Abstract
Abstract
Background
Chronic arsenic exposure in drinking water causes skin lesions and cancers and raises infant, cardiovascular and all-cause mortality in Bangladesh. National surveys tested source water in 2012-13, 2019 and 2025 with different kits, so the trend is uncertain.
Objectives
We estimated how arsenic in source water changed after laboratory adjustment of 2012-13 readings, and how many people remained exposed. We also assessed whether districts high in earlier surveys held most people exposed in 2025.
Methods
We analysed 11,624 households in three Multiple Indicator Cluster Surveys. We counted arsenic strictly above 10 and 50 parts per billion (ppb). Survey-weighted Poisson regression compared rounds, after adjusting 2012-13 readings with the survey’s laboratory cross-check.
Results
Estimates reproduced published 2019 and 2025 tables. After laboratory adjustment of 2012-13, the share above 10 ppb fell from 24.0% to 14.4% (adjusted ratio 0.61, sampling 95% CI 0.55 to 0.67; 0.69 with interview month). Above 50 ppb it fell from 10.9% to 8.6% (0.80, 0.69 to 0.93; 0.93 with month). Most of the fall at 10 ppb occurred within source types. Assuming no false positives, the fall would vanish only if later kits detected 61% of sources above 10 ppb. About 26 million people used such water in 2025 (kit scenarios 23 to 32 million). The 25 districts high in 2012-13 and 2019 held 39% of the population but 72% of those exposed (95% CI 68 to 76). Clustering was evident, but no district was a significant hotspot.
Conclusions
Arsenic above 10 ppb was lower in 2025 than in 2012-13, but the size of the fall depends on unverified later kits. Change at 50 ppb and since 2019 was unclear after month adjustment. One household in seven exceeded the WHO guideline in 2025. Districts found high earlier held most exposed people, supporting prioritised well testing there alongside broader surveillance. Future surveys need one kit with a laboratory cross-check.