Abstract
BACKGROUND: Certain factors, such as race or age, are known to be associated with variation in initiation of adjuvant chemotherapy for stage III colon cancer, but little is known about what factors are associated with completion of adjuvant therapy. To determine whether predictors of initiation also predict completion, we analyzed Surveillance, Epidemiology, and End Results (SEER) program data linked to Medicare claims. We investigated mortality as a means to testing the validity of the completion measure that we created. METHODS: We studied 3193 stage III colon cancer patients whose diagnosis was recorded in 1992-1996 SEER program data linked to 1991-1998 Medicare claims and who initiated adjuvant chemotherapy after colon cancer resection. We defined a measure of adjuvant chemotherapy completion as one chemotherapy administration claim in a month. We tested the validity of the created measure and its relation to 3-year cancer mortality adjusted for demographic, clinical, and environmental variables. We explored the association of patient characteristics and treating physician characteristics with chemotherapy completion by use of multivariable logistic regression modeling. RESULTS: Of the 3193 patients, 2497 (78.2%) completed the course. Risk of cancer-related mortality was statistically significantly lower among those completing chemotherapy (relative risk = 0.79, 95% confidence interval = 0.69 to 0.89) than those with no adjuvant therapy. Patients who were female, widowed, increasingly elderly, rehospitalized, and living in certain regions were less likely to complete adjuvant chemotherapy than other patients. Race and other clinical, environmental, and physician characteristics were not associated with completion of therapy. CONCLUSIONS: Factors associated with incomplete adjuvant chemotherapy may represent physical frailty, treatment complications, and lack of social and psychological support. Interventions to mitigate these influences are a logical next step toward increasing chemotherapy completion rates.