Hostname: page-component-586b7cd67f-t7czq Total loading time: 0 Render date: 2024-11-27T14:38:51.977Z Has data issue: false hasContentIssue false

Prevalence and determinants of double and triple burden of malnutrition among mother–child pairs in Malawi: a mapping and multilevel modelling study

Published online by Cambridge University Press:  21 October 2024

Jessie Jane Khaki*
Affiliation:
Centre for Health Informatics, Statistics and Computing (CHICAS), Lancaster University, Lancaster, UK Malawi Liverpool Wellcome (MLW) Programme, Blantyre, Malawi School of Global and Public Health, Kamuzu University of Health Sciences, Blantyre, Malawi
Peter M Macharia
Affiliation:
Centre for Health Informatics, Statistics and Computing (CHICAS), Lancaster University, Lancaster, UK Department of Public Health, Institute of Tropical Medicine, Antwerp, Belgium Population & Health Impact Surveillance Group, Kenya Medical Research Institute-Wellcome Trust Research programme, Nairobi, Kenya
Lenka Beňová
Affiliation:
Department of Public Health, Institute of Tropical Medicine, Antwerp, Belgium
Emanuele Giorgi
Affiliation:
Centre for Health Informatics, Statistics and Computing (CHICAS), Lancaster University, Lancaster, UK
Aline Semaan
Affiliation:
Department of Public Health, Institute of Tropical Medicine, Antwerp, Belgium
*
Corresponding author: Jessie Jane Khaki; Email [email protected]
Rights & Permissions [Opens in a new window]

Abstract

Objective:

To establish the prevalence of double burden of malnutrition (DBM) and triple burden of malnutrition (TBM) among mother–child pairs in Malawi and explore their geographical distribution and associated multilevel factors.

Design:

Cross-sectional study using secondary data from the 2015–2016 Malawi Demographic and Health Survey using a mixed effects binomial model to identify multilevel factors associated with DBM and TBM. Georeferenced covariates were used to map the predicted prevalence of DBM and TBM.

Setting:

All twenty-eight districts in Malawi.

Participants:

Mother–child pairs with mothers aged 15–49 years and children aged below 60 months (n 4618 pairs) for DBM and between 6 and 59 months (n 4209 pairs) for TBM.

Results:

Approximately 5·5 % (95% confidence interval (CI): 4·7 %, 6·4 %) of mother–child pairs had DBM, and 3·1 % (95 % CI: 2·5 %, 4·0 %) had TBM. The subnational-level prevalence of DBM and TBM was highest in cities. The adjusted odds of DBM were threefold higher (adjusted Odds Ratio, AOR: 2·8, 95 % CI: 1·1, 7·3) with a higher proportion of wealthy households in a community. The adjusted odds of TBM were 60 % lower (AOR: 0·4; 95 % CI: 0·2, 0·8) among pairs where the women had some education compared with women with no education.

Conclusions:

Although the prevalence of DBM and TBM is currently low in Malawi, it is more prevalent in pairs with women with no education and in relatively wealthier communities. Targeted interventions should address both maternal overnutrition and child undernutrition in cities and these demographics.

Type
Research Paper
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2024. Published by Cambridge University Press on behalf of The Nutrition Society

Malnutrition, including micronutrient deficiencies, excesses or imbalances, causes serious and long-lasting adverse outcomes at both individual and community levels. Nearly half (45 %) of all deaths among under-five children globally are due to nutrition-related factors, and the burden is higher in low- and middle-income countries(1). Since 2000, there has been substantial progress in reducing the burden of undernutrition among under-five children(Reference Micha, Mannar and Afshin2). The global prevalence of stunting, which is the most common type of child malnutrition, has reduced from 32·6 % in 2000 to 22·2 % in 2017(Reference Micha, Mannar and Afshin2). However, a recent World Health Organization (WHO) report cited that populations in 88 % of 141 countries experienced multiple types of malnutrition, including child stunting and overweight in women(Reference Micha, Mannar and Afshin2). The report further highlighted that forty-one (29 %) of the countries, thirty of which are in Africa, experience a high burden of child stunting (≥ 20 %) and overweight and obesity in adult women (≥ 35 %).

The coexistence of undernutrition (such as stunting, wasting or micronutrient deficiency) and overnutrition (obesity or overweight) is defined as double burden of malnutrition (DBM)(3). The WHO states that DBM can occur at the individual level (e.g. coexistence of overnutrition with mineral or vitamin deficiencies in one individual), household level (e.g. nutritional anaemia in a child and overnutrition in another member of the household) and population level (e.g. the existence of a burden of undernutrition and overnutrition in the same community such as a village, district or country)(3). DBM among mother–child pairs is defined as the coexistence of undernutrition (wasting, stunting or underweight) in the child and overnutrition (overweight or obesity) in the mother(Reference Sunuwar, Singh and Pradhan4). Furthermore, a mother–child pair can also have a child with overnutrition and undernutrition in the mother. The triple burden of malnutrition (TBM) refers to the coexistence of micronutrient deficiencies and undernutrition in children and maternal overnutrition(Reference Sunuwar, Singh and Pradhan4). Similarly, TBM among mother–child pairs can include overnutrition in the child and the coexistence of undernutrition and micronutrient deficiencies in the mother. Childhood malnutrition is associated with multiple adverse outcomes such as delayed cognitive development and mortality(Reference Vassilakou5). Likewise, overnutrition in adults is associated with an increased risk of acquiring non-communicable diseases such as high blood pressure and diabetes and poor pregnancy outcomes among women(Reference Coimbra, Tavares and Ferreira6). DBM and TBM are, therefore, increasingly being recognised as public health threats because of the risks they pose to both the mother and child and the underlying complexity resulting in the coexistence of different types of malnutrition.

