Skip to main content

Analyzing the impact of an MDG-Fund program on childhood malnutrition in Timor-Leste

Abstract

Background

Timor-Leste’s food insecurity, propelled by political conflicts, a fragile economy and biophysical limitations that characterize mountainous Small Island Developing States (SIDS), is expressed in a high incidence (50%) of stunted children. Hence, the Millennium Development Goals Achievement Fund’s Joint Program’s (MDG-F JP) in 2009 was a timely intervention to reduce prevalence of underweight among under-fives. Since the impact of the program remains largely unclear, the current study investigates the contributions of the MDG-F JP on improving children’s nutritional status in Timor-Leste, in order to inform policymakers on how to make future programs more effective.

Methods

Using bivariate analyses and multiple linear regression models we analyzed Demographic and Health Survey (DHS) data from under-fives in 2009–2010 and 2016, combined with spatially explicit data from geographic information systems (GIS). The analyses generated trends and factors associated with undernutrition, which were used in a quasi-experimental setting to compare districts that received the MDG-F JP with similar districts that did not receive MDG-F JP interventions.

Results

A comprehensive set of factors dependent on seasons, locations, and individuals determine undernutrition in Timor-Leste. A positive impact of the MDG-F JP was found for the average severity of wasting, but not for stunting and underweight.

Conclusions

The findings reinforce the pressing need for integrated and cross-sectoral programs, aimed especially at agricultural workers, mothers, and children. The agricultural challenge is to sustainably select, produce and conserve higher-yield and nutrient-rich crops, and educational enhancement should be aligned with local practices and research.

Introduction

In 2019, 21% of under-fives worldwide, 144 million children, experienced stunted growth, and an additional 47 million were underweight [69]. These figures are appalling, especially since it is well known that childhood undernutrition causes health problems throughout a person’s life, such as limited physical and cognitive development, cardiovascular diseases, and obesity, all of which have a direct effect on educational, labor and social capacities [4, 39]. Despite the focus of the Millennium Development Goals (MDGs) on undernutrition, the target to halve the global prevalence of underweight under-fives between 1990 and 2015 (MDG 1.8) was not achieved [73]. Low- and middle-income countries in particular were far from achieving this goal, mainly due to a lack of investment in effective child nutrition [2, 22, 63].

Among the lower middle-income countries, Timor-Leste (East Timor) is no exception; it failed to meet its target of reducing the prevalence of underweight among under-fives to 31%. Hence, Timor-Leste sadly remains the lower middle-income country with the second-highest child undernutrition rates [677]. The stunting prevalence of 50% among under-fives is surpassed only by Burundi and Eritrea [21, 70]. Demographic and Health Surveys (DHS) and Living Standard Surveys over the last 20 years show that the percentage of stunted and underweight children in Timor-Leste has fallen slightly, though this trend is not statistically significant [31, 35, 54].

The alarming child undernutrition rates in Timor-Leste are to be understood and addressed within the Small Island Developing State’s (SIDS) unique socio-economic, political, institutional, cultural and biophysical context. The agricultural potential of Timor-Leste is restricted by its steep and stony slopes, poor soil suitability and trends of increased rainfall and temperatures due to climate change [5, 50, 54]. The small proportion of workable land and remoteness from major markets translate into high import costs for a limited range of products that are often high in calories, fat and sugar [25, 65]. Timor-Leste’s vulnerability is further characterized by devastating effects of its history of conflicts on educational, political, economic and social capital and infrastructure [27, 46, 54].

In this context, the Millennium Development Goals Achievement Fund’s Joint Program (MDG-F JP), entitled ‘Promoting Sustainable Food and Nutrition Security in Timor-Leste’, was set up. Nearly 90% of the MDG-F JP’s total budget of USD 3.5 million was allocated to improving health and nutrition status of the under-fives, as well as for pregnant and nursing women [44]. Yet, the MDG-F JP’s success was reported to be limited: it proved a challenge to implement sustainable changes that could be continued within the systems and partnerships without further support from MDG-F [44, 52]. In addition, the MDG-F JP’s impact on the nutritional status of Timor-Leste’s population remained largely unexplained because of the absence of baseline surveys, lack of appropriate sample designs and inadequate monitoring.

