.18 For acute medical inpatients, the evidence suggests that hospitalists improve patient flow by decreasing LOS while improving other aspects of quality of care. However, Kuo and Goodwin's findings suggest that the transition of care from inpatient to outpatient settings still requires attention.17

SURGICAL COMANAGEMENT

The Society of Hospital Medicine (SHM) core competencies include perioperative medicine.19, 20 In the 2006 SHM national survey, 85% of hospital medicine groups indicated that they participated in surgical comanagement.21

Hospitalists have improved patient flow and outcomes for orthopedic patients. Hospitalist management of hip fracture patients decreases time to surgery and LOS compared to standard care.2224 Phy and colleagues studied 466 patients for 2 years after the inception of hospital medicine comanagement of surgical patients, and found that care by hospitalists decreased LOS by 2.2 days.22 In a retrospective study of 118 patients, Roy and colleagues found that hospitalist‐managed patients had shorter time to consultation and surgery, decreased LOS, and lower costs.23 In a retrospective cohort study, Batsis looked at mortality in 466 patients with hip fracture, and found no difference between hospitalist management and standard care.24 In patients undergoing elective hip and knee arthroplasty, Huddleston and colleagues reported that patients managed by hospitalists had fewer complications and shorter LOS. The nurses and orthopedic surgeons preferred the hospitalistorthopedist comanagement model.25

The benefits of hospitalist comanagement are not limited to adult patients undergoing orthopedic surgery. For high‐risk patients undergoing lower extremity reconstruction surgery, Pinzur and colleagues noted that LOS was shorter for a cohort of patients managed by hospitalists than for a group of historical controls not treated by hospitalists.26 Simon and colleagues studied comanagement for pediatric spinal fusion patients, and found a decrease in LOS from 6.5 to 4.8 days.27

Several factors should be considered in developing and implementing a successful comanagement program. Since comanagement duties may fall upon hospitalists in order to protect surgeons' time,28 hospital medicine groups should ensure adequate staffing prior to taking on additional services. Clear guidelines to delineate roles and responsibilities of the comanaging groups also need to be developed and implemented.29, 30

Comanaging may also involve additional training. Hospitalists who manage neurologic, neurosurgical, trauma, and psychiatric patients report being undertrained for such conditions.31, 32 Hospital medicine groups need to ensure training needs are met and supported. Given the successes of comanagement and the increasing complexity of surgical patients,33 this practice will likely expand to a greater variety of non‐medical patients.

SHORT STAY UNITS

In 2003, short stay units (SSU) were present in approximately 20% of US hospitals, with 11% of hospitals planning on opening one in the next year.34 SSU are designed to manage acute, self‐limited medical conditions that require brief staysusually less than 72 hours. Approximately 80% of SSU patients are discharged home, avoiding hospitalization.35 Historically, SSU have been under the domain of the ED; however, there is an emerging role for hospitalist‐run SSU.36

Despite demand for SSU, little research has been performed on hospitalist‐led SSU. In 2000, Abenhaim and colleagues showed that a hospitalist‐run SSU at a university‐affiliated teaching hospital had a shorter LOS and lower rates of complications and readmissions when compared to medicine teaching services.37 In 2008, Northwestern Memorial Hospital opened a 30‐bed hospitalist‐run SSU; for those patients, LOS decreased by 2 days.38 In 2010, Leykum and colleagues showed that a hospitalist‐run observation unit can decrease LOS from 2.4 days to 2.2 days.39 Careful selection of SSU patients is needed to obtain these results. Lucas and colleagues found that whether or not SSU patients required assistance of specialists was the strongest predictor of unsuccessful stays (>72 hours or inpatient conversion) in SSU.36

Whether SSU are run by hospital medicine or emergency medicine is decided at an institutional level. Location of SSU in a specifically designated area is crucial, as it allows physicians to round efficiently on patients and to work with staff trained in observation services. Development of admission criteria that include specific diagnoses which match hospitalists' scope of practice is also important (Table 1).32

The protocol‐based and diagnosis‐specific nature of SSU may enhance quality of care through standardization. Future research may delineate the utility of SSU.

CHEST PAIN UNITS

In the United States, in 2004, approximately 6 million patients present annually to EDs with chest pain.40 Cost of care of patients unnecessarily admitted to coronary care units has been estimated to be nearly $3 billion annually.41 Still, as many as 3% of patients with acute myocardial infarction are discharged home.42 Chest pain units (CPU) were developed to facilitate evaluation of patients with chest pain, at low risk for acute coronary syndrome, without requiring inpatient admission. A number of studies have suggested that admission to a CPU is a safe and cost‐effective alternative to hospital admission.4348

CPU have traditionally been staffed by ED physicians and/or cardiologists. In a prepost study, Krantz and colleagues found that a CPU model, incorporating hospitalists at an academic public safety‐net hospital, decreased ED LOS with no difference in 30‐day cardiac event rate.49 Myers and colleagues created a hospitalist‐directed nonteaching service in an academic medical center to admit low‐risk chest pain patients. Patients admitted to the hospitalist service had a statistically significant lower median LOS (23 hours vs 33 hours) and approximately half the median hospital charges than those admitted to teaching services.50 At the same academic medical center, Bayley and colleagues showed that 91% of patients admitted for chest pain waited more than 3 hours for a bed. This adversely affected ED revenue by tying up beds, resulting in an estimated annual loss of $168,300 of hospital revenue. Creation of a hospitalist‐managed service for low‐acuity chest pain patients reduced hospital LOS by 7 hours.51 Somekh and colleagues demonstrated that a protocol‐driven, cardiologist‐run CPU results in a decreased LOS and readmission rate compared to usual care.52 In a non‐peer reviewed case study, Cox Health opened an 8‐bed, hospitalist‐led CPU in 2003. They decreased LOS from 72 to 18 hours, while increasing revenue by $2.5 million a year.53 These studies suggest that hospitalist‐run CPU can decrease LOS, increase revenue, and relieve ED overcrowding.

Development of a successful CPU depends upon clear inclusion/exclusion criteria; close collaboration among ED physicians, hospitalists, and cardiologists; the development of evidence‐based protocols, and the availability of stress testing.

ACTIVE BED MANAGEMENT

As of 2007, 90% of EDs were crowded beyond their capacity.2 ED crowding leads to ambulance diversion,54 which can delay care and increase mortality rates.55 One of the main causes of ED crowding is the boarding of admitted patients.56 Boarded, admitted patients have been shown to have decreased quality of care and patient satisfaction.35

Active bed management (ABM) by hospitalists can decrease ED diversion. Howell and colleagues instituted ABM where hospitalists, as active bed managers, facilitate placement of patients to their inpatient destinations to assist ED flow.57 This 24‐hour, hospitalist‐led, active bed management service decreased both ED LOS and ambulance diversion. The bed manager collaborated real‐time with medicine and ED attending physicians, nursing supervisors, and charge nurses to change patient care status, and assign and facilitate transfer of patients to appropriate units. These hospitalist bed managers were also empowered to activate additional resources when pre‐diversion rounds identified resource limitations and impending ED divert. They found overall ED LOS for admitted patients decreased by 98 minutes, while LOS for non‐admitted patients stayed the same. AMB decreased diversion due to critically ill and telemetry patients by 28% (786 hours), and diversion due to lower acuity patients by 6% (182 hours). This intervention proved cost‐effective. Three full‐time equivalent (FTE) hospitalists' salaries staff 1 active bed manager working 24/7. Nearly 1000 hours of diversion were avoided at an annual savings of $1086 per hour of diversion decreased.

ABM is a new frontier for hospitals in general, and hospitalists in particular. Chadaga and colleagues found that a hospital medicine‐ED team participating in active bed management, while caring for admitted patients boarded in the ED, can decrease ED diversion and improve patient flow. The percentage of patients transferred to a medicine floor and discharged within 8 hours was reduced by 67% (P < 0.01), while the number of discharges from the ED of admitted medicine patients increased by 61% (P < 0.001).58

To decrease initial investment, components of ABM (ED triage, bed assignment, discharge facilitation) can be instituted in parts. Hospital medicine groups with limited resources may only provide a triage service by phone for difficult ED cases. Bedside evaluations and collaboration with nursing staff to improve bed placement may be a next step, with floor and/or intensive care unit (ICU) rounds to facilitate early discharges as a final component.

OTHER AREAS

CONCLUSION

While many hospital medicine groups were started to provide acute inpatient medical care, most have found that their value to hospitals reaches beyond bedside care. With an epidemic of ED diversion and lack of access to hospital beds and services, optimizing throughput has become imperative for hospital systems. While hospital access can be improved with addition of new beds, improving throughput by decreasing LOS maximizes utilization of existing resources.

