Hygienic disposal of stools and risk of diarrheal episodes among children aged under two years: Evidence from the Ghana Demographic Health Survey, 2003–2014

Background Most childhood diarrheal illnesses are a result of the faeco-oral transmission of infected food, water, and unclean fingers. The present paper was conducted to estimate the prevalence of hygienic disposal of stools (HDS) and its associated factors, and further quantify the impact of HDS on diarrheal diseases among children under two years. Methods A cross-sectional design was used to evaluate three rounds of the Ghana Demographic Health Survey (GDHS) from 2003–2014 involving 4869 women with children aged under two years. The outcomes were prevalence of HDS and diarrheal diseases. Poisson regression model was employed to assess risk factors associated with HDS and dominance analysis was used to rank the important risk factors. Inverse Probability Weighting Poisson Regression Adjustment (IPWPRA) with Propensity Score 1:1 density kernel-based matching was employed to assess impact. Results The pooled prevalence rate of HDS was 26.5%(95%CI = 24.6–28.4) and it ranged from 18.7% (95%CI = 16.4–21.2) in 2014 to 38.8%(95%CI = 35.3–42.4) in 2003. Diarrhea diseases pooled prevalence was 17.9%(95%CI = 16.4–19.5) and ranged from 13.3%(95%CI = 11.1–15.9) in 2014 to 25.4%(95%CI = 22.2–28.9) in 2003. The overall growth rate for HDS and prevalence of diarrhea diseases, decreased by 21.6% and 11.4% respectively. The most important risk factors of HDS from dominance analysis included; age of the child, wealth index, and differences in region. From pooled data wealth index, increasing age of the child, and regional disparity constituted approximately 72% of the overall impact (Weighted Standardized Dominance Statistics (WSDS) = 0.30, 0.24, and 0.19 respectively). In 2014, they constituted approximately 79% (WSDS = 0.139, 0.177 and 0.471 respectively). The average prevalence of diarrheal diseases among children of women who practiced HDS reduced over the period of the GDHS compared to those whose mothers did not practice HDS [2008 ATE(95%CI) = -0.09(-0.16–0.02), 2014 ATE(95%CI) = -0.05(-0.09–0.01) and Pooled data ATE(95%CI) = -0.05(-0.09–0.02)]. Conclusion This analysis has provided empirical evidence of the impact of practicing HDS in Ghana from a national household survey. Implementation of the WASH agenda in this low-income setting requires a synergy of interventions and collaborations of actors (government, private and development partners) to improve water and sanitation facilities and to increase hygiene education to prevent the spread of diseases including diarrhea by 2025.