Overall, evaluations of food security programs were conducted with a relatively narrow scope, focused only on output results and impact, and so failing to gain a good understanding of the broad underlying and multifaceted set of determinants that determine the project’s failure or success [22, 45]. For this reason, the research aimed to investigate the contributions of the MDG-F Joint Program to children’s nutritional status in Timor-Leste. The study both compares anthropometric data over time, and also explains the prevalence of malnutrition in Timor-Leste in its dependence of demographic and household characteristics, as well as the prevailing natural endowments. The findings of this causality analysis are used for a conditional comparison between areas with and without the MDG-JF programs.

The major scientific challenge of this study was to attribute and isolate the impact of the program to changes in in the under-fives’ nutritional status among the many confounding factors that influence this process. We therefore formulated the following research question: How did the MDG-F JP improve the nutritional status of under-fives in Timor-Leste? And, what lessons can be learned about context-specific determinants of food security? The remainder of this paper is structured as follows. “Methodology” section presents the methodology. The results are presented in “Results” section. “Conclusions” section discusses the findings and concludes.

Methodology

Dependent and independent data

Our dependent variables are three forms of malnutrition among under-fives: stunting, underweight and wasting, selected from the Demographic and Health Survey (DHSFootnote 1) conducted in 2016 [32] and 2009–2010 [55]. The selection of independent variables as potentially associated determinants was based on their correlation in the existing literature. The independent data that are used to explain the incidence of malnutrition and varying MDG-F JP impacts were based on the same DHS household surveys and on a set of geographical data. DHS surveys provided information on household characteristics, socio-economic conditions and personalized data of women and men of reproductive age and children under the age of five. Furthermore, DHS data include a geo-reference that marks latitude and longitude of households to geographical data sets of administrative units, agronomic characteristics and the prevailing natural resource base. The Mean Vegetation Health Index (MVHI) was used to assess the severity of drought during growing seasons, for each district separately [26]. Additional biophysical information on altitude, crop cover, population density, slope and soils of the household’s place of residence were retrieved from Buchhorn et al. [14]. By combining DHS data with the biophysical environment, we create a comprehensive analytical profile for our food security analysis.

The geographical link

We established the geographical link between DHS data points and attribute maps in three steps. First attribute maps were rasterized to a 30 arcsecond resolution (app. 1 × 1 km) and resampled to the same geo-reference. A straightforward crossing between attribute and rasterized DHS point data retains only one single DHS observation per grid, leaving many DHS observations unattended. Hence, our second step applies a dedicated SAS program that exports gridded attribute maps into three vectors: latitude, longitude and attribute. In a third step we established the geographic link by seeking for each DHS point the nearest neighbor grid cell of the attribute maps.Footnote 2 Removing observations with missing data creates a harmonized full matrix structure with attribute information from the DHS and geographic data. The format allows to conduct uni-, bi- and multi-variate analysis to explain children’s undernutrition in its geographical dependence on household characteristics, administrative entities, and biophysical variables. Furthermore, grouping data by administrative entities allows to conduct a quasi-experiment that assesses the MDG-F JP’s impact among different districts.

Statistical analysis

The stunting, underweight and wasting among under-fives were represented in the DHS dataset by Z-scores lower than − 2.0, indicating the number of standard deviations (SD) below the mean of the WHO international reference standards [36, 66, 74]. Data processing and analysis were conducted with SAS 9.4 software.

