Introduction
Food and water security are critical to child health, nutrition, and well-being. Both are recognized by the United Nations as fundamental human rights that are indispensable to achieving the 2030 Sustainable Development Goals.(1) Household food insecurity (HFI), the limited or uncertain ability to access sufficient safe and nutritious food for an active and healthy life is estimated to impact 156 million or one-quarter of households in Latin America,(2) many with young children.(Reference Tuñón, Salvia, Cimadamore and Cherniak3) Household water insecurity (HWI), the inability to access and benefit from affordable, adequate, reliable, and safe water to support well-being and a healthy life,(Reference Jepson, Wutich and Colllins4) occurs when one or more of these dimensions is affected due to water scarcity, excess, or contamination.(Reference Miller, Workman and Panchang5) Nearly one-quarter (22%) of Latin American households lack access to safely managed drinking water(6) and 50 million reside in water-scarce areas(Reference Vazquez, Serrano and Cestti7) including 12.8 million children.(8) Latin America is ranked second to only Asia as the region most subject to drought, flooding, and other extreme weather events due to the changing climate.(8) This and other challenges (inflation, economic instability, persistent inequality, poor infrastructure, rapid urbanization, increasing pollution) are projected to increase future exposures populations in the region to HFI and HWI.(2,8,9)
Diarrheal (DI) and respiratory illnesses (RI) are highly prevalent and major causes of mortality among under-5 children in Latin America.(10) Anaemia(11) and stunting(12) are also prevalent. These are both documented as major risk factors for morbi-mortality and developmental delays in this age group.(Reference Jonker and van Hensbroek Mël13,14) Furthermore, the rising prevalence of child overweight/obesity and child-level double malnutrition (DBM), i.e. combined stunting and overweight/obesity, are emerging public health challenges.(15,Reference Ghattas, Acharya and Jamaluddine16)
Prior studies have linked HFI to RI (Reference Chakraborty, Armijos and Rosenberg17,Reference Hackett, Melgar-Quiñonez and Alvarez18) and DI (Reference Chakraborty, Armijos and Rosenberg17–Reference Pedraza21) in under-5 Ecuadorian, Colombian, and Brazilian children. HFI has also been associated with anaemia,(Reference Da Silva, Normando and Schincaglia22,Reference Palacios-Rodríguez, Mundo-Rosas and Parra-Cabrera23) stunting,(Reference Hackett, Melgar-Quiñonez and Alvarez18,Reference Psaki, Bhutta and Ahmed24–Reference Rosas, Guendelman and Harley27) or overweight/obesity(Reference MBMdC, Oliveira and Lamounier28) in children from Ecuador, Colombia, Peru, Brazil, and Mexico but others from the region reported null associations for some outcomes.(Reference Gubert, Spaniol and Bortolini19,Reference MBMdC, Oliveira and Lamounier28–Reference Kac, Schlüssel and Pérez-Escamilla32) Fewer authors have examined the relationship between HWI and the health and nutrition of under-5 Latin American children. Some reported positive associations for HWI with RI(Reference Chakraborty, Armijos and Rosenberg17) or DI(Reference Checkley, Gilman and Black33–Reference Colombara, Hernández and McNellan36) for children from Ecuador, Peru, Guatemala, and Panama although others did not.(Reference Chakraborty, Armijos and Rosenberg17,Reference Torres-Slimming, Carcamo and Wright37,Reference Rosinger38) HWI also has been associated with higher blood haemoglobin levels(Reference Choudhary, Brewis and Schuster39) and anaemia(Reference Kothari, Coile and Huestis40,Reference Westgard, Orrego-Ferreyros and Calderón41) in 6–59 months olds and with lower height-for-age Z-scores(Reference Choudhary, Brewis and Schuster39) and stunting(Reference Checkley, Gilman and Black33,Reference Balza, Cuartas and Gomez-Parra42) in under-5 Guatemalan and Bolivian children. In contrast, other studies from the region reported null associations for HWI and some of these outcomes.(Reference Rivadeneira and Cadena30,Reference Kothari, Coile and Huestis40,Reference Yu, Addo and Williams43–Reference Pedreschi, Fontes and Roa48)
The reason for the mixed study findings is uncertain. One reason could be because prior studies did not adjust for whether households with food insecurity also suffered from water insecurity or vice-versa. Another possibility is that most studies were conducted earlier in the nutrition transition of the middle-income Latin American populations and prior to the Covid-19 pandemic with its well-documented negative impacts on food and water security, health, and nutrition.(1,9,Reference Vasconcelos-Neto, Jacob and Tregidgo49–52) In addition, few authors stratified their analyses to account for age-related physiological differences in under-5 children. We also note that a dearth of prior studies makes it uncertain as to whether or not HWI is associated with overweight/obesity or if either HFI or HWI are associated with child-level DBM in young children.
