Residential water hardness and incident chronic kidney disease: A causal inference analysis in a prospective cohort study

• First prospective cohort study linking residential water hardness to incident CKD. • Waterborne Ca and Mg associated with increased CKD risk (HRs 1.12 and 1.08 per IQR). • Nonlinear exposure-response relationships plateau at high mineral concentrations. • GPS- and IPW-based causal models corroborated mineral-specific Ca and Mg associations. • Associations modified by age, smoking, employment status, and comorbidities. Chronic kidney disease (CKD) poses a global health burden, yet the impact of drinking water mineral content—especially hardness—on CKD development is poorly understood and causal evidence is limited. This study investigated the association between long-term exposure to water hardness and incident CKD risk using causal-inference approaches. We followed 488,663 UK Biobank participants enrolled 2006–2010 without CKD for ≥1 year, tracking health outcomes to 2022 via linked records. Residential water hardness (CaCO3 equivalents, Ca, Mg) was quantified by postcode and analyzed as USGS/WHO categories, quartiles, and continuous variables. Incident CKD (ICD-10 N18) associations were assessed using multivariable Cox models adjusted for demographics, lifestyle, clinical factors, and PRS tertiles. Nonlinear trends were evaluated with penalized splines. To probe causality, we implemented causal-inference approaches including generalized propensity score-adjusted and inverse-probability weighting-weighted modeling. Over a median follow-up of 13.7 years, 21,091 of 488,663 participants (4.3%) developed CKD. Compared with the lowest CaCO3 quartile, the highest quartile had a HR of 1.05 (95% CI 1.01–1.10). CaCO3, Ca, and Mg per IQR were associated with HRs of 1.03 (1.00–1.06), 1.12 (1.08–1.16), and 1.08 (1.07–1.09), respectively. USGS moderate (60–120 mg/L) and very hard (>180 mg/L) categories, and WHO-defined hardness (≥200 mg/L), showed elevated CKD hazards. Nonlinear splines indicated plateauing at high concentrations. Causal analyses produced broadly concordant mineral-specific signals, although composite CaCO3 patterns were sensitive to model specification. Stronger associations were observed among older adults, smokers, unemployed participants, non-diabetics, and participants with hypertension, with no modification by sex or genetic risk. Elevated domestic water hardness—particularly Ca and Mg concentrations—was modestly associated with increased CKD risk, suggesting the potential inclusion of water mineral content in renal disease prevention strategies. While causal-inference methods largely corroborated mineral-specific associations, definitive causal claims remain limited by exposure misclassification, potential unmeasured co-contaminants, and the assumptions underlying GPS/IPW. Further studies with individual-level water measurements and broader contaminant panels are needed.