We have reviewed how hospitalists improve patient flow in acute inpatient medicine, surgical comanagement, short stay units, chest pain units, and active bed management. In each instance, the literature supports measures for decreasing LOS while maintaining or improving quality of care. Hinami and colleagues showed physician satisfaction with hospitalist‐provided patient care.75 Most studies have been limited by tracking upstream effects of improved efficiency. As there is now some evidence that decreasing LOS may increase readmissions,17 future studies should incorporate this metric into their outcomes. The effect of formal operations management principles on patient flow and bed efficiency is not well known and should be further examined.

In addition, we have touched on other areas (perioperative clinics, palliative care, geographic rounding, and flexible staffing models) where hospitalists may impact patient throughput. These areas represent excellent opportunities for future research.

Hospitalist participation in many of these areas is in its infancy. Hospital medicine programs interested in expanding their services, beyond acute inpatient care, have the opportunity to develop standards and continue research on the effect of hospital medicine‐led services on patient care and flow.

Anorexia nervosa occurs in 0.9% of women and 0.3% of men in the United States1 and is associated with a prolonged course,2 extensive medical complications that can affect almost every organ system,3, 4 and a 5% mean crude mortality rate9.6 times expected for age‐matched women in the United States.2, 5 Those with anorexia nervosa die as a complication of their illness more frequently than any other mental illness.3 Anorexia nervosa is commonly diagnosed during the adolescent years,2 with almost 25% going on to develop chronic anorexia nervosa.2, 6 Consequently, many patients with severe anorexia nervosa will receive treatment by adult medicine practitioners.

Patients with anorexia nervosa frequently require hospitalization. Published guidelines suggest that those who are 70% or less than ideal body weight, bradycardic, hypotensive, or those with severe electrolyte disturbances warrant admission for medical stabilization.79 Once admitted, however, there are no published guidelines for best practices to medically stabilize patients.7, 10 Although most experts advocate a multidisciplinary approach with weight restoration and medical stability as the goals of hospital admission,8, 9 controversy exists in the literature about how best to achieve these goals.7, 10

It is known, however, that for patients with complicated medical illnesses, such as human immunodeficiency virus (HIV) and sepsis, higher volumes of patient caseloads treated by physicians with disease‐specific expertise has been found to lead to improved outcomes in patients.11, 12 The adult patient with severe anorexia nervosa who requires inpatient medical stabilization may also benefit from a multidisciplinary trained staff familiar with the medical management of anorexia nervosa. Accordingly, we have developed the Acute Comprehensive Urgent Treatment for Eating Disorders (ACUTE) Center.

PROGRAM DESCRIPTION

The ACUTE Center at Denver Health is a 5‐bed unit dedicated to the medical stabilization of patients with severe malnutrition due to anorexia nervosa or severe electrolyte disorders due to bulimia nervosa. ACUTE accepts patients 17 years and older with medical complications related to chronic malnutrition and refeeding.

ACUTE uses a multidisciplinary approach to patient care. The physician team is composed of a hospital medicine attending physician, consultative expertise by an internal medicine specialist in the management of the medical complications of eating disorders, and a psychiatrist specializing in eating disorders. There is a dedicated team of nurses, two dieticians, physical therapists, certified nursing assistants, speech therapists, a psychotherapist, and a chaplain.

ACUTE patients are on continuous telemetry monitoring for the duration of their hospitalization to monitor for arrhythmias as well as signs of covert exercise. As part of the initial intake, a full set of vital signs is obtained, including height and weight. Patients are weighed daily with their back to the scale. There is no discussion of weight fluctuations. Patients may walk at a slow pace around the unit. No exercise is allowed.

Each patient at the ACUTE Center has an individualized meal plan and are started on an oral caloric intake 200 kcal below their basal energy expenditure (BEE). Indirect calorimetry is performed on the first hospital day. Each patient meets on a daily basis with the registered dietician to choose meals that meet their caloric goals.

All patients have a sitter continuously for their first week, and thereafter sitter time may be reduced to supervision surrounding each meal. Patients who fail to finish their prescribed meal are required to drink a liquid supplement to meet caloric goals. Calories are increased weekly until the patient's weight shows a clear pattern of weight increase. 0

Figure 1

Patients are discharged from the ACUTE Center when they have achieved several basic goals: They are consuming greater than 2000 kcal per day, they are consistently gaining 23 pounds per week, their laboratory values have stabilized without electrolyte supplementation, and they are strong enough for an inpatient eating disorder program.

METHODS

Patients admitted to the ACUTE Center between October 2008 and December 2010 for medical stabilization and monitored refeeding were included. Patients with a diagnosis of bulimia nervosa were excluded. Demographic data and laboratory results were obtained electronically from our data repository, whereas weight, height, and other clinical characteristics were obtained by manual chart abstraction. The statistical analysis was conducted in SAS Enterprise Guide v4.1 (SAS Institute, Cary, NC).

RESULTS

In its first 27 months, the ACUTE Center had 76 total admissions, comprising 59 patients. Of the 76 admissions, the 62 admissions for medical stabilization and monitored refeeding of 54 patients with anorexia nervosa were included. Forty‐eight of the 54 (89%) included patients were female. Six patients were hospitalized twice, and 1 patient 3 times. There were 3 transfers to the intensive care unit, and no inpatient mortality. Of the 62 admissions, 11 (18%) discharges were to home, and 51 (82%) were to inpatient psychiatric eating disorder units.

The mean age at admission was 27 years (range 1765 years). The mean percent of ideal body weight (IBW) on admission was 62.2% 10.2%. The mean body mass index (BMI) was 12.9 2.0 kg/m² on admission, and 13.1 1.9 kg/m² upon discharge. The median length of stay was 16 days (interquartile range [IQR] 929 days). Median calculated BEE (1119 [10671184 IQR]) was higher than measured BEE by indirect calorimetry (792 [6341094]), (Table 1).

The majority of admission laboratory values, including serum albumin, blood urea nitrogen (BUN), creatinine, potassium, magnesium, and phosphate levels, were within normal limits. Fifty‐six percent were hyponatremic at admission, with a mean serum sodium level of 133 6 mmol/L (Table 2).

DISCUSSION

Hospital Medicine is currently the fastest growing area of specialization in medicine.13 Palliative care, inpatient geriatrics, short stay units, and bedside procedures have evolved into hospitalist‐led services.1418 The management of the medical complications of severe eating disorders is another potential niche for hospitalists.

The ACUTE Center at Denver Health represents a center in which highly specialized, multidisciplinary care is provided for a rare and extremely ill population of patients. Prior to entering the ACUTE Center, the patients described in our program had each experienced prolonged and unsuccessful stays for medical stabilization in acute care hospitals across the country, after being denied treatment in eating disorder programs due to medical instability.

Patients transferred to ACUTE often received medical care reflecting a lack of specific expertise, training, and exposure. The most common management discrepancy we noted was over‐aggressive provision of intravenous fluids. Consequently, we often diurese 1020 pounds of edema weight, gained during a prior medical hospitalization, before beginning the process of weight restoration. This edema weight artificially increases admission weight and results in less than expected weight gain from admission to discharge.

Even without substantial weight gain, medical stabilization is evidenced by consistent caloric oral intake, and fluid and electrolyte stabilization after initial refeeding. Accordingly, patients who have been treated at the ACUTE Center often become eligible for admission to eating disorder programs at body weights below the typical 70% of ideal body weight that most programs use as a threshold for admission.

From a clinical research perspective, centers such as ACUTE allow for opportunities to better understand and investigate the nuances of patient care in the setting of severe malnutrition. From our cohort of patients to date, we have noted unique issues in albumin levels,19 coagulopathy,20 and liver function,21 among others. As an example, the cohort of patients with anorexia nervosa described here had profoundly low body weight, but relatively normal admission labs. Even the serum albumin, a parameter often used to reflect nutrition in an adult internal medicine setting, is usually normal, reflecting, in an otherwise generally healthy young population, the absence of a malignant, inflammatory, or infectious etiology of weight loss.19

Hospitalists also advocate for their patients by helping to maximize the benefits of their health care coverage. Many health care plans place limits on inpatient psychiatric care benefits. Patients who are severely malnourished from their eating disorder may waste valuable psychiatric care benefits undergoing medical stabilization in psychiatric units while physically unable to undergo psychotherapy. This has become increasingly important as health insurance plans continue to decrease coverage for residential care of patients with anorexia.22

In contrast, the medical benefits of most health plans are more robust. Accordingly, from the patient perspective, medical stabilization in an acute medical unit before admission to a psychiatry unit maximizes their ability to participate in the intensive psychiatric therapy which is still needed after medical stabilization. A recent study from a residential eating disorder program confirmed that a higher discharge BMI was the single best predictor of full recovery from anorexia nervosa.23

In the future, we believe that a continuing concentration of care and experience may also lend itself to the development of protocols and management guidelines which may benefit patients beyond our own unit. Severely malnourished patients with anorexia nervosa, or bulimic patients with complicated electrolyte disorders, are likely to benefit both medically and financially from centers of excellence. Inpatient or residential psychiatric eating disorder programs may act in synergy with medical eating disorders units, like ACUTE, to most efficiently care for the severely malnourished patient. Hospitalists, with the proper training and experience, are uniquely positioned to develop such centers of excellence.