HFI (Reference Maitra53–Reference Abraham, Breeze, Sutton and Lambie-Mumford55) and HWI (Reference Rosinger and Young56–Reference Rosinger58) are hypothesized to directly and/or indirectly influence child health and nutrition through dietary, microbial/chemical, psychosocial/emotional, and other pathways. Emerging evidence indicates that HFI and HWI often co-exist in households(Reference Young, Bethancourt and Frongillo59–Reference Alfonso, Quinonez and García62) including those with under-5 children.(Reference Chakraborty, Armijos and Rosenberg17,Reference Miller, Ocampo Ordóñez and Headley63) It has been suggested that HWI can exacerbate HFI and that joint HFI/HWI can magnify adverse health and nutrition impacts over that of either alone.(Reference Choudhary, Brewis and Schuster39,Reference Young, Bethancourt and Frongillo59,Reference Young, Frongillo and Jamaluddine64–Reference Brewis, Jepson and Rosinger67) Several recent studies have linked joint HFI/HWI to poorer mental(Reference Vuong, Dang and Toze68–Reference Charles, Salinger and Sweeney71) and physical health(Reference Vuong, Dang and Toze68) as well as functional disability in adults.(Reference Wang, Armijos and Commodore72) However, the joint relationship of HFI/HWI with the health and nutrition outcomes of Latin American LMIC children is severely understudied. One of two published studies to report on children linked ‘food issues’, (i.e., low dietary quality or HFI) and ‘water issues’ (i.e. HWI or non-bottled drinking water use) with undernutrition/infection and overweight/cardio-metabolic risk factors in a mixed-age sample of 2–10 year olds and adults surveyed in the Galapagos Island in 2012.(Reference Thompson, Nicholas and Watson73) The other reported that joint HFI/HWI was associated with a higher prevalence of RI and DI but not stunting in under-5 child participants from the ENSANUT 2018 survey (Reference Chakraborty, Armijos and Rosenberg17) but did not examine other outcomes.
The present study used 2023–2024 Ecuadorian National Infant Malnutrition Survey data to explore the independent and joint relationships of HFI and HWI with anaemia, overweight/obesity, and child-level DBM in 6–59 month olds. We also stratified our analyses to account for age-related gastrointestinal function, immunity, metabolism, and other physiological differences in 6–23 month olds vs. 24–59 month olds that influence health and nutrition. The study also explored whether the RI, DI, and stunting findings from the 2018 ENSANUT survey(Reference Chakraborty, Armijos and Rosenberg17) would be similar to our study or not given the severe impact of the covid-19 pandemic and its aftermath on the food and water security, health and nutrition situation in Ecuadorian households. Our working hypothesis was that RI, DI, and anaemia would be more prevalent among children exposed to HFI only, HWI only, and especially joint HFI/HWI. We also posited that these exposures would be associated with a higher prevalence of stunting, overweight/obesity, and DBM especially in 24–59 month olds as undernutrition and overnutrition usually take time to develop. Our study extends the current evidence regarding the impacts of HFI and HWI on the health and nutrition of under-5 Latin American children. It also is the first to specifically examine the relationship of joint HFI/HWI with anaemia, overweight/obesity, and child-level DBM in this group and only the second to do so for RI, DI, and stunting. The evidence from this study is expected to help inform policy and more targeted strategies to reduce child exposures to HFI and HWI and mitigate their adverse health and nutrition consequences.
Methods
Study setting
Ecuador is an upper middle-income Latin American country where 29% of the population lives in poverty and another 13% in severe poverty.(74) One-third (33.3%) of Ecuadorian households were reported to have experienced moderate–severe food insecurity during 2022–2024(2) and 30% lacked access to safely managed water.(75) Respiratory and diarrheal illnesses are highly prevalent and are major causes of morbidity and mortality in under-5 children.(76,Reference Acosta-España, Altamirano-Jara and Arnao-Noboa77) Anaemia affects an estimated 38.6% of 6–59 month olds.(2) More than one-fifth (17.7%) of under-5 children are stunted and 4.7% are overweight.(2) The prevalence of child-level DBM in under-5 children is uncertain due to the lack of published studies.
Study design & sampling
We analysed cross-sectional data from the 2nd Ecuadorian National Infant Malnutrition 2023–2024 Survey (Encuesta Nacional sobre Desnutrición Infantil) (Spanish acronym: ENDI) to investigate the study hypotheses. Information on the ENDI survey design, sampling strategy, data collection, weighting procedures, quality control and other methodological aspects have been described in detail in Ecuadorian National Institute of Statistics and Census (Spanish acronym: INEC) documents.(78–81) Briefly, the probabilistic 2-stage survey collected data from July 2023 to August 2024 to obtain a representative sample of under-5 children living in 23 mainland Ecuadorian provinces. The main ENDI database published by the INEC did not include data on children living in the Galapagos Island province as the survey used a different sampling strategy due to reported complex operational challenges.(78,79) The survey sampling frame was drawn from households registered in pre-census surveys (i.e. VII Censo de Población, VI de Vivienda del Ecuador) along with INEC cartographic updates.(78) The survey had an overall coverage of 90.4% and a low refusal rate (0.7%).(78,79)
A total of 20,347 households with 6–59 month old children were potentially available for analysis. Our analyses were restricted to 18,258 children aged 6–59 months, one from each household, with complete data on the HFI and HWI exposure and child illness, anaemia, and malnutrition outcomes of interest. Of these, 32.5% were aged 6–23 months and 67.5% were 24–59 month olds. The INEC approved survey procedures involving child and maternal caregiver participants. The survey obtained written informed consent from maternal caregivers for the themselves and their 6–59 month old child participants.(78,79) This secondary analysis of fully de-identified ENDI survey data published by INEC is classified by the IRB of our institution as research not subject to human subject regulations.
Study measures
Exposures
Food insecurity The Spanish-language household Food Insecurity Experience Scale (FIES) was administered to maternal caregivers in face-to-face interviews conducted by the trained survey personnel.(79,81) The survey modified the language for the 8-question scale to focus on HFI among under-5 children living in the household. The FIES questions reflected increasing levels of difficulty by households in obtaining food for their children.(82) These ranged from the least severe, i.e. caregiver worry about not being able to feed their children enough food caused by a lack of money or other resources to the most severe situation, i.e. children were unable to eat for a whole day due to the lack of money or other resources. Affirmative and negative responses to the questions were coded as one or zero. The summed responses produced raw scores ranging from 0 to 8. We used the FIES online App, a tool based on the Rasch model, to validate the data and identify cut-off scores for food insecurity levels.(83) Briefly, the infit statistics ranged from 0.71 to 1.52, and Rasch reliability was 0.79. Joint correlations between all item pairs were 0.41 or lower, indicating acceptable model fit and reliability. Marginal HFI was defined as a raw score of 1–3 and moderate–severe HFI as a score of 4–8 on the FIES scale. Child households with a score of 1–8 score, i.e. any food level of insecurity were classified with HFI. Those with a zero score were classified as food secure.
