Modeling Advanced Practice Provider Productivity in the Emergency Department
Abstract
Objectives: In this study, we examine APP productivity to determine if a similar pattern applies to that of residents and attendings, with the hypothesis that hourly productivity decreases after the first few hours of the shift.
Background: APP productivity follows a similar pattern to previously described behaviors in both residents and attendings. This further confirms the results of prior studies that productivity is a dynamic process that needs to be considered when adjusting staffing models. Additionally, this pattern by APPs at a community hospital provides additional validation of this model outside of academic institutions or training environments.
Methods: This is a retrospective cohort study from 7/1/21 through 6/30/21 at a single suburban community hospital in the northeast. APPs work ten hour shifts from 10AM to 8PM, nearly every day. APPs also provide coverage for approximately 5 shifts per month from 3PM to 11PM. Timestamps of initial patient contact are automatically logged by the electronic health record, and then analyzed to determine in which hour of the shift this occurred. A mixed linear model was performed with the hour as a categorical variable, and day of the week, month and year as covariates. Data was grouped by individual shifts.
Results: A total of 345 10-hour shifts were worked by five APPs over the one year studied. There were 64 additional afternoon shifts which were excluded due to the shorter length and lower frequency. Two APPs worked the vast majority of the shifts, and the other three provided per-diem coverage. A mean of 13.3 patients (SD 2.7) were seen per shift. In the first hour, APPs saw an average of 2.67 patients (95% CI 2.59-2.76). Each hour demonstrated a statistically significant decrease relative to the first hour (p< 0.001), with the highest magnitude over the second (-0.58 (95% CI -0.69 — -0.47)) and third (-0.98 (95% CI -1.09 — -0.86)) hours. This downward trend continued throughout the rest of the shift; however, the magnitude of this difference after the sixth hour was not significant. There was no effect by day of week, month or year.
Conclusion: APP productivity follows a similar pattern to previously described behaviors in both residents and attendings. This further confirms the results of prior studies that productivity is a dynamic process that needs to be considered when adjusting staffing models. Additionally, this pattern by APPs at a community hospital provides additional validation of this model outside of academic institutions or training environments.