Novel surgical performance evaluation approximates Standardized Incidence Ratio with high accuracy at simple means

Background: Excess adverse events may be attributable to poor surgical performance but also to case-mix, which is controlled through the Standardized Incidence Ratio (SIR). SIR calculations can be complicated, resource consuming, and unfeasible in some settings. This article suggests a novel method for SIR approximation. Methods: In order to evaluate a potential SIR surrogate measure we predefined acceptance criteria. We developed a new measure - Approximate Risk Index (ARI). " Number Needed for Event" (NNE) is the theoretical number of patients needed " to produce" one adverse event. ARI is defined as the quotient of the group of patients needed for n_o observed events G_e by total patients treated G_a. Our evaluation compared 2500 surgical units and over 3 million heterogeneous risk surgical patients that were induced through a computerized simulation. Surgical unit's data were computed for SIR and ARI to evaluate compliance with the predefined criteria. Approximation was evaluated by correlation analysis and performance prediction capability by Receiver Operating Characteristics (ROC) analysis. Results: ARI strongly correlates with SIR (r² = 0.87, p < 0.05). ARI prediction of excessive risk revealed excellent ROC (Area Under the Curve > 0.9) 87% sensitivity and 91% specificity. Discussion and conclusions: ARI provides good approximation of SIR and excellent prediction capability. ARI is simple and cost-effective as it requires thorough risk evaluation of only the adverse events patients. ARI can provide a crucial screening and performance evaluation quality control tool. The ARI method may suit other clinical and epidemiological settings where relatively small fraction of the entire population is affected.