Many hospitals rely on improved throughput to maintain high volume in the face of limited inpatient capacity. Earlier discharges may free up inpatient beds, decrease the strain on a crowded Emergency Department (ED), and allow inpatient services to provide timely care to patients by improving throughput from the ED. Although early discharge times are often assumed to be beneficial, their impact is unproven and may have possible untoward effects such as hasty discharge preparation. We describe our efforts to increase the proportion of discharge orders placed before 11 am on the inpatient Medicine Service and its effect on hospital throughput metrics.


A quality improvement project was implemented to incentivize residents and hospitalists to place discharge orders before 11am. At our institution hospitalists care for patients on both the Teaching and Non–teaching Services. The major components of the initiative were: (1) discussion and recording of discharge order times at resident report each morning, (2) weekly distribution of physician–level data to housestaff and hospitalists, and (3) monthly gift certificate awards to the top two performing housestaff teams. We assessed whether the initiative was successful by measuring discharge order times and two “downstream” throughput metrics: the time when the patient actually leaves the hospital and length of stay (LOS). We used July 2011 as the baseline period (N = 741 discharges) and tracked the impact over a 4 month period from August to November 2011 (N = 3,200 discharges).


Discharge orders before 11am improved from 29.5 to 56.0% (P < 0.0001) and the mean discharge order time improved by 94 minutes (Figure 1). Patients left the hospital an average of 22 min earlier and percent of patients leaving before 3 pm increased from 32.2 to 38.1% (P < 0.001). Mean LOS was unchanged (Figure 2). These findings were consistent for both the Teaching and the Non–teaching Services.


An intervention providing daily feedback to residents and weekly metrics to housestaff and hospitalists improved the timeliness of discharge order placement. However, the ultimate impact on throughput was small. The mild benefit despite the substantially earlier discharge order time likely reflects the complexities of the discharge process, including patient and caregiver factors. Stressing early discharge order time alone is, therefore, of limited value. Further improvements in throughput may be achieved by engaging other members of the caregiver team and optimizing each step in the discharge process after order entry.

Figure 1Mean discharge order times (bottom graphs) and corresponding mean discharge times (top graphs) for patients discharged by physicians on the teaching medicine service (includes general medicine, GI/liver, cardiology) and non–teaching medicine service (includes general medicine and GI).

Figure 2Mean length of stay (LOS) did not decrease and instead increased by approximately 0.25 days as compared to July 2011.