Anorexia nervosa occurs in 0.9% of women and 0.3% of men in the United States1 and is associated with a prolonged course,2 extensive medical complications that can affect almost every organ system,3, 4 and a 5% mean crude mortality rate9.6 times expected for age‐matched women in the United States.2, 5 Those with anorexia nervosa die as a complication of their illness more frequently than any other mental illness.3 Anorexia nervosa is commonly diagnosed during the adolescent years,2 with almost 25% going on to develop chronic anorexia nervosa.2, 6 Consequently, many patients with severe anorexia nervosa will receive treatment by adult medicine practitioners.

Patients with anorexia nervosa frequently require hospitalization. Published guidelines suggest that those who are 70% or less than ideal body weight, bradycardic, hypotensive, or those with severe electrolyte disturbances warrant admission for medical stabilization.79 Once admitted, however, there are no published guidelines for best practices to medically stabilize patients.7, 10 Although most experts advocate a multidisciplinary approach with weight restoration and medical stability as the goals of hospital admission,8, 9 controversy exists in the literature about how best to achieve these goals.7, 10

It is known, however, that for patients with complicated medical illnesses, such as human immunodeficiency virus (HIV) and sepsis, higher volumes of patient caseloads treated by physicians with disease‐specific expertise has been found to lead to improved outcomes in patients.11, 12 The adult patient with severe anorexia nervosa who requires inpatient medical stabilization may also benefit from a multidisciplinary trained staff familiar with the medical management of anorexia nervosa. Accordingly, we have developed the Acute Comprehensive Urgent Treatment for Eating Disorders (ACUTE) Center.

PROGRAM DESCRIPTION

The ACUTE Center at Denver Health is a 5‐bed unit dedicated to the medical stabilization of patients with severe malnutrition due to anorexia nervosa or severe electrolyte disorders due to bulimia nervosa. ACUTE accepts patients 17 years and older with medical complications related to chronic malnutrition and refeeding.

ACUTE uses a multidisciplinary approach to patient care. The physician team is composed of a hospital medicine attending physician, consultative expertise by an internal medicine specialist in the management of the medical complications of eating disorders, and a psychiatrist specializing in eating disorders. There is a dedicated team of nurses, two dieticians, physical therapists, certified nursing assistants, speech therapists, a psychotherapist, and a chaplain.

ACUTE patients are on continuous telemetry monitoring for the duration of their hospitalization to monitor for arrhythmias as well as signs of covert exercise. As part of the initial intake, a full set of vital signs is obtained, including height and weight. Patients are weighed daily with their back to the scale. There is no discussion of weight fluctuations. Patients may walk at a slow pace around the unit. No exercise is allowed.

Each patient at the ACUTE Center has an individualized meal plan and are started on an oral caloric intake 200 kcal below their basal energy expenditure (BEE). Indirect calorimetry is performed on the first hospital day. Each patient meets on a daily basis with the registered dietician to choose meals that meet their caloric goals.

All patients have a sitter continuously for their first week, and thereafter sitter time may be reduced to supervision surrounding each meal. Patients who fail to finish their prescribed meal are required to drink a liquid supplement to meet caloric goals. Calories are increased weekly until the patient's weight shows a clear pattern of weight increase. 0

Figure 1

Patients are discharged from the ACUTE Center when they have achieved several basic goals: They are consuming greater than 2000 kcal per day, they are consistently gaining 23 pounds per week, their laboratory values have stabilized without electrolyte supplementation, and they are strong enough for an inpatient eating disorder program.

METHODS

Patients admitted to the ACUTE Center between October 2008 and December 2010 for medical stabilization and monitored refeeding were included. Patients with a diagnosis of bulimia nervosa were excluded. Demographic data and laboratory results were obtained electronically from our data repository, whereas weight, height, and other clinical characteristics were obtained by manual chart abstraction. The statistical analysis was conducted in SAS Enterprise Guide v4.1 (SAS Institute, Cary, NC).

RESULTS

In its first 27 months, the ACUTE Center had 76 total admissions, comprising 59 patients. Of the 76 admissions, the 62 admissions for medical stabilization and monitored refeeding of 54 patients with anorexia nervosa were included. Forty‐eight of the 54 (89%) included patients were female. Six patients were hospitalized twice, and 1 patient 3 times. There were 3 transfers to the intensive care unit, and no inpatient mortality. Of the 62 admissions, 11 (18%) discharges were to home, and 51 (82%) were to inpatient psychiatric eating disorder units.

The mean age at admission was 27 years (range 1765 years). The mean percent of ideal body weight (IBW) on admission was 62.2% 10.2%. The mean body mass index (BMI) was 12.9 2.0 kg/m² on admission, and 13.1 1.9 kg/m² upon discharge. The median length of stay was 16 days (interquartile range [IQR] 929 days). Median calculated BEE (1119 [10671184 IQR]) was higher than measured BEE by indirect calorimetry (792 [6341094]), (Table 1).

The majority of admission laboratory values, including serum albumin, blood urea nitrogen (BUN), creatinine, potassium, magnesium, and phosphate levels, were within normal limits. Fifty‐six percent were hyponatremic at admission, with a mean serum sodium level of 133 6 mmol/L (Table 2).

DISCUSSION

Hospital Medicine is currently the fastest growing area of specialization in medicine.13 Palliative care, inpatient geriatrics, short stay units, and bedside procedures have evolved into hospitalist‐led services.1418 The management of the medical complications of severe eating disorders is another potential niche for hospitalists.

The ACUTE Center at Denver Health represents a center in which highly specialized, multidisciplinary care is provided for a rare and extremely ill population of patients. Prior to entering the ACUTE Center, the patients described in our program had each experienced prolonged and unsuccessful stays for medical stabilization in acute care hospitals across the country, after being denied treatment in eating disorder programs due to medical instability.

Patients transferred to ACUTE often received medical care reflecting a lack of specific expertise, training, and exposure. The most common management discrepancy we noted was over‐aggressive provision of intravenous fluids. Consequently, we often diurese 1020 pounds of edema weight, gained during a prior medical hospitalization, before beginning the process of weight restoration. This edema weight artificially increases admission weight and results in less than expected weight gain from admission to discharge.

Even without substantial weight gain, medical stabilization is evidenced by consistent caloric oral intake, and fluid and electrolyte stabilization after initial refeeding. Accordingly, patients who have been treated at the ACUTE Center often become eligible for admission to eating disorder programs at body weights below the typical 70% of ideal body weight that most programs use as a threshold for admission.

From a clinical research perspective, centers such as ACUTE allow for opportunities to better understand and investigate the nuances of patient care in the setting of severe malnutrition. From our cohort of patients to date, we have noted unique issues in albumin levels,19 coagulopathy,20 and liver function,21 among others. As an example, the cohort of patients with anorexia nervosa described here had profoundly low body weight, but relatively normal admission labs. Even the serum albumin, a parameter often used to reflect nutrition in an adult internal medicine setting, is usually normal, reflecting, in an otherwise generally healthy young population, the absence of a malignant, inflammatory, or infectious etiology of weight loss.19

Hospitalists also advocate for their patients by helping to maximize the benefits of their health care coverage. Many health care plans place limits on inpatient psychiatric care benefits. Patients who are severely malnourished from their eating disorder may waste valuable psychiatric care benefits undergoing medical stabilization in psychiatric units while physically unable to undergo psychotherapy. This has become increasingly important as health insurance plans continue to decrease coverage for residential care of patients with anorexia.22

In contrast, the medical benefits of most health plans are more robust. Accordingly, from the patient perspective, medical stabilization in an acute medical unit before admission to a psychiatry unit maximizes their ability to participate in the intensive psychiatric therapy which is still needed after medical stabilization. A recent study from a residential eating disorder program confirmed that a higher discharge BMI was the single best predictor of full recovery from anorexia nervosa.23

In the future, we believe that a continuing concentration of care and experience may also lend itself to the development of protocols and management guidelines which may benefit patients beyond our own unit. Severely malnourished patients with anorexia nervosa, or bulimic patients with complicated electrolyte disorders, are likely to benefit both medically and financially from centers of excellence. Inpatient or residential psychiatric eating disorder programs may act in synergy with medical eating disorders units, like ACUTE, to most efficiently care for the severely malnourished patient. Hospitalists, with the proper training and experience, are uniquely positioned to develop such centers of excellence.

