Menu
Menu
Menu
Menu
Background: Diarrheal disease is the second leading cause of disease in children less than 5 y of age. Poor water, sanitation, and hygiene conditions are the primary routes of exposure and infection. Sanitation and hygiene interventions are estimated to generate a 36% and 48% reduction in diarrheal risk in young children, respectively. Little is known about whether the number of households sharing a sanitation facility affects a child’s risk of diarrhea. The objective of this study was to describe sanitation and hygiene access across the Global Enteric Multicenter Study (GEMS) sites in Africa and South Asia and to assess sanitation and hygiene exposures, including shared sanitation access, as risk factors for moderate-to-severe diarrhea (MSD) in children less than 5 y of age. Methods/Findings: The GEMS matched case-control study was conducted between December 1, 2007, and March 3, 2011, at seven sites in Basse, The Gambia; Nyanza Province, Kenya; Bamako, Mali; Manhiça, Mozambique; Mirzapur, Bangladesh; Kolkata, India; and Karachi, Pakistan. Data was collected for 8,592 case children aged <5 y old experiencing MSD and for 12,390 asymptomatic age, gender, and neighborhood-matched controls. An MSD case was defined as a child with a diarrheal illness 93%) had access to a sanitation facility, while 70% of households in rural Kenya had access to a facility. Practicing open defecation was a risk factor for MSD in children <5 y old in Kenya. Sharing sanitation facilities with 1–2 or ≥3 other households was a statistically significant risk factor for MSD in Kenya, Mali, Mozambique, and Pakistan. Among those with a designated handwashing area near the home, soap or ash were more frequently observed at control households and were significantly protective against MSD in Mozambique and India. Conclusions: This study suggests that sharing a sanitation facility with just one to two other households can increase the risk of MSD in young children, compared to using a private facility. Interventions aimed at increasing access to private household sanitation facilities may reduce the burden of MSD in children. These findings support the current World Health Organization/ United Nations Children's Emergency Fund (UNICEF) system that categorizes shared sanitation as unimproved.
Written informed consent was obtained from caretakers of enrolled children. The scientific and ethical review committees of each participating organization, including in-country ethics approval, and the Institutional Review Board of the University of Maryland, Baltimore, approved the protocol and consent forms (S1 Table). The seven GEMS sites included Basse Sante Su, The Gambia; Nyanza Province, Kenya; Bamako, Mali; Manhiça, Mozambique; Mirzapur, Bangladesh; Kolkata, India; and Karachi (Bin Qasim Town), Pakistan [33]. Two sites are located in urban centers (Mali and India), and four sites are in rural settings (The Gambia, Mozambique, Kenya, and Bangladesh), whereas the study villages in Pakistan, located on the coast approximately 20 km outside Karachi, are considered periurban. Each GEMS site was linked to a defined population under a demographic surveillance system (DSS) that visited every household 2–3 times per year to record births, deaths, and migrations. The GEMS is a matched case-control study in which cases were children <5 y old seeking care for MSD at one of the sentinel health centers serving the DSS at each site (S2 Table). MSD was defined as passing three or more loose stools within 24 h, in conjunction with clinical signs of moderate-to-severe dehydration (sunken eyes, loss of skin turgor, or administration of IV fluids), dysentery, or admission to a health facility. Stool specimens were collected from all children at enrollment. Control children without diarrhea were randomly selected from the DSS population within 14 d of presentation of the case and matched to the case by age, sex, and neighborhood. Detailed GEMS clinical and epidemiologic methods have been published [33,34]. Case and control enrollment into GEMS took place over 36 mo from December 1, 2007, to March 3, 2011. Demographic information collected about the case or control and his/her household (defined as a group of people who share a cooking fire) included maternal education and household size (including the number of children <5 y old). Building materials and household possessions were documented as potential indicators for constructing a wealth index for each site [33]. WASH data were collected at enrollment from the caretakers of case children presenting at health facilities and at home for matched control children by means of a standardized questionnaire. Approximately 60 (range: 50–90) days after enrollment, a trained field worker visited the household of each case and control to collect follow-up health information and record WASH observations. Information on water sources, facilities to dispose of human fecal waste, and handwashing and other hygiene practices was collected at enrollment, with additional information (including direct observation of hygiene practices, latrines, and toilet facilities) recorded at the 60-d follow-up home visit. Five sociodemographic variables were considered in this analysis as potential confounders (Table 1). A wealth index quintile (WIQ) variable was generated by principal component analysis of 13 household assets. This method has been described elsewhere [33,35]. Access to an improved water source was defined as the main source of drinking water for the household at follow-up as a public or private piped water tap, tube well, borehole, protected dug well, protected spring, or rainwater that was available every day, with a round trip time of 30 min or less to fetch water [16]. Abbreviations: GEMS, Global Enteric Multicenter Study; VIP, ventilated improved pit. *Denotes five wealth index quintiles, with 1 representing the poorest households and 5 representing the wealthiest households. Eight sanitation and hygiene variables were explored in this paper (Table 1), including three self-reported or observed sanitation variables, two directly observed fecal contamination variables, and two directly observed handwashing variables. Sanitation variables included facility type, facility access and sharing, and disposal of child’s feces as reported by respondent. Because of the skewed distribution of the number of households sharing facilities and for comparison across sites, the highest category for numbers of households sharing a sanitation facility was categorized based on the overall median of 3. Data analysis was limited to subjects for whom complete data were available on sanitation access at enrollment and follow-up. Data were analyzed using SAS version 9.3 (SAS Institute, Cary, NC). Descriptive statistics for sociodemographic and exposure variables were reported as proportions, medians, and ranges. We aimed to describe site-to-site variability in effects; therefore, we present all results stratified by site. The modeling strategy involved estimating unadjusted effects of association between sanitation and hygiene exposures and MSD and assessing two-way interactions between exposures and age, followed by selecting and including consistent sociodemographic confounders across all sites for reporting adjusted estimates of sanitation and hygiene exposures. Site-specific univariable conditional logistic regression models were used to evaluate the relationship between sanitation exposure variables and MSD. Unadjusted matched odds ratios (mORs) and 95% confidence intervals (CIs) are reported. Since risk factors are likely to be different for infants, we assessed two-way interactions between risk factors and age. There were no significant interactions with age; thus, only main effects are reported, and all models still account for the age-, sex-, and geography-matched case-control design. For many of the primary sanitation variables of interest, there were low exposure frequencies, which limited the number of variables that could be included in multivariable conditional logistic models. Therefore, we ran separate multivariable conditional logistic models for each sanitation and hygiene exposure of interest. We considered the following variables as potential confounders: WIQs, caretaker education, parental residence in the household, other young children in the household, and access to an improved water source. We assessed for confounding one at a time in each of the models by identifying significant associations with MSD and effect size changes in our estimates of sanitation and hygiene variables. Based on previous research, we considered a priori that wealth was an important epidemiological factor associated with MSD and should be included in the multivariable models to produce adjusted estimates of sanitation and hygiene exposures [22]. We aimed to present consistent results across sites, so we adjusted for these same parameters in all site-specific multivariable conditional logistic regression models. Multivariable models were assessed for collinearity using condition index diagnostics.
