Linking cancer registry data to health insurance claims databases could significantly improve the capture of cancer treatment data, investigators report.
“As the number of new chemotherapy agents and targeted drugs approved for cancer treatment increases, estimating the population-based survival differences related to these treatments is increasingly important,” wrote Mia Hashibe, PhD, from the Utah Cancer Registry, and coauthors in JCO Cancer Clinical Informatics.
In this study, researchers identified 13,533 reportable cancer diagnoses in the Utah Cancer Registry between January 2013 and June 2014, of which 10,759 (79.1%) had health claims data in the Utah all-payer claims database.
Among these 10,759 patients, 24.1% had identifiable claims for chemotherapy. By linking the registry with the health claims database, researchers were able to identify an additional 497 patients in the registry who received chemotherapy, 590 treated with hormone therapy, 326 treated with radiation therapy, and 1,190 treated with immunotherapy.
The addition of the health claims data increased the proportion of patients treated with chemotherapy to 27.6%, the proportion of patients treated hormone therapy increased from 14.1% to 18.8%, immunotherapy increased from 4.3% to 13.2%, and radiation therapy increased from 24.9% to 27.5%.
The health claims data was particularly comprehensive when it came to information about hormone therapy for breast cancer, chemotherapy for lung or colorectal cancer, radiation therapy in patients with lung or prostate cancer, and biologic therapy in patients with melanoma.
The health claims data was also able to provide more information about the chemotherapy agents used and the duration of treatment, the authors reported.
“Even after augmenting with APCD [all-payer claims data], there was an indication of under-reporting of chemotherapy for breast cancer by the cancer registry variable compared with abstraction,” the authors wrote. “A factor contributing to cancers that were coded as treated through the augmented cancer registry variable, but not from abstraction, was when the therapy was determined by the abstractor to not be the first course.”
However, the cancer registry did have some data about patient therapy that was not found in the health claims database.
The authors noted that this was a pilot study to see whether claims data could improve information on cancer treatment and link that data to information in cancer registries.
“Next steps will include routine linkage for additional years of diagnosis and incorporation of APCD information into registry treatment variables.”
The study was supported by the National Cancer Institute, the Centers for Disease Control and Prevention, the Huntsman Cancer Foundation, and the University of Utah, Salt Lake City. One author declared institutional funding from a pharmaceutical company. No other conflicts of interest were declared.
SOURCE: Hashibe M et al. JCO Clin Cancer Inform. 2019 Oct. doi: 10.1200/CCI.19.00027.