Sanitary inspection, microbial water quality analysis, and water safety in handpumps in rural sub-Saharan Africa

In sub-Saharan Africa (SSA) 42% of people lack access to a basic water supply, defined as an improved water source accessible within a 30 min fetching time. However, even people using a basic water supply may still be drinking contaminated water, as water quality is not considered directly. While improved water sources generally tend to produce water with better microbial quality than unimproved sources, a systematic review shows that many improved sources are still contaminated. A 2012 study estimated that 52% of the SSA population was exposed to fecally contaminated drinking water, and another reported that 31% of boreholes/tubewells in SSA exceeded the WHO guideline of no detectable fecal indicator bacteria per 100-mL sample. Ensuring water safety is especially difficult in rural SSA, where many people rely on small, community-managed systems for drinking water that tend to have poorer water quality than urban systems. Risk-based preventative management is recommended to protect and improve water safety in all settings through comprehensive risk assessment and risk management approaches. Sanitary inspection is a visual survey of risk factors that may contribute to the likelihood of fecal contamination in water systems and is considered an effective, low-cost risk assessment tool. WHO-type forms typically comprise 9–12 yes/no questions representing the presence/absence of sanitary risk factors; the sum yields a sanitary risk score often used to compare risk levels across systems. Several studies have explored the relationship between sanitary inspection and water quality. Despite a conceptual link, some studies found no significant association, leading some professionals to question sanitary inspection’s effectiveness. Prior assessments often assumed a positive, linear association between sanitary risk score and E. coli concentration and used standard statistical tests accordingly. However, sanitary risk score is a simplified output intended to identify observable risk factors and guide corrective action, not a comprehensive representation of system risk at a point in time. This study examines the relationship between sanitary inspection and microbial water quality from a new perspective using data from 1028 boreholes with handpumps in rural SSA, to address: (1) the nature of the relationship between sanitary risk score and E. coli occurrence, (2) how best to quantify and model this relationship, and (3) how delivery and monitoring of drinking-water service can be improved based on these findings.

Study population and setting: Data were collected as part of an evaluation of World Vision (WV) water, sanitation, and hygiene (WaSH) programs. Rural communities in 14 countries (Ethiopia, Ghana, Honduras, India, Kenya, Malawi, Mali, Mozambique, Niger, Rwanda, Tanzania, Uganda, Zambia, Zimbabwe) were included in data collection; Honduras and India were excluded from this analysis to focus on sub-Saharan Africa (SSA), leaving 12 SSA countries. Only boreholes with handpumps were analyzed for comparability. Sampling design: In each country, subnational administrative units were stratified into WV and comparison areas. For each country, 56 WV and 56 comparison primary sampling units (PSUs) were randomly selected using probability proportional to size. Large PSUs (>200 households) were subdivided into secondary sampling units (SSUs), with one SSU randomly selected per PSU. Enumerators mapped all households and water sources. In each sampling unit, 25 households were randomly selected for survey. All unprotected and improved water sources were surveyed and received sanitary inspections. Up to five functional, non-surface water sources per sampling unit were randomly selected for microbial water quality testing; if fewer than five eligible sources existed, all were tested. Sanitary inspection: An adapted WHO sanitary inspection form for boreholes with handpumps (10 yes/no items) was used. Each source received a sanitary risk score from 0–10 (count of risk factors present) and was categorized into risk classes: low (0–2), intermediate (3–5), high (6–8), very high (9–10). Microbial water quality: E. coli was assessed in 100-mL samples; presence/absence and categories (WHO risk categories) were used for modeling. Summary statistics described E. coli MPN/100 mL and sanitary inspection scores. Established statistical models: Following prior literature, analyses included: (1) logistic regression of E. coli presence/absence (binary) on sanitary risk score (ordinal) and on individual sanitary risk factors (binary); (2) ordinal logistic regression comparing WHO water quality risk category (ordinal) with sanitary risk score (ordinal) and with individual sanitary risk factors (binary); (3) Pearson’s chi-square test for independence of E. coli presence/absence and sanitary risk score. Alternative causal-pathway model: Sanitary risk factors were grouped into three categories reflecting contamination pathways: sources (e.g., latrine within 10 m, uphill latrine, other nearby pollution sources), carriers (e.g., stagnant water within 2 m; in an expanded model, recent rainfall within past two days added as a carrier), and barrier breakdowns (e.g., cracks in apron, damaged or broken drainage channel). For each category, a binary indicator was set to present if any risk factor in that category was present. Logistic regression modeled E. coli presence/absence as a function of all combinations of source, carrier, and barrier breakdown presence, with the a priori hypothesis that contamination would be most associated when all three are present. Model assessment: Models were compared using Akaike’s Information Criterion (AIC), with lower AIC indicating better fit; pseudo-R² values were also reported. Analyses were conducted in Stata/SE 13.1. Data availability: data available on request from the corresponding author due to privacy considerations.

