Pharmacy Interventions on Medication Orders Rise With Time on Shift Regardless of Resident Seniority

Physicians in the emergency department (ED) experience high patient volumes and cognitive demand, contributing to increased decision fatigue throughout shifts. Little has been studied on how this affects residents, whose training is subject to controlled duty hours. At our health system, pharmacists review all ED medication orders and intervene as needed (e.g., to correct errors). This study will use intervention rates as a proxy for resident decision fatigue to determine how the association between time-on-shift and fatigue evolves across the residency training period. Our objective is to test whether the odds of a pharmacy intervention increase with time-on-shift, and whether this differs by post-graduate year (PGY).

Medication orders placed by emergency medicine residents at an urban academic medical center from 2022-2023 were queried from institutional data and linked to resident shift schedules. The outcome was pharmacist intervention on an individual resident medication order (yes/no). Multivariate logistic regression models were used to identify associations between PGY, time-on-shift, and pharmacy intervention rate. Covariates were overnight vs. daytime shifts, medication therapeutic class, patient age (<65 vs. >=65), and the National Emergency Department Overcrowding Scale score. Sensitivity analyses controlled for resident caseload (number of patients carried by a resident at a given time).

Ninety-one residents placed 83,484 medication orders with a pharmacy intervention rate of 2.9%. Overall intervention rates did not differ by PGY with power=0.87 to detect an odds ratio [OR] of 0.93 per PGY increase. Odds of a pharmacy intervention increased 2% per hour with resident time-on-shift (OR 1.02/hr, 95% CI 1.01-1.04, p=0.002). The interaction between time-on-shift and PGY was not significant, with power=0.84 to detect an OR of 0.99 per hour-on-shift * PGY increase. Results were unchanged after additionally controlling for caseload.

Time-on-shift was associated with higher odds of pharmacist intervention; this association was not modified by PGY. While interventions are an operational proxy (not a direct fatigue measure), results suggest that time-related risk signals persist across residency training. Next steps include stratifying by intervention types and testing mitigation strategies (e.g., breaks, clinical decision support).