The burden of DBM and TBM varies between countries. Although the global mother–child pair prevalence of DBM and TBM is unknown, the prevalence of household-level DBM ranges between 3 % and 35 % across 126 low- and middle-income countries, with the highest prevalence reported in Southern Africa, South America and Asia(Reference Seferidi, Hone and Duran7,Reference Popkin, Corvalan and Grummer-Strawn8) . For Southern Africa, a study carried out in 2021 in twenty-three countries of the region estimated the prevalence of household-level DBM and TBM to be 8 % and 5 %, respectively(Reference Christian and Dake9).

Malawi has been experiencing DBM. The most recent 2015–2016 Malawi Demographic and Health Survey (MDHS) reported that the prevalence of stunting (37 %), underweight (12 %), overweight (5 %) and wasting (3 %) among under-five children declined compared with the 2010 MDHS (stunting = 47 %, underweight = 13 %, overweight =8 % and wasting = 4 %)(10,11) . However, the prevalence of overweight among women of reproductive age increased from 17 % in 2010 to 21 % in 2015–2016. Further, the prevalence of anaemia among under-five children decreased from 73 % in 2004 to 63 % in 2010 and remained at that level in 2015–2016, suggesting a stall in the reduction. Although there has been extensive research in Malawi on factors associated with various forms of malnutrition among children and overnutrition among women of reproductive age, research looking at the co-occurrence of undernutrition, micronutrient deficiencies among children and overnutrition among their mothers in Malawi is scarce(Reference Ntenda and Kazambwe12Reference Machira and Chirwa14).

There is limited research on geographical disparities in DBM and TBM among mother–child pairs in low- and middle-income countries. Tarekegn et al. used the Anselin Local Moran’s I test to identify hotspots for DBM and TBM among mother–child pairs in Ethiopia(Reference Tarekegn, Assimamaw and Atalell15). In Kenya, Kasomo et al. used a Bayesian geoadditive regression model to identify factors associated with DBM among women and to identify areas with a high burden of DBM(Reference Kasomo and Gayawan16). Both studies showed that the distribution of DBM and TBM varies across space. In addition to the spatial studies, other previous research used multilevel analyses to determine factors associated with TBM(Reference Andriani, Friska and Arsyi17).

The objectives of the present study were to (1) estimate the prevalence of DBM and TBM among mother–child pairs, (2) examine the variability in the prevalence of DBM and TBM geographically and (3) identify individual, household and community-level factors associated with DBM and TBM among mother–child pairs, in Malawi.

Methods

Data

The most recent 2015–2016 MDHS data were used for this study(10). The Demographic and Health Surveys (DHS) are nationally representative cross-sectional household surveys that are carried out in low- and middle-income countries for tracking health and demographic indicators(18). Respondents for the 2015–2016 MDHS were sampled using a two-stage process, which was guided by a sampling frame generated from the 2008 Malawi Population and Housing Census(10). Details on the MDHS sampling methodology can be found in other reports(10). In summary, the 2015–2016 MDHS sampled 850 enumeration areas (EA) in all the twenty-eight districts of Malawi in the first sampling stage. In the second stage, thirty-three households in each rural cluster and thirty households in each urban cluster were selected. Enumerators used a global positioning system to identify the central point of each EA and to collect coordinates (latitude and longitude). The DHS programme displaces the coordinates by up to 2 km in urban areas and up to 5 km in 99 % of the rural areas and by 10 km in the remaining 1 % of EA in rural areas for anonymization purposes(Reference Burgeer, Colston and Roy19). Coordinate displacements ensure that points remain within the same administrative boundaries of the EA(Reference Burgeer, Colston and Roy19).

Study population

The population of interest in the study was living together in a household mother–child pairs where the child was less than 60 months (5 years) old at the time of the survey. We used the women’s and children’s recode datasets that contain information on women of reproductive age (15–49 years), their most recent birth and their children. Anthropometry and biomarker data were collected from members of a sub-sample of the interviewed households. Height and weight measurements were taken from women aged between 15 and 49 years and children aged between 0 and 59 months in the eligible households(10). To measure hemoglobin (Hb) levels for determining anaemia status, blood specimens were collected from a sub-sample of households eligible for anthropometry data collection.

The analysis sample for DBM included women of reproductive age (15–49 years) with all their under-five children residing in the same household. The analysis of TBM was restricted to a subset of women of reproductive age and their children aged between 6 and 59 months because the DHS collected Hb levels for this child age group only(10). Pregnant women at the time of the survey or who gave birth in the two months before the survey were excluded from the sample to avoid their pregnancy weight biasing their body mass index (BMI)(10,Reference Croft, Aileen and Courtney20) . We also excluded women and children whose anthropometric measurements were not recorded(Reference Croft, Aileen and Courtney20). Furthermore, we excluded children whose dates of birth were missing or unknown and also excluded children with anthropometric measurements outside of plausible ranges as defined by the Guide to DHS Statistics(Reference Croft, Aileen and Courtney20). The sample inclusion flow chart is provided in the online supplementary material, Supplemental Information I.

Outcome and independent variables definition

This study had two outcome variables: double burden of malnutrition (DBM) and triple burden malnutrition (TBM) among mother–child pairs. The operational definitions for DBM and TBM were adapted from previous studies and are presented in Table 1, along with definitions of malnutrition indicators such as stunting and wasting available in the MDHS(Reference Sunuwar, Singh and Pradhan4,Reference Tarekegn, Assimamaw and Atalell15) .

Table 1. Outcome variable definition as adapted from previous DBM and TBM studies(Reference Sunuwar, Singh and Pradhan4,Reference Tarekegn, Assimamaw and Atalell15,Reference Kumar, Chauhan and Patel21)

The study explored associations between the outcomes and independent variables at the individual, household and community levels. An initial selection of variables was based on the WHO conceptual framework for the DBM (see online supplementary material, Supplemental Information II) and previous research looking at the prevalence and burden of DBM and TBM(Reference Sunuwar, Singh and Pradhan4,Reference Tarekegn, Assimamaw and Atalell15,Reference Kumar, Chauhan and Patel21,Reference Kurian and Subi22) . These variables are listed in Table S1 in the online supplementary material, Supplemental Information II, along with the variable label. The community-level variables were aggregated from the individual/household level to the EA (i.e. cluster) level. For instance, we generated a new variable and assigned a one to the households in the middle, rich or richest wealth quintiles and a zero to the households in the poor/poorest quintiles. The variable capturing the percentage of households in at least the middle wealth quintile in a cluster was, therefore, computed by taking the sum of the new variable divided by the total number of households sampled in that cluster.

