Predicting Daily Surgical Volume for an Academic Medical Center

Abstract Background Surgical volume is highly variable and perioperative managers expend substantial time and energy adjusting resources in the days before surgery to accommodate unexpected low or high volume days. Last-minute changes lead to increased workload, adjustment of human and material resources, and potential for errors and delays which can impact employee morale and retention. Our objective was to improve forecasting of daily surgical volume weeks in advance of the day of surgery to allow perioperative managers adequate time to adjust staffing levels and material resources. Methods Accumulating surgery schedule and block release data were collected from July 2017 to October 2018 from two hospital-based operating room campuses at an academic medical center. Regression analyses were performed on accumulating schedule data 7, 14, and 21 business days before the day of surgery to create predictive analytic models. Results Assuming average daily volume of 32 cases for Campus 1 and 43 cases for Campus 2, the 7-day-out-model had an average error of 11.0% (standard deviation (SD) 8.6%) and 9.9% (SD 6.6%), respectively. Although the 21-day-out model had an average error of 14.5% (SD 10.9%) for Campus 1 and 12.2% (SD 9.1%) for Campus 2, cross-validation results indicated predicted volumes were within 7 cases 81.4% of days and 68.6% of days. Conclusion Building on previously published research, a predictive analytic model was created that improved estimated daily surgical volume up to 4 weeks before day of surgery. Achieving face validity with perioperative management is paramount in order to make use of predicted volume estimates.