Anorexia nervosa occurs in 0.9% of women and 0.3% of men in the United States1 and is associated with a prolonged course,2 extensive medical complications that can affect almost every organ system,3, 4 and a 5% mean crude mortality rate9.6 times expected for age‐matched women in the United States.2, 5 Those with anorexia nervosa die as a complication of their illness more frequently than any other mental illness.3 Anorexia nervosa is commonly diagnosed during the adolescent years,2 with almost 25% going on to develop chronic anorexia nervosa.2, 6 Consequently, many patients with severe anorexia nervosa will receive treatment by adult medicine practitioners.

Patients with anorexia nervosa frequently require hospitalization. Published guidelines suggest that those who are 70% or less than ideal body weight, bradycardic, hypotensive, or those with severe electrolyte disturbances warrant admission for medical stabilization.79 Once admitted, however, there are no published guidelines for best practices to medically stabilize patients.7, 10 Although most experts advocate a multidisciplinary approach with weight restoration and medical stability as the goals of hospital admission,8, 9 controversy exists in the literature about how best to achieve these goals.7, 10

It is known, however, that for patients with complicated medical illnesses, such as human immunodeficiency virus (HIV) and sepsis, higher volumes of patient caseloads treated by physicians with disease‐specific expertise has been found to lead to improved outcomes in patients.11, 12 The adult patient with severe anorexia nervosa who requires inpatient medical stabilization may also benefit from a multidisciplinary trained staff familiar with the medical management of anorexia nervosa. Accordingly, we have developed the Acute Comprehensive Urgent Treatment for Eating Disorders (ACUTE) Center.

PROGRAM DESCRIPTION

The ACUTE Center at Denver Health is a 5‐bed unit dedicated to the medical stabilization of patients with severe malnutrition due to anorexia nervosa or severe electrolyte disorders due to bulimia nervosa. ACUTE accepts patients 17 years and older with medical complications related to chronic malnutrition and refeeding.

ACUTE uses a multidisciplinary approach to patient care. The physician team is composed of a hospital medicine attending physician, consultative expertise by an internal medicine specialist in the management of the medical complications of eating disorders, and a psychiatrist specializing in eating disorders. There is a dedicated team of nurses, two dieticians, physical therapists, certified nursing assistants, speech therapists, a psychotherapist, and a chaplain.

ACUTE patients are on continuous telemetry monitoring for the duration of their hospitalization to monitor for arrhythmias as well as signs of covert exercise. As part of the initial intake, a full set of vital signs is obtained, including height and weight. Patients are weighed daily with their back to the scale. There is no discussion of weight fluctuations. Patients may walk at a slow pace around the unit. No exercise is allowed.

Each patient at the ACUTE Center has an individualized meal plan and are started on an oral caloric intake 200 kcal below their basal energy expenditure (BEE). Indirect calorimetry is performed on the first hospital day. Each patient meets on a daily basis with the registered dietician to choose meals that meet their caloric goals.

All patients have a sitter continuously for their first week, and thereafter sitter time may be reduced to supervision surrounding each meal. Patients who fail to finish their prescribed meal are required to drink a liquid supplement to meet caloric goals. Calories are increased weekly until the patient's weight shows a clear pattern of weight increase. 0

Figure 1

Patients are discharged from the ACUTE Center when they have achieved several basic goals: They are consuming greater than 2000 kcal per day, they are consistently gaining 23 pounds per week, their laboratory values have stabilized without electrolyte supplementation, and they are strong enough for an inpatient eating disorder program.

METHODS

Patients admitted to the ACUTE Center between October 2008 and December 2010 for medical stabilization and monitored refeeding were included. Patients with a diagnosis of bulimia nervosa were excluded. Demographic data and laboratory results were obtained electronically from our data repository, whereas weight, height, and other clinical characteristics were obtained by manual chart abstraction. The statistical analysis was conducted in SAS Enterprise Guide v4.1 (SAS Institute, Cary, NC).

RESULTS

In its first 27 months, the ACUTE Center had 76 total admissions, comprising 59 patients. Of the 76 admissions, the 62 admissions for medical stabilization and monitored refeeding of 54 patients with anorexia nervosa were included. Forty‐eight of the 54 (89%) included patients were female. Six patients were hospitalized twice, and 1 patient 3 times. There were 3 transfers to the intensive care unit, and no inpatient mortality. Of the 62 admissions, 11 (18%) discharges were to home, and 51 (82%) were to inpatient psychiatric eating disorder units.

The mean age at admission was 27 years (range 1765 years). The mean percent of ideal body weight (IBW) on admission was 62.2% 10.2%. The mean body mass index (BMI) was 12.9 2.0 kg/m² on admission, and 13.1 1.9 kg/m² upon discharge. The median length of stay was 16 days (interquartile range [IQR] 929 days). Median calculated BEE (1119 [10671184 IQR]) was higher than measured BEE by indirect calorimetry (792 [6341094]), (Table 1).

The majority of admission laboratory values, including serum albumin, blood urea nitrogen (BUN), creatinine, potassium, magnesium, and phosphate levels, were within normal limits. Fifty‐six percent were hyponatremic at admission, with a mean serum sodium level of 133 6 mmol/L (Table 2).

DISCUSSION

Hospital Medicine is currently the fastest growing area of specialization in medicine.13 Palliative care, inpatient geriatrics, short stay units, and bedside procedures have evolved into hospitalist‐led services.1418 The management of the medical complications of severe eating disorders is another potential niche for hospitalists.

The ACUTE Center at Denver Health represents a center in which highly specialized, multidisciplinary care is provided for a rare and extremely ill population of patients. Prior to entering the ACUTE Center, the patients described in our program had each experienced prolonged and unsuccessful stays for medical stabilization in acute care hospitals across the country, after being denied treatment in eating disorder programs due to medical instability.

Patients transferred to ACUTE often received medical care reflecting a lack of specific expertise, training, and exposure. The most common management discrepancy we noted was over‐aggressive provision of intravenous fluids. Consequently, we often diurese 1020 pounds of edema weight, gained during a prior medical hospitalization, before beginning the process of weight restoration. This edema weight artificially increases admission weight and results in less than expected weight gain from admission to discharge.

Even without substantial weight gain, medical stabilization is evidenced by consistent caloric oral intake, and fluid and electrolyte stabilization after initial refeeding. Accordingly, patients who have been treated at the ACUTE Center often become eligible for admission to eating disorder programs at body weights below the typical 70% of ideal body weight that most programs use as a threshold for admission.

From a clinical research perspective, centers such as ACUTE allow for opportunities to better understand and investigate the nuances of patient care in the setting of severe malnutrition. From our cohort of patients to date, we have noted unique issues in albumin levels,19 coagulopathy,20 and liver function,21 among others. As an example, the cohort of patients with anorexia nervosa described here had profoundly low body weight, but relatively normal admission labs. Even the serum albumin, a parameter often used to reflect nutrition in an adult internal medicine setting, is usually normal, reflecting, in an otherwise generally healthy young population, the absence of a malignant, inflammatory, or infectious etiology of weight loss.19

Hospitalists also advocate for their patients by helping to maximize the benefits of their health care coverage. Many health care plans place limits on inpatient psychiatric care benefits. Patients who are severely malnourished from their eating disorder may waste valuable psychiatric care benefits undergoing medical stabilization in psychiatric units while physically unable to undergo psychotherapy. This has become increasingly important as health insurance plans continue to decrease coverage for residential care of patients with anorexia.22

In contrast, the medical benefits of most health plans are more robust. Accordingly, from the patient perspective, medical stabilization in an acute medical unit before admission to a psychiatry unit maximizes their ability to participate in the intensive psychiatric therapy which is still needed after medical stabilization. A recent study from a residential eating disorder program confirmed that a higher discharge BMI was the single best predictor of full recovery from anorexia nervosa.23

In the future, we believe that a continuing concentration of care and experience may also lend itself to the development of protocols and management guidelines which may benefit patients beyond our own unit. Severely malnourished patients with anorexia nervosa, or bulimic patients with complicated electrolyte disorders, are likely to benefit both medically and financially from centers of excellence. Inpatient or residential psychiatric eating disorder programs may act in synergy with medical eating disorders units, like ACUTE, to most efficiently care for the severely malnourished patient. Hospitalists, with the proper training and experience, are uniquely positioned to develop such centers of excellence.

