water · humanitarian · family: the context you didn’t design for
the filter works, a third use it
Point-of-use water treatment adoption stalls at 36% despite proven efficacy
Problem statement
Household water treatment (HWT) technologies — solar disinfection, chlorination, biosand filters, ceramic filters — demonstrably reduce waterborne disease, yet adoption across Sub-Saharan Africa is stuck at a pooled prevalence of only 36.31% (95% CI: 27.64–44.98%). A systematic review analyzing 147 articles identified 77 distinct barriers and 76 enablers. The dominant barriers are psychosocial (37.7%) and technology-related (28.6%), not economic — meaning cost reduction alone will not solve the problem. Technology-domain barriers outnumber technology enablers nearly 3:1 (28.6% vs. 10.5%), indicating that product usability, maintenance burden, and perceived reliability are the primary failure modes.
Why this matters
Unsafe drinking water causes an estimated 485,000 diarrheal deaths annually, primarily among children under 5 in low- and middle-income countries. Point-of-use treatment is the WHO-recommended intervention where piped infrastructure is unavailable, yet nearly two-thirds of households in the pooled Sub-Saharan studies were not practising it. The adoption gap perpetuates a preventable disease burden concentrated in the world's poorest communities.
What’s been tried and why it hasn’t worked
The 3:1 technology barrier-to-enabler asymmetry reveals that current PoU devices generate user frustration exceeding their perceived benefits. Specific failures include: taste and odor changes from chlorination that make treated water unpalatable; flow rate limitations in ceramic and biosand filters that make treated water unavailable when needed; breakage without local repair pathways; and maintenance complexity requiring knowledge that is not transmitted alongside the devices. Adoption varies enormously by country (Nigeria 71.4% vs. South Africa 17.0%), suggesting context-specific barriers that one-size-fits-all designs cannot address. Households with formal education were 2.38 times as likely to practise treatment, and household heads trained in water treatment 2.25 times as likely (pooled odds ratios).
What would unlock progress
Human-centered design approaches that prioritize usability testing in actual households — rather than laboratory performance — would address the technology barrier asymmetry. Co-design with local communities could resolve taste/odor concerns (the #1 barrier to chlorination adoption). Modular, locally repairable designs with visual indicators of treatment status would reduce the maintenance burden. Pairing technology distribution with training in water treatment would build on the documented 2.25-times higher odds of practice among trained household heads.
Entry points for student teams
A team could conduct a structured usability evaluation of 2–3 commercially available PoU treatment devices with a target community, documenting barriers using the COM-B (Capability, Opportunity, Motivation — Behavior) framework from the review. A design team could prototype a ceramic filter with a built-in visual indicator of flow rate and treatment status, addressing the maintenance knowledge gap. Relevant disciplines: public health, industrial design, water engineering, behavioral science.
Genes
Structural cousins: stuck the same way, in other fields
Sources
Chaúque, B.J.M. et al., "Why do low-cost point-of-use water treatment technologies succeed or fail in combating waterborne diseases in the field? A systematic review," Journal of Environmental Chemical Engineering, 11(5), 110575, 2023, Atalay, Y.A. et al., "Household Water Treatment Practice and Associated Factors in Sub-Saharan Africa: A Systematic Review and Meta-Analysis," Environmental Health Insights, 18, 2024, accessed 2026-02-20 go to source 1 ↗ go to source 2 ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
Meta-analysis of SSA household water treatment (Atalay et al.) retrieved 927 records and included 28 studies with 11,960 participants; the 147-article count in the Problem Statement is attributed to the other source, the Chaúque et al. systematic review, which could not be read in this round. The 3:1 barrier-to-enabler ratio in the technology domain is the key actionable finding. Related briefs: water-electrochemical-arsenic-removal (unviable economics of water treatment), water-field-pathogen-detection (pathogen detection challenge). The wrong-stakeholder tag applies because interventions target the household adopter when the binding constraint is the product design context. The rural paradox (rural 41.3% vs. urban 25.8% adoption) contradicts assumptions about infrastructure-dependent adoption.
