Menu
Menu
Menu
Menu
Young children in low-and middle-income countries (LMICs) are vulnerable to adverse effects of household microenvironments. The UN Sustainable Development Goals (SDGs)—specifically SDG 3 through 7—urge for a comprehensive multi-sector approach to achieve the 2030 goals. This study addresses gaps in understanding the health effects of household microenvironments in resourcepoor settings. It studies associations of household microenvironment variables with episodes of acute respiratory infection (ARI) and diarrhoea as well as with stunting among under-fives using logistic regression. Comprehensive data from a nationally representative, cross-sectional demographic and health survey (DHS) in Uganda were analysed. We constructed and applied the multidimensional energy poverty index (MEPI) and the three-dimensional women empowerment index in multivariate regressions. The multidimensional energy poverty was associated with higher risk of ARI (OR = 1.32, 95% CI 1.10 to 1.58). Social independence of women was associated with lower risk of ARI (OR= 0.91, 95% CI 0.84 to 0.98), diarrhoea (OR = 0.93, 95% CI 0.88 to 0.99), and stunting (OR = 0.83, 95% CI 0.75 to 0.92). Women’s attitude against domestic violence was also significantly associated with episodes of ARI (OR = 0.88, 95% CI 0.82 to 0.93) and diarrhoea (OR = 0.89, 95% CI 0.84 to 0.93) in children. Access to sanitation facilities was associated with lower risk of ARI (OR = 0.55, 95% CI 0.45 to 0.68), diarrhoea (OR = 0.83, 95% CI 0.71 to 0.96), and stunting (OR = 0.64, 95% CI 0.49 to 0.86). Investments targeting synergies in integrated energy and water, sanitation and hygiene, and women empowerment programmes are likely to contribute to the reduction of the burden from early childhood illnesses. Research and development actions in LMICs should address and include multi-sector synergies.
In Figure 1, we present a simplified conceptual framework to show possible links between energy poverty, access to clean water, sanitation, hygiene, women empowerment, and health outcomes of children. Our conceptual approach incorporated long-standing accepted linkages between household microenvironments, women empowerment, and health outcomes of children. We also incorporated recently developed indices for environmental exposure [39] and women empowerment [41] to have a better understanding of the context-specific relationship between the exposures and health outcomes. The women empowerment index is useful to identify what types of empowerment are linked with health outcomes of children. Conceptual framework: household environment (including social dimension) and health outcomes of children below 5 years of age. The bolder lines show key associations that we aim to test while the lighter lines show control variables. The data analysed in this study are from the 2016 Uganda Demographic and Health Surveys (UDHS), collected by the Uganda Bureau of Statistics with technical support from ICF international [42]. The UDHS is a nationally representative survey that provides comprehensive data about households, health outcomes of children, and maternal characteristics. The survey was carried out from 20 June to 16 December 2016 on key demographic and health indicators, including nutritional status of children and women and gender-related variables. A stratified and multistage sampling method was used in the 2016 UDHS to collect key information on child and maternal health indicators, which is nationally representative. A detailed description of methods, design, collected data, study participants and other important information is documented in the 2016 Demographic and Health Survey of Uganda [42]. The survey has rich data including information about children’s morbidity, though most of them are symptomatic data. It also has rich data on household and parents’ socioeconomic and demographic characteristics. Therefore, the UDHS survey data are attractive for rigorous quantitative analysis to establish an association between morbidity among children under 5 years of age and household microenvironments and women empowerment. The key health outcome variables in this study came from the mother’s responses to questions on episodes of various child morbidity within two weeks before the survey date. As indicated in Figure 1, the child health outcomes in this study are ARI, diarrhoea, and nutritional status of children using stunting as a key indicator. In the 2016 DHS surveys, symptoms of ARI are defined as short, rapid breathing which was chest-related and/or difficult breathing which was chest-related [43]. Following this definition, the children were categorised into two groups: those who experienced ARI symptoms and those who did not within 2 weeks before the survey. A limitation of this indicator is that it is based on the mothers’ perception of the morbidity, not a definitive diagnosis. The DHS data also contains diarrhoea prevalence by asking mothers whether the child had diarrhoea in the two weeks preceding the survey. This health outcome is dichotomous, identifying children who suffered diarrhoea and those who did not. Stunting in children below 5 years of age is the other health outcome examined in this study. Following the WHO Multicentre Growth Reference guideline [44,45], a child is stunted if the height-for-age z-score (HAZ) is below -2SD of the median for their age, including both mildly and severely stunted children. Key predictors of interest in this study are a comprehensive set of variables related to household microenvironments and women empowerment, as presented in Figure 1. Most household-environment-related variables in the DHS are standardized in the recode files and often used as they are with moderate modifications; for example, see [46]. In this study, however, indices were constructed and globally set standards were used to categorize key variables of interest. The multidimensional energy poverty index [39] was constructed and used as an indicator for household air pollution. We also estimated an women empowerment index, relevant in African settings. To include water quality and sanitation and hygiene facilities, we used the revised standard ladder by WHO/UNICEF Joint Monitoring Program [8]. We discuss these measurements and standards in the following sections. An indicator for household air pollution is a key predictor in this study. Previous studies often considered households’ consumption of solid fuel as an indicator of household air pollution to explain childhood morbidity [18,21,47,48,49,50]. In this study, we constructed a multidimensional energy poverty index (MEPI) at the household level to explain childhood morbidity. The MEPI captures a set of energy deprivations that affect a person or household [39,51]. The MEPI provides a framework to identify the categories of households left behind on access to clean, safe, and sustainable household energy [52]. Methodologically, the MEPI is derived from