M. T. Berrigan1, B. R. Beaulieu-Jones1,2, S. Odom1, A. Gupta1, C. Parsons1, A. Seshadri1, C. Cook1, K. Crowell1, T. Cataldo1, E. Messaris1, G. A. Brat1,2 1Beth Israel Deaconess Medical Center, Department Of Surgery, Boston, MA, USA 2Harvard Medical School, Department Of Biomedical Informatics, BOSTON, MASSACHUSETTS, USA
Introduction: The American Society of Anesthesiologists (ASA) physical classification status is a subjective assessment generated by the anesthesiologist as part of the routine pre-operative evaluation, and used as an easy-to-interpret, standardized way to classify and communicate the overall health of a patient prior to surgery. The ASA score has been shown to be closely related to the rate of post-operative complications, but the score does not integrate intra-operative events. Having previously shown that immediate post-surgery surgeon intuition is valuable in predicting post-operative complications, we sought to combine ASA score with a surgeon-generated assessment of risk that takes into account intra-operative findings to generate a more accurate risk assessment.
Methods: Surgeons at one academic center were surveyed before and immediately after surgery to evaluate perceived risk of post-surgery complication. ASA score and clinical outcomes were abstracted from an institutional registry. Binomial regression models predicting post-surgery complications, as defined by ACS-NSQIP, were trained using the ASA score assigned by the anesthesia provider combined with surgeon intuition data.
Results: Pre- and immediate post-surgery surgeon risk estimates were collected for 486 patients with the following ASA scores: 9 (1.9%) ASA 1, 180 (37%) ASA2, 220 (45.3%) ASA 3, 76 (15.65%) ASA 4, 1 (0.2%) ASA 5. The correlation between ASA score and pre-surgery surgeon risk estimate was 0.185, suggesting surgeons prioritize factors other than overall pre-surgery health status when predicting risk of post-operative complication. The overall complication rate was 29.4%. Model performance is shown in Table 1.
Conclusion: Anesthesiologists and surgeons use medical knowledge and clinical experience to assess overall health and stratify their risk of post-operative complication. The best performing model combines ASA and surgeon intuition. ASA score alone has high sensitivity for predicting post-operative complication – in general, patients with higher ASA scores are more likely to develop a complication. The addition of surgeon input to ASA score improves model specificity suggesting that surgeons reliably recognize when an operation went well and can identify patients who will recover without complication. The combination of these judgement-derived assessments can be used to predict a patient’s risk. Further analysis is needed to better understand the optimal way to combine these factors and under what circumstances this risk assessment is most valuable.
