Primary payer data reported by cancer registries are subject to measurement error and can result in biased estimates of insurance-related policy effects.
Primary payer data reported by cancer registries are subject to measurement error and can result in biased estimates of insurance-related policy effects, according to a study by the National Cancer Institute (NCI) presented during the American Society of Clinical Oncology 2022 Annual Meeting.
Findings from the study showed that reporting of primary payer status collected by cancer registries at the time of diagnosis is underreported 38% of individuals enrolled in Medicaid and 42% enrolled in Medicare. Additionally, registry data concordantly identified 61% of patients enrolled only in Medicaid and 58% of patients enrolled only in Medicare.
“This study was motivated by the need to better understand the role of health insurance coverage on cancer-related care and outcomes, and the role of federal and state policy on insurance coverage,” Amy J. Davidoff, PhD, MS, wrote in an email to OncologyLive®.
“With new availability of the Medicaid enrollment data, which we consider to be the gold standard, we had the opportunity to examine whether the primary payer measure provided through registry abstraction was subject to errors. We also wanted to assess factors associated with errors in insurance assignment, to try to facilitate improvements or at the minimum, to help guide data users.”
Davidoff is a health economist and program director in the Healthcare Assessment Research Branch of the Healthcare Delivery Research Program at the National Cancer Institute in Rockville, Maryland.
The study used data from the Surveillance, Epidemiology, and End Results (SEER)Medicaid and -Medicare enrollment databases. Investigators examined adult patients aged 19 to 64 years who received a new diagnosis of cancer between 2007 and 2011 (N = 896,031). Study authors also linked in Medicare enrollment data and several state-year policy and insurance pattern measures pulled from the Kaiser Family Foundation, the American Community Survey, or Centers for Medicare & Medicaid Services reports.
The distribution of primary payer at diagnosis was mostly private (68%) among patients in the overall study population. Investigators also reported that patients in the overall population had Medicare (7.8%), Medicaid (10.1%), unknown insurance status (5.6%), uninsured (5.1%), and other government (3.3%) listed as the primary payer at diagnosis (Table).
Additional data from the study showed that older patients had a greater rate of underreporting of Medicaid enrollment compared with younger patients. Patients aged 60 to 64 years underreported at a rate of 58% compared with 46% of patients aged 19 to 39 years. Underreporting was also more common among patients who were eligible through a poverty-related or 1115 waiver expansion (64%) and among men (56%).
“For this age group [19-64 years], there was substantial underreporting or misclassification of both Medicaid and Medicare in the primary payer variable,” Davidoff said. “The misclassification of Medicaid, in particular, was more common for individuals with demographics less commonly covered by Medicaid pre–Affordable Care Act [ACA], [including] men and older adults. Underreporting was more common for Medicaidenrolled individuals who were eligible through a poverty-related or 1115 waiver expansion, and in state-years with heavy reliance on capitated managed care enrollment. This has important implications for using the primary payer variable to examine effects of the ACA on insurance coverage among individuals with newly diagnosed cancer.”
In the poster, the study authors concluded that strategies to enhance registry capture or to link enrollment data may help mitigate concerns of biased estimates of insurance-related policy effects.
Investigators have already examined data for patients younger than 19 years and older than 65 years and will pull together those results for dissemination, Davidoff said, adding that the NCI is extending the SEER-Medicaid data linkage through 2019.
“We expect to use the Medicaid and Medicare enrollment data to impute coverage, potentially editing the primary payer measure for our research,” Davidoff said. “With that new information, as well as a host of other area-level measures of social risk factors [a different research endeavor] we expect to revisit some of the questions regarding [effects] of the ACA on insurance coverage, and the effects of insurance coverage on cancer-related care and outcomes.”
Davidoff AJ, Enewold L, Williams C, Bhattacharya M, Sanchez JI. Reliability of cancer registry primary payer information and implications for policy research. J Clin Oncol. 2022;40(suppl 16):1587. doi:10.1200/JCO.2022.40.16_suppl.1587