Abstract LB-156: New-onset cancer cases in FDA’s Sentinel System: A large distributed system of US electronic healthcare data

Abstract

Abstract Background: Stemming from the US Cancer Moonshot initiative and 21st Century Cures Act, efforts to harness information technology to evaluate safety and effectiveness of cancer treatment are of high public health interest. Pharmacoepidemiologic cancer etiology and safety evaluations of cancer treatment are predicated upon sufficient number of patients and adequate follow-up time. Objective: To investigate whether Sentinel System’s electronic healthcare data is useful for these purposes, we assessed observable enrollment surrounding a member’s new-onset cancer diagnosis. Methods: Using administrative claims paid by Medicare and 16 other insurers, we identified incident cases of 19 cancer types among 292.5 million members between January 2000 and August 2017. International Classification of Diseases (ICD) diagnosis codes were used to define the cancers, after confirming diagnostic code mappings across the 2015 switch from ICD-9 to ICD-10 era. Study inclusion required 1 year of enrollment without claims for each respective diagnostic code. For each cancer, we summed annual incident cases by sex, age group, medical and drug coverage type, and Medicare or other insurer. Prior to and following diagnosis, we examined median observable time and duration (≥1 yr, ≥2, ≥3, ≥4 yrs). End of observable time was defined by member disenrollment, death, or end of insurer data. Results: We identified 7,926,450 members with medical coverage and an incident cancer diagnosis. Medicare data constituted 49.7% of the identified cases, with similar enrollment duration trends as the 16 commercial insurers. Across cancer types, median duration of observable time prior to and following diagnosis ranged from 2.5-3.8 and 0.5-2.5 years, respectively; 1,701,600 (21% of the cases) had at least 4 years of observable time following cancer diagnosis. Observable time following diagnosis varied greatest by cancer type, as expected. Observable time ended most frequently because insurer data ended; death was the least common reason. When drug coverage was additionally required, the number of incident cancers reduced to 4,705,968 (41% decrease). Range in median observable time did not change markedly (0.5-2.3 years post-diagnosis), albeit the sex differential increased after requiring drug coverage (with 54% women and 46% men having ≥4 years post-diagnosis). Conclusion: A distributed system with routinely updated data and standardized analysis tools provides opportunities for Sentinel System to contribute rapid and large-scale assessments of cancer incidence. Results suggest observable time following cancer diagnosis may provide sufficient duration of follow-up for drug-related adverse events of relatively short latency. Consideration of study-specific eligibility criteria and utilization of specific drug products is warranted prior to performing in-depth pharmacoepidemiologic studies. Citation Format: Nicole R. Haug, Anita K. Wagner, Katherine A. McGlynn, Charles E. Leonard, Michael D. Nguyen, Jacqueline M. Major. New-onset cancer cases in FDA’s Sentinel System: A large distributed system of US electronic healthcare data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr LB-156.