First, univariate analyses were conducted to explore the distribution of undernutrition and other covariates among under-fives. The bivariate relationships between stunting, underweight and wasting and various covariates (as listed in Additional file 1) were tested with independent samples t tests for continuous variables and with MANOVA for categorical variables, using data from 2009 to 2010. For continuous determinants, scatterplots were generated beforehand, in order to assess distribution of these variables and opt for application of either the Pearson's correlation coefficient or the Spearman’s correlation coefficient [37]. Correlation was reported for variables having a correlation coefficient between (−) 0.10 and 1.0, in combination with statistical significance of p < 0.05 [17, 61]. To facilitate interpretation of the ranges of these continuous variables a categorical version of these variables was computed and included in the later performed multivariate regression analyses [41, 51]. Only if the bivariate analyses showed no statistically significant correlation with the (novel) categorical variable was the original continuous variable included. For bivariate regression analysis of dichotomous variables, the Folded F-test (α = 0.05) was applied to determine if the variances were indeed equal. If not (p < 0.05), the Satterthwaite output of the independent samples t test was examined instead of the Pooled output. For categorical determinants, the General Linear Model (GLM) procedure was used to perform multivariate analysis of variance (MANOVA), as the GLM procedure can also analyze unbalanced data. For each categorical variable, the p-value of the F-test was reported to assess whether the expected means of the variable’s subcategories differed significantly from each other. The R-square was reported to indicate how much variance was explained by each variable. By including an interaction term to each individual MANOVA of potential effect modifiers, relationships between independent variables were also examined.

Multiple linear regression models were generated using the General Linear Model Selection (GLMSELECT) procedure to find out which set of variables uniquely explained variance in stunting, underweight and wasting among the under-fives. In order to meet the multiple regression analysis linearity assumption, only biologically plausible and strong potential predictors (p < 0.05) based on bivariate analyses were included. A stepwise traditional backwards selection procedure was applied until only predictors with p < 0.05 were included, with a minimum number of 10–15 participants, while allowing split classification variables to leave the model independently. The resulting functional equation was tested for stability among different circumstances through sensitivity analyses [13, 15]. Model selection for HAZ-, WHZ-, and WAZ-scores was repeated on a total of 100 resampled subsets of the input data, each containing a randomly drawn 75% of the total sample, drawn without replacement [53]. Subsequently, average models were generated by averaging the parameter estimates of the selected models. Effects were dropped if they were not selected in at least 20% of the samples.

Impact analysis

Based on our findings from the previous analytical exercise, we illustrate if and how changes have appeared over the years of the MDG-F JP interventions overall, stratified by the main determinants that were identified in the bi-variate and multivariate analyses. We analyzed trends, absolute scores and relative scores reported for 2009–2010 and 2016, as resulting from independent samples t tests and descriptive statistics. Subsequently, in order to analyze the MDG-F JP’s impact despite the unequal baseline scores, relative changes in HAZ, WAZ and WHZ means between 2009–2010 and 2016 among populations living in MDG-F JP districts were compared to those of populations not living in MDG-F JP districts. Within these independent populations, stratification was applied to create uniform groups that could be considered similar at the baseline, on the assumption of both resemblance of confounding structures and quasi-random exposure [16, 43]. There was no full random exposure, as the districts Aileu, Baucau, Manatuto and Oecusse were selected for the intervention due to their potential for optimizing the support and complementarities of existing programs and relationships. Baucau, Manatuto and Oecusse have previously been reported to be particularly susceptible to food insecurity and poor nutritional status [48]. More detailed information about the MDG-F JP can be found in Additional file 2.

Results

Sample characteristics

Anthropometric data, reported as height and weight, as well as HAZ, WAZ and WHZ-scores, were available for 7378 of the 9806 children under the age of five included in the DHS data file of 2009–2010 (missing data of 24.8%) and for 5081 of the 7221 children in the DHS data file of 2016 (missing data of 29.6%). A flow chart describing the study population selection process is provided as Additional file 3 and detailed descriptive statistics of the final study sample can be found in Additional file 4. Anthropometric characteristics are displayed in Table 1. As a result of natural variation, malnutrition occurs among 2.3% of under-fives in a well-fed population [69]. After subtraction of this percentage, the excess prevalence of stunting, underweight and wasting among under-fives in Timor-Leste is respectively 59%, 28% and 49% in 2009, and 51%, 41% and 29% in 2016. In our sample, 76% suffered from any form of malnutrition in 2009–2010, and 69% in 2016.

Table 1 Anthropometric characteristics of children in the sample

Determinants of childhood undernutrition

For policy purposes we aimed to identify the main determinants that explain the status of malnutrition of under-fives. In the next sections we use bivariate and multivariate analyses for the detection of statistically significant explanatory variables.

