Child undernutrition remains a persistent global health concern, significantly affecting millions of children worldwide. It manifests in different forms, such as stunted growth, acute wasting and underweight. The global estimate indicates 148·1 million under-5 children experienced stunted growth and 45 million suffered from acute wasting in 2022(1). A significant share of this burden falls on low- and middle-income countries (LMICs)(2,3) , particularly in South Asia, where the prevalence of stunting and wasting is very high; 31·4 % and 14·8 % among under-5 children, respectively(1). The consequences of child undernutrition are multifaceted, including severe health risks, hindered cognitive and physical growth and the perpetuation of intergenerational malnutrition cycles(Reference De Sanctis, Soliman and Alaaraj4,Reference Swaminathan, Hemalatha and Pandey5) . Undernutrition contributes to 45 % of global under-5 deaths, although this burden is not evenly distributed, with a significant portion occurring in LMICs(6). In response to this ongoing challenge, Sustainable Development Goal (SDG) 2·2 has set the ambitious target of eliminating under-5 stunting and wasting by 2030(1–3).
In LMICs, child undernutrition emerges from a complex interplay of factors, including socio-economic conditions, maternal nutritional status, children’s age, birth weight, birth order and family size(Reference Katoch7,Reference Chowdhury, Rahman and Billah8) . It is also influenced by inadequate access to nutritious food, poor breastfeeding, dietary and caregiving practices(Reference Clarke, Zuma and Tambe9,Reference Megersa, Haile and Kitron10) , parents’ inadequate knowledge about healthy rearing of children(Reference Clarke, Zuma and Tambe9,Reference Forh, Apprey and Frimpomaa Agyapong11) and compromised health care(Reference Elhady, Ibrahim and Abbas12). In light of these multifaceted determinants, reducing child undernutrition, that is, stunting, wasting and underweight, in LMICs requires a holistic approach that not only addresses these known factors but also delves into the less explored aspects of this challenge.
The effect of poor housing and its environment on child nutrition receives inadequate focus. The household environment is defined by specific quality indicators within a dwelling, including factors such as the availability of water, sanitation and hygiene (WASH), the construction materials of the house and the presence of potential pollutants(Reference Morakinyo, Adebowale and Obembe13,Reference Husseini, Darboe and Moore14) . An estimated 494 million people worldwide still practice open defaecation(15), and 2 billion rely on unsafe drinking water(16), with these statistics being particularly pronounced in LMICs like sub-Saharan Africa and South Asia(15,16) . Besides, three in ten people globally lack proper handwashing facilities(17) and the burden being higher in South Asia, where two in five people lack the access(17). Around 2·4 billion people, mainly in LMICs, still use solid fuels for cooking, and the percentage is much higher in rural areas (52%) compared with urban areas (14%)(18). All these indicators set a benchmark of poor household environmental conditions (HECs) that amplify the risks of diseases like diarrhoea, tuberculosis and acute respiratory infections that may cause death and hinder the healthy growth of the children(Reference Raju, Siddharthan and McCormack19,Reference Shrestha, Six and Dahal20) . These factors often intertwine, forming a complex network of influences that leads to severe child undernutrition and adverse health outcomes(Reference Katoch7,Reference Clarke, Zuma and Tambe9,Reference Forh, Apprey and Frimpomaa Agyapong11) .
