Chronic kidney disease and household behaviors in Sri Lanka: Historical choices of drinking water and agrochemical use

This paper examines whether there are systematic differences in the historical behaviors of households that are affected and unaffected by chronic kidney disease (CKD) in Sri Lanka pertaining to their water source choices, water treatment practices, and agrochemical use. This analysis is motivated by the Sri Lankan government’s largest policy response to this epidemic – to encourage communities to switch from untreated well water to publicly provided alternatives. We use recall methods to elicit information on the drinking water source and treatment choices of households over an 18-year period from 2000–2017. Our analysis is based on a survey of 1497 rural ground-water dependent households in the most CKD-affected areas of the 10 districts of Sri Lanka with the highest prevalence of CKD. Our main findings are that (a) households that have ever used a pump to extract (typically deep) drinking water from a household well are more likely to be affected by CKD; (b) we fail to find a relationship between disease status and households’ use of buckets to extract (typically shallow) groundwater from their wells; and (c) those who have ever treated their shallow well water by boiling it are less likely to be affected by CKD. We also find that a greater share of CKD affected households historically used agrochemicals, used wells that were geographically removed from surface water sources, and displayed lower proxies of wealth. The implications of these findings are fourfold. First, since the systematic differences in the historical patterns of water sources and treatments used by CKD affected and non-affected households are modest, the sources of water and the treatment practices themselves may not be the sole risk factors in developing CKD. Second, although we find a negative association between boiling water and the probability of CKD, it is not obvious that a public policy campaign to promote boiling water is an appropriate response. Third, the hydrochemistry of deep and shallow well water needs to be better understood in order to shed light on the positive relationship between deep well water and disease status, and on why boiling shallow but not deep well water is associated with a lower probability of CKD. Fourth, there is a need for a deeper understanding of other risk factors and of the efficacy of preventative programs that provide alternative sources of household drinking water.