Purpose: We examined the agreement between patient reported cancer treatment to treatment reported by the Surveillance Epidemiology and End Results (SEER) registry for breast and prostate cancer patients.

Methods: Patients (n=495 breast patients and n=294 prostate patients) were identified from the Metropolitan Detroit Cancer Surveillance System. We compared cancer treatment documented in both sources using kappa statistics and further examined the potential influence of demographic and cancer characteristics on overall treatment agreement measured as a dichotomous variable for chemotherapy, radiation, surgery and hormone treatments.

Results: Patients generally reported obtaining more treatment than what was recorded in the SEER registry. Breast cancer patients had moderate levels of agreement for receipt of chemotherapy (k=0.51), radiation (k=0.58) and surgery (k=0.48). In contrast, prostate cancer patient and SEER reports achieved nearly perfect concordance for radiation therapy (k=0.84), substantial agreement for surgery (k=0.77) and moderate levels of agreement for hormone therapy (k=0.51). Gender significantly influenced agreement for radiation (p<0.01) and surgery (p<0.05) therapies. Men were more likely to have agreement for radiation reporting but less likely to have agreement on surgical therapy. Age and comorbid conditions significantly influenced hormone therapy agreement (both p<0.05). Perceived health status significantly influenced surgery agreement (p<0.05). Household income significantly influenced radiation agreement (p<0.05). Cancer stage significantly influenced chemotherapy (p<0.01), surgery (p<0.05) and hormone therapy agreement (p<0.05).

Conclusions: Possible explanations for treatment discrepancies between patient-report and the SEER registry include recall issues, lack of patient understanding of treatment, poor communication between patient and interviewer, and/or incomplete or inaccurate SEER data. These results suggest that neither patient report nor SEER registry should be used as a sole source of data. Patients generally report obtaining more therapy than SEER registry documents. These data do not demonstrate the concordance that other studies have described between SEER registry and Medicare claims files. This analysis suggests that while patients accurately report their cancer treatments to a moderate degree, the most accurate information would likely result from supplementation by additional data sources such as medical record audit or health care claims information. As a long-term objective, more investment needs to be made in bolstering patient understanding of treatment received and cancer registry completeness.