Bangladesh, similar to many other LMICs, faces a higher burden of child undernutrition. Recent statistics suggest approximately 3·9 million children experience stunted growth and 1·4 million experience acute wasting in the country in 2022(1). The factors contributing to this situation in Bangladesh are similar to those found in other LMICs(Reference Katoch7,Reference Chowdhury, Rahman and Billah8,Reference Das and Gulshan21–Reference Chowdhury, Rahman and Billah25) . Besides, in terms of WASH indicators, in Bangladesh, approximately 68·3 million people lack access to safe water, 103 million lack proper sanitation and 61·7 million lack proper hygiene(26). Furthermore, solid fuels are used for cooking in around 80 % of households, causing moderate to severe household air pollution (HAP)(27). Unfortunately, the overall impact of these crucial indicators on child undernutrition is largely unexplored in the Bangladeshi context. Although a few studies considered some of these factors sporadically(Reference Raju, Siddharthan and McCormack19,Reference Shrestha, Six and Dahal20,Reference Ghosh, Kabir and Islam28) , they rarely conducted comprehensive assessments of overall household environmental quality indicators and/or did not include nationally representative data in their analyses. Also, it is crucial to acknowledge that these HECs indicators and their types, along with other factors, vary significantly across urban and rural areas of Bangladesh(Reference Srinivasan, Zanello and Shankar29). Yet, a significant research gap remains as to how HEC indicators individually and collectively affect child undernutrition. This study aims to investigate the relationship between HEC indicators and undernutrition among under-5 children in Bangladesh and to compare the magnitude of the associations in rural and urban areas.
Methods
Data source
In this study, we analysed cross-sectional survey data from the eighth round of the Bangladesh Demographic Health Survey (BDHS), which was conducted in 2017/18. The National Institute of Population Research and Training, under the Ministry of Health and Family Welfare, conducts this nationally representative survey every 3 years. The survey aimed to provide information on the socio-demographic, health and nutritional aspects of women, infants and children(27). The BDHS 2017/18 used a multistage random sampling technique to collect nationally representative data. In the initial stage, 675 clusters or enumeration areas (EAs) were randomly selected using probability proportional to size (PPS) method, consisting of 250 urban and 425 rural areas; however, after excluding three EAs due to flood, a total of 672 EA were finally chosen. These EAs, taken from the list of 293 579 EA listed in the 2011 National Population Census, typically represent city blocks or villages, with around 120 households on average. In the second stage, thirty households were randomly selected from each EA, yielding a list of 20 160 households. The survey covered 19 457 households and identified 20 376 eligible respondents who met the following criteria: (i) ever-married women aged 15–49 years and (ii) staying in the selected households on the night preceding the survey. A total of 20,127 women were finally interviewed, with a response rate of 98·8 %(27). Additional sampling details can be found in the survey reports(27).
Study participants
Among the women interviewed in BDHS 2017/18, a total of 7,562 women gave birth to 8,759 children during the 5 years prior to the survey. We analysed the data of 8,057 under-5 children who were eligible for anthropometric measurements (Fig. 1). For this study, the following criteria were used for selecting study participants: (i) children who were under the age of 5 at the time of the survey and (ii) whose anthropometric measurements were taken.
Sampling strategy of the Bangladesh Demographic Health Survey 2017/18 and selection of study participants using STROBE guidelines.
Nutritional measurements
The BDHS collected height and weight data for under-5 children. Health technicians, both male and female, received training, including standardisation exercises, to ensure measurement accuracy. Weight was measured using electronic SECA 878U scales, and height was measured using ShorrBoard®(27). Children aged under 24 months were measured lying down (recumbent length), while older children were measured standing up using the mentioned tool(27).
Outcome variables
The primary outcome variables of this study were child stunting (low height-for-age), wasting (low weight-for-height) and underweight (low weight-for-age). Basic anthropometric measures were calculated using age, height and weight, which were then converted into Z-scores. Under-5 children were classified as stunted if their height-for-age Z-score was less than −2 s d from the median of the WHO Child Growth Standards (median −2 s d or median −3 sd). Similarly, the children were classified as wasted if their determined Z-score was less than 2 sd from the median of the WHO Child Growth Standards (median −2 sd or median −3 sd) for weight-for-height, and as underweight if their Z-score was less than 2 sd from the median of the WHO Child Growth Standards (median −2 sd or median −3 sd) for weight-for-age(30).
As a secondary outcome of the study, the Composite Index of Anthropometric Failure (CIAF) was used to summarise multiple indicators of child nutritional status. It combines measures of stunting, wasting and underweight into a single index(Reference Islam and Biswas22,Reference Permatasari and Chadirin31) . We further categorised them into seven groups, each representing a specific combination of nutritional failures, such as no failure, wasted only, stunted only, underweight only, wasted with underweight, stunted with underweight and stunted, wasted and underweight. The details can be found in the online Supplemental Table 1.