The following gridded raster covariates were used in the mapping analysis: precipitation, nightlights, elevation, temperature, aridity, antenatal visits during pregnancy, female literacy and percentage of children who had received all basic vaccinations. These variables have been shown to be associated with nutrition-related indicators and have been included in previous modelling and mapping studies(Reference Andriani, Friska and Arsyi17,Reference Ahinkorah, Amadu and Seidu23Reference Khan, Islam and Faisal28) . The variables were also selected based on the WHO conceptual framework (see online supplementary material, Supplemental Information II) of DBM and conceptual frameworks for mother–child pair DBM from previous research(29,Reference Masibo, Humwa and Macharia30) . Details on the sources of the variables and their spatial and temporal resolution are given in the online supplementary material, Supplemental Information II.

Statistical analyses

Outcome and independent variables were explored by tabulating frequencies for the categorical variables and computing the mean and standard deviations for the continuous variables. Bivariate analyses were carried out using a Chi-square test for categorical variables and t-test for continuous variables. To avoid including multiple independent variables that were highly correlated, we fitted a logistic regression model to the outcome variables and computed the variance inflation factors of the independent variables. All the variables had a variance inflation factor of less than 3 and were included in the multilevel models(31,Reference Wariri, Utazi and Okomo32) .

The multivariable analysis first considered a binomial mixed model (three-level multilevel model where mother–child pairs reside within a household that is in a community/EA) to investigate multilevel risk factors of TBM and DBM. Mixed models are valuable for analysing data with hierarchical or nested structures, allowing for the incorporation of both fixed and random effects(Reference Bates, Mächler and Bolker33). The clustering variables that were used in the study were the household number and the EA, which is coded as a cluster number in the DHS data. A three-level multivariable binomial mixed model was fitted to investigate the determinants of DBM and TBM at the individual, household and cluster levels where mother–child pairs (individual, Level 1) are in households (Level 2), which are clustered within an EA (community level, Level 3).

Spatial modelling

To map DBM and TBM, we adapted a methodology utilised in estimating and mapping other health-related outcomes that are derived from multiple indicators(Reference De Silva and Hall34,Reference Mogaji, Johnson and Adigun35) . Each of the indicators contributing to DBM and TBM (child stunting, child wasting, child underweight, child overnutrition, child anaemia, maternal undernutrition, maternal short height, maternal anaemia and maternal overnutrition) was individually considered in our approach(Reference De Silva and Hall34). We examined multicollinearity and selected between closely associated spatial covariates as described above(Reference Wariri, Utazi and Okomo32,Reference Giorgi, Fronterrè and Macharia36) . We then fitted binomial generalised mixed models to each of the nine indicators. Random effects extracted from the binomial mixed models were used to fit variograms, which confirmed the absence of spatial correlation in all nine indicators, as depicted in Fig. S4 and Fig. S5 in the online supplementary material. We, therefore, mapped DBM and TBM using the non-spatial binomial mixed models as done in other non-spatial mapping studies since spatial dependence is required when using geospatial methods(Reference Giorgi, Fronterrè and Macharia36Reference Deribe, Mbituyumuremyi and Cano38). We computed the predicted prevalence of each of the nine indicators and combined them to produce the predicted prevalence of DBM and TBM at the pixel level (3 km × 3 km grid) and district level(Reference De Silva and Hall34). We utilised the multivariate Gaussian approximation of the maximum likelihood estimator to compute confidence intervals (CIs) for the estimates as employed in previous non-spatial mapping studies(Reference Deribe, Mbituyumuremyi and Cano38).

We generated new level-weights according to the recent guidance by the DHS methodology team(Reference Elkasabi, Ren and Pullum39). All analyses adjusted for the weights and were carried out at a 5 % significance level using Stata version 15 and R. We further adjusted for the stratification. Statistical details for the multilevel model and mapping analyses are available in the online supplementary material.

Results

We analysed data for a total of 4618 for DBM and 4209 for TBM (Level 1 data) mother–child pairs. The DBM mother–child sample was from 3661 households (Level 2) within 848 communities (Level 3). The TBM mother–child sample was from 3442 unique households (Level 2) within 844 communities (Level 3). Among the 4618 mother–child pairs in the DBM sample, there were 4473 (97 %) unique women, and in the TBM sample, 4089 (97 %) were unique women, meaning that less than 3 % of children in the analysis pairs shared the same mother. The distribution of the DBM and TBM samples is displayed in Table 2. Briefly, about a fifth of the children in both samples (DBM 21 %, TBM 23 %) were aged between 12 months and 23 months, and a slight majority were female (DBM 52 %, TBM 52 %). Furthermore, three-quarters of the mothers were aged between 20 and 34 years (DBM 74 %, TBM 74 %), and two-thirds had attended up to primary school education at the time of the survey (DBM 66 %, TBM 66 %).

Table 2. Socio-demographic characteristics among mother–child pair analysis samples for DBM (n 4618) and TBM (n 4209) from the Malawi 2015–2016 Demographic and Health Survey

DBM, double burden of malnutrition; TBM, triple burden of malnutrition; N, total sample size across outcome variable; n, column sample size across independent variable; ANC, antenatal care; %, percentage.

