LAUSR

Learning Objectives: APP utilization has been advocated to increase 24/7 coverage in the ICU. The APP workforce model has been demonstrated in a number of institutions, however, the optimal number of patients per provider has not been studied. Methods: This was a retrospective study at 3 hospitals (1 academic, 2 community/academic hybrid) with 126 ICU beds staffed with 24/7 APP coverage. Billing data from June 2016 to May 2017 was extracted for each APP along with patient data and shifts where billing occurred (night vs. day). This data was analyzed to obtain total billing and patients seen per day. The data was then indexed to provide hours per patient categorized by patients seen per day. The data was normalized using the log of hours per patient per day for each category. The inflection point was defined as the point nearest to zero (where there was a change in the direction of the curve). The change in y-axis per category was calculated to arrive at the inflection point for optimal patient load using the formula Δy/Δx. T test was used to compare patient loads between day and night shifts. Results: There were a total of 9167 billed patient encounters during the 1 year period, 6936 during days and 2,231 during nights. Overall, there was an average of 5.3 hours billed per APP per day. There was a significant difference in hours billed by days (5.7) vs nights (4.1) p = 0.0001. For the overall data the inflection point was 8 patients per day (0.012). For day shifts the inflection point was 6 patients (0.02). For nights the inflection point was 11 patients (-0.006). In analyzing night time patient loads there was no significant difference in APP hours per night between large ICUs (≥ 18 beds) with an average of 8 and small ICUs (< 18 beds) with an average of 7.3. When comparing ICUs with residents at night there was a numerical difference in APP hours billed per night between resident ICUs (6.3) and non-resident ICUs (9) but not significant (p = 0.1). Conclusions: In our Critical Care Center, optimal APP patient load was calculated at 6 for daytime hours and 11 overnight with possible limitations due to resident presence.