Communication failures are well‐recognized as causes of medical errors.1, 2 Specifically, handoffs of patient care responsibilities, which are increasingly prevalent in academic medical centers,3 have been cited as the most frequent cause of teamwork breakdown resulting in the harmful medical errors found in malpractice claims.1 The Institute of Medicine has recently identified patient handoffs as the moment where patient care errors are most likely to occur.4 A survey of 125 U.S. medical schools, however, found that only 8% specifically taught students how to hand off patient care.3

In July 2003, the American Council of Graduate Medical Education (ACGME) mandated that residency programs decrease resident work hours to improve patient care and safety by reducing fatigue,5 and a recent Institute of Medicine report suggests that they be decreased even further.4 Studies examining outcomes during the first 2 years after reducing duty hours did not find reductions in risk‐adjusted mortality.68 One proposed explanation for this lack of improvement is that the reduction in fatigue‐related medical errors is being offset by discontinuity of care with due to the increased number of patient handoffs resulting from shortened duty hours,911 one recent study found that omission of key information during patient sign outs frequently resulted in adverse patient care outcomes.12

In 2007, the Joint Commission developed a new National Patient Safety Goal that requires organizations to improve communication between caregivers.13 We recently developed an approach by which Internal Medicine residents hand off patient care using a structured process, written and verbal templates, formal training about handoffs, and direct attending supervision.14 Because fourth‐year medical students perform the duties of interns when working as subinterns, we recognized that education about handoffs should occur prior to the time students became interns. Accordingly, we developed a course designed to teach patient handoffs to medical students at the transition between their third and fourth years of training.

Setting

The Handoff Selective was developed by faculty of Denver Health and the University of Colorado Denver School of Medicine.

Program Description

The Selective was first offered in April 2007 as part of an Integrated Clinician's Course (ICC), a 2‐week course for students beginning their fourth year, which starts in April at the University of Colorado. The ICC includes both mandatory and selective sessions that are focused on developing clinical skills and preparing them for their subinternships. The Handoff Selective was conducted in a computerized teaching laboratory, lasted a total of 2 hours and consisted of 2 parts. Each of the 5 Denver Health Hospital Medicine faculty members versed in handoff education taught 2 sessions of 6 to 8 students.

Part 1: Didactic

During the first hour of class, the faculty presented a lecture that summarized the relevant literature on handoffs and explained the importance of the topic. The objectives of the didactic were to: (1) understand the importance of handoffs; (2) explore different communication elements and structures; (3) gain exposure to handoffs outside of healthcare; and (4) learn a structure for handoffs of patient care in hospitalized patients.

We used 3 video clips of handoffs from 2 football games to demonstrate the importance of practice, training, and 2‐way communications in handoffs. The first video clip showed a runner trying to make a spontaneous handoff while being tackled. The receiver was not expecting the handoff and was preoccupied with blocking another player. This attempted handoff resulted in a fumble, which we related to an adverse patient event.

The next 2 video clips showed 2 complex, seldom used, but well‐known football handoffsthe hook and lateral and the Statue of Liberty. Both handoffs were successfully executed presumably as a result of education, practice and the active participation of both players (handing off and receiving) in the process. We then related the teaching and practicing of complex communication to the Joint Commission on Accreditation of Healthcare Organizations (JCAHO; now simply the Joint Commission) data suggesting that most sentinel events have their root cause in communication and training failures.2

Basic communication elements and process structures were then explored using scenarios from everyday life and evidence from fields outside of medicine. We emphasized that structures for communication (modes, vehicles, and settings) must be chosen according to the occasion and that handoffs are common and important in all occupations. In discussing modes (verbal, written, or nonverbal), vehicles (paper, telephone, or e‐mail), and settings (face‐to face, virtual, or disconnected), we emphasized that the most effective structures for communication (verbal, face‐to face meetings, with written materials and other visual aids at the patient's bedside) were also the most time‐consuming (Figure 1). While our standard for resident handoffs is a face‐to‐face verbal interaction with preprinted written materials as an aid, we also emphasized that for complex patients (eg, mental status changes, concern for an acute abdomen) more robust communication is often needed. Accordingly, a more time‐consuming bedside handoff with simultaneous, focused physical exam and history‐taking by both oncoming and off‐going providers may be most appropriate.

Figure 1

As real‐life examples, we asked our students to communicate a happy birthday wish to their mother, who lives in another state. Almost uniformly, in addition to a written aid (birthday card), they choose the telephone as a vehicle for their verbal mode in a virtual setting with 2‐way communication possible. In contrast, when asked to propose marriage to a significant other in another state, students felt that a face‐to‐face meeting with verbal and nonverbal (ie, ring) modes was appropriate. This time‐consuming mode of communication was felt to be necessary to create a sentiment of importance and avert any possible miscommunication.

The didactic session concluded by demonstrating how to use standardized written and verbal templates for handoffs of the care of a hospitalized patient. We explore the differentiation between written and verbal handoffs in our discussion below.

Program Evaluation

We developed a 2‐part survey to evaluate the effectiveness of the Selective and to solicit feedback about the didactic and practicum portions of the course. The first part of the survey (Table 2) contained 16 items to assess the students' knowledge of, and attitudes toward handing off patient care, along with their comfort with the handoff process. Responses to this section were scored using a 5‐point Likert scale with 1 indicating strongly disagree and 5 indicating strongly agree. This part of the survey was administered both prior to and after the Selective.

The second part (Table 3) contained 12 items and was designed to evaluate the perceived usefulness of the different components of the class. This section was only administered at the end of the Selective. It utilized a 4‐point Likert scale with 1 indicating that the component was not useful at all, and 4 indicating that it was extremely useful. The first 6 items of the second section allowed students to evaluate the didactic portion of the handoff. The second 6 items allowed students to evaluate the practicum. Responses to all 12 items were then combined to determine an overall composite usefulness for the Selective.

The Selective was also evaluated qualitatively through the use of open‐ended, written comments that were solicited at the end of the survey. All surveys were administered anonymously.

Results

More students chose the Selective than we had capacity to accommodate (60 of a class of 150). The pre‐ and postcourse survey response rate was 56 of 60 (93%) and 58 of 60 (97%), respectively. After the Selective, the mean score in response to whether handoffs are well taught in medical school increased from 1.6 to 3.5 (P < 0.003). Our students' self‐perceived skills and knowledge about handoffs improved after the Selective (Table 2). The greatest changes in perceived knowledge occurred in questions regarding the what, how, and when of read‐backs, and the knowledge of standard verbal and written handoff structures. The responses to the survey elements which assessed our students' attitudes regarding the importance of standardization and whether they felt handoffs were safer with faculty supervision did not change after the Selective (Table 2).

A total of 92% of the students felt that the course was extremely useful or useful. The role‐playing activity was thought to be more helpful than the didactic, but 84% of the students still rated the didactic portion as useful or extremely useful (Table 3). The element which was the least well received in the didactic portion was the use of video clips to demonstrate successful and unsuccessful (fumbled) college football handoffs, although the majority (64%) of students still found it useful.

The major theme generated from the comments section of the survey was that the Selective should be a required course.

Discussion

We know of no previously published literature that has addressed teaching handoffs to medical students. Horwitz et al.15 developed a sign‐out curriculum for Internal Medicine residents and found that none of their house‐staff had any previous training in handoffs during medical school, consistent with the finding that only 8% of U.S. medical schools provided formal instruction on handoffs.3 Prior to taking the Selective, our students had no knowledge of verbal or written templates for patient handoffs, although both before and after the course they felt that standardization was an important component of the process.

A number of verbal structures for handing off patient care have been described in the literature and there is not a consensus as to which functions best. Perhaps the most cited verbal communication format is SBAR (ie, situation, background, assessment and recommendation).16, 17 This tool was developed by Leonard et al.18 specifically for use by nurses to provide 1‐way communication to physicians pertaining to a change in patient status. We considered teaching the SBAR approach to the students but felt that it did not provide a suitable structure for handoffs because the transfer of care is not generally an event‐based situation and the literature on handoffs indicates that an optimal verbal system includes 2‐way communication.

