PS2-35: Variation in Capture of Chemotherapy Data: Preliminary Findings From the CRN Economic Rx Burden Pilot Study

Abstract

Background: A number of high-cost chemotherapy agents, including targeted biologic and genetic therapies, have been introduced to the U.S. in recent years. Much of the literature on use and costs of these new agents has been derived from clinical trials and Medicare/SEER-based studies. To date, few studies have been published describing diffusion and costs of new cancer therapies within or across health plan populations. Aims: To describe capture of chemotherapy use (antineoplastics, hormone therapies, immune therapies, infusion time and site, ancillary therapeutics) within and across four Cancer Research Network (CRN) health plans for older (age 50+) cancer patients. Methods: Using data from health plan pharmacy systems and the Virtual Data Warehouse (VDW) Tumor Registry (TR), Pharmacy (Rx), and Procedure (PROC) files, we created dichotomous variables from each file separately denoting capture of chemotherapy use for each cancer patient aged 50 and older for the years 2000–2007. Variables were created using chemotherapy-specific national drug codes (NDCs) stratified by type of product (antineoplastic, hormone, immune) and route of administration (oral, infused/injectable) (Rx), procedure codes (CPT4, HCPCS, CPT4 Q codes, and revenue center codes) (PROC), and chemotherapy treatment flags (TR). Data were then stratified by health plan, year, cancer site, and stage. Chi-square analyses were employed to measure variations over time, within and across the four health plans’ Group Health, KPCO, KPNW, and HFHS. Results: Preliminary findings show significant variation in chemotherapy capture by data source over time within and across the four health plans, and by cancer site and stage. While the VDW TR has complete capture of abstracted data that denote whether or not chemotherapy was recommended in the initial treatment plan, other VDW-based sources may not be as complete with respect to detailed capture of the products by NDC code, dosing, and infusion time and site. Conclusion: Additional analyses are required to investigate how missing data affect the use of VDW chemotherapy data for research and to develop the steps to improve data capture. Algorithms may be needed to impute the average cost of some therapies for subsets of cancer cases across the four health plans. This project complements and builds on several components of the CRN infrastructure projects, and will leverage the expansion the VDW with respect to the inclusion of chemotherapy services.