Exposure variables
The main exposure variables in our analysis were the HEC indicators, comprising WASH indicators and the use of solid fuel for cooking and materials used for building the house. WASH indicators are essential metrics used in WASH programmes and include household members’ access to clean drinking water, improved sanitation facilities and proper hygiene(27,Reference Schwemlein, Cronk and Bartram32) . Respondents were classified as unexposed to HAP if they used clean fuels, such as LPG or biogas, for cooking, moderately exposed if they used solid fuels for cooking in a separate building or outdoors and highly exposed if they used solid fuels for cooking inside their homes. This classification was done based on prior research conducted in LMICs(Reference Ranathunga, Perera and Nandasena33–Reference Rana, Islam and Khan36). The term ‘housing material’ refers to the building materials utilised in constructing the roofs, floors and walls of houses. The rationale for such inclusion is that in the Bangladeshi context, none of the literature considered the overall HEC, and the selection of HEC indicators was based on previous literature available for LMICs(Reference Morakinyo, Adebowale and Obembe13,Reference Husseini, Darboe and Moore14) . Their operational definitions and categorisations are detailed in online Supplemental Table 1.
Calculation of poor household environmental condition score
A HEC score was computed to evaluate the overall quality of the household environment, serving as another primary exposure variable in this study. This score was derived using participants’ responses to questions on house-building materials (natural or rudimentary materials), HAP from cooking, water sources, safe drinking water, sanitation and handwashing facilities. Each category adds to the overall score by assigning a value of 1 for each indicator of poor HEC. These individual values were then added together to create a composite index, resulting in a poor household environment score ranging from 0 to 6 for each household(Reference Brown, Ravallion and van de Walle37). For instance, if a household had three poor HEC characteristics, it received a score of 3. A higher score signifies that the environment of that household was worse compared with households with lower scores. The calculation of poor HEC score was based on existing literature(Reference Brown, Ravallion and van de Walle37), and it allowed us to measure how the effect size changes with each incremental increase in the score in terms of poor HEC indicators.
Other variables/covariates
Other variables included in this study were identified through literature searches and identified based on existing evidence from LMICs, including Bangladesh(Reference Katoch7,Reference Chowdhury, Rahman and Billah8,Reference Das and Gulshan21,Reference Islam and Biswas22,Reference Chowdhury, Khan and Mondal24,Reference Chowdhury, Rahman and Billah25,Reference Permatasari and Chadirin31) . The covariates we included were child’s age in months (continuous), child’s sex (male or female), religion (Muslim or non-Muslim), sex of the household head (male or female), education level of the child’s mother (no formal education, primary, secondary or higher), education level of the child’s father (no formal education, primary, secondary or higher), employment status of the child’s mother (unemployed or employed), household size (1–5 members, 6–10 members or 10+ members), place of residence (urban or rural) and administrative divisions (Barishal, Chattogram, Dhaka, Mymensingh, Khulna, Rajshahi, Rangpur or Sylhet).
Statistical analysis
We used descriptive statistics to estimate the prevalence of stunting, wasting and underweight, as well as the distribution of HEC indicators for the entire study population and among rural–urban subgroups. Subsequently, we employed bivariate analysis to observe the distribution of child undernutrition indicators across HEC indicators and other covariates. The statistical significance of the bivariate analysis was assessed using the Pearson χ 2 test. Multilevel mixed-effect generalised linear models (GLM) modified with a Poisson regression approach were used to examine the association between HEC indicators and different types of undernutrition among under-5 children. The rationale for choosing multilevel regression was to account for the hierarchical structure of the BDHS data, where children are nested within households (level 1) and households are nested within clusters (level 2). We chose glm modified with Poisson regression to address the high prevalence of the outcome variable (>10%). Previous studies have found that ordinary logistic regression produces less precise results under these conditions(Reference Khan, Rahman and Islam38,Reference Zou39) . We therefore ran two levels of multilevel modelling (household and cluster) for each outcome variable. Each model was run separately, with adjustments made for covariates, and the models estimated prevalence ratios to assess the strength of the associations after assessing multicollinearity. Additionally, in order to examine the impact of HEC indicators on anthropometric failure, a multilevel mixed-effect multinomial logistic regression was used. This was also done after adjusting for covariates, allowing for a direct assessment of risk ratios (RRs). We excluded the wealth quintile from the adjusted variables since it was calculated using household characteristics and other assets. Its inclusion caused multicollinearity with HEC indicators in the model. All analyses took into consideration the complex survey design and sampling weights. Results were reported with a 95% CI and a significance level of P < 0·05. The data were analysed using the statistical programme STATA, version 15·1 SE (Release 15; StataCorp LLC).
