Data and sample selection
This study used a sample of 4,815 under-five children from the Myanmar Demographic and Health Survey (MDHS 2015-16). MDHS used a two-stage stratified clustered sampling framework. The sampling design contained 76,990 primary sampling units (PSUs). Out of 30 sampling strata and 441 clusters, a selection of 30 households was made by using probability proportional to size rule (PPS). A total of 12,885 women aged 15 to 49 years were interviewed from 13,230 households, and data was compiled in women’s file (KR), yielding a response rate of 98%. The information on socio-demographic factors, maternal and children’s health, fertility, reproductive health, family planning, pregnancy, postnatal care, and immunization were collected using the Standard DHS tools. MDHS can be accessed freely from the publicly available data repository via the “DHS Program” website (https://www.dhsprogram.com/data/).
Data Processing
Ethical Review Committee of the Myanmar Ministry of Health and Sports, Department of Medical Research granted ethical approval for the implementation of MDHS [23]. The data collection held from December 7, 2015, through July 7 2016, by 19 trained field teams. The field editors used computer-assisted field editing (CAFE) procedures to enter the completed paper questionnaires soon after data collection. Completed questionnaires were entered twice to check for the inconsistencies using CSPro software.
Outcome variable
We examined three childhood illnesses; cough, fever, and diarrhea as dependent variables. MDHS enquired women aged 15-49 whether their children had diarrhea and/or fever, and/or cough in the past two weeks preceding the survey. The answers were recorded as ‘Yes’ or ‘No’. The definition of diarrhea was read to the mothers to ensure that the mothers understood diarrhea and validate the accuracy of responses. This paper used the information on cough, fever, and diarrhea as reported by mothers to form dichotomous (0/1) outcome variables; where 1 implies the under-five child suffered by the disease, and 0 means not-experienced.
Explanatory variables
We used three WASH-related variables as explanatory variables: water, sanitation (toilet facility), and child feces disposal by households. The standard guidelines for the grouping of improved and unimproved sanitation, “as recommended by the WHO/UNICEF Joint Monitoring Program (JMP) for Water Supply and Sanitation, were used (Table 1) [24]”.
Table 1 WHO classification of improved sanitation and water supply
|
Unimproved
|
Improved
|
Water
|
Unimproved-drinking-water sources: Unprotected dug well, unprotected spring, cart with small tank/drum, surface water (river, dam, lake, pond, stream, canal, irrigation channels), and bottled water
|
Other improved drinking-water sources: Public taps or standpipes, tube wells or boreholes, protected dug wells, protected springs or rainwater collection. Piped water on premises: Piped household water connection located inside the user’s dwelling, plot or yard
|
Sanitation and child’s excreta disposal facilities
|
Unimproved sanitation facilities: do not ensure hygienic separation of human excreta from human contact. Unimproved facilities include pit latrines without a slab or platform, hanging latrines and bucket latrines.
|
Improved sanitation facilities: ensure hygienic separation of human excreta from human contact. They are use of the following facilities: Flush/pour flush to: piped sewer system, septic tank, pit latrine; Ventilated improved pit (VIP) latrine, Pit latrine with slab, Composting toilet.
|
Source: WHO/UNICEF Joint Monitoring Programme for Water Supply and Sanitation.
Households’ access to different water sources was regrouped into safe drinking water or unsafe water. The responses of households interviewed about the type of toilet were combined into two responses; unimproved latrine, and improved latrine. Similarly, the practices of child feces disposal were organized into two categories, as follows, safe disposal of child feces and unsafe disposal. The binary variable (0/1) was used to represent each WASH- related indicator, 1 implies unimproved, and 0 means improved.
Other independent variables included child characteristics, such as child sex (categorical: male or female); age in months (categorical: 1-12, 13-24, 25-36, 37-48 and 49-59); child birth size (categorical: very small, average or smaller but not very small, and above average or very large); and nutritional status (binary: stunting, wasting, and underweight). The parent characteristics used in this study included mother’s age in years (categorical: less than 20, 20-29, 30-39, and 40-49); mother’s education (categorical: no education, primary, secondary or higher), and father’s education (categorized as mother’s education). The household-related variables comprised type of residence (categorical: urban or rural), and children in the households (categorical: 1-2, 3-4, and more than 4).
MDHS 2015-16 collected detailed information on durable assets and housing features. It used Principle Component Analysis (PCA) to derive the factor scores of household wealth based on the obtained data. The Household wealth in the MDHS was categorized in terms five quintiles: “poorest, poorer, middle, richer, and richest”. We re-organized the wealth index by combining first three wealth quintiles “(poorest, poorer, and middle)” into “poor” and last two quintiles “(richer and richest)” into rich.
Data Analysis
Cross-tabulation between socio-demographic characteristics and WASH-related indicators were presented as frequency distribution and in percentage. Since this study involved mostly categorical variables, we employed Pearson Chi-square test to determine the relationship between each explanatory variable and dependent variable. Three levels of significance, “p<0.001, p<0.05, p<0.01”, were used.
This study analyzed the prevalence of each cough, fever, and diarrhea across childhood characteristics, parent-related, and household-related factors. The univariate and bivariate statistical analyses were employed over a weighted sample. The bivariate relationships between each explanatory variable and each childhood illness were appraised at a 95% confidence interval. A p-value<0.05 indicates a statistically significant relationship. The multivariate logistic regression was used to assess the association between WASH indicators and each childhood disease; diarrhea, fever, and cough. The unadjusted odd ratios (reference category=use of adequate WASH) showed the odds of cough, fever and diarrhea for unsafe drinking water, unimproved toilet, and hazardous child feces disposal separately. The adjusted odds ratios showed the likelihood of diarrhea, cough, and fever for each WASH indicator while controlling for other indicators. Sampling weights were used while estimating the logistic model to adjust for the sampling errors and complex sampling design. The analysis was performed in Stata, version 15.