Water insecurity The survey interviewers collected data from maternal caretakers on the source, accessibility, and availability of household drinking water. In addition, survey technicians collected and analysed household drinking water samples for coliform bacteria at the point of consumption using a standardized protocol and quality control measures.(79,81) We used these data to construct a household water insecurity indicator based on the WHO/UNICEF Joint Monitoring Programme for Water Supply, Sanitation, and Hygiene (JMP) criteria for safely managed water, i.e. drinking water that comes from an improved source, that is accessible on the premises, that is available when needed, and is free from contamination.(6) In the case of the latter indicator, only microbial contamination was considered in the assessment of water safety as the survey did not test for toxic chemical contaminants in household drinking water, e.g. arsenic, cadmium, PFAS.
Households were classified as water secure if they met all four criteria, i.e. their drinking water came from an improved source, water was accessible on the premises, a sufficient quantity of water was available when needed in the prior 2 weeks, and the microbial analysis showed the absence of coliform bacteria. They were classified as water insecure if their household drinking water did not meet one or more criteria, i.e. water came from an unimproved source, it was not accessible on household property, it was not always available when needed during previous 2 weeks and if the findings from the microbial analysis performed by the trained survey technicians confirmed the absence of coliform bacteria water obtained at the point of consumption.
4-category HFI and HWI composite indicator
We constructed a 4-category composite indicator to assess food and water insecurity status in the child households. The four categories were defined as follows: (1) joint HFI/HWI (households with both HFI and HWI), (2) HFI only (households with HFI but not HWI), (3) HWI (households with HWI but not HFI), (4) food and water secure (households with no HFI or HWI), the reference group.
Health and nutrition outcomes
Recent respiratory illness Maternal caregivers reported on child respiratory illness symptoms during the past 2 weeks (i.e. cough, runny nose, shortness of breath, flu, sore throat, flu).(79) Children whose caregivers reported in the affirmative were classified as positive for respiratory illness symptoms and negative if none was reported.
Recent diarrheal illness Maternal caretakers reported on whether their under-5 children experienced diarrhoea during the prior two weeks.(79) Children whose caregivers responded in the affirmative were classified as positive for diarrhoea and negative if none was reported.
Anaemia Trained survey technicians collected capillary blood samples obtained by fingerstick to measure child blood haemoglobin levels.(79,81) The blood samples were analysed using a calibrated HemoCue photometer (Hemocue AB, Sweden). The survey used age-and altitude-adjusted blood haemoglobin (Hb) values to identify anaemia as per the revised WHO 2024 guidelines. Children aged 6–23 months and 24–59 months were classified as having anaemia if their respective Hb was ≤10.5 g/dL or ≤11 g/dL.
Anthropometric indicators Following a standardized protocol,(79,81) trained survey anthropometrists measured recumbent length in children under the age of 2 years using a calibrated infantometer. They measured standing height in children aged 2 years and older with a calibrated stadiometer. Weight was measured using a calibrated electronic weighing scale. Child length/height and weight measurements were each performed twice and recorded. In cases where the first two height measurements differed by ±0.5 cm, a third measurement was made and the average value from the two closest measurements was used. Children were classified as stunted if their length/height-for-age z-scores were <2 standard deviations (SD) below the median.(84) They were classified as overweight if their BMI-for-age scores were >2 SD below the median and obese if they were >3 SD below the median according to WHO international age-and sex-specific growth reference.(84) In this study, we combined overweight and obesity into a single indicator due to the relatively low frequency of both. Children identified with concomitant stunting and overweight/obesity were classified with individual-level DBM.(Reference Ghattas, Acharya and Jamaluddine16)
Covariates
The sociodemographic, maternal caregiver, and child variables considered for inclusion as covariates in the multivariable statistical models were based on those identified by prior studies, conceptual frameworks, and their established associations with the child outcomes. These were child sex (male, female), maternal ethnoraciality (indigenous, afro-descendant, montubio, mestizo, white/other), maternal education (none/primary/basic education, middle school/high school, post-baccalaureate), residence (urban, rural), and region of residence (Andean highlands, Pacific Coast, Ecuadorian Amazon), and household sanitation (access to basic/improved sanitation, other), and household income poverty (yes, no) defined as household per capita income that was below the governmental poverty line cutoff which was US＄ 90–92 during the study period. The other covariates were child micronutrient supplementation during the past 12 months (yes, no), antiparasitic drugs in past 6 months (yes, no), and whether the child received at least one dose of five childhood vaccines (noted in health record or by maternal report) relevant to diarrheal and respiratory illnesses and growth, i.e. BCG, rotavirus, MMR, DPT, influenza, pneumococcal pneumonia.
Data analysis
Data were analysed using StataSE19.5 (College Station, TX: StataCorp LLC, USA). Sample weights provided by the 2023–2024 ENDI survey were used to produce representative estimates for the analyses. We used the svy prefix command for survey data to calculate standard errors which adjusted for household-level clustering and stratification. Summary statistics are reported as frequencies with weighted percentages for categorical variables and weighted means ± SE for continuous variables. We analysed categorical and mean differences in sample characteristics by household food and water security status using X 2, students’ t-test, or GLM as appropriate.