the multidimensional poverty measures developed by Alkire and Foster [53]. Literature on methodological developments of multidimensional poverty have their root in Amartya Sen’s discussion of deprivations and capabilities [54] which argues for the need to focus on the absence of opportunities and choices for living a basic human life. The MEPI is composed of five dimensions representing basic energy services with six indicators. More specifically, it is composed of indicators of a household using modern cooking fuel and cooking places, having access to electricity for lighting, having a refrigerator, having a TV or radio for entertainment and education, and having a phone or mobile phone for communication (see Table 1). Multidimensional energy poverty dimensions and respective variables with cut-offs, including relative weights (in parenthesis). Source: taken from Nussbaumer, Bazilian, and Modi [39]. Following previous studies [39,51,52] and the relative importance of indicators to human health, we unequally assigned weights to the various dimensions and indicators. This reflects the relative importance of the various energy poverty variables considered in household pollution and human health. We refer readers to Nussbaumer, Bazilian, and Modi [39] for further understanding of dimensions, indicators, and weights used in MEPI construction. A household is identified as energy poor if the respective set of deprivation scores (Ci) exceeds a predefined threshold, k. Previous studies in LMICs used a multidimensional energy poverty cut-off score at k=0.3 [39,52]. Nussbaumer, Nerini, Onyeji, and Howells [51] further categorized a household multidimensional energy poor level as acute when the MEPI exceed 0.7, moderate between 0.3 and 0.7, and low below 0.3. We also followed these cut-off points. However, in our analysis, very few (0.16%) of the observations fell in the low multidimensional energy poor category. Consequently, we combined the ‘low’ and ‘moderate’ energy poor category and coded as ‘moderate’ multidimensional energy poor. Therefore, households were categorised into ‘moderate’ and ‘acute’ multidimensional energy poor. The MEPI captures information on both the incidence and the intensity or severity of energy poverty. We computed the poverty headcount as H=qn, where q is the number of energy poor households (where ci>k) and n the total number of households. The severity of poverty indicates the average proportion of indicators in which multidimensional energy poor households is obtained as A=∑i=1nci (k)/q. Finally, the MEPI is obtained as the product of the multidimensional energy poverty headcount ratio (H) and multidimensional energy poverty intensity (A): MEPI=H×A. In coding quality and source of household water and sanitation, previous studies used the dichotomous improved vs. unimproved or safe vs. unsafe definitions [17,48,55]. As our focus in this study is household microenvironments, we opted for definitions and categorization that are clearer and more distinct. We followed the revised water and sanitation ladder by WHO/UNICEF joint monitoring programme (JMP) [8] to define quality of water and sanitation facilities but with some modifications. We defined household access to a hygiene facility following the WHO/UNICEF joint monitoring programme ladder for hygiene: basic, limited, and no facility. The WHO/UNICEF defines and categorizes quality of water and sanitation facilities into five levels. This was not possible in the Uganda DHS dataset due to less clarity in wording used in the questionnaire to match with WHO/UNICEF JMP ladders for water and sanitation. We categorised sanitation facilities, slightly deviating from the WHO/UNICEF JMP ladder for sanitation, into three: no facility, unimproved, and improved. We categorized the quality of the source of drinking water into two: improved and unimproved. We adopted the women empowerment index construction method developed by [41] using DHS data from Africa. This composite index consists of three domains of empowerment: attitude to violence, social independence, and decision making. These empowerment dimensions overlaps with most of the dimensions considered by [56] focusing on sub-Saharan Africa, particularly East African countries. Similar empowerment dimensions were considered in other studies in LMICs [29]. The three empowerment dimensions comprise various information. ‘Attitude to violence’ is composed of information related to the respondent’s opinion about whether wife-beating is justified or not in various scenarios. ‘Social independence’ includes items related to education, frequency of information consumption (reading), age at first cohabitation and first childbirth, and differences between age and their years of schooling of the woman and her husband. The ‘decision making’ domain is comprised of information related to a woman’s involvement in household decisions and labour force participation. In this study, the three dimensions were weighed following [41]. In addition to the household microenvironments and women empowerment-related predictors, other relevant predictors such as individual and parental variables are included in the analysis. Potential predictors associated with the health of children below 5 years of age were included, considering their relevance in previous studies [16,21,23,25,29,30,46,47,57,58] as control variables. A summary of definitions of key predictors used in the analysis is presented in (Table 2). Description of key predictors used in the analysis. The statistical analyses used in this study include descriptive statistics, bivariate analyses, and logistic regression models. The DHS sampling weights are applied in all analyses to account for the complex survey design. We used descriptive analyses, percentages, and numbers to show the distribution of childhood morbidity and nutritional status by predictor variables. Associations between childhood health outcomes and predictors were first analysed using bivariate analyses, the χ2 tests, before fitting the regression models. These analyses were carried out to compare the prevalence of childhood morbidities and stunting among the levels of the selected predictors and to inform further analyses using regression models. The dependent variables considered in this study are dichotomous variables. Therefore, binary response econometric models are the natural choice. Logistic regression models were estimated to evaluate the association between key predictors and health outcomes of children considered for this study. The logistic regression model we estimate as: where pij is a dichotomous health outcome for child i in household j, β denotes vector of coefficients estimated, Χ denotes a set of values of predictors: household energy poverty, water, sanitation and hygiene, women empowerment, and control variables. We used the Stata 15 software package for all data analyses and logistic regressions, and we reported odds ratios.