The prevalence of DBM among mother–child pairs in Malawi was 5·5 % (95 % CI: 4·7, 6·4), and the prevalence of TBM was 3·1 % (95 % CI: 2·5 %, 4·0 %). Figure 1 shows the prevalence of the components of DBM and TBM. Anaemia prevalence was 63·4 % among children 6–59 months, and among child undernutrition indicators (child stunted, child underweight and child wasted), stunting was highest at 36·8 %.

Figure 1. Prevalence of measures of malnutrition among mother–child pairs included in the analysis in Malawi (double burden of malnutrition (DBM): n 4618, triple burden of malnutrition (TBM): n 4209).

Factors associated with double burden of malnutrition and triple burden of malnutrition

The distribution of characteristics of mother–child pairs according to whether they had DBM or TBM is displayed in Table 3. In bivariate analysis, DBM was associated with older ages of the child and mother, parity and highest level of education of the mother. On the community level, the percentage of households in at least the middle wealth quintile was associated with higher DBM, whilst the proportion of women who attended fewer than four antenatal care (ANC) visits during pregnancy was also associated with lower DBM. In a bivariate analysis of TBM, only parity and maternal education were significantly associated with TBM. Mothers with some education had lower odds of mother–child TBM.

Table 3. Bivariate and multivariable analyses (from the multilevel logistic regression model) of the individual, household and community-level variables associated with mother–child pair DBM (n 4618) and TBM (n 4209), 2015–2016 Malawi Demographic and Health Survey

DBM, double burden of malnutrition; TBM, triple burden of malnutrition; n, column sample size across independent variable (DBM = yes, TBM = yes); P-value = Chi-square P-value for categorical variables and t-test P-value for continuous (cluster-level) variables; AOR, adjusted OR; Ref, reference category, AOR, 1; ANC, antenatal care; %, percentage; bold, statistical significance in the multilevel modelling (P-value < 0·05).

In the adjusted multilevel logistic regression, an increase in the child’s age was associated with an increase in adjusted odds of DBM. Being a child between age 24 and 35 months is associated with four times higher odds of DBM (adjusted Odds Ratio, AOR: 4·3; 95 % CI: 1·6, 11·8) compared with children aged below 12 months. Similarly, being a child between age 48 and 59 months is associated with four times higher odds of DBM (AOR: 4·1, 95 % CI: 1·4, 11·7) compared with children aged below 12 months. Conversely, the odds of having a mother–child pair with DBM was 60 % lower (AOR: 0·4; 95 % CI: 0·2, 0·8) among mother–child pairs with women aged between 20 and 34 years compared with women aged 35 years and above. In the multilevel logistic regression for TBM among mother–child pairs, the mother’s level of educational attainment was the only factor statistically significantly associated with the outcome (P-value < 0·01). The odds of TBM were 60 % lower (AOR: 0·4; 95 % CI: 0·2, 0·8) among mother–child pairs where the women had primary education and 80 % lower (AOR: 0·2; 95 % CI: 0·1, 0·6) among mother–child pairs where the women had secondary or tertiary education, compared with mother–child pairs where the mother had no education.

Sex of the child, parity, mother’s marital status, mother’s employment status, number of attended ANC visits, household wealth quintile and community level of ANC attendance were not associated with DBM or TBM (P> 0·05). DBM among mother–child pairs was significantly associated with household wealth in the community. A one-unit increase in the percentage of households in the middle wealth quintile or wealthier in the community was associated with a threefold increase in the adjusted odds of DBM (AOR: 2·8 95 % CI: 1·1, 7·3). No statistically significant community-level effects were seen in the TBM model.

In the spatial analysis, an increase in the nightlight value in an area, which is a proxy for the wealth index, was associated with decreased odds of child-level outcomes such as child stunting, anaemia and underweight. On the other hand, an increase in nightlight value in an area was associated with increased odds of maternal overnutrition (overweight and obese). Likewise, an increase in the proportion of literate women was associated with reduced odds of childhood stunting, wasting, underweight and anaemia. With respect to maternal-level outcomes, an increase in the proportion of literate women was associated with increased odds of maternal overnutrition. Climatic variables such as precipitation, temperature and aridity index were not significantly associated with any of the child-level and maternal-level outcomes.

Geographic distribution of double burden of malnutrition and triple burden of malnutrition

The predicted prevalence of DBM was heterogeneous, ranging from 1·2 % to 8·2 % across the pixels in Malawi (Fig. 2). The predicted prevalence of TBM ranged from 0·9 % to 3·5 % at the pixel level. The highest prevalence of mother–child pairs with DBM and TBM was estimated in cities (Fig. 3). All the four cities in Malawi (Blantyre, Lilongwe, Mzuzu and Zomba) had a DBM prevalence of greater than 5·1 %, and a TBM predicted prevalence greater than 2·4 %. Maps illustrating the uncertainties in the predicted prevalence of DBM and TBM are included in the online supplementary material, Supplemental Information II.

Figure 2. Predicted prevalence of the double burden of malnutrition (DBM) and triple burden of malnutrition (TBM) among mother–child pairs in Malawi.

Figure 3. District-level predicted prevalence of the double burden of malnutrition (DBM) and triple burden of malnutrition (TBM) among mother–child pairs in Malawi.