We used modified Poisson regression analysis to estimate the association of the joint HFI/HWI composite indicator with the child health (i.e. DI, RI, anaemia) and nutrition outcomes of interest in the study (i.e. stunting, overweight/obesity, DBM). We also stratified our analyses by child age (i.e. 6–23 months, 24–59 months) to account for age-related physiological variations in nutritional requirements, immune development, growth, and development. The model findings are reported as unadjusted and adjusted prevalence ratios with their respective CIs. Adjusted Model 1 included child age, maternal ethnicity, maternal education, household income poverty, urbanicity, region of residence, and household sanitation. Adjusted Model 2 included Model 1 variables as well as child micronutrient supplementation during the past 12 months, child antiparasitic medications during the prior 6 months, and childhood vaccine history. We assessed for potential multicollinearity using Variance Inflation Factors. All VIF values were below 2 (range: 1.05–1.45) suggesting that multicollinearity was not a significant concern. We did not adjust for multiplicity as the planned analyses in this study were exploratory and not meant to support causal inference. The statistical significance level for 2-sided tests was set at p < 0.05.
In addition to the main full sample and age-stratified analyses, we conducted additional sensitivity analyses where we restricted food insecurity cases to those exposed to moderate–severe HFI, a more serious form of food insecurity characterized by decreased food quality and/or quantity. In the sensitivity analyses, child households with moderate–severe HFI (i.e. a raw score of 4–8 on the FIES) were defined as food insecure while households with 0–3 scores were defined as food secure. Thus, the sensitivity analysis considered moderate–severe HFI but not marginal HFI for the HFI only and joint HFI/HWI indicators. Findings from the sensitivity analyses are synopsized further below and detailed in the Supplemental Analyses section.
Results
The study identified a high overall prevalence of any household food insecurity, i.e. marginal to moderate–severe (69.8%) and any water insecurity (33.8%). The 4-category indicator showed that 43.4% of children lived in households with mild to moderate–severe HFI but no water insecurity (i.e. HFI only), 7.4% in households with water insecurity but no food insecurity (i.e. HWI only), and 26.4% in households with both food and water insecurity (i.e. joint HFI/HWI). The other 22.8% resided in fully food and water secure households. One-third of children were aged of 6–23 months and the other two-thirds were aged 24–59 months. Approximately one-half (50.1%) of children were male. Table 1 shows that although household food and water security status did not differ by child sex, it did differ by other sample characteristics including child age, maternal and household attributes, and other child characteristics (i.e. micronutrient supplementation, antiparasitic drug and vaccination status).
Sample characteristics by household food and water security status
Table 1. Long description
The table presents sample characteristics by household food and water security status. It has 18 rows and 7 columns. The columns are labeled as Total sample, Food & water secure, HFI only, HWI only, and Joint HFI/HWI, with each column providing weighted percentages or means with standard errors. The rows include child age, child sex, maternal ethnoraciality, maternal education, household income poverty, urban area residence, region of residence, household sanitation, child micronutrient supplements, child antiparasitic medication, and child vaccination status. Each row provides detailed data for each category, showing variations in household food and water security status across different characteristics.
HFI: Household food insecurity (marginal or moderate–severe HFI).
HWI: Household water insecurity.
PR = prevalence ratio.
a No. of common childhood vaccines received (at least one dose): BCG, rotavirus, influenza, measles-mumps-rubella, DPT, pneumococcal pneumonia.
Maternal caregivers reported that 51% of their children had suffered from a respiratory illness during the prior 2 weeks and 18% from a diarrheal illness during the prior two weeks. Nearly one-third (32.2%) had blood haemoglobin levels indicative of anaemia. Anthropometry identified 18% with stunting, 4.8% with overweight/obesity, and 0.8% with double-malnutrition (i.e. combined stunting and overweight/obesity).
Table 2 reports the modified Poisson regression analysis results examining the unadjusted and adjusted associations of joint HFI/HWI, HFI only, and HWI only with child health and nutrition indicators in the full 6–59 month old sample. The final adjusted model (Model 2) findings showed that children exposed to joint HFI/HWI had a higher prevalence of recent RI (aPR = 1.29; 95% CI = 1.17,1.43) and recent DI (aPR = 1.36; 95% CI = 1.13,1.72) compared to the food and water secure reference group. Exposure to HFI only was also associated with a higher prevalence of recent RI (aPR = 1.21; 95% CI = 1.12,1.31), recent DI (aPR = 1.17; 95% CI = 1.12,1.67), stunting (aPR = 1.21; 95% CI = 1.04,1.40) and overweight/obesity (aPR = 1.34; 95% CI = 1.03,1.74). Furthermore, exposure to joint HFI/HWI was associated with a higher prevalence of recent RI or DI compared to HFI only exposure. The study also found that exposure to HWI only was associated with higher overweight/obesity prevalence (aPR = 1.91; 95% CI = 1.17, 3.10) but a lower anaemia prevalence (aPR = 0.77; 95% CI = 0.64, 0.92).
Association of household food and water insecurity status with health and nutrition outcomes in 6–59 month old children (n = 18,258)
Table 2. Long description
The table presents the association of household food and water insecurity with health and nutrition outcomes in 6-59 month old children. It has 10 rows and 5 columns. The columns are labeled as No. (weighted %), Unadjusted model PR (95% C.I.), Model 1 adjusted PR (95% C.I.)a, and Model 2 adjusted PR (95% C.I.)b. The rows are grouped under different health and nutrition outcomes: Recent RI, Recent DI, Anaemia, Stunting, Overweight/obesity, and Double malnutrition. Each group has subcategories: Joint HFI/HWI, HFI only (no HWI), HWI only (no HFI), and Food and water secure. The table provides prevalence ratios (PR) and confidence intervals (C.I.) for each category and outcome. Notable trends include higher prevalence ratios for recent RI and DI in children exposed to joint HFI/HWI and HFI only compared to the food and water secure reference group. Exposure to HWI only is associated with higher overweight/obesity prevalence but lower anaemia prevalence.