Additional mnemonics for handoffs found in the literature include SIGNOUT (ie, Sick or DNR, Identifying information, General hospital course, New events of the day, Overall health status, Upcoming possibilities with plan, and Tasks to complete),14 I PASS the BATON (ie, Introduction, Patient, Assessment, Situation, Safety, Background, Actions, Timing, Ownership, Next)19 and the SAIF‐IR system (see boxed text).14

We developed the SAIF‐IR mnemonic to maximize efficiency and effectiveness while differentiating the verbal portion of the handoff from the written and incorporating 2‐way communication into its structure. In the Summary statement, we emphasize that this is not a history of present illness. We ask our students to summarize, in 1 to 3 sentences, the patient's presentation and working diagnosis. When discussing patient issues, we ask our students to only verbalize Active issues, although the written template has inactive, chronic issues listed. Here, we also ask our students to express their level of concern for the active issues and patient in general. If‐then's and Follow‐ups are usually verbalized together. Based on the offgoing provider's knowledge of the patient, we encourage the offgoing provider to anticipate potential problems and advise the oncoming provider on potential responses. Much of this advice is difficult to express in the written format and thus may not be found on the written handoff when the verbal handoff occurs. We encourage oncoming providers to take notes on the preprinted handoff sheet as part of the handoff process.

Through Interactive questioning and Read‐backs, we train our students and house‐staff to use the active listening techniques used outside of healthcare, in settings such as nuclear power plants and National Aeronautics and Space Administration mission control, where poor handoff communication may also result in safety concerns and adverse events.20 Interactive questioning allows the oncoming provider to correct or clarify any information given by the off‐going provider. Read‐backs are a method of confirming follow‐up activity or contingency plans. Together, the SAIF‐IR mnemonic builds a 2‐way communication structure into the patient handoff with both offgoing and oncoming providers having predefined roles.

Much of the information on our written handoff (patient identifying information, medications, language preference, code status, admission date) is not verbalized unless it is part of the active issues or the if‐then, follow‐ups (ie, medication titration for a patient admitted with an acute coronary syndrome or cor status in a patient newly made comfort care). By not reading extraneous information, we seek to emphasize the Active issues as well as the If‐then, Follow‐ups. We feel this emphasis maximizes the effectiveness of the handoff, while the purposeful nonverbalization of written materials such as identifying information maximizes its efficiency. Future work may examine which verbal and written structures for patient handoffs most benefit patient care and workflow through standard communication.

While our students found the Handoff Selective to be useful and to improve their self‐perceived ability to perform handoffs, we were not able to determine whether our program affected downstream outcomes such as adverse events relating to failures in handoff communication. Additionally, since we only taught and evaluated our Selective at the University of Colorado Denver School of Medicine, the response of our students may not generalize to other medical schools. Multicentered, prospective, randomized controlled trials may determine whether handoff education programs are successful in reducing patient adverse events related to transfers of care.

While handoffs occur frequently and are increasingly recognized as a vulnerable time in patient care, little is known about how to effectively teach handoffs to medical students during their clinical years. We developed a formal course to teach the importance of handoffs and how the process should be conducted. Our students reported that the Handoff Selective we developed improved their knowledge about the process and their perception of their ability to perform handoffs in a time‐appropriate and effective manner. In response to the feedback we received from our students, the Handoff Selective is the only course in the ICC that has been made mandatory for all students.

Communication failures are well‐recognized as causes of medical errors.1, 2 Specifically, handoffs of patient care responsibilities, which are increasingly prevalent in academic medical centers,3 have been cited as the most frequent cause of teamwork breakdown resulting in the harmful medical errors found in malpractice claims.1 The Institute of Medicine has recently identified patient handoffs as the moment where patient care errors are most likely to occur.4 A survey of 125 U.S. medical schools, however, found that only 8% specifically taught students how to hand off patient care.3

In July 2003, the American Council of Graduate Medical Education (ACGME) mandated that residency programs decrease resident work hours to improve patient care and safety by reducing fatigue,5 and a recent Institute of Medicine report suggests that they be decreased even further.4 Studies examining outcomes during the first 2 years after reducing duty hours did not find reductions in risk‐adjusted mortality.68 One proposed explanation for this lack of improvement is that the reduction in fatigue‐related medical errors is being offset by discontinuity of care with due to the increased number of patient handoffs resulting from shortened duty hours,911 one recent study found that omission of key information during patient sign outs frequently resulted in adverse patient care outcomes.12

In 2007, the Joint Commission developed a new National Patient Safety Goal that requires organizations to improve communication between caregivers.13 We recently developed an approach by which Internal Medicine residents hand off patient care using a structured process, written and verbal templates, formal training about handoffs, and direct attending supervision.14 Because fourth‐year medical students perform the duties of interns when working as subinterns, we recognized that education about handoffs should occur prior to the time students became interns. Accordingly, we developed a course designed to teach patient handoffs to medical students at the transition between their third and fourth years of training.

Setting

The Handoff Selective was developed by faculty of Denver Health and the University of Colorado Denver School of Medicine.

Program Description

The Selective was first offered in April 2007 as part of an Integrated Clinician's Course (ICC), a 2‐week course for students beginning their fourth year, which starts in April at the University of Colorado. The ICC includes both mandatory and selective sessions that are focused on developing clinical skills and preparing them for their subinternships. The Handoff Selective was conducted in a computerized teaching laboratory, lasted a total of 2 hours and consisted of 2 parts. Each of the 5 Denver Health Hospital Medicine faculty members versed in handoff education taught 2 sessions of 6 to 8 students.

Part 1: Didactic

During the first hour of class, the faculty presented a lecture that summarized the relevant literature on handoffs and explained the importance of the topic. The objectives of the didactic were to: (1) understand the importance of handoffs; (2) explore different communication elements and structures; (3) gain exposure to handoffs outside of healthcare; and (4) learn a structure for handoffs of patient care in hospitalized patients.

We used 3 video clips of handoffs from 2 football games to demonstrate the importance of practice, training, and 2‐way communications in handoffs. The first video clip showed a runner trying to make a spontaneous handoff while being tackled. The receiver was not expecting the handoff and was preoccupied with blocking another player. This attempted handoff resulted in a fumble, which we related to an adverse patient event.

The next 2 video clips showed 2 complex, seldom used, but well‐known football handoffsthe hook and lateral and the Statue of Liberty. Both handoffs were successfully executed presumably as a result of education, practice and the active participation of both players (handing off and receiving) in the process. We then related the teaching and practicing of complex communication to the Joint Commission on Accreditation of Healthcare Organizations (JCAHO; now simply the Joint Commission) data suggesting that most sentinel events have their root cause in communication and training failures.2

Basic communication elements and process structures were then explored using scenarios from everyday life and evidence from fields outside of medicine. We emphasized that structures for communication (modes, vehicles, and settings) must be chosen according to the occasion and that handoffs are common and important in all occupations. In discussing modes (verbal, written, or nonverbal), vehicles (paper, telephone, or e‐mail), and settings (face‐to face, virtual, or disconnected), we emphasized that the most effective structures for communication (verbal, face‐to face meetings, with written materials and other visual aids at the patient's bedside) were also the most time‐consuming (Figure 1). While our standard for resident handoffs is a face‐to‐face verbal interaction with preprinted written materials as an aid, we also emphasized that for complex patients (eg, mental status changes, concern for an acute abdomen) more robust communication is often needed. Accordingly, a more time‐consuming bedside handoff with simultaneous, focused physical exam and history‐taking by both oncoming and off‐going providers may be most appropriate.

Figure 1

As real‐life examples, we asked our students to communicate a happy birthday wish to their mother, who lives in another state. Almost uniformly, in addition to a written aid (birthday card), they choose the telephone as a vehicle for their verbal mode in a virtual setting with 2‐way communication possible. In contrast, when asked to propose marriage to a significant other in another state, students felt that a face‐to‐face meeting with verbal and nonverbal (ie, ring) modes was appropriate. This time‐consuming mode of communication was felt to be necessary to create a sentiment of importance and avert any possible miscommunication.

The didactic session concluded by demonstrating how to use standardized written and verbal templates for handoffs of the care of a hospitalized patient. We explore the differentiation between written and verbal handoffs in our discussion below.

Program Evaluation

We developed a 2‐part survey to evaluate the effectiveness of the Selective and to solicit feedback about the didactic and practicum portions of the course. The first part of the survey (Table 2) contained 16 items to assess the students' knowledge of, and attitudes toward handing off patient care, along with their comfort with the handoff process. Responses to this section were scored using a 5‐point Likert scale with 1 indicating strongly disagree and 5 indicating strongly agree. This part of the survey was administered both prior to and after the Selective.

