Predicting the Risk of Postoperative Respiratory Failure in Elective Abdominal and Vascular Operations Using the National Surgical Quality Improvement Program (NSQIP) Participant Use Data File

Objective: This study aims to develop a Respiratory Failure Risk Score (RFRS) with good predictability for elective abdominal and vascular patients to be used in the outpatient setting for risk stratification and to guide preoperative pulmonary optimization. Summary Background Data: Postoperative respiratory failure (RF), defined as ventilator dependency for more than 48 hours or unplanned reintubation within 30 days, is associated with increased mortality and hospital costs. Many tools have been previously described for risk stratification, but few target elective surgical candidates. Methods: Our training sample included patients undergoing inpatient, non-emergent general and vascular procedures sampled for the American College of Surgeon National Surgical Quality Improvement Program 2012 Participant Use File. Multivariable logistic regression identified independent preoperative risk factors associated with RF, used to derive a weighted RFRS. We then determined goodness-of-fit and optimal cutoff values through receiver operator characteristic analysis and Youden indices to evaluate internal and external validity with a retrospective institutional validation sample (2013 and 2014). Results: Multivariable analysis of 151,700 patients from the National Surgical Quality Improvement Program Participant Use File identified 12 variables independently associated with RF. The RFRS showed good external prediction in the validation sample with a c-statistic of 0.73 (95% confidence interval, 0.68-0.79). With the highest Youden index, 30 was determined to be the optimal cutoff value with a sensitivity 0.62 and specificity of 0.75. Additional cutoff values of 15 and 40 optimized sensitivity (>0.80) and specificity (>0.80), respectively. Conclusions: In the preoperative setting, the RFRS can effectively stratify patients into low (<15), moderate low (15-29), moderate high (30-39), and high risk (>39) to assist in patient counseling and guide application of perioperative pulmonary optimization measures.