HFI: Household food insecurity (marginal or moderate–severe HFI).
HWI: Household water insecurity.
PR = prevalence ratio.
a Model 1 adjusted for maternal ethnoraciality, maternal education, income poverty, urbanicity, and region of residence.
b Model 2 adjusted for Model 1 variables + basic/improved home sanitation, child micronutrient supplementation (past 12 mos.), child antiparasitic treatment (past 6 months), and no. of common childhood vaccines received (at least one dose): BCG, rotavirus, influenza, measles-mumps-rubella, DPT, pneumococcal pneumonia.
* p < 0.001;
^ p < 0.01;
# p < 0.05; all other p-values ≥ 0.05.
The age-stratified analyses displayed in Table 3 found that exposure to joint HFI/HWI was associated with recent RI in 6-23 month old (aPR = 1.27; 95% CI = 1.08,1.49) and 24–59 month old children (aPR = 1.30; 95% CI = 1.17,1.46). Likewise, HFI only exposure was associated with recent RI in both the younger (aPR = 1.23; 95% CI = 1.07,1.41) and older age groups (aPR = 1.20; 95% CI = 1.11,1.31). Recent DI also was associated with exposure to joint HFI/HWI (aPR = 1.43; 95% CI = 1.11,1.84) and HFI only in 23 month olds (aPR = 1.32; 95% CI = 1.05,1.66) but not the 24–59 month old age group. Exposure to joint HFI/HWI (aPR = 0.84; 95% CI = 0.72, 0.99) or HWI only (aPR = 0.76; 95% CI = 0.61, 0.95) was associated with a lower prevalence of anaemia in 24–59 month olds. In 24–59 month olds, HFI only was associated with a higher stunting prevalence (aPR = 1.26; 95% CI = 1.05,1.52) whereas HWI was associated with a higher overweight/obesity prevalence (aPR = 1.97; 95% CI = 1.05, 3.71).
Association of household food and water insecurity status with child health and nutrition outcomes: analysis stratified by child age 6–23 (n = 5,933) and 24–59 months (n = 12,325)
Table 3. Long description
The table presents data on the association of household food and water insecurity with child health and nutrition outcomes, stratified by child age groups 6-23 months and 24-59 months. It includes columns for unadjusted model PR, Model 1 adjusted PR, and Model 2 adjusted PR, with rows categorized by joint HF/HW, HFI only, HWI only, and food and water secure. The table has 6-23 month children and 24-59 month children sections, each with subcategories for recent DI, anemia, stunting, overweight/obesity, and double malnutrition. Each row provides weighted percentages and adjusted prevalence ratios (PR) with 95% confidence intervals (CI) for different health and nutrition outcomes. Notable trends include higher prevalence ratios for recent RI and DI in children exposed to joint HFI/HWI and HFI only, and lower prevalence of anemia in older children exposed to joint HFI/HWI or HWI only. Additionally, HFI only is associated with higher stunting prevalence, and HWI with higher overweight/obesity prevalence in older children.
HFI: Household food insecurity (marginal or moderate-severe HFI).
HWI: Household water insecurity.
PR= prevalence ratio.
a Model 1 adjusted for maternal ethnoraciality, maternal education, income poverty, urbanicity, and region of residence.
b Model 2 adjusted for Model 1 variables + basic/improved home sanitation, child micronutrient supplementation (past 12 mos.), child antiparasitic treatment (past 6 months), and no. of common childhood vaccines received (at least one dose): BCG, rotavirus, influenza, measles-mumps-rubella, DPT, pneumococcal pneumonia.
* p < 0.001;
^ p < 0.01;
# p < 0.05; all other p-values > 0.05.
Sensitivity analysis
The sensitivity analyses restricted the definition of food insecurity to child households with a FIES score of 4–8 (i.e. moderate–severe HFI). With this definition, 22% of the child households had HFI only, 18.9% had HWI only, and 14.9% had joint HFI/HWI. The magnitude and direction of association for the food and water security indicators with child health and nutrition in the full sample was similar overall to the main analysis except for the relationship of HFI only and HWI only with child overweight/obesity which was not evident (Table S1). The age-stratified sensitivity analysis findings for recent RI were similar to those reported for main analyses for both age groups although in 24–59 month olds, recent DI was also more prevalent among those exposed to HFI only and HFI/HWI (Table S2). In 24–59 month olds, the previously identified relationships of joint HFI/HWI with anaemia and HWI only with overweight/obesity were no longer apparent. However, DBM was more prevalent among those with exposed to HFI only or joint HFI/HWI (Table S2).
Discussion
This study is the first to report on the association of joint food and water insecurity with anaemia, overweight/obesity, and child-level DBM in under-5 Latin American or other LMIC children and only the second to do so for RI, DI, and stunting. It recorded a high prevalence of food and water insecurity. Specifically, 43.4% of the 6–59 month old Ecuadorian children lived in food insecure but not water insecure households, 7.4% in water insecure but not food insecure households, and 26.4% in food and water insecure households. Only 22.8% of the children were from fully food and water secure households. In this study, recent RI, recent DI, stunting and overweight/obesity were more prevalent in children from households with HFI only. Overweight/obesity was more prevalent in those from HWI only households although anaemia was less prevalent. Both RI and DI were more prevalent in children from households with joint HFI/HWI. After stratification by age, RI and DI were found to be more prevalent in 6–23 month olds from HFI only or joint HFI/HWI households. In 24–59 month olds, RI and stunting were more prevalent in those from HFI only households. Overweight/obesity was more prevalent in this age group of children who lived in HWI only households but anaemia was less prevalent. Those from joint HFI/HWI households had a higher RI prevalence and a lower prevalence of anaemia. Furthermore, when food insecurity was restricted to moderate–severe HFI cases in the sensitivity analysis, both DI and DBM were more prevalent in 24–59 month olds from HFI only or joint HFI/HWI households.
