Study Selection Bias and Racial or Ethnic Disparities in Estimated Age at Onset of Cardiometabolic Disease Among Midlife Women in the US

IMPORTANCE: Racial disparities in cardiometabolic health are consistently observed in cohort studies. However, most studies neither evaluate differences in age at onset nor account for systematic exclusion stemming from “weathering” (accelerated health declines for minoritized groups due to structural social and economic marginalization). OBJECTIVE: To evaluate racial or ethnic disparities in age at onset of 4 cardiometabolic outcomes (hypertension, isolated systolic hypertension [ISH], insulin resistance [IR], and diabetes), accounting for multiple forms of potential selection bias. DESIGN, SETTING, AND PARTICIPANTS: This cohort study used data from the Study of Women’s Health Across the Nation longitudinal cohort (1995-2016) and a cross-sectional screening sample (1995-1997). Data were analyzed from July 2019 to October 2021. Participants were eligible for the cohort if they were aged 42 to 52 years, had not received hormone therapy in the past 3 months, were not pregnant, had an intact uterus and at least 1 ovary, and were premenopausal or early perimenopausal (most recent menses ≤3 months). EXPOSURES: Self-reported racial or ethnic group (Black, Chinese, Hispanic, Japanese, or White). MAIN OUTCOMES AND MEASURES: The main outcomes were hypertension (systolic blood pressure [BP] ≥140 mm Hg and diastolic BP ≥90 mm Hg or use of antihypertensive medication), ISH (systolic BP ≥140 mm Hg and diastolic BP <90 mm Hg or use of antihypertensive medication), IR (homeostasis model assessment for IR value >5.9 or insulin use), and diabetes (fasting serum glucose level ≥126 mg/dL [to convert to mmol/L, multiply by 0.0555], use of insulin or oral antidiabetic medication, or physician diagnosis). Selection into the cohort was addressed via inverse probability weighting and interval-censored survival models and selection out via multiple imputation. Accelerated failure time models were used to examine racial or ethnic differences in age at disease onset and estimate the median age at onset. RESULTS: A total of 3302 women were included in the study, with a median age of 46.2 years (range, 42-52 years) at baseline. In the sample, 42.6% had a bachelor’s degree or higher and 36.3% self-rated their health as “very good” at baseline; 23.9% had hypertension, 43.7% had ISH, 13.5% had IR, and 4.6% had diabetes at baseline. Hypertension occurred a median of 5.0 years (95% CI, 5.4-5.5 years) earlier and metabolic outcomes (diabetes and IR) a median of 11.3 years (95% CI, 9.7-12.9 years) earlier for Black and Hispanic women vs White women; ISH occurred a median of 7.7 years (95% CI, 7.3-8.1 years) earlier for Black women vs White women. Adjustment for selection was associated with a mean 20-year decrease in estimated median age at onset, with greater decreases among Black and Hispanic women. CONCLUSIONS AND RELEVANCE: In this multiracial cohort of midlife women, failure to account for selection biases, especially at study onset, was associated with falsely high estimates of age at cardiometabolic onset, with greater misestimation among Black and Hispanic women. The results suggest that hypertension and metabolic interventions, particularly for Black and Hispanic women, should be targeted to women aged as young as 30 years for hypertension and 40 years for metabolic interventions. Considering the timing of disease and fully addressing inherent selection biases in research are critical to understanding aging and disease risk, especially for racial and ethnic minoritized populations.