Result
Background characteristics
The study analysed the data of 8057 under-5 children of whom 52·2% were male, 92·0% were Muslim and 86·6% belonged to male-headed households. Most of the mothers had a primary (28·8%) or secondary level of education (48·4%). The majority of children were from poor households (41·8%), and 40·1% were from households with 6–10 members. The sample comprised 72·5% rural residents, with the highest proportions in Dhaka (25·8%) and Chattogram division (20·8%) (online Supplemental Table 2).
Household environmental condition indicators
Table 1 presents the results of the descriptive analysis. Most households had finished roofs (99·0%), and walls (87·2%); however, majority of the floors were constructed with natural or rudimentary materials (63·8%). Approximately, 78·3% households had moderate exposure to HAP from cooking, 98·4% had improved drinking water sources and 89·9% lacked proper water treatment facilities. Around 55·7% and 61·7% lacked proper sanitation and handwashing, respectively, and over a third (33·2%) had ≥ 5 poor HEC characteristics (Table 1).
Distribution of household environmental condition indicators of the study participants (n 8,759)
All are column percentages. *missing = 317. †Calculated using composite scoring. All values are weighted. HEC refers to household environmental condition.
Prevalence of child undernutrition
As presented in Table 2, the prevalence of stunting, wasting and underweight among under-5 children were 30·7%, 8·4% and 21·8%, respectively. Rural–urban differences were evident, with higher rates of stunting (32·7% in rural vs. 25·3% in urban) and underweight (22·8% rural vs. 19·1% urban) in rural areas (Table 2). Geographically, child undernutrition indicators were most prevalent in the Sylhet division (52·0%), while the Mymensingh (49·1%) and Barishal (42·5%) divisions had the highest percentages of households with poor environmental quality (online Supplemental Table 3).
Prevalence of child undernutrition among under-5 children and their anthropometric failure index (n 8,057)
All are column percentages. Survey weight was applied.
Distribution of child undernutrition across household environmental condition indicators
Table 3 presents bivariate associations between HEC indicators and undernutrition among under-5 children. There was a significant association observed between stunting and children living in households made of unimproved housing materials (35·5%), exposed to HAP from cooking (45·4%), lacking adequate water treatment facilities (31·6%), proper sanitation (34·8%), and handwashing facilities (35·3%). A similar pattern of associations was obseved for underweight children as well. The prevalence of stunting and underweight gradually increased with the number of poor HEC characteristics of a household (Table 3).
Bivariate analyses of under-5 children’s stunting, wasting and underweight across household environment condition indicators
All are row percentages. All results are weighted. HEC refers to household environmental conditions.
Association between household environmental condition indicators and child undernutrition
Table 4 presents how the HEC indicators are associated with stunting and underweight in under-5 children. A higher likelihood of stunting was found among under-5 children who lived in households constructed with unimproved housing materials (aPR: 1·17; 95% CI: 1·04, 1·32), having moderate (aPR: 1·16; 95% CI: 1·00, 1·35) or high HAP exposure from cooking (aPR: 1·37; 95% CI: 1·01, 1·70), using unimproved drinking water source (aPR: 1·28; 95% CI: 1·04, 1·59) and having poor handwashing facilities (aPR: 1·24; 95% CI: 1·13, 1·37), compared with their counterparts. Likewise, under-5 children who lived in households constructed with unimproved materials (aPR: 1·17; 95% CI: 1·02, 1·35), lacked proper facilities for drinking water treatment (aPR: 1·21; 95% CI: 1·01, 1·45), with poor sanitation facilities (aPR: 1·16; 95% CI: 1·06, 1·30) and having poor hand washing facilities (aPR: 1·18; 95% CI: 1·05, 1·33) were more likely to experience under-5 underweight compared with those who did not. (Table 4).