Several direct and indirect pathways exist that could explain the independent and joint relationships of food and water insecurity with the child outcomes identified in this study. One of these is through diet/nutrition. For example, starting around 6 months, children need to consume a diverse, high-quality diet to support their increasing nutritional requirements as breastmilk or artificial milk alone is insufficient to meet these.(85) While caregivers may try to buffer their children against HFI or HWI,(Reference Rhue, Torrico and Amuzie57,Reference Schuster, Butler and Wutich66,Reference Brewis, Jepson and Rosinger67) this is not always possible. In the case of HFI, they may be forced to substitute preferred weaning, complementary, or other foods for cheaper, less diverse, and lower quality foods or engage in other coping strategies, e.g. serving smaller portions, spacing feedings(Reference Maitra53–Reference Abraham, Breeze, Sutton and Lambie-Mumford55,Reference Miller, Ocampo Ordóñez and Headley63) Caregivers from households with insufficient, unreliable, or unsafe water may try to manage by extending breastfeeding duration, giving their child with bottled water or pre-packaged sugar-sweetened beverages (SSBs), or by rationing drinking water.(Reference Miller, Workman and Panchang5,Reference Miller, Ocampo Ordóñez and Headley63,Reference Schuster, Butler and Wutich66,Reference Lee, Whitney and Blum86–Reference Deshpande, Huanca and Conde88) In addition, HWI can exacerbate HFI by making it harder for caregivers to serve children nutritious foods requiring water for their preparation.(Reference Miller, Ocampo Ordóñez and Headley63,Reference Young, Frongillo and Jamaluddine64,Reference Schuster, Butler and Wutich66,Reference Brewis, Jepson and Rosinger67)
Child exposure to food-or water-borne microbial or chemical pathogens is another potential pathway. For instance, caregivers may try to cope with HFI by serving expired foods, or removing spoiled sections or insects from contaminated foods. (Reference Anater, McWilliams and Latkin89,Reference Kempson, Keenan and Sadani90) Insufficient or contaminated household water makes it difficult to adequately wash baby bottles, spoons, bowls, and other feeding utensils (Reference Rhue, Torrico and Amuzie57,Reference Miller, Ocampo Ordóñez and Headley63,Reference Schuster, Butler and Wutich66) and can also negatively impact caregiver and child hand hygiene.(Reference Workman, Stoler and Harris91) The adverse effects of HFI and HWI on caregiver health is another plausible pathway. Both HFI (Reference Maitra53–Reference Abraham, Breeze, Sutton and Lambie-Mumford55) and HWI (Reference Rhue, Torrico and Amuzie57,Reference Schuster, Butler and Wutich66,Reference Brewis, Jepson and Rosinger67) are well-documented psychosocial stressors associated with lower maternal self-efficacy, anxiety, depression, and poorer overall mental health. Exposure to joint HFI/HWI is also reported to worsen the mental and physical health of maternal caretakers and other adults. (Reference Vuong, Dang and Toze68–Reference Charles, Salinger and Sweeney71) HFI has been specifically linked to depressive symptoms,(Reference Weigel, Armijos and Racines92,Reference Weigel and Armijos93) psychological stress manifested as ethnospecific illness and poorer physical health in Ecuadorian maternal caregivers.(Reference Weigel and Armijos93) In Ecuadorian adults, joint HFI/HWI has been linked to functional disability.(Reference Wang, Armijos and Commodore72) Experiencing HFI or HWI may cause caregivers to be less responsive to child food and water feeding cues or to carry out other important child care (e.g. hygiene, physical/cognitive/emotional stimulation),(Reference Rhue, Torrico and Amuzie57,Reference Miller, Ocampo Ordóñez and Headley63,Reference Schuster, Butler and Wutich66,Reference Weigel and Armijos93) all of which can place young children at risk for suboptimal health and nutrition. Other potential pathways include the disruptive impact of HFI or HWI on household food production by caregivers(Reference Workman, Stoler and Harris91) and the physical burden and opportunity costs of seeking out food, water, or both. This can also negatively impact child feeding, hygiene, and other essential caregiving.(Reference Rhue, Torrico and Amuzie57,Reference Schuster, Butler and Wutich66,Reference Brewis, Jepson and Rosinger67)
RI and DI are among the most common causes of morbi-mortality and hospitalization in under-5 Ecuadorian children. (Reference Buñay-Morocho, Álvarez and Zurita94–Reference Vargas-Gaibor, Rendón-Viteri and Alvarado-Villa96) The higher RI and DI prevalence identified for children exposed to HFI only or joint HFI/HFI is consistent with our hypothesis. Our findings generally align with those from the only other study to use a comparable 4-category food and water insecurity indicator to also explore these relationships in an under-5 LMIC group, in this case, 0–59 month old Ecuadorian children.(Reference Chakraborty, Armijos and Rosenberg17) We also found that the magnitude of RI and DI prevalence estimates among those from households with joint HFI/HWI was higher than those from HFI only households.(Reference Chakraborty, Armijos and Rosenberg17) We surmise that the higher RI and DI prevalence identified for children from HFI only or joint HFI/HWI households could be due to greater exposures to food-and/or water-borne pathogens or reduced malnutrition-associated immunity. RI and DI are well-documented players in the malnutrition-infection cycle. Both can worsen child nutritional status through their deleterious effects on child appetite and nutrient intake, absorption, excretion, and metabolism. In turn, malnutrition can adversely impact child immune response, leading to more frequent or severe infections and over time, to chronic malnutrition such as stunting.(Reference Sinha and Guerrant97)
Ecuador has one of the highest under-5 child stunting rates in the Latin American region.(98) The higher stunting prevalence identified for HFI only-exposed children in our study, especially 24–59 month olds, was consistent with our hypothesis although we also had expected to find positive associations for joint HFI/HWI and HWI only. Our finding differs from the null association reported by the other Ecuadorian study which used the 4-category HFI and HWI indicator.(Reference Chakraborty, Armijos and Rosenberg17) It is possible that differences in when the data from the two studies were conducted could explain this discrepancy. Specifically, data from the prior study were collected two years before the onset of the 2020–2022 Covid-19 pandemic in Ecuador whereas ours was gathered from July 2023 to August 2024. Child stunting usually develops over time as a consequence of chronically inadequate diets, repeated bouts of enteric and other infections that increase energy requirements and/or impair nutrient absorption, and diet-infection interactions.(12,14) Thus, it is possible that the well-documented adverse impacts of the pandemic on food security, diet, and health (1,Reference Maitra53,Reference Frongillo, Adebiyi and Boncyk54,Reference Collins, Mock and Chaparro87) may have influenced the prenatal and/or post-natal linear growth trajectories of some children conceived or birthed during that period.
