Linked insurance-tumor registry database for health services research

OBJECTIVE. Breast cancer screening and treatment data are often limited to restricted populations, including women older than 65 years old, The goal of this project was to develop procedures to link tumor registry and insurance claims databases on women younger than 65 years old with breast cancer and to assess the accuracy and validity of the linked dataset, METHODS. Iowa Cancer Registry (ICR) and Wellmark Blue Cross/Blue Shield of Iowa (BC/BS) membership files of women with incident in situ or invasive breast cancer from 1989 to 1996 were linked. An automated deterministic match was followed with visual inspection from three independent reviewers applying a matching protocol. Matched and overall registry data were compared to assess population representativeness. Claims from BC/BS for incident cases during 1994 were examined for coding of a recent breast cancer diagnosis or treatment. RESULTS, The final dataset included 4,397 matched cases of patients aged 21 years and older from 1989 to 1996. The sociodemographic and tumor characteristics of the ICR population younger than 65 years old (n = 7,469) with breast cancer or carcinoma in situ were nearly identical with those of the matched patients younger than 65 years old (n = 3,449), Nearly all (96%) of the 445 matched incident cases in 1994 had claims data (CPT, DRG, or ICD-9 code) indicative of breast cancer. Treatment patterns varied by data source, with agreement ranging from 76% to 82% CONCLUSIONS. The validity and generalizability of these data demonstrate their potential for further health services research among younger insured women with breast cancer. Additionally the process outlined may be useful for developing other datasets to study other cancers in the population younger than 65 years old.