Association between stunting, wasting and underweight among under-5 children, and household environment condition indicators, adjusted for household-level socio-demographic characteristics: assessed using generalised linear model modified with Poisson regression
All the models were run separately for each type of household environment condition characteristics and were adjusted for the child’s age, child’s sex, religion, sex of the household head, education level of the child’s mother, education level of the child’s father, employment status of the child’s mother, household size, place of residence and administrative division. Values with superscript asterisks *, ** and *** indicate P < 0·05, P < 0·01 and P < 0·001, respectively. (ref), reference category; PR, prevalence ratio; HEC refers to household environmental condition.
There is an incremental relationship between HEC scores and the likelihood of children being stunted and underweight (Table 4). For instance, children under-5 in households with 1 to 5 or more poor HEC characteristics were 1·83 to 2·44 times more likely to be stunted. Similarly, compared to children living in houses with no poor HEC, those who were living in houses with 1, 2, 3, 4 and 5 poor HEC were 1·44, 1·54, 1·90, 1·79 and 2·12 times likely to be underweight, respectively (Table 4).
Urban–rural variation in the association between household environmental condition indicators and child undernutrition
Table 5 presents urban–rural variations in the association of HEC indicators with under-5 stunting and underweight. In urban areas, only households lacking proper drinking water treatment facilities (aPR: 1·36; 95% CI: 1·04, 1·78) and having inadequate handwashing facilities (aPR: 1·39; 95% CI: 1·16, 1·65) had higher likelihoods of stunted children. Conversely, in rural settings, households constructed with unimproved materials (aPR: 1·36; 95% CI: 1·16, 1·58), exposed to high HAP from cooking (aPR: 1·52; 95% CI: 1·07, 2·16), poor sanitation (aPR: 1·16; 95% CI: 1·05, 1·28) and with poor handwashing facilities (aPR: 1·29; 95% CI: 1·16, 1·44) demonstrated higher likelihoods of stunting compared to their counterparts. In urban areas, there were significant associations between underweight and under-5 children living in households constructed with unimproved materials (aPR: 1·22; 95% CI: 1·02, 1·48), exposed to moderate (aPR: 1·27; 95% CI: 1·02, 1·57) and highly exposed to HAP from cooking (aPR: 2·12; 95% CI: 1·02, 4·37), lack of proper drinking water treatment facility (aPR: 1·43; 95% CI: 1·13, 1·83) and inadequate handwashing facility (aPR: 1·38; 95% CI: 1·12, 1·69). On the contrary, in rural areas, the likelihood of underweight significantly increased among under-5 children residing in households constructed with unimproved materials (aPR: 1·23; 95% CI: 1·04, 1·46), those with poor sanitation facilities (aPR: 1·25; 95% CI: 1·10, 1·41) and inadequate handwashing facilities (aPR: 1·17; 95% CI: 1·02, 1·33), when compared to their counterparts.
Urban-rural variations in the effect size of the association between stunting, wasting and underweight in under-5 children, and household environment condition indicators
All models were run separately for each type of household environment condition characteristics and were adjusted for the child’s age, child’s sex, religion, sex of the household head, education level of the child’s mother, education level of the child’s father, employment status of the child’s mother, household size, place of residence and administrative division. Values with superscript asterisks *, ** and *** indicate P < 0·05, P < 0·01 and P < 0·001, respectively. (ref), reference category; PR, prevalence ratio; HEC refers to household environmental condition.
As the number of poor HEC characteristics increases, the likelihood of stunting and underweight also increases gradually. There were substantial urban–rural variations in the effect size of their association with child undernutrition. For instance, the effect size of stunting in urban households with five or more poor HEC characteristics was 1·87 (95% CI: 1·12, 2·83), and in rural areas, it was 8·11 (95% CI: 1·20, 54·77). Similarly, poor HEC scores demonstrated a gradual rise in the likelihood of under-5 underweight in urban areas, whereas no significant associations were observed in rural areas (Table 5).