Discussion and Conclusion

Discussion

This study examined the prevalence of and spatial variation in DBM and TBM, as well as individual, household and community-level correlates of DBM and TBM among mother–child pairs in Malawi. Our study contributes to the current literature by investigating the multilevel factors associated with DBM and TBM and generating high spatial resolution maps of DBM and TBM in Malawi. The prevalence of DBM and TBM among mother–child pairs were 5·5 % and 3·1 %, respectively. These prevalence values are higher than the reported DBM and TBM prevalence in Ethiopia (3·1 % and 1·6 %, respectively) and similar to the prevalence reported in Nepal (6·6 % and 7·0 %) and India (TBM, 5 %), respectively(Reference Sunuwar, Singh and Pradhan4,Reference Tarekegn, Assimamaw and Atalell15,Reference Kumar, Chauhan and Patel21) . Although the overlap between child undernutrition and maternal overweight was small, two-fifths of children were malnourished (39 %), and 19 % of women were overweight or obese. Whilst there was an absence of spatial correlation in our data, our findings suggest that despite the relatively low prevalence, the geographic distribution of mother–child DBM and TBM varied greatly within Malawi, thereby suggesting that environmental conditions conducive for DBM and TBM exist throughout Malawi, particularly in cities. We found a strong positive association between the increased age of a child and the odds of having mother–child pair DBM and TBM. This result is consistent with results from other studies where children older than 24 months had higher odds of being in a mother–child pair with DBM and TBM compared with younger ones(Reference Sunuwar, Singh and Pradhan4,Reference Christian and Dake9,Reference Tarekegn, Assimamaw and Atalell15,Reference Kumar, Chauhan and Patel21,Reference Ahinkorah, Amadu and Seidu23,Reference Das, Fahim and Islam40) . Regarding age, our results also revealed that mother–child pairs where the mother was aged between 20 and 34 had reduced odds of DBM compared with women aged at least 35 years. This finding is consistent with previous studies where the odds of household-level DBM and TBM were higher among older women than younger women due to the older women being overweight and obese(Reference Christian and Dake9,Reference Chilot, Belay and Merid41) . Some studies have postulated that sedentary lifestyles and the increased likelihood of overnutrition (overweight and obesity) among older women might be some of the leading drivers of DBM and TBM(Reference Sunuwar, Singh and Pradhan4,Reference Christian and Dake9,Reference Tarekegn, Assimamaw and Atalell15) .

Educational attainment was another predictor of TBM in our study. Mothers with at least primary and secondary education had reduced odds of TBM. This result further corresponds with the spatial modelling where an increase in the proportion of literate women was associated with a reduced burden of childhood malnutrition (underweight, stunting and wasting). These results align with previous evidence that women with higher education have reduced odds of household-level DBM and TBM as higher education attainment is associated with reduced risk of childhood malnutrition(Reference Christian and Dake9,Reference Ahinkorah, Amadu and Seidu23) .

Our results show that an increase in the proportion of wealthy households at the community level increased the odds of DBM, thereby suggesting a higher burden in relatively wealthier areas(Reference Popkin, Corvalan and Grummer-Strawn8,Reference Kumar, Chauhan and Patel21) . The spatial analysis further decomposes this finding by showing that an increase in the nightlights value (a proxy for the level of wealth) is associated with reduced odds of child-level outcomes and increased odds of maternal overnutrition. No significant community-level effects were seen in the TBM model. Previous work has postulated that higher income levels may enhance the ability to buy food, thereby shaping dietary habits and preferences(Reference Chilot, Belay and Merid41). Furthermore, previous studies have hypothesised that inadequate physical activity, sedentary lifestyles and westernised dietary behaviours adopted by wealthier households and communities drive adult overnutrition(Reference Popkin, Corvalan and Grummer-Strawn8,Reference Kumar, Chauhan and Patel21,Reference Chilot, Belay and Merid41,Reference Mahumud, Sahle and Owusu-Addo42) . Previous reports have shown that these behaviours tend to be more common in urban settings, which could be reflected in our results(Reference Crush, Frayne and McLachlan43). Further research is needed in Malawi to identify more community-level determinants of DBM and TBM.

The absence of spatial correlation in our data suggests that the spatial correlation in our DBM and TBM samples can be explained by the environmental covariates that were used in the models. This notwithstanding, the maps generated in this study suggest that DBM and TBM are not homogeneously distributed in Malawi. These results underscore the potential existence of spatial drivers influencing the observed patterns of DBM and TBM, which should be further investigated. In studies where the data exhibit spatial correlation, the mixed effects models used in this study could be further extended to explicitly incorporate the spatial correlation using geostatistical methods(Reference Giorgi, Fronterrè and Macharia36,Reference Diggle and Giorgi37) .

Limitations and strengths of the study

The main limitation of this study is that the DHS was a cross-sectional study; therefore, no causal associations could be inferred from the study. Second, our analysis included categorising continuous variables such as wasting and stunting that have been shown to lead to some loss of information on the variable(Reference Kyomuhangi and Giorgi44). An additional constraint pertains to the use of outdated DHS data. Nevertheless, this research remains valuable as it demonstrates the application of mixed effects models for estimating factors at multiple levels and mapping health outcomes like DBM and TBM. Another limitation of DHS data is the unavailability of most of the variables in the WHO DBM conceptual framework. Furthermore, the DHS data lack information on dietary behaviours that are immediately related to nutritional outcomes. This shortcoming creates the need to have more nutrition-related datasets in Malawi to better understand the determinants of nutrition-related issues such as DBM and TBM in the country. Another important limitation is the lack of micronutrient data. Thus, we used anaemia status as a proxy for estimating TBM prevalence. We, therefore, recommend collecting nationally representative data on micronutrient deficiencies and not necessarily through DHS.

Given that common household-level risk factors for child undernutrition are often protective for maternal overweight and vice versa, most of the risk factors identified for DBM and TBM in this study were driven either by strong positive associations with child undernutrition (and weaker or null negative associations with maternal overnutrition) or by strong positive associations with maternal overnutrition (and weaker or null negative associations with child undernutrition). These limitations might explain why the prevalence of DBM and TBM appears possibly lower than what would be observed by chance. These methodological challenges are, nevertheless, common when studying DBM and TBM.

