Summary
An analysis of three Bangladesh household surveys estimates that the share of drinking-water sources above 10 ppb arsenic fell between 2012–13 and 2025. Different test kits used across the surveys make the size of the decline uncertain.
An analysis of 11,624 households in Bangladesh’s 2012–13, 2019 and 2025 Multiple Indicator Cluster Surveys estimates that the share using drinking-water sources with arsenic above 10 parts per billion (ppb) fell between the earliest and latest surveys. After laboratory adjustment of the 2012–13 readings, the estimated share declined from 24.0% to 14.4%.
The study is a medRxiv preprint, not yet peer reviewed. It compares surveys that used different test kits, so the authors caution that the size of the change depends on how those kit differences are accounted for. The 10 ppb threshold is the WHO drinking-water guideline; the analysis concerns source-water exposure, not health outcomes in the surveyed households.
The decline depends on how kit differences are handled
For readings above 10 ppb, the adjusted ratio between the 2025 and 2012–13 estimates was 0.61, with a sampling 95% confidence interval of 0.55 to 0.67. When the analysis also accounted for interview month, the ratio was 0.69. A ratio below 1 indicates a lower estimate in 2025.
At the higher threshold of more than 50 ppb, the estimated share fell from 10.9% in 2012–13 to 8.6% in 2025. The adjusted ratio was 0.80 (sampling 95% confidence interval 0.69 to 0.93), but it was 0.93 after accounting for interview month. The authors conclude that the change at this higher level, and the change since 2019, is unclear after that adjustment.
The authors’ sensitivity analysis highlights why the test kits matter: assuming no false-positive readings, the decline above 10 ppb would disappear only if later kits detected 61% of sources that were actually above the threshold. Their estimate for 2025 is that about 26 million people used water above 10 ppb, with kit scenarios ranging from 23 million to 32 million. In household terms, the paper reports that about one in seven exceeded the WHO guideline.
Earlier high-arsenic districts held most of the 2025 exposure
The analysis also found that the 25 districts identified as high in both the 2012–13 and 2019 surveys contained 39% of Bangladesh’s population but 72% of people estimated to be exposed above 10 ppb in 2025. The bootstrap 95% interval for the exposed population share was 60% to 76%. The study found geographic clustering, although no individual district was a statistically significant hotspot.
The authors say this pattern supports prioritising well testing in districts that were high in earlier surveys, alongside broader surveillance. They also recommend using a consistent test kit in future surveys with a laboratory cross-check, which would make changes over time easier to interpret.