The second part (Table 3) contained 12 items and was designed to evaluate the perceived usefulness of the different components of the class. This section was only administered at the end of the Selective. It utilized a 4‐point Likert scale with 1 indicating that the component was not useful at all, and 4 indicating that it was extremely useful. The first 6 items of the second section allowed students to evaluate the didactic portion of the handoff. The second 6 items allowed students to evaluate the practicum. Responses to all 12 items were then combined to determine an overall composite usefulness for the Selective.

The Selective was also evaluated qualitatively through the use of open‐ended, written comments that were solicited at the end of the survey. All surveys were administered anonymously.

Results

More students chose the Selective than we had capacity to accommodate (60 of a class of 150). The pre‐ and postcourse survey response rate was 56 of 60 (93%) and 58 of 60 (97%), respectively. After the Selective, the mean score in response to whether handoffs are well taught in medical school increased from 1.6 to 3.5 (P < 0.003). Our students' self‐perceived skills and knowledge about handoffs improved after the Selective (Table 2). The greatest changes in perceived knowledge occurred in questions regarding the what, how, and when of read‐backs, and the knowledge of standard verbal and written handoff structures. The responses to the survey elements which assessed our students' attitudes regarding the importance of standardization and whether they felt handoffs were safer with faculty supervision did not change after the Selective (Table 2).

A total of 92% of the students felt that the course was extremely useful or useful. The role‐playing activity was thought to be more helpful than the didactic, but 84% of the students still rated the didactic portion as useful or extremely useful (Table 3). The element which was the least well received in the didactic portion was the use of video clips to demonstrate successful and unsuccessful (fumbled) college football handoffs, although the majority (64%) of students still found it useful.

The major theme generated from the comments section of the survey was that the Selective should be a required course.

Discussion

We know of no previously published literature that has addressed teaching handoffs to medical students. Horwitz et al.15 developed a sign‐out curriculum for Internal Medicine residents and found that none of their house‐staff had any previous training in handoffs during medical school, consistent with the finding that only 8% of U.S. medical schools provided formal instruction on handoffs.3 Prior to taking the Selective, our students had no knowledge of verbal or written templates for patient handoffs, although both before and after the course they felt that standardization was an important component of the process.

A number of verbal structures for handing off patient care have been described in the literature and there is not a consensus as to which functions best. Perhaps the most cited verbal communication format is SBAR (ie, situation, background, assessment and recommendation).16, 17 This tool was developed by Leonard et al.18 specifically for use by nurses to provide 1‐way communication to physicians pertaining to a change in patient status. We considered teaching the SBAR approach to the students but felt that it did not provide a suitable structure for handoffs because the transfer of care is not generally an event‐based situation and the literature on handoffs indicates that an optimal verbal system includes 2‐way communication.

Additional mnemonics for handoffs found in the literature include SIGNOUT (ie, Sick or DNR, Identifying information, General hospital course, New events of the day, Overall health status, Upcoming possibilities with plan, and Tasks to complete),14 I PASS the BATON (ie, Introduction, Patient, Assessment, Situation, Safety, Background, Actions, Timing, Ownership, Next)19 and the SAIF‐IR system (see boxed text).14

We developed the SAIF‐IR mnemonic to maximize efficiency and effectiveness while differentiating the verbal portion of the handoff from the written and incorporating 2‐way communication into its structure. In the Summary statement, we emphasize that this is not a history of present illness. We ask our students to summarize, in 1 to 3 sentences, the patient's presentation and working diagnosis. When discussing patient issues, we ask our students to only verbalize Active issues, although the written template has inactive, chronic issues listed. Here, we also ask our students to express their level of concern for the active issues and patient in general. If‐then's and Follow‐ups are usually verbalized together. Based on the offgoing provider's knowledge of the patient, we encourage the offgoing provider to anticipate potential problems and advise the oncoming provider on potential responses. Much of this advice is difficult to express in the written format and thus may not be found on the written handoff when the verbal handoff occurs. We encourage oncoming providers to take notes on the preprinted handoff sheet as part of the handoff process.

Through Interactive questioning and Read‐backs, we train our students and house‐staff to use the active listening techniques used outside of healthcare, in settings such as nuclear power plants and National Aeronautics and Space Administration mission control, where poor handoff communication may also result in safety concerns and adverse events.20 Interactive questioning allows the oncoming provider to correct or clarify any information given by the off‐going provider. Read‐backs are a method of confirming follow‐up activity or contingency plans. Together, the SAIF‐IR mnemonic builds a 2‐way communication structure into the patient handoff with both offgoing and oncoming providers having predefined roles.

Much of the information on our written handoff (patient identifying information, medications, language preference, code status, admission date) is not verbalized unless it is part of the active issues or the if‐then, follow‐ups (ie, medication titration for a patient admitted with an acute coronary syndrome or cor status in a patient newly made comfort care). By not reading extraneous information, we seek to emphasize the Active issues as well as the If‐then, Follow‐ups. We feel this emphasis maximizes the effectiveness of the handoff, while the purposeful nonverbalization of written materials such as identifying information maximizes its efficiency. Future work may examine which verbal and written structures for patient handoffs most benefit patient care and workflow through standard communication.

While our students found the Handoff Selective to be useful and to improve their self‐perceived ability to perform handoffs, we were not able to determine whether our program affected downstream outcomes such as adverse events relating to failures in handoff communication. Additionally, since we only taught and evaluated our Selective at the University of Colorado Denver School of Medicine, the response of our students may not generalize to other medical schools. Multicentered, prospective, randomized controlled trials may determine whether handoff education programs are successful in reducing patient adverse events related to transfers of care.

While handoffs occur frequently and are increasingly recognized as a vulnerable time in patient care, little is known about how to effectively teach handoffs to medical students during their clinical years. We developed a formal course to teach the importance of handoffs and how the process should be conducted. Our students reported that the Handoff Selective we developed improved their knowledge about the process and their perception of their ability to perform handoffs in a time‐appropriate and effective manner. In response to the feedback we received from our students, the Handoff Selective is the only course in the ICC that has been made mandatory for all students.

Communication failures are well‐recognized as causes of medical errors.1, 2 Specifically, handoffs of patient care responsibilities, which are increasingly prevalent in academic medical centers,3 have been cited as the most frequent cause of teamwork breakdown resulting in the harmful medical errors found in malpractice claims.1 The Institute of Medicine has recently identified patient handoffs as the moment where patient care errors are most likely to occur.4 A survey of 125 U.S. medical schools, however, found that only 8% specifically taught students how to hand off patient care.3

In July 2003, the American Council of Graduate Medical Education (ACGME) mandated that residency programs decrease resident work hours to improve patient care and safety by reducing fatigue,5 and a recent Institute of Medicine report suggests that they be decreased even further.4 Studies examining outcomes during the first 2 years after reducing duty hours did not find reductions in risk‐adjusted mortality.68 One proposed explanation for this lack of improvement is that the reduction in fatigue‐related medical errors is being offset by discontinuity of care with due to the increased number of patient handoffs resulting from shortened duty hours,911 one recent study found that omission of key information during patient sign outs frequently resulted in adverse patient care outcomes.12

In 2007, the Joint Commission developed a new National Patient Safety Goal that requires organizations to improve communication between caregivers.13 We recently developed an approach by which Internal Medicine residents hand off patient care using a structured process, written and verbal templates, formal training about handoffs, and direct attending supervision.14 Because fourth‐year medical students perform the duties of interns when working as subinterns, we recognized that education about handoffs should occur prior to the time students became interns. Accordingly, we developed a course designed to teach patient handoffs to medical students at the transition between their third and fourth years of training.

Setting

The Handoff Selective was developed by faculty of Denver Health and the University of Colorado Denver School of Medicine.

Program Description

The Selective was first offered in April 2007 as part of an Integrated Clinician's Course (ICC), a 2‐week course for students beginning their fourth year, which starts in April at the University of Colorado. The ICC includes both mandatory and selective sessions that are focused on developing clinical skills and preparing them for their subinternships. The Handoff Selective was conducted in a computerized teaching laboratory, lasted a total of 2 hours and consisted of 2 parts. Each of the 5 Denver Health Hospital Medicine faculty members versed in handoff education taught 2 sessions of 6 to 8 students.

Part 1: Didactic

During the first hour of class, the faculty presented a lecture that summarized the relevant literature on handoffs and explained the importance of the topic. The objectives of the didactic were to: (1) understand the importance of handoffs; (2) explore different communication elements and structures; (3) gain exposure to handoffs outside of healthcare; and (4) learn a structure for handoffs of patient care in hospitalized patients.

We used 3 video clips of handoffs from 2 football games to demonstrate the importance of practice, training, and 2‐way communications in handoffs. The first video clip showed a runner trying to make a spontaneous handoff while being tackled. The receiver was not expecting the handoff and was preoccupied with blocking another player. This attempted handoff resulted in a fumble, which we related to an adverse patient event.

