Household illness and associated water and sanitation factors in peri-urban Lusaka, Zambia, 2016–2017

Rapid urbanization is concentrating population growth in urban and peri-urban areas of low-income countries, where basic water and sanitation infrastructure often lags behind demand. In 2015, billions lacked safely managed drinking water and sanitation, contributing to high burdens of diarrheal and respiratory diseases, especially among children under five. Zambia’s capital, Lusaka, exemplifies these challenges: legacy infrastructure serves a population far larger than it was designed for, and most peri-urban residents rely on communal water sources and pit latrines. Against this backdrop—and coinciding with a planned large-scale infrastructure investment by the Millennium Challenge Corporation—the CDC undertook a 2016–2017 household survey in peri-urban Lusaka. The study aimed to characterize household water and sanitation conditions, estimate recent household-level diarrheal and respiratory disease burden, and identify socio-demographic and WASH factors associated with these illnesses to inform policy and infrastructure planning.

Study design: A cross-sectional household survey was conducted from October 17, 2016 to October 26, 2017 across peri-urban areas (PUAs) of Lusaka, Zambia, spanning a full year to capture seasonal variation in outcomes. Sampling: A two-stage cluster sampling approach used Standard Enumeration Areas (SEAs) from the Zambia Central Statistics Office as primary sampling units. A total of 275 peri-urban SEAs were selected using probability proportionate to size. Within each SEA, a fixed number of households were randomly selected so that overall selection probabilities were equal irrespective of SEA size. Within each SEA, 11 households were additionally selected to receive testing of stored household water. Participants: Eligible respondents were residents aged ≥18 years identified as the head of household or spouse. Written informed consent was obtained. Ineligible households (e.g., absent after three visits, non-consent, residence <2 months, respondent <18 years, unavailable during field hours) were replaced. Data collection: The questionnaire (English, translated to Nyanja and Bemba) captured demographics, socio-economic measures (including detailed expenditure), WASH access and practices, illness among all household members in the prior 7 days, and water collection time. Outcomes: Three binary household-level outcomes were defined for the 7 days before interview: (1) waterborne illness (any household member with diarrhea and/or respiratory/flu-like illness), (2) diarrheal illness (any member with diarrhea), and (3) respiratory illness (any member with fever plus cough and/or sore throat). Diarrhea was defined as ≥3 loose/watery stools in 24 hours. Water quality testing: For selected households, approximately 120 mL of stored household drinking water (from jerry cans, buckets, or other containers) was collected aseptically into sterile, thiosulfate-treated bags, transported on ice, and analyzed same-day via membrane filtration to enumerate Escherichia coli (indicator of fecal contamination). Free chlorine residual (FCR) in stored water was measured on site. Per WHO guidance, FCR was dichotomized as ≥0.2 vs <0.2 mg/L; E. coli as present (≥1 CFU/100 mL) vs absent (<1 CFU/100 mL). Exposures: Candidate predictors included demographics (gender and education of household head, household size, presence of any child <5 years), socio-economic status (home ownership: rent vs own; household expenditure dichotomized at the 3rd quartile), and WASH indicators (drinking water source per JMP improved vs unimproved; daily per-capita water consumption for all purposes ≤20 vs >20 L/day; daily water availability <7, 7–11, >11 hours; storage container coverage some/none vs all; stored water treatment yes/no; self-reported handwashing with soap yes/no; heavy toilet sharing ≥18 vs <18 people). Sanitation facilities were classified per JMP as improved (pit latrine with slab, VIP, flush toilet) vs unimproved (pit latrine without slab, bucket/chamber pot, no facility/bushes/plastic bags). Statistical analysis: Three multivariable logistic regression models (one per outcome) with Generalized Estimating Equations accounted for clustering of households within SEAs and SEAs within PUAs (compound symmetry correlation structure). Model selection among correlated variables used quasi information criterion. A secondary analysis restricted to households with water quality testing added FCR and E. coli variables to the same multivariable framework. Ethics: Approved by the University of Zambia Biomedical Research Ethics Committee; CDC determined the activity to be non-research program evaluation.

