Under-Promising and Over-Delivering to Improve Patient Satisfaction at Emergency Departments : Evidence from a Field Experiment Providing Wait Information

Overcrowded Emergency Departments (EDs) across locations struggle to improve patient experience while dealing with long waits, which erodes medical and financial performance. We investigate whether and how managers could improve patient satisfaction by communicating waits to patients.We conduct a field experiment at an urban ED. We develop a machine learning-based wait time prediction application and implement it within the electronic medical records system. Our treatments provide patients with personalized estimated waits with no overestimation (the median), moderate overestimation (70th-percentile), or high overestimation (90th-percentile). Patients report higher satisfaction when receiving their estimated waits, but the effect vary widely depending on the announcement. Drawing on Prospect theory, we hypothesize that the announced wait acts as a reference point against which patients compare the actual wait and that patients are lossaverse (end effect): Waits longer than announced will lower satisfaction more than waits shorter than announced will increase satisfaction. Overestimating waits will then improve satisfaction. At the same time, however, rising the announced wait will reduce satisfaction initially and while waiting (initial effect), and this effect will dominate over the end effect for high levels of overestimation. Accordingly, we hypothesize and show that patients are more satisfied when they are told an estimate based on the 70 or 90th-percentiles, with the benefit being the largest for the 70thpercentile announcement. Wait estimates based on the median have a null effect.Despite the benefits from wait announcements in settings where queues are unobservable, less is known about their effects in EDs, where queues are partially observable. With the Centers for Medicare & Medicaid Services tying reimbursements to patients’ ratings, our research suggests a cost-effective lever to improve patients’ satisfaction and hospitals’ financial performance: under-promising and over-delivering by providing moderately overestimated wait information