{"doi":"10.1111/jgs.18739","title":"Comparison of health and retirement study participants with and without linkage to Medicare claims","abstract":"The health and retirement study is a longitudinal, nationally representative study of older adults that has been collecting data since 1992.1 Participants are selected using a multistage probability sampling design and surveyed every two years in topics ranging from health, to finances, to cognition and physical function. The HRS data with a linkage to Medicare claims are particularly valuable, facilitating many high-impact health services studies.2 In any HRS wave, approximately 80%–90% of HRS participants consent to have their HRS survey linked to their Medicare claims. Given this high consent rate, adjustments for non-consent to Medicare linkage are not typical among claims-based analyses because there is an assumption of similarity with those who consent. However, systematic differences between participants who consent and those who do not could introduce bias. Understanding these differences could provide insights for possible adjustments of statistical analyses. Therefore, we aimed to examine the differences between HRS participants with and without a Medicare data linkage. We used data from 2018 HRS participants age 65+ with Medicare insurance coverage (N = 8439). Our primary outcome was participants' consent to having their Medicare claims linked with their HRS data. Consent was obtained during the core HRS interview, and all study participants had the option to consent or refuse consent. Sociodemographic variables included age and age categories (65–74, 75–84, ≥85), gender, self-reported race and ethnicity (White, Black, Hispanic, Another Race), marital status, education levels (less than high school, high school graduate, some college or higher), and household wealth. Health variables included the presence of chronic health conditions and difficulty with activities of daily living (ADLs). Insurance-related characteristics included current employment status and private insurance coverage. We used chi-squared statistics to determine how consent to Medicare linkage varied by sociodemographic, health, and insurance-related characteristics. Next, we estimated standardized mean differences (SMD) for these variables between those who did and did not consent to Medicare linkage.3 Finally, we assessed the differences between the two groups in two more ways: by restricting the analysis to only those age 70+, and by including the inverse probability of consent weighting (IPCW) in the analysis, where the probability of consent to linkage was estimated using age, race and ethnicity, and marital status. Participants had an average age of 75 years (SD: 8), 56% were women, 9% were Black, and 8% were Hispanic. Overall, 13% did not consent to link their Medicare claims to their HRS survey. Participants who did and did not consent to Medicare linkage differed across several sociodemographic and health characteristics (Table 1). Those who did not consent to Medicare linkage were younger (Non-consenting: 70 vs. Consenting: 75, p < 0.001), more often Black (13% vs. 9%, p = 0.004), married or partnered (70% vs. 59%, p < 0.001), and employed (30% vs. 19%, p < 0.001). Non-consenting participants were less likely to report cancer diagnoses (14% vs. 22%, p < 0.001), heart disease (22% vs. 34%, p < 0.001) or meet HRS dementia criteria (4% vs. 6%, p = 0.005), and reported less difficulty with walking (6% vs. 9%, p = 0.025), dressing (8% vs. 11%, p = 0.026), and bathing (5% vs. 8%, p = 0.005). In the analysis restricted to those 70+, only age, race and ethnicity and marital status remained significant (Supplemental Table 1). In the full sample, standardized mean differences for all variables except for age were 0.2 or less. When including IPCW in the analysis, all SMD values except for heart disease were less than 0.1 (Figure 1), indicating small difference between those who do and do not consent to Medicare linkage.4 Our study found that HRS participants who do and do not consent to Medicare data linkage differ in several characteristics. Specif","journal":"Journal of the American Geriatrics Society","year":2024,"id":472933,"datarank":0.10397207708399181,"base_score":0.6931471805599453,"endowment":0.6931471805599453,"self_citation_contribution":0.10397207708399181,"citation_network_contribution":0.0,"self_endowment_contribution":0.10397207708399181,"citer_contribution":0.0,"corpus_percentile":22.178386323199504,"corpus_rank":9377,"citation_count":1,"citer_count":1,"citers_with_citation_signal":0,"citers_with_endowment":0,"datacite_reuse_total":0,"is_dataset":true,"is_dataset_confidence":0.7895,"is_data_producer":false,"deposit_databanks":null,"is_oa":true,"file_count":0,"downloads":0,"has_version_chain":false,"published_date":"2024-01-01","fair_score":null,"fair_percentile":null,"algorithm_id":"datarank_citation_only_1hop_v6","ranking_scope":"data_only","authors":[{"id":273909,"name":"W. John Boscardin","orcid":"0000-0003-3121-9526","position":1,"is_corresponding":false},{"id":447298,"name":"Ashwin Kotwal","orcid":"0000-0002-6137-8512","position":2,"is_corresponding":false},{"id":275985,"name":"Matthew J. Miller","orcid":"0000-0002-1301-7149","position":3,"is_corresponding":false},{"id":444932,"name":"Irena Cenzer","orcid":"0000-0002-5839-1070","position":0,"is_corresponding":true}],"reference_count":1,"raw_metadata":null,"created_at":"2026-07-19T02:05:57.032095Z","pmid":"38206864","pmcid":"PMC11090725","fwci":null,"citation_percentile":null,"influential_citations":0,"oa_status":null,"license":null,"views":0,"total_file_size_bytes":0,"version_count":0,"fair_f":null,"fair_a":null,"fair_i":null,"fair_r":null,"fair_zscore":null,"fair_rationale":null,"fair_model":null,"fair_agent_version":null,"fair_fulltext_source":null,"fair_has_llm":null,"fair_computed_at":null,"clinical_trials":[],"software_tools":[],"db_accessions":[],"linked_datasets":[],"topics":[]}