How collaboration between surgeons and medical oncologists affects outcomes for patients with pancreatic cancer.

Abstract

694 Background: The number of patients shared between medical providers can serve as a marker of collaboration, which may improve clinical outcomes. We evaluated the association of surgeon and medical oncologist collaboration with pancreatic cancer patient survival and Medicare expenditures. Methods: The SEER-Medicare linked database was queried for pancreatic cancer patients who underwent resection (2014-2019). Provider pairs were created by linking patient’s index surgeon and medical oncologist 6 months pre- / post-surgery. The 95 th percentile of patients shared by provider pairs was the exposure: ≥5 patients (high collaboration), <5 patients (low collaboration). Kaplan-Meier survival analysis and Cause-specific hazard modeling for pancreatic cancer mortality were performed. Quantile regression assessed Medicare spending at 1- and 6-months post-surgery. P<0.01 was considered significant. Results: Among 4,836 patients, 6.7% (n=325) had high collaboration of which a majority were white race (87.7%), frail (91.4%), had early-stage disease (74.8%), were female (54.8%), while few had rural residence (7.7%). Patients with high compared to low collaboration more often were top 50 th percentile for income (83.7% vs 76.7%, p=0.0038) and hospital volume (76.9% vs 48.6%, p<0.0001). The 5-year OS rate significantly increased with high collaboration (23.4% [17.8-29.4%] vs 17.9% [16.5-19.3%]; p<0.0001). Hazards for pancreatic cancer death decreased among patients with early-stage disease and treated at hospitals in the top 50 th percentile for volume (Table). Medicare spending at 1 and 6 months did not significantly differ based on collaboration among patients regardless of expenditure quartile. Conclusions: Collaboration between surgeons and medical oncologists is associated with lower mortality without increased Medicare spending among patients with pancreatic cancer. Facilitating collaboration between specialists can benefit patients undergoing treatment with curative intent. Hazards for pancreas cancer specific death among early-stage patients at hospitals in the top 50 th percentile for volume. Variable HR 95% CI p-value ≥5 Patients Shared (Ref: <5) 0.71 0.58-0.87 0.0010* Age ≥74 Years (Ref: <74) 1.34 1.19-1.50 <0.0001* Male (Ref: Female) 0.96 0.86-1.08 0.5294 White (Ref: Other) 1.00 0.82-1.21 0.9736 Top 50 th Income Percentile (Ref: Bottom) 0.87 0.75-1.01 0.0764 Any Comorbidity a (Ref: None) 1.16 1.02-1.31 0.0212 CoC b or NCI c Facility (Ref: Not) 0.87 0.76-1.00 0.0439 *p <0.01; a per Charlson Comorbidity Index; b CoC = Commission on Cancer; c NCI = National Cancer Institute. Frailty/rurality were excluded from cause-specific death analysis given small counts.