This cross-sectional study used pooled 2023–2024 National Health Interview Survey (NHIS) Adult Sample data to examine whether current self-reported immunosuppressive treatment was associated with self-reported lifetime stroke history among United States adults. The authors hypothesized that adults reporting current immunosuppressive treatment would have higher odds of reporting prior stroke after adjustment for demographic, socioeconomic, behavioral, and cardiometabolic factors.
The NHIS is a nationally representative, stratified, multistage probability survey of the civilian, noninstitutionalized US population conducted by the National Center for Health Statistics. Public-use 2023 and 2024 adult sample files were pooled, and sampling weights were halved to account for two years of data. The final complete-case analytic sample included 60,026 adults with valid survey weights and complete data on exposure, outcome, and selected covariates. Participants with missing covariate data (2,008 of 62,034 eligible adults) were excluded, representing about 3.2% of the eligible sample. The authors used survey design variables (weights, strata, primary sampling units) for estimation and variance calculation.
The exposure was current self-reported use of medications that weaken the immune system, as ascertained by the NHIS item MEDRXTRT_A. The primary exposure was coded as yes versus no. The NHIS item does not capture medication class, dose, indication, or disease severity; therefore, the exposure should be interpreted as a broad marker of immunosuppressive treatment status rather than a specific therapeutic agent.
The outcome was self-reported lifetime stroke history, defined using the NHIS item STREV_A. The NHIS provides a single self-report item without physician confirmation or stroke subtype classification; thus, “stroke” in this analysis refers broadly to any self-reported lifetime cerebrovascular event.
Covariates selected a priori included age, sex, race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, Other), hypertension, type 2 diabetes, smoking status (never, current every day, current some day, former), coronary heart disease, and income-to-poverty category. Other established stroke-related factors (body mass index, hyperlipidemia, physical activity, atrial fibrillation, chronic kidney disease) were not included in the primary adjustment set and were discussed as potential sources of residual confounding.
All analyses accounted for the NHIS complex survey design. Descriptive statistics used survey-weighted proportions and means. Survey-weighted logistic regression estimated odds ratios (ORs) and 95% confidence intervals (CIs). Sequential models were fit: unadjusted; Model 1 adjusted for age, sex, and race/ethnicity; Model 2 added hypertension, diabetes, and smoking; Model 3 (primary fully adjusted model) additionally included coronary heart disease and income-to-poverty category. Exploratory interaction tests examined effect modification by sex, age group, race/ethnicity, diabetes status, and hypertension status. No correction for multiple comparisons was applied for exploratory subgroup tests.
In the weighted sample representing approximately 250 million US adults, 2,975 participants reported current immunosuppressive treatment, corresponding to a weighted prevalence of 4.6%. Participants receiving immunosuppressive treatment were older (weighted mean age 52.4 vs 47.9 years) and had greater comorbidity burden: higher weighted prevalences of hypertension (42.1% vs 31.5%), diabetes (15.5% vs 9.6%), and coronary heart disease (9.4% vs 4.7%). Stroke history was more common among adults receiving immunosuppressive treatment than among those not receiving treatment (6.0% vs 2.8%, p < 0.001).
In unadjusted survey-weighted logistic regression, current immunosuppressive treatment was associated with higher odds of self-reported lifetime stroke history (OR 2.24, 95% CI 1.86–2.70, p < 0.001). Adjustment for age, sex, and race/ethnicity reduced the estimate (OR 2.00, 95% CI 1.65–2.44). Further adjustment for hypertension, diabetes, and smoking yielded OR 1.84 (95% CI 1.51–2.24). In the fully adjusted model (including coronary heart disease and income-to-poverty), the association remained statistically significant (adjusted OR 1.71, 95% CI 1.40–2.08, p < 0.001).
In the fully adjusted model, older age, hypertension, diabetes, current and former smoking, coronary heart disease, and lower income-to-poverty category were independently associated with higher odds of stroke history. Coronary heart disease (OR 2.41, 95% CI 2.10–2.78) and hypertension (OR 2.25, 95% CI 1.95–2.59) showed particularly strong independent associations.
Exploratory interaction analyses suggested possible effect modification by age and race/ethnicity. The association between immunosuppressive treatment and stroke history was stronger among adults aged 55 years or younger (OR 2.67, 95% CI 1.76–4.06) than among adults older than 55 years (OR 1.35, 95% CI 1.10–1.67), with an interaction p = 0.018. Possible interaction by race/ethnicity was observed (interaction p = 0.041). No significant interaction was found by sex, diabetes, or hypertension. These subgroup findings were labeled exploratory because multiple tests were performed without correction.
A three-level exposure sensitivity analysis (definite, uncertain, none) produced nearly identical results for definite immunosuppressive treatment (OR 1.71, 95% CI 1.41–2.09). Participants with uncertain treatment status also had higher odds of stroke history (OR 2.03, 95% CI 1.03–4.02). Additional sensitivity checks addressing uncertain stroke responses and inclusion of body mass index and hyperlipidemia did not materially change the primary association.
Key limitations include the cross-sectional design, which prevents causal inference; the broad exposure definition without medication class, dose, indication, or disease severity; and reliance on self-reported stroke history without clinical confirmation or subtype data. Potential confounding by indication is important because the exposure likely reflects both medication use and the underlying conditions that necessitate immunosuppression. Several established stroke risk factors (e.g., atrial fibrillation, physical activity) could not be included in primary models due to NHIS data limitations.
The authors interpret the findings as indicating that current immunosuppressive treatment status identifies a medically complex population with higher cerebrovascular disease burden rather than proving that immunosuppressive medications cause stroke.
In this nationally representative sample of US adults, current self-reported immunosuppressive treatment status was associated with higher odds of self-reported lifetime stroke history after multivariable adjustment. The association likely reflects a combination of medication exposure and underlying disease burden; findings should be viewed as associative and hypothesis-generating rather than causal.