Despite these limitations, to our knowledge, this study is the first to establish the prevalence of mother–child pair DBM and TBM in Malawi. Our study also presents an important finding that the coexistence of undernutrition in children and overnutrition in mothers is associated with both individual and community-level factors. Furthermore, our study is the first to illustrate the geographical disparities in the distribution of mother–child DBM and TBM in Malawi. It shows that the burden of DBM and TBM is higher in cities than in other areas. Future studies could further extend this work by building a spatiotemporal model to assess whether the burden of DBM and TBM has also been higher in cities than in other areas over time. This is because a recent study has shown a shifting spatiotemporal trend in DBM from metropolitan areas to other regions in Guatemala(Reference Sagastume, Peñalvo and Ramírez-Zea45). The findings from the spatiotemporal modelling could help the government and implementing partners anticipate which districts/areas in Malawi might get an increasing burden of DBM and TBM in the future. The study results could also inform interventions about specific areas where attention is needed to target DBM and TBM and control the burden of these conditions. The findings from this study could also be used to inform a new hypothesis of shifting spatial DBM and TBM trends in other countries.

Conclusion

This study highlighted that the mother–child pair prevalence of DBM and TBM in Malawi was relatively low, with the highest burden in cities. It further showed that the increased age of the child and mother also increased the odds of a mother–child pair having DBM. Additionally, individual-level maternal educational attainment was shown to have a protective effect against TBM. Our study also highlighted that there are community-level determinants of DBM such as household wealth that are associated with increased odds of DBM. These results emphasise the need to not neglect wealthier communities in disseminating and implementing adult-related nutrition-related interventions in Malawi. These results could be explicitly adopted and translated into national documents such as the ‘Malawi National Multi-Sector Nutrition Policy 2018–2022’, which acknowledges the existence of both undernutrition and overnutrition but does not provide any recommendations on addressing their coexistence such as DBM and TBM(46).

Through the revised Malawi National Multi-Sector Nutrition Policy, the government of Malawi can tackle DBM and TBM among mother–child pairs in Malawi by implementing comprehensive, integrated nutrition programmes that simultaneously address undernutrition in children and overnutrition in mothers. The government can also promote a multisectoral response to dealing with mother–child DBM and TBM in Malawi. A multisectoral approach can ensure that policies across multiple sectors, such as health and agriculture, work together to ensure access to supplements and nutritious foods, address food insecurity and promote healthy habits among women. These interventions could, as a priority, begin to focus on the Malawian cities where DBM and TBM among mother–child pairs are the highest.

Supplementary material

For supplementary material accompanying this paper, visit https://doi.org/10.1017/S1368980024002064.

Acknowledgements

The authors would like to thank the DHS Program for granting them access to the data. We would also like to gratefully acknowledge the women who participated in the DHS.

Financial support

J. J. K. received a partial scholarship from the Belgian Directorate-General for Development Cooperation and Humanitarian Aid to participate in the training on writing a manuscript using the DHS data at the Institute of Tropical Medicine in Antwerp. The sponsors had no role in analysing data or interpreting the study’s results.

Competing interests

The authors declare that they have no financial and non-financial conflicts of interest.

Authorship

J. J. K. conceptualised the study and extracted the DHS data and spatial covariates. A. S. and L. B. supervised the development of the project proposal. J. J. K. carried out the multilevel and geographic analyses. P. M. M. and E. G. supervised the geographic analysis, whilst A. S. supervised the multilevel modelling and proposal implementation. J. J. K. wrote the first draft of the manuscript. All authors reviewed the draft manuscript and approved the final version of the manuscript.

Ethics of human subject participation

This study used secondary 2015–2016 MDHS data. The 2015–2016 MDHS was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the National Health Sciences Research Committee in Malawi and ICF Institutional Review Board. Verbal consent was sought from all the DHS study participants. Parents and guardians of minors also consented on behalf of their wards. The verbal consent was witnessed and formally recorded. Permission to use the DHS data was sought from and approved by the DHS programme. To protect the anonymity of the study participants, the DHS displaced geospatial data of participants’ locations, as discussed in the methods section. No ethical approval was required to conduct this secondary data analysis by the institution with which the first author is affiliated.

