Background: Audit and feedback improves clinical care by highlighting the gap between current and ideal practice. Electronic health record (EHR) data can provide contemporaneous data for quality improvement but has not yet been studied extensively. We conducted a randomized trial to determine whether audit and feedback leveraging EHR data with modern web-based dashboards could improve quality outcomes in an academic medical center compared to usual feedback.
Methods: Eight medicine teams were randomly assigned to receive usual versus intensive feedback from February – June 2016. Usual feedback (ongoing for the 19 months prior to the intervention) consisted of a twice-monthly email with graphical feedback of each team’s performance on QI metrics. The intensive feedback consisted of both access to a team-based, real-time online data dashboard (Figure 1) and weekly in-person review of team data, which we referred to as “STAT rounds.” STAT rounds was a 15 minute session of facilitated discussion among intervention teams to review performance of metrics and identify roadblocks and solutions to meet them. The primary outcome was the team’s performance on a composite measure of three discharge metrics: a completed medication reconciliation in the EHR prior to discharge, a high-quality after-visit-summary (AVS), and a discharge summary completed within 24 hours of discharge. We used generalized linear models to assess the adjusted effect of the intervention on the primary outcome.
Results: 24 medicine teams (12 intervention and 12 control) comprised of 84 housestaff and 48 attendings participated in the trial over 3 months. The dashboard was accessed an average of 56 times per month. STAT rounds were conducted 12 times with at least one physician representative from each team 98% of the time.
The primary outcome, completion of all three metrics, was achieved 79.3% (426/537) in the intervention group compared to 63.2% (326/516) in the control group. Adjusted for possible confounders, composite performance of metrics was significantly increased in the intervention group (OR=2.40, (95% CI 1.66-3.45). Out of the components, the lowest performing metric — the high-quality after visit summary, demonstrated the strongest improvement. (63% vs 79%, p<0.0001) (Table 1)