Ecuador has a persistently high child burden of infectious illnesses and chronic undernutrition including stunting. At the same time, it is undergoing rapid demographic, nutritional, and epidemiologic transition and along with it, has rising rates of overweight/obesity, DBM, and diet-related non-communicable conditions (i.e. metabolic syndrome, diabetes, hypertension).(Reference Chis Ster, Alvarez-Mena, Zurita-Loma, Gualan, Rojeab-Bravo, Cooper and Romero-Sandoval99–Reference Ramírez-Luzuriaga, Belmont and Waters101) This transition is characterized by a shift away from traditional diets to one characterized by a higher consumption of energy-dense, ultra-processed foods and SSBs, lower fruit and vegetable intakes, changes in gastrointestinal microbiota, decreased physical activity levels, and increased sedentarism.(Reference Weigel, Armijos and Racines92,Reference Chis Ster, Alvarez-Mena, Zurita-Loma, Gualan, Rojeab-Bravo, Cooper and Romero-Sandoval99,Reference Micoanski, Izquierdo-García, Huacho-Jácome, Trelis and Gozalbo102,Reference Chakraborty, Armijos and Weigel103) We note that Ecuador has one of the highest per capita annual consumption of sugars and sweeteners in Latin America (78.5 kg/yr).(104) The higher overweight/obesity prevalence identified for the full sample of 6–59 month olds from HFI only or HWI only households was consistent with our hypothesis as was that for 24–59 month olds from HWI only households. We posit that the excess prevalence of overweight/obesity identified in Ecuadorian children by our study could be due to a higher consumption of SSBs which can replace water intake. Indirect support for this mechanism comes a recent study reporting that 2–17 year olds from HFI and HWI households in the U.S. had higher SSB intakes than their non-exposed counterparts. It also reported that HWI was more strongly associated with SSB intake than HFI.(Reference Rosinger and Young105) The findings from the other study also found that 2–5 years who did not drink plain water had higher average SSB intake.(Reference Rosinger, Bethancourt and Francis106) This topic needs to be further investigated due to its important paediatric and public health implications.
Consistent with our hypothesis, there was a higher prevalence of child-level DBM (i.e. combined overweight/obesity and stunting) identified by the sensitivity analysis for 24–59 month olds from households with moderate–severe HFI only or joint moderate–severe HFI/HWI. Double malnutrition is thought to originate from in-utero and/or early postnatal nutritional insult, manifested as stunting.(Reference Wells, Sawaya and Wibaek107) This finding is of high public health concern as DBM has been linked to metabolic dysregulation, impaired insulin signalling, gut microbiome imbalance, and other perturbations that promote the accumulation of fat deposits instead of lean tissue when children are subsequently exposed to energy-dense, nutrient-poor diets.(Reference Wells, Sawaya and Wibaek107)
The lower prevalence of altitude-adjusted anaemia identified for the 24–59 month old children exposed to HWI only or joint HFI/HWI was unexpected and differs from our hypothesis. It also differs from other authors reporting higher blood haemoglobin levels(Reference Choudhary, Brewis and Schuster39) and anaemia(Reference Kothari, Coile and Huestis40,Reference Westgard, Orrego-Ferreyros and Calderón41) in 6–59 months or null associations in Latin American children from water insecure homes.(Reference Yu, Addo and Williams43,Reference da Silva, Fawzi and Cardoso44) The reason for this unanticipated finding is not immediately evident but it is possible that unmeasured factors such as dehydration could have played a role. Dehydration has been previously linked to HWI in Bolivian 2–12 year olds.(Reference Rosinger38) The fluid intake of most 24–59 month olds comes from drinking water and other foods and beverages other than breastmilk. Insufficient fluid intake and/or excessive loss caused by enteric/other infections and/or thermal stress can cause hypovolemia leading to hemoconcentration.(Reference Daley and Avva108) Thus, it is possible that haemoglobin levels of some water insecure children were artificially elevated resulting in anaemia misclassification. For this reason, future studies should consider measuring hydration status when assessing the HWI-anaemia relationship in children.