References

World Health Organisation (2017) Double-Duty Actions for Nutrition: Policy Brief Technical Report. Published Online 2017. https://www.who.int/publications/i/item/WHO-NMH-NHD-17.2 (accessed April 2023).Google Scholar
Micha, R, Mannar, V, Afshin, A et al. (2020) 2018 Global Nutrition Report: Action on Equity to End Malnutrition. Published Online 2020. https://globalnutritionreport.org/reports/global-nutrition-report-2018/ (accessed April 2023).Google Scholar
The World Health Organisation (WHO) (2016) The Double Burden of Malnutrition: Policy Brief. Published Online 2016. chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://apps.who.int/iris/bitstream/handle/10665/255413/WHO-NMH-NHD-17.3-eng.pdf?sequence=1 (accessed May 2023).Google Scholar
Sunuwar, DR, Singh, DR & Pradhan, PMS (2020) Prevalence and factors associated with double and triple burden of malnutrition among mothers and children in Nepal: evidence from 2016 Nepal demographic and health survey. BMC Public Health 20, 405. doi: 10.1186/s12889-020-8356-y.Google Scholar
Vassilakou, T (2021) Childhood malnutrition: time for action. Children 8, 103. doi: 10.3390/children8020103.Google Scholar
Coimbra, CE, Tavares, FG, Ferreira, AA et al. (2021) Socioeconomic determinants of excess weight and obesity among Indigenous women: findings from the First National Survey of Indigenous People’s Health and Nutrition in Brazil. Public Health Nutr 24, 19411951. doi: 10.1017/S1368980020000610.Google Scholar
Seferidi, P, Hone, T, Duran, AC et al. (2022) Global inequalities in the double burden of malnutrition and associations with globalisation: a multilevel analysis of Demographic and Health Surveys from 55 low-income and middle-income countries, 1992–2018. Lancet Glob Health 10, e482e490. doi: 10.1016/S2214-109X(21)00594-5.Google Scholar
Popkin, BM, Corvalan, C & Grummer-Strawn, LM (2020) Dynamics of the double burden of malnutrition and the changing nutrition reality. Lancet 395, 6574. doi: 10.1016/S0140-6736(19)32497-3.Google Scholar
Christian, AK & Dake, FA (2022) Profiling household double and triple burden of malnutrition in sub-Saharan Africa: prevalence and influencing household factors. Public Health Nutr 25, 15631576. doi: 10.1017/S1368980021001750.Google Scholar
Malawi National Statistical Office (NSO) (2016) The 2015–2016 Malawi Demographic adn Health (MDHS) Survey. Published Online 2017. chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://dhsprogram.com/pubs/pdf/FR319/FR319.pdf (accessed January 2023).Google Scholar
Malawi National Statistical Office (NSO) The 2010 Malawi Demographic and Health (MDHS) Survey. Published Online 2011. dhsprogram.com/pubs/pdf/fr247/fr247.pdf (accessed April 2024).Google Scholar
Ntenda, PAM & Kazambwe, JF (2019) A multilevel analysis of overweight and obesity among non-pregnant women of reproductive age in Malawi: evidence from the 2015–2016 Malawi Demographic and Health Survey. Int Health 11, 496506. doi: 10.1093/inthealth/ihy093.Google Scholar
Ngwira, A (2022) Shared geographic spatial risk of childhood undernutrition in Malawi: an application of joint spatial component model. Public Health Pract 3, 100224. doi: 10.1016/j.puhip.2021.100224.Google Scholar
Machira, K & Chirwa, T (2020) Dietary consumption and its effect on nutrition outcome among under-five children in rural Malawi. PLOS ONE 15, e0237139. doi: 10.1371/journal.pone.0237139.Google Scholar
Tarekegn, BT, Assimamaw, NT, Atalell, KA et al. (2022) Prevalence and associated factors of double and triple burden of malnutrition among child-mother pairs in Ethiopia: spatial and survey regression analysis. BMC Nutr 8, 34. doi: 10.1186/s40795-022-00528-5.Google Scholar
Kasomo, JM & Gayawan, E (2021) Spatial location, temperature and rainfall diversity affect the double burden of malnutrition among women in Kenya. SSM - Popul Health 16, 100939. doi: 10.1016/j.ssmph.2021.100939.Google Scholar
Andriani, H, Friska, E, Arsyi, M et al. (2023) A multilevel analysis of the triple burden of malnutrition in Indonesia: trends and determinants from repeated cross-sectional surveys. BMC Public Health 23, 1836. doi: 10.1186/s12889-023-16728-y.Google Scholar
The DHS Program The Demographic and Health Surveys (DHS) Program. https://www.dhsprogram.com/ (accessed January 2023).Google Scholar
Burgeer, C, Colston, J, Roy, T et al. (2013) Geographic Displacement Procedure and Georeferenced Data Release Policy for the Demographic and Health Surveys. Published Online 2013. https://dhsprogram.com/publications/publication-SAR7-Spatial-Analysis-Reports.cfm (accessed May 2024).Google Scholar
Croft, T, Aileen, MJ, Courtney, KA et al. (2020) Guide to DHS Statistics: DHS-7 (Version 2). Published Online 2020. https://dhsprogram.com/data/Guide-to-DHS-Statistics/Guide_to_DHS_Statistics_DHS-7.htm (accessed January 2023).Google Scholar
Kumar, P, Chauhan, S, Patel, R et al. (2021) Prevalence and factors associated with triple burden of malnutrition among mother-child pairs in India: a study based on National Family Health Survey 2015–2016. BMC Public Health 21, 391. doi: 10.1186/s12889-021-10411-w.Google Scholar
Kurian, O & Subi, S (2019) Weighed down by the Gains: India’s Twin Double Burdens of Malnutrition and Disease. Vol 193. Observer Research Foundation. https://www.researchgate.net/publication/338139478_Weighed_Down_by_the_Gains_India%27s_Twin_Double_Burdens_of_Malnutrition_and_Disease (accessed August 2023).Google Scholar
Ahinkorah, BO, Amadu, I, Seidu, AA et al. (2021) Prevalence and factors associated with the triple burden of malnutrition among mother-child pairs in Sub-Saharan Africa. Nutrients 13, 2050. doi: 10.3390/nu13062050.Google Scholar
Kanu, F, Jeffers, ME, Williams, AM et al. (2023) Association between hemoglobin and elevation among school-aged children: a verification of proposed adjustments. Am J Clin Nutr 118, 114120.Google Scholar
Tusting, LS, Bradley, J, Bhatt, S et al. (2020) Environmental temperature and growth faltering in African children: a cross-sectional study. Lancet Planet Health 4, e116e123. doi: 10.1016/S2542-5196(20)30037-1.Google Scholar
Cooper, MW, Brown, ME, Hochrainer-Stigler, S et al. (2019) Mapping the effects of drought on child stunting. Proc Natl Acad Sci 116, 1721917224. doi: 10.1073/pnas.1905228116.Google Scholar
Amegbor, PM, Zhang, Z, Dalgaard, R et al. (2020) Multilevel and spatial analyses of childhood malnutrition in Uganda: examining individual and contextual factors. Sci Rep 10, 20019. doi: 10.1038/s41598-020-76856-y.Google Scholar
Khan, JR, Islam, MM, Faisal, ASM et al. (2023) Quantification of urbanization using night-time light intensity in relation to women’s overnutrition in Bangladesh. J Urban Health 100, 562571. doi: 10.1007/s11524-023-00728-9.Google Scholar
World Health Organisation (2016) Strategic Action Plan to Reduce the Double Burden of in the South-East Asia Region 2016–2025. Published Online 2016. https://apps.who.int/iris/bitstream/handle/10665/253377/Strategic%20Action%20Plan%20to%20reduce%20the%20double%20burden%20of%20malnutrition%20in%20SEAR%202016-2025.pdf?sequence=1&isAllowed=y (accessed February 2023).Google Scholar
Masibo, PK, Humwa, F & Macharia, TN (2020) The double burden of overnutrition and undernutrition in mother−child dyads in Kenya: demographic and health survey data, 2014. J Nutr Sci 9, e5. doi: 10.1017/jns.2019.39.Google Scholar
Advanced Research Computing: Statistical Methods and Data Analytics Regression with STATA Chapter 2: Regression Diagnostics. Checking for Multicollinearity. https://stats.oarc.ucla.edu/stata/webbooks/reg/chapter2/stata-webbooksregressionwith-statachapter-2-regression-diagnostics/ (accessed May 2023).Google Scholar
Wariri, O, Utazi, CE, Okomo, U et al. (2023) Mapping the timeliness of routine childhood vaccination in The Gambia: a spatial modelling study. Vaccine 41, 56965705. doi: 10.1016/j.vaccine.2023.08.004.Google Scholar
Bates, D, Mächler, M, Bolker, B et al. (2015) Fitting linear mixed-effects models using lme4. J Stat Softw 67, 148. doi: 10.18637/jss.v067.i01.Google Scholar
De Silva, N & Hall, A (2010) Using the prevalence of individual species of intestinal nematode worms to estimate the combined prevalence of any species. PLoS Negl Trop Dis 4, e655. doi: 10.1371/journal.pntd.0000655.Google Scholar
Mogaji, HO, Johnson, OO, Adigun, AB et al. (2022) Estimating the population at risk with soil transmitted helminthiasis and annual drug requirements for preventive chemotherapy in Ogun State, Nigeria. Sci Rep 12, 2027. doi: 10.1038/s41598-022-06012-1.Google Scholar
Giorgi, E, Fronterrè, C, Macharia, PM et al. (2021) Model building and assessment of the impact of covariates for disease prevalence mapping in low-resource settings: to explain and to predict. J R Soc Interface 18, 20210104. doi: 10.1098/rsif.2021.0104.Google Scholar
Diggle, PJ & Giorgi, E (2019) Model-Based Geostatistics for Global Public Health: Methods and Applications. Boca Raton, FL: CRC Press.Google Scholar
Deribe, K, Mbituyumuremyi, A, Cano, J et al. (2019) Geographical distribution and prevalence of podoconiosis in Rwanda: a cross-sectional country-wide survey. Lancet Glob Health 7, e671e680. doi: 10.1016/S2214-109X(19)30072-5.Google Scholar
Elkasabi, M, Ren, R & Pullum, T (2020) Multilevel Modelling using DHS Surveys: A Framework to Approximate Level Weights. Published Online 2020. chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://www.dhsprogram.com/pubs/pdf/MR27/MR27.pdf (accessed May 2023).Google Scholar
Das, S, Fahim, SM, Islam, MS et al. (2019) Prevalence and sociodemographic determinants of household-level double burden of malnutrition in Bangladesh. Public Health Nutr 22, 14251432. doi: 10.1017/S1368980018003580.Google Scholar
Chilot, D, Belay, DG, Merid, MW et al. (2023) Triple burden of malnutrition among mother–child pairs in low-income and middle-income countries: a cross-sectional study. BMJ Open 13, e070978. doi: 10.1136/bmjopen-2022-070978.Google Scholar
Mahumud, RA, Sahle, BW, Owusu-Addo, E et al. (2021) Association of dietary intake, physical activity, and sedentary behaviours with overweight and obesity among 282 213 adolescents in 89 low and middle income to high-income countries. Int J Obes 45, 24042418. doi: 10.1038/s41366-021-00908-0.Google Scholar
Crush, J, Frayne, B & McLachlan, M (2011) Rapid Urbanization and the Nutrition Transition in Southern Africa. Published Online 2011. https://fsnnetwork.org/resource/rapid-urbanization-and-nutrition-transition-southern-africa (accessed December 2023).Google Scholar
Kyomuhangi, I & Giorgi, E (2022) A threshold-free approach with age-dependency for estimating malaria seroprevalence. Malar J 21, 1. doi: 10.1186/s12936-021-04022-4.Google Scholar
Sagastume, D, Peñalvo, JL, Ramírez-Zea, M et al. (2024) Dynamics of the double burden of malnutrition in Guatemala: a secondary data analysis of the demographic and health surveys from 1998–2015. Public Health 229, 135143. doi: 10.1016/j.puhe.2024.01.035.Google Scholar
Health, Ministry of the Malawi (2018) National Multi-Sector Nutrition Policy 2018–2022. Published Online April 2018. https://leap.unep.org/countries/mw/national-legislation/national-multi-sector-nutrition-policy-2018-2022 (accessed 30 July 2023).Google Scholar
Figure 0

