Variability of NSQIP-assessed surgical quality based on age and disease process

Recent national attention has focused on improving upon the surgical quality of hospitals across the United States. The American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database includes expected morbidity probability for each surgical patient. We sought to investigate the accuracy of this probability across the spectrum of general surgical operations and assess the variability based on the age and disease process.

Using the 2008 ACS-NSQIP database, we identified 190,929 operations that would be in the scope of practice of a modern general surgeon; the four most common included breast resection (n = 22,175; 11.6%), colon resection (n = 21,363; 11.2%), cholecystectomy (n = 20,889; 10.9%), and inguinal hernia repair (n = 11,709; 6.1%). We calculated the surgical observed versus expected morbidity rates (O/E) of each operation type and compared them by decile of patient age. We then determined the effect of case mix and patient age on theoretical hospitals performing at the NSQIP average.

There is substantial variability in O/E ratios when comparing these disease processes across deciles of age. For patients undergoing breast resections, 67.2% of morbidities were solely attributed to 30-d reoperations; colon resections had an O/E ratio greater than 1 for all age deciles except over 90 y old. For cholecystectomies and the majority of patients undergoing inguinal hernia repairs, there was a lower morbidity rate than expected. Case mix and patient age were found to independently affect assessment of hospital quality.

It is conceivable that general surgery case mix and patient age could independently affect the quality assessment of a hospital. This variability may have implications for overall quality measures.