Interhospital Variability in Quality Outcomes of Pancreatic Surgery.

BACKGROUND: Assessment of optimal patient outcomes from health care delivery is critical for success amidst current reform. We developed a composite index of quality for pancreaticoduodenectomy (PD) and compared high and low performers nationwide. METHODS: We performed a retrospective analysis of 17,220 patients undergoing elective PD between October 2010 and June 2014 using the Vizient database. A quality index score (QIS) was developed from five variables associated with optimal outcomes: postoperative complication rate, length of stay, 30-d readmission rate, mortality rate, and hospital volume. Value was defined as hospital-based QIS divided by mean hospital charges. High-value centers (top quintile) were compared to low-value centers (bottom quintile). RESULTS: The majority of high-value centers (79%) achieved top performer status in 1-2 of five quality categories though only 11% were low performer in at least one category. Conversely, 41% of low-value centers were top performers in at least one category, although rarely more than one (8%); 63% of low-value centers were low performers in two or more categories. There was no significant association between QIS and hospital charges (-570, 95% CI -1308 to 168, P = 0.13). CONCLUSIONS: High-value centers infrequently provided high quality surgical care across all five metrics but instead excelled in a few quality metrics while avoiding low performance in any quality metric. Although low-value centers could achieve excellence in one quality metric, they were frequently low performers in two or more outcomes. Improvements in value of PD can be achieved by a consistent effort across all quality metrics rather than efforts at constraining financial expenditures of health care delivery.