D. Stewart1, C. Janowak1, A. Liepert1, A. O’Rourke1, H. Jung1, S. Agarwal1 1University Of Wisconsin,Surgery,Madison, WI, USA
Introduction: American Society of Anesthesiologists-Physical Status (ASA-PS) classification assesses pre-anesthesia surgical risk. Numerous studies correlate higher ASA-PS classification with increased perioperative mortality. As the number of comorbidities in a traumatically injured patient is correlated to mortality rate, we evaluated if ASA-PS was an indicator of mortality risk for adult trauma patients.
Methods: Our prospectively collected and internally validated database at an academic Level I trauma center was retrospectively reviewed for adult patients for 2009-2013. ASA-PS scores were assigned based on patient comorbidities. Three different methods were used to reflect a lack of concordance on the consideration of patient age in establishing ASA-PS. In all three methods, NTDB-defined comorbidities were assigned an ASA-PS value and summed for each risk level. Patients with no comorbidities were considered PS1, while PS2 consisted of those with a single PS2 condition. Multiple PS2 conditions were considered multi-system disease, elevating a patient’s risk to PS3. Presence of 3+ PS3 conditions led to a PS4 classification. We then evaluated mortality rates as a primary outcome for each ASA-PS class using receiver operating characteristic (ROC) and Pearson Chi-Square analysis. Discharge disposition and major complications were assessed as secondary outcomes.
Results: Model 1 (ASA), considered patient age >70 as a PS2 comorbidity, yielded an ROC of 0.619 for predicting mortality. Model 2, not including age as a factor in ASA-PS (ASA–w/o Age), produced an ROC of 0.615. Model 3, Age-Modified ASA (AM-ASA), produced an ROC of 0.648 (p<0.001). Cross-tabulation revealed mortality rates of 2.4%, 2.4%, 4%, and 13.2%, for PS1, PS2, PS3, and PS4, respectively. ASA–w/o Age (2.4%, 2.7%, 3.9%, and 13.2%) showed a similar trend, as did AM-ASA (2.4%, 1.9%, 2.9%, 10.2%), albeit with a dip in mortality rate for PS2. All three ASA models had two-sided p<0.001 under Pearson Chi-Square analysis of mortality rates. For discharge disposition (ASA ROC=0.668; ASA–w/o Age ROC=0.650; AM-ASA ROC=0.693) and major complications (ASA ROC=0.648; ASA–w/o Age ROC=0.653; AM-ASA ROC=0.641) all three models showed moderate predictive power.
Conclusion: ASA-PS classification models show an association between higher risk status and increasing mortality rate. ASA-PS is moderately predictive of mortality, discharge disposition, and major complications per ROC analysis. AM-ASA performed significantly better for mortality and discharge disposition, indicating that age can serve as an adjustment to the codified system to improve accuracy in the trauma population.