Association between household environmental condition scores and anthropometric failure
Table 6 presents the association between under-5 children’s anthropometric failure index and HEC score, adjusted for household socio-demographic traits. Multinomial analysis indicated that the chances of Failure C (stunted only) rose gradually from 3·21 (95% CI: 1·69, 6·09) to 4·10 (95% CI: 2·19, 7·66) with an increase in poor HEC characteristics from 1 to 5 compared to its counterpart. Similarly, the chance of Failure E (wasted with underweight) among under-5 children increased from 4·54 (95% CI: 1·17, 17·58) to 5·78 (95% CI: 1·46, 22·99) with 3 to 5 poor HEC characteristics, respectively. A similar trend was observed for Failure F (stunted with underweight). The likelihood of Failure G (stunted, wasted and underweight) was 3·66 (95% CI: 1·04, 12·93) times higher with five or more poor HEC characteristics compared with their counterparts. (Table 6).
Association of under-5 children’s anthropometric failure index with household environmental quality score adjusted for household-level socio-demographic characteristics using multinomial logistic regression analysis, n 8,057
All models were run separately for each type of household environment condition characteristics and was adjusted for child’s age, child’s sex, religion, sex of the household head, education level of child’s mother, education level of child’s father, employment status of the child’s mother, household size, place of residence and administrative division. Values with superscript asterisks *, ** and *** indicate P < 0·05, P < 0·01 and P < 0·001, respectively. (ref), reference category; RRR, relative risk ratio; HEC refers to household environmental condition.
Failure A: No anthropometric failure (reference category); Failure B: Wasted only; Failure C: Stunted only; Failure D: Underweight only; Failure E: Wasted with underweight; Failure F: Stunted with underweight; Failure G: Stunted, wasted and underweight.
Discussion
This study explored the relationship between undernutrition in children under-5 and HEC indicators. In Bangladesh, approximately 30·7% of children experienced stunting, 8·4% suffered from wasting and 21·8% were underweight. Furthermore, notable variations between urban and rural areas were observed in the prevalence of stunting and underweight among under under-5 children. Around one-third of the total households analysed reported the presence of five or more poor HEC among the eight indicators considered. We found an increased likelihood of stunting among children living in houses built with unimproved materials, highly exposed to HAP from cooking, with unimproved drinking water sources and with inadequate handwashing facilities. Similarly, children residing in houses constructed with unimproved materials, using drinking water from unsafe sources and lacking proper sanitation and handwashing facilities were also linked to underweight conditions. The likelihood of stunting and being underweight increased gradually as the HEC score increased, and the results highlighted substantial urban–rural variations in the association with child undernutrition. Compared to the children living in rural areas, those who were living in urban areas had higher likelihoods of being underweight with poor HEC scores.
We reported that one in every three households in Bangladesh has five or more HEC out of the eight indicators considered, which cover various domains such as household roofs, walls, floor, cooking fuels and sanitation facilities. While the score we generated aligns with previous literature on LMICs, we were unable to validate our findings with existing evidence in Bangladesh due to a lack of relevant studies. Following prior literature, we classified each indicator as either poor or good, though there may be intermediate conditions that we could not account for. Such binary classification may lead to conflicting estimates of HEC, with a risk of over- or underestimation. Although these issues are likely random, they could affect the associations reported in this study. Addressing this limitation would require surveys with sufficient variables to classify HEC across more nuanced levels, which are currently lacking. Therefore, the findings of this study should be interpreted with this limitation in mind.
