Background: Surgical margin involvement is an important outcome after pancreatic cancer surgery; however, variation in pathologic review practices may limit its use as a quality indicator. Our objectives were to assess variation in hospital performance and the reliability of margin involvement after pancreatic cancer surgery.
Methods: From the National Cancer Data Base, patients who underwent pancreatic resection for stage I to III adenocarcinoma were identified. Risk-adjusted surgical margin involvement was evaluated using hierarchical regression methods, and variation in hospital performance and reliability was determined.
Results: From 1,002 hospitals, 14,889 patients underwent pancreatic resection for adenocarcinoma, and 3,573 (24.0 %) had an involved surgical margin (R1 22.8 %; R2 1.2 %). The strongest predictors associated with margin involvement were T stage [T3: odds ratio (OR) 2.08, 95 % confidence interval (CI) 1.68-2.59; T4: OR 7.26, 95 % CI 5.50-9.60; vs. T1] and tumor size (2-3.9 cm: OR 1.66, 95 % CI 1.39-1.98, ≥ 4 cm: OR 2.28, 95 % CI 1.90-2.74; vs. <2 cm). Factors associated with a decreased likelihood of margin involvement were the use of neoadjuvant therapy and hospital type (academic and National Cancer Institute-designated comprehensive cancer centers vs. community). At the hospital level, the mean risk-adjusted surgical margin involvement rate was 25.9 % and ranged 10.1 to 50.5 %. Twenty-one (2.1 %) hospitals had lower-than-expected and 17 (1.7 %) had higher-than-expected margin involvement. A minimum acceptable reliability of 0.4 was met after 13 cases and was achieved by 249 hospitals that performed 79 % of pancreatic resections assessed.
Conclusions: Despite differences in pathologic evaluation practices, hospitals can be feasibly and reliably provided comparative data on surgical margin status after resection for pancreatic cancer.