Personalised estimates of dosage within a population-based cohort suggest limited evidence that drinking water chemistry modulates health outcomes

Access to clean drinking water is crucial for human health, but there are concerns that consumption levels of particular solutes may be linked to negative health outcomes. Many studies use data aggregated to area-level to assess consumption-health associations, but these do not account for differences in behaviour between individuals. Therefore, we combined publicly reported tap water chemistry from compliance data with estimated drinking water consumption within the home calculated from a water habits survey for 1970 UK adults from the TwinsUK cohort. The resulting average daily dose (ADD) estimate of key solutes (including nitrate, hardness, chlorine, selenium), multiple solutes dimensionally reduced to k-means clusters, and ingestion rate (IngR) were used as predictors in nested regression analysis, including stratification for employment status (as a proxy for time spent and home) and adjustment for age. Health outcomes included cardiovascular disease (CVD), CVD risk factors, eczema, sarcopenia, frailty, gastrointestinal disease and cancer. We found estimation of ADD significantly changed the distribution of solutes (ks-tests p < 0.001). Overall, we identified few associations with health outcomes, with mixed or inconsistent signals across nested models. There were some exceptions, including increased odds of hardness predicting CVD risk factors in our employment strata (Hardness ADD OR:1.25, 1.09-1.48, p < 0.002; age-adjusted OR: 1.22, 1.05-1.4, p < 0.002) and eczema negatively associated with IngR in non-stratified models (IngR OR: 0.86, 0.77-0.96, p < 0.01). We found no evidence of difference in ADD in twins discordant for health outcome. The lack of significant findings in solute-health pairings could be inferred as a positive result that supports UK drinking water as a safe source of hydration and therefore our results represent a positive public health outcome. Our study demonstrates an effective approach to estimate personalised exposure to water solutes that can inform researchers designing similar studies.