Establishment of quality of care among hospitals in digestive surgery - E-PASS study

I. Results for short-term outcomes In this study, we compared our prediction model E-PASS and its modified form mE-PASS with Western models, American Society of Anesthesiologists (ASA) status-based model and Portsmouth modification of Physiologic and Operative Severity Score for the enUmeration of Mortality and morbidity (P-POSSUM) in elective gastrointestinal surgery. The area under the receiver operative characteristic curve (AUC, 95% CI) values were 0.86 (0.79-0.93) for E-PASS, 0.86 (0.79-0.92) for mE-PASS, 0.81 (0.75-0.88) for P-POSSUM and 0.73 (0.63-0.83) for the ASA status-based model. Subsequently, we quantified the ratio of observed- to-estimated in-hospital mortality rates (OE ratio) as a measure of quality. The OE ratios for mE-PASS among large volume hospitals significantly correlated with those for other existing models. II. Results for long-term outcome gastric cancer resection We generated a model to predict overall survival (OS) using Cox hazard regression analysis in gastric cancer resection. This model exhibited a better discrimination power (AUC, 95% CI: 0.89, 0.86-0.91) than that for the UICC stage (0.81, 0.77-0.84). Subsequently, we quantified the OE ratio of 5-year OS rates as a measure of quality. The OE ratios among the participating hospitals revealed no significant variation between 0.74 and 1.1. III. Results for long-term outcome colorectal cancer resection We generated a model to predict OS using Cox hazard regression analysis in colorectal cancer resection. This model exhibited a better discrimination power (AUC, 95% CI: 0.87, 0.85-0.90) than that for the UICC stage (0.80, 0.76-0.83). Subsequently, we quantified the OE ratio of 5-year OS rates as a measure of quality. The OE ratios among the participating hospitals revealed no significant variation between 0.90 and 1.1.