Bivariate analyses

The output of the Pearson’s and Spearman’s tests to explore correlation between HAZ-, WHZ-, and WAZ-scores and independent continuous variables is displayed in Additional file 5a, providing correlation coefficients and 95% confidence intervals. Weak statistically significant correlation (r > 0.1 and p < 0.05) was found between HAZ-scores and the MVHI of the first growing season (r = − 0.236; 95% CI – 0.259 to – 0.212); and birth weight (r = − 0.107; 95% CI 0.056 to 0.158). For WAZ-scores, birth weight (r = − 0.150; 95% CI 0.099 to 0.200); respondent's weight (r = − 0.148; 95% CI 0.124 to 0.172); the MVHI of season 1 (r = − 0.138; 95% CI − 0.162 to − 0.113); and respondent’s BMI (r = − 0.118; 95% CI 0.094 to 0.142) were weakly correlated. No significantly correlating continuous variables were found for wasting. Next, independent samples t tests showed whether Z-score population means varied significantly across groups composed on the basis of dichotomous variables. A brief overview of the results is shown in Table 2, with a focus on reporting categories with lower Z-scores than others, indicating increased severity of undernutrition. The complete output of population means, t-values, p-values and equality of variances is given in Additional file 5b.

Table 2 Bivariate analyses of dichotomous variables

Table 3 displays the categorical classification variables which were statistically significant associated with either HAZ-, WAZ- or WHZ-score population means, sorted by their statistical significance (see Additional file 5c). None of the classification variable estimates is estimable on their own. Lastly, by including an interaction term to each individual MANOVA of potential effect modifiers, relationships between the independent variables were examined, as displayed in Additional file 6. Child age was significantly related (p < 0.05) to 50% of other variables when predicting stunting, and to 75% of other variables when predicting wasting and underweight. Wealth was an important effect modifier only for underweight (75%).

Table 3 Bivariate analyses of categorical variables

Multiple linear regression models

Explanatory modelling entailed a backwards selection procedure and multivariate regression analysis for each dependent variable, whilst pre-selecting variables that were reported to be related to HAZ-, WHZ-, and WAZ-scores in bivariate models. The relationships between independent variables and HAZ-, WAZ-, and WHZ-scores from multiple linear regression analysis are shown in Table 4, and entirely in Additional file 7. The adjusted R-Square of the regression model for HAZ-scores was 0.1091, for WAZ-scores was 0.1859 and for WHZ-scores was 0.1071. During the repeated model selection procedure, each covariate in the main models appeared to have been selected in at least 20% of the samples (as shown in Additional file 8) and main model parameter estimates did not deviate significantly from the averaged samples’ estimates. Covariates in the main models were therefore considered robust explanatory variables.

Table 4 Explanatory regression model of HAZ-, WAZ- and WHZ-scores

MDGF-JP impact

At baseline, in 2009–2010, average HAZ-scores were higher in MDG-F JP districts (diff. between means: − 0.778; 95% CI − 0.885 to − 0.671), indicating less severe stunting, whereas average WHZ-scores were lower in MDG-F JP districts (diff. between means 0.661; 95% CI 0.579 to 0.743), indicating more severe wasting. No significant difference with non-MDG-F JP districts was found for WAZ scores, indicating underweight. At end line, after the MDG-F JP, in 2016, no statistically significant difference in mean HAZ-, WHZ-, and WAZ-scores between MDGF-JP and non-MDG-F JP districts appeared. As we know from our absolute descriptive results, overall, in MDGF-JP and non-MDG-F JP districts together, average severity of stunting and underweight decreased between 2009 and 2016, but wasting severity increased.