In the LMIC contexts, living in houses constructed with poor housing materials and exposure to HAP are widely known global risks of under-5 stunting and align greatly with our findings(Reference Ranathunga, Perera and Nandasena33,Reference Tusting, Gething and Gibson40,Reference Sinharoy, Clasen and Martorell41) . Conversely, child underweight is associated with living in houses constructed with poor housing materials but not with HAP exposure. Poor housing conditions are characterised by low-quality roofs, walls, floors and inadequate insulation and ventilation, thereby exposing children to extreme temperatures and pollutants(Reference Ranathunga, Perera and Nandasena33,Reference Sinharoy, Clasen and Martorell41,Reference Vardoulakis, Dimitroulopoulou and Thornes42) . Besides, HAP, which often stems from the use of solid fuels such as wood or biomass for cooking and heating, emits harmful particulate matter and toxic gases that pollute households’ indoor environments(Reference Raju, Siddharthan and McCormack19). These heighten the vulnerability to infections, especially acute respiratory infections(Reference Raju, Siddharthan and McCormack19), subsequently contributing to severe chronic undernutrition(Reference Ranathunga, Perera and Nandasena33,Reference Sinharoy, Clasen and Martorell41,Reference Nasanen-Gilmore, Saha and Rasul43) . Similarly, other poor HEC indicators, that is, poor-quality drinking water, inadequate sanitation and insufficient handwashing facilities, also play a significant role in contributing to underweight or stunted growth in under-5 children in the context of LMIC, which strongly supports our findings as well(Reference Shrestha, Six and Dahal20,Reference Bekele, Rahman and Rawstorne44–Reference Sahiledengle, Petrucka and Kumie46) . These conditions are well-established as contributors to waterborne diseases and infections(Reference Katoch7,Reference Shrestha, Six and Dahal20,Reference Gizaw and Worku45,Reference Tofail, Fernald and Das47) , promoting pathogen transmission, causing inflammation, disrupting nutrient absorption and ultimately hindering children’s growth and nutrition(Reference Katoch7,Reference Shrestha, Six and Dahal20,Reference Gizaw and Worku45,Reference Tofail, Fernald and Das47) .
We found no significant association between wasting and HEC. The underlying reasons for this, despite the significant associations of stunting and underweight with HEC, are unknown and require further exploration. However, this might be linked to the government’s focus on reducing child undernutrition through several programmes, with wasting often receiving priority due to its ease of detection and growing community concern. Additionally, methodological issues may have contributed to the lack of a significant association. Wasting is a measure of acute malnutrition and usually indicates recent and severe weight loss because a person has not had sufficient food intake and/or has had an infectious disease, such as diarrhoea, resulting in rapid weight loss(6). However, our results suggest that the detrimental effects of HEC primarily manifest over a long period of time, impacting the growth and development of children with little or no immediate effect on their short-term nutritional status. The lack of relevant data on the duration that the households maintained improved conditions may also explain the insignificant association we found, for instance, between the use of unimproved sources of drinking water and wasting. However, our results are consistent with that in the existing literature(Reference Bekele, Rahman and Rawstorne44,Reference Sahiledengle, Petrucka and Kumie46) .
The study revealed a dose–response relationship between HEC scores and the likelihood of child undernutrition such as stunting and underweight. While each HEC indicators independently contribute to child undernutrition, their combined impact is expected to be amplified. For instance, a child living in a household with poor sanitation facilities might already face an increased likelihood of stunting and underweight due to the potential exposure to diseases and inadequate nutrient absorption. If this household also lacks proper ventilation, highly exposed to HAP produced from the use of solid fuels and has substandard water sources, the combined impact of these factors is likely to be greater than the impact of individual HEC factors.
The underlying reasons for child undernutrition varied between rural and urban areas. There may be different underlying reasons for these differences. In urban settings, high population density often restricts the access to clean water and increases the risk of diseases that impede child growth(Reference Rosenberg, Kano and Ludford48,Reference Neiderud49) . Besides, in rural areas, low-quality housing, HAP exposure and poor water and sanitation increase waterborne diseases and respiratory issues, leading to an increased prevalence of stunting(Reference Ghosh, Kabir and Islam28,Reference Tofail, Fernald and Das47) . A range of factors, including poor socio-economic conditions, poor maternal health and nutrition, frequent illness and/or inappropriate feeding and inadequate care in early life, are likely to cause child undernutrition, apart from genetic factors, if there are any. In rural areas, these unfavourable social determinants of health are also prevalent. It is possible that the effect of relatively poor HEC aggravates undernutrition, especially stunting, caused by the unfavourable social determinants of health. Further research is recommended to examine this association.
