Ethan Kuperman

Background: Our institution used an automated provider alert for venous thromboembolism (VTE) prophylaxis. This alert had an option to “snooze,” generating a user-specific delay before repeat firing. Providers snoozed 67% of VTE alerts. Our aim was to decrease alert burden by 25% through lengthening the snooze delay. Methods: This initiative was conducted at a tertiary care academic medical center using an integrated electronic health record. Data was collected with snooze delay time set at 1, 2 and 3 hours. Our primary outcome was the number of alerts per 100 admissions. Secondary outcomes included percentage of alerts snoozed, percent of patients receiving pharmacoprophylaxis, and percent of admissions satisfying VTE core measures. Results: There was no change in alerts per 100 admissions (168 vs. 176, p=0.375) or percentage of patients receiving pharmacoprophylaxis (50.9% vs. 50.9%, p=0.997). There was a small reduction in snooze percentage (67.0% to 63.2%, p=<0.001) and a small increase in core measure satisfaction (86.0% to 88.0%, p=0.025). On post-hoc analysis, the intervention increased the median interval between same-patient, same-provider VTE alerts (3.4 to 8.3 hours). Internal medicine hospitalists were responsible for 22% of alerts, selecting snooze 78% of the time. Conclusion: Increasing snooze delay time did not decrease alert burden. Manipulating alert parameters is unlikely to decrease alert burden without simultaneous changes in provider response culture.