- Sample: 1028 handpumps across 12 SSA countries; 805 (78%) had no detectable E. coli in a 100-mL sample. - Sanitary risk scores ranged 0–9; mean 3.4; roughly normal distribution. Most common risk: missing fence (78%); least common: latrine within 10 m (1%). - Established models (using overall sanitary risk score/class): No significant association between E. coli occurrence and sanitary risk score or risk class at 95% confidence. Best fit among these was logistic regression with binary E. coli outcome (AIC ≈ 1079). - Individual sanitary risk factors (multivariable logistic regression): Several factors associated with E. coli presence at 95% confidence: latrine on higher ground (OR = 0.60, p = 0.046), broken drainage channel (OR = 1.8, p = 0.005), cracks in the apron (OR = 0.56, p = 0.002). Model fit AIC ≈ 1074. - Alternative causal-pathway model (grouping factors as source, carrier, barrier breakdown): Overall weak explanatory power (AIC ≈ 1079; R² ≈ 0.0086). The scenario where only barrier breakdown was present (without identified source or carrier) was significantly associated with E. coli presence (OR = 3.6, p = 0.019). Two other scenarios showed associations at 90% confidence, including the expected scenario with source, carrier, and breakdown present. - Adding recent rainfall (past two days) as an additional carrier improved model fit (AIC improved from ~1079 to ~1070; R² increased from ~0.0086 to ~0.019), but no risk scenarios remained significant at 95% confidence. - Overall, sanitary inspections and microbial tests convey distinct information; a strong correlation between aggregate sanitary risk score and single-sample E. coli presence was not observed and is not expected.

The study shows that while sanitary inspection and microbial water quality are conceptually linked, aggregate sanitary risk scores do not significantly predict single-sample E. coli occurrence across diverse rural SSA settings. This aligns with prior literature and reflects both the design intent of sanitary inspections (risk identification and management) and the high spatial-temporal variability of microbial contamination. The alternative causal-pathway model provides a more nuanced framing by distinguishing contamination sources, carriers, and barrier breakdowns. Results suggest barrier breakdowns may be more predictive of contamination than other categories, potentially because infrastructure failures directly enable ingress regardless of whether sources or carriers were captured by the inspection form. The modest improvement when rainfall was included as a carrier underscores the importance of contextual and environmental data in explaining contamination events. These findings imply that sanitary inspections and microbial testing should be viewed as complementary. Sanitary inspections are stable over time, require less technical capacity, and yield actionable interventions (e.g., repairing cracks, drainage, and fencing), making them especially valuable where routine testing is infeasible. For policy and practice, enhancing sanitary inspection forms to better capture evidence-based sources and carriers and integrating external contextual data (e.g., rainfall, seasonality) could improve their diagnostic utility. For research, larger longitudinal datasets linking sanitary risks, meteorology, water quality, and health outcomes are needed, along with exploration of setting-specific factor importance and interactions. Ultimately, applying frameworks grounded in causal pathways and environmental determinants can better inform timely, effective risk mitigation.

This work clarifies that sanitary inspection scores and microbial water quality measurements provide distinct, complementary insights into water safety at rural handpumps in SSA. Established models show no significant association between overall sanitary risk scores and E. coli occurrence, whereas an alternative causal-pathway model offers a slightly better fit and highlights the role of barrier breakdowns and contextual factors like rainfall. The study supports prioritizing routine sanitary inspections for proactive management and repair, particularly where water quality testing is infrequent. Future research should refine sanitary inspection tools (including weighting and additional risk factors), integrate environmental and contextual data, and develop larger, longitudinal datasets to validate and enhance predictive models and inform targeted interventions.

- Data aggregated from 12 diverse SSA countries; sample size limited disaggregation by country/region, potentially masking context-specific relationships. - Models were relatively simple and did not include many potentially relevant covariates (e.g., well depth, soil/subsurface geology, community/rurality factors, population density, land use, sanitation coverage, intermittent vs continuous supply), which may improve associations if incorporated. - Single-sample microbial measurements may not represent temporal variability or central tendency of water quality; E. coli presence varies widely over short times and space. - Sanitary inspection forms do not weight risk factors; factor importance may vary by setting, season, and construction quality. Some risk factors were difficult to classify uniquely (e.g., damaged drainage channel could be a barrier breakdown or a carrier). - Carrier category in the base model included few items (e.g., stagnant water within 2 m), potentially undercapturing transport mechanisms; adding rainfall improved fit but did not yield strong associations.