The study has several strengths and limitations that should be taken into account while interpreting the results. First, the use of a hierarchical regression model allowed us to account for potential clustering effects within urban and rural settings, enhancing the robustness of our results. The use of nationally representative data on HEC and child nutritional status, along with a large, diverse sample from both urban and rural areas, increases the generalisability, reliability, and validity of our findings. By focusing on urban–rural disparities, our study shed light on variations that might have important policy implications. Lastly, the HEC composite scoring helped to critically illustrate and understand the associations. However, the study also has several limitations. The cross-sectional nature of BDHS data hampers establishing causality, and reliance on self-reported variables may have introduced recall and social desirability biases. While constructing the HEC variables and scores, we were unable to cover all aspects of housing quality, such as ventilation, insulation, heating/cooling facilities and tenure security. Additionally, the data allowed us to classify the indicators as either poor or good, though there may be intermediate conditions that we were unable to account for. Moreover, using equal weighting to all components in constructing HEC score is a potential limitation of our analysis. Another limitation of BDHS nutritional measurements for children is their reliance on anthropometric data, which may not capture all aspects of nutritional health, such as micronutrient deficiencies or dietary quality and can also be influenced by measurement errors. The lack of dietary data limits our ability to fully explore how household environments impact overall nutritional status. Future research should include comprehensive dietary assessments to better understand and address the multifaceted nature of undernutrition in Bangladesh.
Conclusion
The prevalence of stunting, wasting and underweight among under-5 children is relatively high in Bangladesh, with variations observed across urban and rural areas. The study highlights the critical role of HEC indicators, exposure to HAP from cooking, access to safe potable water sources and handwashing facilities in influencing the nutritional status of children. To address these challenges and reduce the burden of malnutrition, there is a need for improved housing infrastructure, reduced HAP, access to clean water and proper handwashing facilities across the countries and especially in rural areas. Bangladesh has made significant progress in improving its sanitation sectors in recent years, but still there are areas to make singificant improvements in HEC indicators. Thus, while ensuring improved housing materials may be a distant goal, augmenting the ongoing programmes of access to clean water and sanitation should be an achievable goal.
Supplementary material
For supplementary material accompanying this paper, visit https://doi.org/10.1017/S1368980024002325.
Acknowledgements
We extend our gratitude to MEASURE DHS for their valuable data support. Additionally, we acknowledge the Maternal and Child Health Division and the Health Systems and Population Studies Division of icddr,b, which acknowledges the support of the Government of Bangladesh, Canada, Sweden and the UK for providing core/unrestricted support for its operations and research, where the data for this study being analysed and the manuscript are written. The authors also recognise the support the Department of Population Science of Jatiya Kabi Kazi Nazrul Islam University, where the study was designed jointly.
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Competing interests
There are no conflicts of interest.
Authorship
MMAK and MNK conceptualised the study. MMAK, MAB and KF conducted the data analysis, while MNK supervised the process. MMAK, MAB, KF and SJK contributed to the initial manuscript draft. BKS, AB-T, MMI. and MNK provided critical review and editing of earlier manuscript versions. All authors have approved the final version of the manuscript.
Ethics of human subject participation
The study data were sourced from the MEASURE DHS Archive, initially gathered by Macro in Calverton, USA. The data collection procedure received approval from the ORC Macro Institutional Review Board. Prior to enrolment, informed consent was obtained from all participants.
Data associated in this study are freely available in https://dhsprogram.com/.
The dataset can be downloaded after registering with the MEASURE DHS at http://dhsprogram.com/data/Using-DataSets-for-Analysis.cfm.
Open access
Target article
Child undernutrition and its association with household environmental conditions in Bangladesh
Related commentaries (1)
In response to letter to the editor: the nexus of child undernutrition and household environmental conditions in Bangladesh: implications for public health and societal productivity