Reconciliation 2026-09-30: Round-6 read of the whole write-up against its two sources. The second source (Atalay et al. 2024) was read in full on 2026-09-30 from https://pmc.ncbi.nlm.nih.gov/articles/PMC11421405/. The first source could not be read: the publisher page returns 403 to scripts, the Internet Archive holds only a publisher error page, and Unpaywall, OpenAlex, Europe PMC, Semantic Scholar and CORE list no open copy or abstract. Only its citation record was fetched, from https://api.crossref.org/works/10.1016/j.jece.2023.110575 (also confirmed at https://api.openalex.org and at https://ouci.dntb.gov.ua/en/works/98QvmbR7/). (1) The Source line named the first author as "Faria, N.R." The record for the linked article gives Beni Jequicene Mussengue Chaúque as first author (with Issufo, Benitez, Cossa, Chaúque, Stauber, Benetti and Rott), Journal of Environmental Chemical Engineering, volume 11, issue 5, article 110575, 2023. A search of Crossref found no paper with that title by an author named Faria. The Source line now reads "Chaúque, B.J.M. et al." with the article number added; title, journal, volume, issue, year and link are unchanged. (2) "Two-thirds of households that could benefit are not using it" went beyond the source, which reports a pooled prevalence of practice of 36.31% across 28 cross-sectional studies; the sentence now says nearly two-thirds of households in the pooled studies were not practising it. (3) "Education is a 2.38x adoption multiplier and structured training a 2.25x multiplier — but these enablers are rarely paired with technology distribution" was reworded: the source gives pooled odds ratios of 2.38 for formal education and 2.25 for household heads who received training about water treatment (it does not call the training "structured"), and it says nothing about how often training is paired with distribution; that clause is removed. (4) In the same spirit, "structured training programs would capture the documented 2.25x adoption multiplier" now says training in water treatment would build on the 2.25-times higher odds of practice among trained household heads. (5) The working note "covering 11,982 screened records, 147 included" did not describe the meta-analysis it was attached to: Atalay et al. retrieved 927 records and included 28 studies with 11,960 participants. The note now says so, and says that the 147-article count in the Problem Statement is attributed to the Chaúque et al. review, which could not be read in this round. Confirmed in Atalay et al. and left as written: the pooled prevalence 36.31% (95% CI 27.64–44.98); Nigeria 71.40% and South Africa 16.98% (written as 17.0%); rural 41.31% and urban 25.77%; the Environmental Health Insights, volume 18, 2024 citation; and point-of-use treatment as the option where sources are not upgraded. NOT checked, because they rest on the first source, which could not be read: the 147 articles, the 77 barriers and 76 enablers, the 37.7% psychosocial and 28.6% technology-related barrier shares, the 10.5% technology-enabler share and the "nearly 3:1" ratio, the list of specific device failures (taste and odour, flow rate, breakage, maintenance), taste and odour as "the #1 barrier to chlorination adoption", the COM-B framework, and the 485,000 diarrhoeal deaths a year, which is in neither readable text. Two framings are left as the write-up's own reading and are noted for Beth: "stuck"/"stalls" (the source reports one pooled level, not a trend), and "not economic", since Atalay et al. also find higher income associated with practice (pooled odds ratio 2.12). Titles unchanged. Genome tags untouched. A second reader went over these repairs on the same day (2026-09-30), reading Atalay et al. in full from the same PMC page and trying the publisher, the Internet Archive, Crossref, OpenAlex, Unpaywall, Semantic Scholar, Europe PMC, CORE, the Elsevier article and Scopus abstract endpoints and repository searches for the first source (no readable copy or abstract found; no Crossref record puts an author named Faria on this title); changes (1)–(4) were confirmed as written and the wording of (5) was tightened as above.