Determination of Minimum Surgical Caseloads for Major Oncologic Resections Using a Population-Attributable Fraction Model of Observational Data in Germany.

Objective: There is a volume-outcome association in cancer surgery; fulfillment of minimum surgical caseloads (MSCs) is known to be associated with reduced in-hospital mortality. To our knowledge, to date, there is no evidence-based approach to determine MSC with regard to in-hospital mortality.

Methods: Hospital billing data of pulmonary, esophageal, gastric, pancreatic, colon, and rectal cancer resections were analyzed. Nonfulfillment of annual caseloads of 5-100 procedures was defined as a risk factor of in-hospital mortality in a population-attributable fraction (PAF) model adjusting for age, sex, resection extent, and comorbidity. MSCs were obtained using a linear-trend approach. The primary end point was the fraction of attributable deaths due to nonfulfillment of MSCs. Driving distances to the treating hospital and closest MSC-fulfilling hospital were obtained using geocoding.

Results: A total of 824,535 patient records were analyzed. Resulting MSCs were 50 in pulmonary, 31 in esophageal, 31 in gastric, 48 in pancreatic, 28 in colon, and 43 per year in rectal resections. The PAF of nonfulfillment of the MSC was lowest in colon resections (8.8%, 95% CI, 1.0% to 16.5%) and highest in pancreatic resections (30.6%, 95% CI, 22.8% to 38.5%). The median difference in the driving distance (to the treating hospital v to MSC-fulfilling hospital) ranged between -3.5 km (IQR, -16.2 km to +0.2 km) in colon resections and +39.1 km (IQR, +0.3 km to +89.5 km) in rectal resections.

Conclusions: A PAF model is feasible in determining MSCs in cancer surgery with regard to in-hospital mortality; differences in driving distances to MSC fulfilling hospitals can be assessed using geocoding.