Monetary resident incentives: effect on patient satisfaction in an academic emergency department

Patient satisfaction most be a priority in emergency departments (EDs). The care provided by residents forms much of the patient contact in academic EDs.

To determine if monetary incentives for emergency medicine (EM) residents improve patient satisfaction scores on a mailed survey.

The incentive program ran for nine months, 1999-2000. Press-Ganey surveys responses from ED patients in 456 hospitals; 124 form a peer group of larger, teaching hospitals. Questions relate to: 1) waiting time, 2) taking the problem seriously, 3) treatment information, 4) home care concerns, 5) doctor's courtesy, and 6) concern with comfort. A 5-point Likert scale ranges from "very poor" (0 points) to "very good" (100). Raw score is the weighted mean, converted to a percentile vs. the peer group. Incentives were three-fold: a year-end event for the EM residents if 80th percentile results were achieved; individual incentives for educational materials of $50/resident (50th percentile), $100 (60th), $150 (70th), or $200 (80th); discount cards for the hospital's espresso cart. These were distributed by 11 EM faculty (six cards/month) as rewards for outstanding interactions. Program cost was <$8,000, from patient-care revenue. Faculty had similar direct incentives, but nursing and staff incentives were ill defined and indirect.

Raw scores ranged from 66.1 (waiting time) to 84.3 (doctor's courtesy) (n=509 or ∼7.2% of ED volume). Corresponding percentiles were 20th-43rd (mean=31st). We found no difference between the overall scores after the incentives, but three of the six questions showed improvement, with one, "doctors' courtesy," reaching 53(rd) percentile. The faculty funded the 50(th) percentile reward.

Incentives are a novel idea to improve patient satisfaction, but did not foster overall Press-Ganey score improvement. We did find a trend toward improvement for doctor-patient interaction scores. Confounding variables, such as increasing patient census, could account for inability to demonstrate a positive effect.