Limitations
This study had a large sample size, standardized data collection, quality control procedures, and high coverage (∼86%) and low non-response (∼1%) rates. It specifically collected data on the food security situation of under-5 children rather than all household members. The composite indicator we used to assess joint HFI/HWI differentiated between children exposed to both vs. those who were fully food and water secure. It used four JMP safely managed household drinking water indicators (i.e. source, accessibility, availability, safety) to assess HWI rather than single measures and water safety was objectively measured (i.e. presence/absence of coliforms) at the point of consumption. The child anaemia and anthropometric indicators also were objectively measured. In addition, the study collected data on a number of child, caregiver, and household variables allowing for adjustment of potential confounders in the multivariable models.
One of the study limitations was its cross-sectional design which allows for inference but does not establish causality or temporality. Another was that caregiver recall or other reporting bias could have affected RI and DI estimates although the short 2-week period assessed somewhat reduces this possibility. Although our study measured drinking water coliforms, it did not assess other potential waterborne or foodborne microbial contaminants or chemicals. Using an experience-based instrument would have been useful such as the Water Insecurity Experiences Scale. This could help to provide insights into the lived experiences of water insecure caretakers, permit a more comprehensive evaluation of HWI, and better align with the experience-based FIES instrument used in our study to assess HFI. It could also facilitate cross-population comparisons. We suggest that future ENDI and other Ecuadorian national surveys consider adopting this cross-culturally validated instrument.
Another potential limitation is that some of reference periods differed. Two of the four HWI indictors had no specific reference period (i.e. water source, accessibility. The time frame for water availability was the prior 2 weeks as was the reference period for DI and RI. The blood haemoglobin measurements used to assess anaemia were collected on the day of the survey. Likewise, the anthropometric indicators used to identify stunting and overweight/obesity were collected on the same day although these usually develop over time. The anaemia assessment did not differentiate between that caused by nutrients other than iron nor inflammation. It is also possible that the relatively low prevalence of overweight/obesity and DBM could have affected the findings especially for the smaller 6–23 month old age group.
Although joint HFI/HWI was common, we point out that the relatively low prevalence of HWI only in the sample could have reduced the power to identify associations with some child outcomes. The multivariable analyses adjusted for a number of child, maternal, and household characteristics but it is possible that confounding could have occurred due to unmeasured variables. It is also important to measure other types of insecurity (e.g. energy, housing, transportation, crime) as these also can impact both HFI and HWI status. Future studies should collect detailed data on these variables to better understand the pathways connecting HFI and HWI with child health and nutrition. It is important for these to specifically identify the causal linkages to inform more effective policies and interventions. Finally, the findings also may be limited regarding their generalizability to child populations outside of Ecuador.
Conclusions
The study findings showed that food and water insecurity were common in the Ecuadorian mainland households with 6–59 month old children. They also suggest that these can act singularly or together to influence child health and nutrition. The findings indicated that joint exposure to HFI/HWI further raised the prevalence of child RI and DI above that of their independent contributions. The study also found that the independent and joint influences of food and water insecurity varied by age, ostensibly due to age-related exposure and physiological differences between 6–23 month olds vs. 24–59 month olds. Furthermore, exposure to any level of food insecurity, including marginal HFI, was associated with poorer child health and nutrition. However, only exposure to more intense food insecurity, i.e. moderate–severe HFI or to joint moderate–severe HFI/HWI was associated with child-level double malnutrition. At the national levels, Ecuador has sufficient food to feed its population and abundant water to drink. However, access to nutritious food and safe drinking water is not evenly distributed among the population especially in low-income households, rural residents, ethnoracial minorities, immigrants/refugees and other vulnerable groups.
Implementation of an overall approach that more tightly integrates policy and interventions across multiple sectors, (i.e. social welfare, agriculture, WASH, health, and environment) and at multiple levels (i.e. national, provincial, canton, local) could help to improve access to nutritious diets and safe drinking water and ultimately, better the health and nutrition of young children and their households.(Reference Miller, Ocampo Ordóñez and Headley109) Examples of specific strategies include the expansion of existing conditional/nonconditional cash transfer and other programmes to cover a larger proportion of low-income child households, the strengthening and enforcement of existing laws and policies on foods and beverages including tightening up the nutrition labelling system to reduce current commercial sector loopholes and advertising for ultra-processed food and beverage products targeted at children,(Reference López-Gil, Quiroz-Cárdenas and López-Moreno110) and increasing the current SSB and VAT taxes on ultra-processed foods. Another is enacting strict policies on the participation of corporate ultra-processed food and beverage company interests in public policy spaces. These interests are reported to have an adverse impact on Ecuadorian national nutrition policy and programme implementation and resource allocation including that focused on reducing child malnutrition.(Reference Torres, Villalba and López-Cevallos111) Furthermore, investing in water and sanitation infrastructure could improve the water security and child health including scaling up local watershed protection and other conservation efforts, expanding and improving rural community water systems, and upgrading ageing municipal water and wastewater systems. This is especially important as Ecuador has been experiencing extreme weather events/climate change and environmental degradation which have caused water shortages, decreased water quality, crop failure, and recurrent, widespread electrical system blackouts.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/jns.2026.10151.
Data availability statement
The de-identified secondary data used for the analyses is available from the Ecuadorian Instituto Nacional de Estadística y Censos (INEC). These data are freely available from the following public domain resource: Encuesta Nacional sobre Desnutrición Infantil |
Author contributions
Drs. Weigel and Armijos both contributed to the conception and design of the study. Dr. Weigel obtained and analysed the data and also wrote the first draft of the manuscript. Both authors participated in the review and writing of subsequent versions. Both also read and approved the final manuscript.
Funding statement
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Competing interests
The author(s) declare none.
Human subjects
The analysis of this publicly available data is classified as exempt research not subject to human subject regulations by the Institutional Review Board (IRB) of the author’s academic institution.
Open access