The next 2 video clips showed 2 complex, seldom used, but well‐known football handoffsthe hook and lateral and the Statue of Liberty. Both handoffs were successfully executed presumably as a result of education, practice and the active participation of both players (handing off and receiving) in the process. We then related the teaching and practicing of complex communication to the Joint Commission on Accreditation of Healthcare Organizations (JCAHO; now simply the Joint Commission) data suggesting that most sentinel events have their root cause in communication and training failures.2

Basic communication elements and process structures were then explored using scenarios from everyday life and evidence from fields outside of medicine. We emphasized that structures for communication (modes, vehicles, and settings) must be chosen according to the occasion and that handoffs are common and important in all occupations. In discussing modes (verbal, written, or nonverbal), vehicles (paper, telephone, or e‐mail), and settings (face‐to face, virtual, or disconnected), we emphasized that the most effective structures for communication (verbal, face‐to face meetings, with written materials and other visual aids at the patient's bedside) were also the most time‐consuming (Figure 1). While our standard for resident handoffs is a face‐to‐face verbal interaction with preprinted written materials as an aid, we also emphasized that for complex patients (eg, mental status changes, concern for an acute abdomen) more robust communication is often needed. Accordingly, a more time‐consuming bedside handoff with simultaneous, focused physical exam and history‐taking by both oncoming and off‐going providers may be most appropriate.

Figure 1

As real‐life examples, we asked our students to communicate a happy birthday wish to their mother, who lives in another state. Almost uniformly, in addition to a written aid (birthday card), they choose the telephone as a vehicle for their verbal mode in a virtual setting with 2‐way communication possible. In contrast, when asked to propose marriage to a significant other in another state, students felt that a face‐to‐face meeting with verbal and nonverbal (ie, ring) modes was appropriate. This time‐consuming mode of communication was felt to be necessary to create a sentiment of importance and avert any possible miscommunication.

The didactic session concluded by demonstrating how to use standardized written and verbal templates for handoffs of the care of a hospitalized patient. We explore the differentiation between written and verbal handoffs in our discussion below.

Program Evaluation

We developed a 2‐part survey to evaluate the effectiveness of the Selective and to solicit feedback about the didactic and practicum portions of the course. The first part of the survey (Table 2) contained 16 items to assess the students' knowledge of, and attitudes toward handing off patient care, along with their comfort with the handoff process. Responses to this section were scored using a 5‐point Likert scale with 1 indicating strongly disagree and 5 indicating strongly agree. This part of the survey was administered both prior to and after the Selective.

The second part (Table 3) contained 12 items and was designed to evaluate the perceived usefulness of the different components of the class. This section was only administered at the end of the Selective. It utilized a 4‐point Likert scale with 1 indicating that the component was not useful at all, and 4 indicating that it was extremely useful. The first 6 items of the second section allowed students to evaluate the didactic portion of the handoff. The second 6 items allowed students to evaluate the practicum. Responses to all 12 items were then combined to determine an overall composite usefulness for the Selective.

The Selective was also evaluated qualitatively through the use of open‐ended, written comments that were solicited at the end of the survey. All surveys were administered anonymously.

Results

More students chose the Selective than we had capacity to accommodate (60 of a class of 150). The pre‐ and postcourse survey response rate was 56 of 60 (93%) and 58 of 60 (97%), respectively. After the Selective, the mean score in response to whether handoffs are well taught in medical school increased from 1.6 to 3.5 (P < 0.003). Our students' self‐perceived skills and knowledge about handoffs improved after the Selective (Table 2). The greatest changes in perceived knowledge occurred in questions regarding the what, how, and when of read‐backs, and the knowledge of standard verbal and written handoff structures. The responses to the survey elements which assessed our students' attitudes regarding the importance of standardization and whether they felt handoffs were safer with faculty supervision did not change after the Selective (Table 2).

A total of 92% of the students felt that the course was extremely useful or useful. The role‐playing activity was thought to be more helpful than the didactic, but 84% of the students still rated the didactic portion as useful or extremely useful (Table 3). The element which was the least well received in the didactic portion was the use of video clips to demonstrate successful and unsuccessful (fumbled) college football handoffs, although the majority (64%) of students still found it useful.

The major theme generated from the comments section of the survey was that the Selective should be a required course.

Discussion

We know of no previously published literature that has addressed teaching handoffs to medical students. Horwitz et al.15 developed a sign‐out curriculum for Internal Medicine residents and found that none of their house‐staff had any previous training in handoffs during medical school, consistent with the finding that only 8% of U.S. medical schools provided formal instruction on handoffs.3 Prior to taking the Selective, our students had no knowledge of verbal or written templates for patient handoffs, although both before and after the course they felt that standardization was an important component of the process.

A number of verbal structures for handing off patient care have been described in the literature and there is not a consensus as to which functions best. Perhaps the most cited verbal communication format is SBAR (ie, situation, background, assessment and recommendation).16, 17 This tool was developed by Leonard et al.18 specifically for use by nurses to provide 1‐way communication to physicians pertaining to a change in patient status. We considered teaching the SBAR approach to the students but felt that it did not provide a suitable structure for handoffs because the transfer of care is not generally an event‐based situation and the literature on handoffs indicates that an optimal verbal system includes 2‐way communication.

Additional mnemonics for handoffs found in the literature include SIGNOUT (ie, Sick or DNR, Identifying information, General hospital course, New events of the day, Overall health status, Upcoming possibilities with plan, and Tasks to complete),14 I PASS the BATON (ie, Introduction, Patient, Assessment, Situation, Safety, Background, Actions, Timing, Ownership, Next)19 and the SAIF‐IR system (see boxed text).14

We developed the SAIF‐IR mnemonic to maximize efficiency and effectiveness while differentiating the verbal portion of the handoff from the written and incorporating 2‐way communication into its structure. In the Summary statement, we emphasize that this is not a history of present illness. We ask our students to summarize, in 1 to 3 sentences, the patient's presentation and working diagnosis. When discussing patient issues, we ask our students to only verbalize Active issues, although the written template has inactive, chronic issues listed. Here, we also ask our students to express their level of concern for the active issues and patient in general. If‐then's and Follow‐ups are usually verbalized together. Based on the offgoing provider's knowledge of the patient, we encourage the offgoing provider to anticipate potential problems and advise the oncoming provider on potential responses. Much of this advice is difficult to express in the written format and thus may not be found on the written handoff when the verbal handoff occurs. We encourage oncoming providers to take notes on the preprinted handoff sheet as part of the handoff process.

Through Interactive questioning and Read‐backs, we train our students and house‐staff to use the active listening techniques used outside of healthcare, in settings such as nuclear power plants and National Aeronautics and Space Administration mission control, where poor handoff communication may also result in safety concerns and adverse events.20 Interactive questioning allows the oncoming provider to correct or clarify any information given by the off‐going provider. Read‐backs are a method of confirming follow‐up activity or contingency plans. Together, the SAIF‐IR mnemonic builds a 2‐way communication structure into the patient handoff with both offgoing and oncoming providers having predefined roles.

Much of the information on our written handoff (patient identifying information, medications, language preference, code status, admission date) is not verbalized unless it is part of the active issues or the if‐then, follow‐ups (ie, medication titration for a patient admitted with an acute coronary syndrome or cor status in a patient newly made comfort care). By not reading extraneous information, we seek to emphasize the Active issues as well as the If‐then, Follow‐ups. We feel this emphasis maximizes the effectiveness of the handoff, while the purposeful nonverbalization of written materials such as identifying information maximizes its efficiency. Future work may examine which verbal and written structures for patient handoffs most benefit patient care and workflow through standard communication.

While our students found the Handoff Selective to be useful and to improve their self‐perceived ability to perform handoffs, we were not able to determine whether our program affected downstream outcomes such as adverse events relating to failures in handoff communication. Additionally, since we only taught and evaluated our Selective at the University of Colorado Denver School of Medicine, the response of our students may not generalize to other medical schools. Multicentered, prospective, randomized controlled trials may determine whether handoff education programs are successful in reducing patient adverse events related to transfers of care.

While handoffs occur frequently and are increasingly recognized as a vulnerable time in patient care, little is known about how to effectively teach handoffs to medical students during their clinical years. We developed a formal course to teach the importance of handoffs and how the process should be conducted. Our students reported that the Handoff Selective we developed improved their knowledge about the process and their perception of their ability to perform handoffs in a time‐appropriate and effective manner. In response to the feedback we received from our students, the Handoff Selective is the only course in the ICC that has been made mandatory for all students.