Background and objectives: The purpose of this study was to compare breast cancer patients' self-report and surveillance, epidemiology, and end results (SEER) abstract data regarding type of treatment received (radiation, chemotherapy, and hormonal therapies).
Methods and design: Patients 65 years of age or older diagnosed during 1999-2001 with stage I-II breast cancer and treated with conserving surgery were identified from the Iowa SEER registry; 307 (41% of those eligible) completed telephone interviews. SEER-registry abstract data also were obtained.
Results: Agreement between self-reports and SEER data varied by type of treatment, with almost perfect agreement for chemotherapy (kappa = 0.93) and moderate to substantial agreement for ever use of hormonal therapy (kappa = 0.61), receipt of radiation therapy (kappa = 0.60), and current use of hormonal therapy (kappa = 0.54). If the SEER data are assumed to be the "gold standard," the sensitivity was generally high (>87%) for all types of treatment. Specificity varied according to type of treatment: highest for chemotherapy (98.4%) and lowest for radiation therapy (49.0%). Predictive values positive and negative were above 75% across type of treatment.
Conclusion: Using self-reported data was an acceptable alternative to reviewing medical records for documenting some types of breast cancer treatment.