Table 1. Outcome variable definition as adapted from previous DBM and TBM studies(4,15,21)

Figure 1

Table 2. Socio-demographic characteristics among mother–child pair analysis samples for DBM (n 4618) and TBM (n 4209) from the Malawi 2015–2016 Demographic and Health Survey

Figure 2

Figure 1. Prevalence of measures of malnutrition among mother–child pairs included in the analysis in Malawi (double burden of malnutrition (DBM): n 4618, triple burden of malnutrition (TBM): n 4209).

Figure 3

Table 3. Bivariate and multivariable analyses (from the multilevel logistic regression model) of the individual, household and community-level variables associated with mother–child pair DBM (n 4618) and TBM (n 4209), 2015–2016 Malawi Demographic and Health Survey

Figure 4

Figure 2. Predicted prevalence of the double burden of malnutrition (DBM) and triple burden of malnutrition (TBM) among mother–child pairs in Malawi.

Figure 5

Figure 3. District-level predicted prevalence of the double burden of malnutrition (DBM) and triple burden of malnutrition (TBM) among mother–child pairs in Malawi.

Supplementary material: File

Khaki et al. supplementary material 1

Khaki et al. supplementary material
Download Khaki et al. supplementary material 1(File)
File 1.1 MB
Supplementary material: File

Khaki et al. supplementary material 2

Khaki et al. supplementary material
Download Khaki et al. supplementary material 2(File)
File 41.8 KB