Background: Hospitalists vary in their ordering practices related to labs, imaging, and discharge order times. Variations in these practices has been associated with over-utilization of services and inefficient clustering of discharges later in the day. Prior studies suggest that peer comparison can be effective to change provider behavior and reduce unnecessary variation in ordering practices, however, sharing data in a way that is transparent, timely, and actionable is resource-intensive. Automation of peer feedback that is easily accessible to providers on their mobile device can help improve engagement and reduce practice variation.
Purpose: We leveraged an automated text-based platform (Agathos) to pursue 2 objectives: first, to measure hospitalist engagement with unblinded peer comparison on four pre-specified process metrics: daily lab orders, CT orders, echocardiogram orders, and discharge orders placed before 10am; second, to evaluate the impact of engagement on change in the process metrics. We hypothesized that if hospitalists engaged with at least 50% of weekly text messages, we would observe reduction in frequency of orders for repeat daily labs, CTs, echocardiograms, and an increase in discharge orders before 10am.
Description: The Agathos platform uses EHR data including orders, notes and results to attribute individual days of care to different hospitalists caring for the same patient. Individual hospitalists are also able to easily click into their performance data to review and validate the data accuracy and review specific cases contributing to a given metric. All physicians receive a weekly text message on the same date and at the same time to spur peer-to-peer conversation and larger team discussions. Hospitalist engagement with the platform is defined as clicking the link sent in weekly text to review their individual data compared to their peers. Over the first 4 months of this 6 month intervention, 57 hospitalists at 2 large teaching hospitals have received weekly audit and feedback via a text message with unblinded peer comparison via the platform. An example of text and linked peer comparison data is shown in Figure 1. Overall weekly engagement averaged 51% and overall monthly engagement averaged 78%. As shown in Figure 2, engagement varied by metric: engagement with CT and consult utilization data started high and trended down; engagement with echocardiogram started low and trended up; engagement with discharge orders was the most consistent across 4 months. Preliminary analysis of hospitalist orders by metric during the first 4 months of the intervention compared to the prior 12 months (pre-intervention) shows a decrease in echocardiogram orders (from 7.6 to 6.5 per 100 cases) and increase in discharge orders before 10am (from 8.0 to 9.1 per 100 cases); there was no significant change in orders for daily labs or CT scans. Full analysis of the 6 months will be completed in January, 2025.
Conclusions: Hospitalist engagement with a novel text and web-based platform was sustained at a high level over several months with variation and directional trends for each metric. Hospitalist orders for echocardiograms and early discharge orders suggest changes from pre-intervention baseline but orders for daily labs and CT scans showed no change.
