Abstract
BACKGROUND: Hypertension disproportionately affects older adults in rural settings. Labour-type physical activity is commonly assumed to protect cardiovascular health, yet it may not replicate the physiological stimulus of structured exercise. Evidence remains limited regarding how labour intensity and blood pressure-free obesity-metabolic phenotypes relate to hypertension in cold-climate rural populations. METHODS: This community-based cross-sectional study analysed 2,191 adults aged ≥65 years from a government health examination programme in a cold-climate agricultural county in Northeast China. Blood pressure-free obesity-metabolic phenotypes were defined using BMI, triglycerides (TG), and HDL-C only. Physical activity was quantified using a cohort-specific PA Index (weekly frequency × duration) and categorised as inactive (0), low active (0-<180), and high active (≥180; median among active participants). We fitted multivariable logistic regression models for prevalent hypertension, excluding blood pressure (BP) variables to avoid circularity, and developed leakage-free prediction models using non-haemodynamic predictors. RESULTS: The prevalence of hypertension was 75.9%. Hypertension prevalence differed across BP-free phenotypes (χ(2) = 28.07, p < 0.001), ranging from 75.1% (MHNO) to 89.7% (MUO). Compared with inactive participants, the odds of prevalent hypertension were higher in the low active group (OR = 1.85, 95% CI 1.32-2.58) and the high active group (OR = 2.00, 95% CI 1.61-2.49) (both p < 0.001). In leakage-free prediction, the primary LASSO model achieved a test-set ROC-AUC of 0.664 and PR-AUC of 0.841, with good calibration (slope 1.10). CONCLUSION: In this cold-climate rural cohort, BP-free metabolic-obesity phenotypes and higher labour-type activity volume were independently associated with higher odds of prevalent hypertension, which is not consistent with the assumption that "work equals exercise." Leakage-free prediction using routinely collected screening variables may help inform outreach prioritisation and follow-up planning in resource-limited agricultural communities, pending external validation and implementation evaluation.