Sample and prevalence: 12,511 households (representing 60,575 individuals, including 8,079 under-5s) were interviewed. At the household level, 26.2% reported any illness (diarrhea and/or respiratory) in the prior 7 days; 12.8% reported diarrhea; 15.9% respiratory illness. Individual-level 7-day prevalence: 3% diarrhea (1,827/60,575), 4% respiratory illness (2,370/60,575). Among under-5s, 8% had diarrhea and 10% respiratory illness. WASH context: 16–17% used unimproved drinking water sources; median water availability at source was 9 hours/day; median per-capita water consumption for all purposes was ~20 L/day. Nearly all households (98.8%) stored drinking water; 82.7% covered all storage containers; only 12.6–13% treated stored water on the interview day. Sanitation: 18–19% had flush toilets, but only ~1% were sewer-connected; 70% used pit latrines with slab; 21% reported heavy toilet sharing (≥18 people). Among tested households (n=3,147), 74.8% of stored water samples had E. coli present; 91.9% had FCR <0.2 mg/L. Associations in full sample (multivariable, aOR [95% CI]): - Larger household size (per additional member): illness 1.05 (1.03–1.08); diarrhea 1.05 (1.02–1.08); respiratory 1.06 (1.03–1.09). - Any child <5 years: illness 1.50 (1.36–1.66); diarrhea 1.68 (1.47–1.92); respiratory 1.39 (1.22–1.57). - Renting vs owning: illness 1.46 (1.29–1.65); diarrhea 1.28 (1.09–1.50); respiratory 1.54 (1.32–1.79). - Household expenditure >3rd quartile: illness 1.20 (1.08–1.34); diarrhea 1.17 (1.03–1.34); respiratory 1.23 (1.07–1.41). - No soap for handwashing: illness 1.20 (1.10–1.32); diarrhea 1.18 (1.05–1.34). - Storage containers not all covered: diarrhea 1.25 (1.06–1.46). - Unimproved sanitation: diarrhea 1.37 (1.06–1.77). - Heavy toilet sharing (≥18 people): illness 1.18 (1.06–1.31); diarrhea 1.33 (1.17–1.53). - Water consumption ≤20 vs ≥20 L/day: diarrhea 0.87 (0.77–0.98) (protective). - Education of head (secondary vs ≤primary): lower odds across outcomes; tertiary vs ≤primary protective for illness and respiratory outcomes. Associations in subset with water testing (n=3,147): - Child <5 years: illness 1.73 (1.42–2.10); diarrhea 1.92 (1.47–2.51); respiratory 1.36 (1.09–1.70). - Household size (per member): illness 1.07 (1.02–1.12); diarrhea 1.07 (1.01–1.13); respiratory 1.09 (1.04–1.15). - Renting: respiratory 1.55 (1.17–2.06); illness 1.35 (1.07–1.71). - Heavy toilet sharing: diarrhea 1.42 (1.07–1.88). - Unimproved sanitation: diarrhea 1.62 (1.00–2.63) (borderline p=0.050). - Water consumption ≤20 vs ≥20 L/day: diarrhea 0.70 (0.55–0.89) (protective). - Water availability <7 vs >11 h/day: illness 1.58 (1.00–2.49); 7–11 vs >11: illness 1.46 (1.03–2.05). - Detectable FCR ≥0.2 mg/L: lower odds of waterborne illness 0.65 (0.45–0.94). - E. coli presence in stored water: directionally higher odds but not statistically significant after adjustment. Overall interpretation: Socioeconomic and crowding factors, suboptimal hygiene and sanitation (heavy sharing, unimproved facilities), and water storage practices are linked to higher odds of recent household diarrhea and waterborne illness. Detectable chlorine and greater water availability are protective, while widespread contamination of stored water likely contributes substantially to risk.

The study addressed whether household socio-demographic and WASH conditions in peri-urban Lusaka are associated with recent diarrheal and respiratory illness. Findings indicate that larger households, presence of young children, renting, and higher expenditures are linked to greater illness burden, underscoring social determinants of health. For diarrheal and overall waterborne illness, several modifiable WASH factors—lack of soap use for handwashing, not covering all water storage containers, heavy sharing of toilets, and using unimproved sanitation—were associated with increased risk. Despite most households using JMP-defined improved sanitation, reliance on pit latrines and heavy sharing likely drives exposure, particularly in a high groundwater table context where latrines can contaminate shallow wells. Contrary to expectations from some prior literature, higher per-capita water consumption for all purposes was associated with greater diarrhea risk, plausibly reflecting extensive fecal contamination of stored household water and multiple exposure pathways in dense urban environments. The protective associations of detectable free chlorine residual in stored water and more hours of water availability highlight the health value of reliably chlorinated, continuous water service and the risks introduced by storage. Together, the results reinforce the potential health gains from transitioning peri-urban communities away from shared latrines toward safely managed sanitation, and from intermittent, storage-dependent water use to continuous, safely managed piped supply, aligning with SDG 6 targets.

In peri-urban Lusaka, recent household diarrhea and waterborne illness are common and are associated with both socioeconomic factors (household size, presence of children under five, renting) and modifiable WASH-related factors (lack of soap use, uncovered storage, heavy toilet sharing, unimproved sanitation). Widespread contamination of stored household water and limited free chlorine residual suggest that intermittent water service and storage contribute substantially to risk, whereas detectable chlorine and greater daily water availability are protective. The study underscores the need for investments that expand access to safely managed sanitation (including sewer connections) and reliable, chlorinated piped water to reduce reliance on storage and shared facilities. Future work should include rigorous impact evaluations of infrastructure upgrades on health outcomes, longitudinal studies to disentangle seasonal and causal relationships, and interventions targeting hygiene behavior and safe storage while infrastructure is being improved.

- Self-reported outcomes (diarrhea and respiratory illness) and several covariates (e.g., water consumption, household expenditure) are subject to recall bias; a 7-day recall window and standardized training were used to mitigate this. - Social desirability bias may affect responses on behaviors such as handwashing and water treatment. - Non-experimental, cross-sectional design limits causal inference and temporality; unmeasured confounding may remain. - While water quality was objectively measured in a subset, E. coli presence did not reach statistical significance in adjusted models, and findings may not generalize beyond the sampled peri-urban SEAs.