Background: The influence of distance traveled for treatment on short- and long-term cancer outcomes is unclear.
Methods: Patients with colon, esophageal, liver, and pancreas cancer from 2003 to 2006 were identified from the National Cancer Data Base (NCDB). Distance traveled for surgical treatment was estimated using zip code centroids. Propensity scores were generated for probability of traveling farther for treatment. Mixed effects logistic regression for 90-day mortality and Cox regression for 5-year mortality were compared between patients treated regionally and those traveling from farther away.
Results: The mean distance traveled for all patients for surgical resection was 30.0 ± 227 miles, with a median distance of 7.5 (interquartile range 14.4) miles. Patients who were aged ≥80 years, on Medicaid, or African American were less likely to be in the fourth quartile of distance (Q4) traveled for surgery. Patients who were in Q4 had a lower risk-adjusted 90-day mortality compared to Q1 for colon [odds ratio (OR) 0.89, 95 % confidence interval (CI) 0.82-0.96], liver (OR 0.49, 95 % CI 0.3-0.78), and pancreatic (OR 0.74, 95 % CI 0.56-0.98) cancer. Similarly, patients in Q4 for all tumor types had a lower risk-adjusted 5-year mortality compared to patients in Q1; colon (hazard ratio (HR) 0.96, 95 % CI 0.93-0.99), esophagus (HR 0.84, 95 % CI 0.75-0.94), liver (HR 0.75, 95 % CI 0.62-0.89), and pancreas (HR 0.87, 95 % CI 0.80-0.95).
Conclusions: Greater travel distance for surgical resection of gastrointestinal cancers is associated with lower 90-day and 5-year mortality outcomes. This distance bias has implications for regionalization and reporting of cancer outcomes.