This analysis used cross-sectional study data from the Ghana Demographic and Health Survey (GDHS), which was conducted across the country in the then ten administrative regions. The GDHS is a nationally representative household surveys that offer data for a variety of population, health, and nutrition monitoring and impact evaluation variables. Data used for this analysis was from the fourth to sixth rounds of the GDHS. The major goals of the GDHS were to collect data on fertility and family planning behaviour, infant and child mortality, breastfeeding, antenatal care, children’s immunizations, and childhood diseases, nutritional status of mothers and children, use of maternal and child health services, and awareness and behaviour regarding AIDS and other STIs. The fourth round of GDHS was conducted in 2003, and approximately 6,600 households were selected nationwide. The fifth round GDHS (2008) selected 12,000 households while the sixth round (2014) selected 12,810 households across the country. Both 2003 and 2008 used 412 Enumeration Areas (EAs) selected from the 2000 Ghana Population and Housing Census (GPHC) used as a frame for the sample whiles 2014 used 427 EAs selected from the 2010 GPHC. The frame was first stratified into the 10 administrative regions in the country, then into rural and urban EAs. All the study rounds adopted a two-stage stratified cluster sampling method to obtain the sample for each survey year. In the first stage, EAs were selected with probability proportional to the EA size and with independent selection in each sampling stratum. The second stage entailed taking systematic sampling from a list of households in each of the EAs that had been selected. The study considered diarrhea disease as the primary outcome which GDHS measured subjectively. Participants living with a child under five years were asked whether the child had diarrhea diseases during the past two weeks preceding the survey. Answers included “Yes”, “No” and “Don’t know”. In this analysis, children aged two years or more and ‘Don’t know responses were excluded. The denominator has changed over time from children under age 5 to children under two in recent times [18]. HDS was the secondary outcome considered in this study and was generated in two steps. First, we generated safe disposal of child’s stool from GDHS data. GDHS asked women with children under age two years the manner of disposal of the child’s last stool ‘Used toilet/latrine’, ‘pot/rinsed in toilet/latrine’, pot/rinsed into drain or ditch’, ‘throw into garbage’, ‘buried’, ‘rinsed away’, ‘use disposable diapers’, ‘use washable diapers’, ‘left in the open/not disposed off’, and ‘other’ were options provided. By GDHS definition, disposal is safe if the child used the toilet or latrine, stools are rinsed into the toilet or latrine, or stools are buried. Secondary, improved toilet facility was also generated and based on the definition by the WHO/UNICEF Joint Monitoring Programme (JMP) for Water Supply and Sanitation [18]. The toilet facility was classified as improved if not shared, flush to-piped sewer system, septic tank, pit latrine, and unspecified; pit latrine—ventilated improved pit, with slab and composting toilet. From the above two approaches, HDS was generated if a woman with a child(ren) under two years practiced safe disposal and had access to an improved toilet facility (coded as 1 and otherwise as 0). The first approach to data analysis was to denormalize the individual women sampling weight since authors merged GDHS data from 2003–2014. In the present analysis, women aged 15–49 years at the time of the survey were used to estimate the sampling fraction. Analysis adjusted for the nature of the design of the GDHS, thus, adjusting for the denormalized sampling weights, stratification, and the primary sampling unit. Descriptive and test of independence analysis were performed by adopting the Rao-Scott test of independence to test the association of covariates with GDHS year of study among participants (S1 Table). Based on previous literature and additional variables in the dataset, 25 variables were identified a priori. Authors then employed Poisson regression method to assess the factors associated with HDS. Poisson regression was employed to estimate the adjusted prevalence (PR) ratio rather than odd ratio (OR). In a cross-sectional survey with prevalence rate 10% or more, the PR is the preferred choice for risk analysis [19]. After identifying significant factors associated with the secondary outcome variable from the Poisson regression, the authors employed a weighted dominance analysis (DA) to estimate the relative importance of significant factors associated with the secondary outcome variable. The Logit model in DA was employed which relies on estimating the coefficient of determination (R2) values of all possible combinations of explanatory variables (EVs) and measures the relative importance by adopting pairwise comparisons of all EVs in the model as they relate to the secondary outcome. DA is a statistical technique for comparing the relative importance of a predictor variable over another which is associated with an outcome variable [20]. Due to the cross-sectional design of the GDHS, a matching procedure was used to assess the impact of having access to hygienic disposal of stools on diarrhea disease. In order to study the effects of the exposure (hygienic disposal of stools), a matching procedure was used to select controls in a sample with the same covariate values as the treated sample [21]. The propensity score 1:1 number of matching per observation was adopted and the Epanechnikov kernel function was obtained to assess the impact. We then estimated the average treatment effect (ATE) of HDS on diarrhea disease using logit. The conceptual framework defining the analytical process adopted for achieving the study objective can be found in S1 Fig. All analyses were performed using Stata 16.1 and a p-value <0.05 was deemed significant. The study relied on the “Strengthening of Epidemiological Observational Research Report” (STROBE) statement [22] in writing the manuscript as presented in S1 Table. In addition, we estimated the overall growth rate of HDS and diarrheal diseases from 2003–2014 by adopting the formula r=(P1P0n)−1; where r = growth rate, P1 the current estimates in 2014, and P0 = the past estimate in 2003. Therefore, HDS r=(18.738.83)−1; thus SISD r = -0.2159 and Diarrhea r=(13.319.13)−1; thus Diarrhea r = -0.1136. The rate of percentage change within the periods was calculated by adopting the formula c=(x2−x1x1)*100; where c represents the relative change while x2 and x1 denote current and initial values respectively. The GDHS protocol was reviewed and approved by the Ghana Health Service Ethical Review Committee and the ICF Institutional Review Board examined. The ICF IRB guarantees that the survey follows all U.S. regulations. Regulations for the protection of human subjects issued by the Department of Health and Human Services (45 CFR 46). Individual women’s written consent was obtained during the data collection process for all participants. Privacy and confidentiality were strictly adhered to.

• tetteh, john
• adomako, isaac
• udofia, emilia asuquo
• yarney, elom
• quansah, henry
• yawson, anita o.
• essuman, akye
• yawson, alfred e.