With regard to change of Z-scores over the years in MDG-F JP and non-MDG-F JP districts, outcomes are displayed in Tables 5 and 6, and more deliberately in Additional file 9. With regard to stunting, in general, living in an MDG- F JP district led to an unfortunate average decrease of HAZ-scores (Abs.: − 21; rel. − 14%), whereas not living in MDG-F JP districts led to a fortunate increase of HAZ-scores (Abs. + 57; rel. + 25%). Within groups classified by each independent variable, positive impact of the MDG-F JP on Z-scores was only found for children whose mother had overweight (No JP: + 35.729 (17%) and JP: + 39.830 (20%)), with an absolute difference in Z-score increase of 4.101. Positive impact was also reported for children whose father was an agricultural employee (No JP: + 99.847 (36%) and JP: + 115.412 (59%), with an absolute difference), or a skilled manual (No JP: + 27.272 (13%) and JP: + 59.055 (25%), with an absolute difference of 31.78%), but the populations in the comparison were too small to reliably report differences (< 50).

Table 5 Impact of the MDG-F JP-overall
Table 6 Impact of the MDG-F JP-specified

For underweight, living in MDG-F JP districts had a weaker positive impact on the whole population’s WAZ-scores than not living in MDG-F JP districts. Only within some specific groups a positive impact of MDG-F JP appeared, namely in groups of children whose household did not own livestock, herds or farm animals; girls; children living in urban regions; children who were 1, 2 or 3 years old; children of mothers who were underweight or overweight; children in the two richest wealth quintiles; children whose parents own more than three hectares of land; children whose mother had primary, incomplete secondary, or complete secondary education; and lastly children who were smaller than average at birth. The groups of children whose father was a skilled or unskilled manual worker also seemed to have profited from living in an MDG-F JP district. However, groups were too small to report such differences. A positive impact of MDG-F JP for wasting was found for the overall population, as well as in every uniform sub-population. For groups with missing measurements or insufficient population size (agricultural land and partner’s occupation), no positive impact could be reported.

Discussion

Contributions of the MDG-F Joint Program

For the evaluation of the impact of the MDG-F Joint Program we compared the nutritional status of the study population before and after the program. In districts in which the MDGF-JP had been implemented, the baseline situation was better in terms of stunting than in non-JP districts (Diff. HAZ-score means: + 0.8). After the program ended, HAZ-score means were similar to those in non-MDG-F JP districts. The fact that HAZ-scores increased in non-JP districts (+ 25%) but decreased in JP districts (− 14%) implies a negative and inverse effect of the MDG-F joint program, potentially caused by persistent factors causing stunting, and/or time- and place-specific adverse environmental factors [9].

For underweight, the situation at baseline and end line was approximately similar in JP and non-JP districts, as no statistically significant differences were found in 2009 and in 2016. For the sample as a whole, the mean WAZ-score increased between 2009 and 2016. No evidence of either a positive or negative impact of the MDG-F JP on reducing population-wide underweight was found. Non-MDG-F JP districts had an HAZ-increase of 0.14 (+ 8%) and MDG-F JP districts had an increase of 0.12 (+ 7%) over the years. However, when moving from a population-wide scope to a comparison of similar groups across intervention and control districts, it is striking to notice an increase of WAZ-scores in specific groups within the MDG-F JP districts that is greater than among similar groups in non-MDG-F JP districts. Specifically, increased WAZ-scores were found for children whose mothers had primary, incomplete secondary, or complete secondary education; children living in urban areas; and for children in the two richest wealth quintiles. The fact that only a limited group seems to have profited from the program might very well indicate that farmers’ groups and mother support groups and the new agriculture and nutrition practices and methods that are taught and provided by the MDG-F program are initially adopted by the better educated and wealthier groups. Reaching non-literate and remote farmers and mothers appears to be a major challenge, which has also been acknowledged in an earlier MDG-JP evaluation report [44]. There is a need for further examination regarding the high variation in the impact of the MDG-F JP’s on different populations’ WAZ-scores.

In terms of wasting, the situation was much more problematic in JP districts than in non-JP districts (WHZ-scores in JP districts: − 1.32; non-JP districts: − 0.7; diff: 0.7) at baseline, but similar at end line (both − 0.96). In the overall population, WHZ-scores fell significantly, representing increased severity of wasting. Despite the average decrease of WHZ-scores, the absolute prevalence of children classified as wasting has slightly increased in the overall population. Regarding the different trends, a negative WHZ-score trend was observed in non-MDG-F JP districts (abs. − 0.3; rel. − 45%) while a positive trend was observed in MDG-F JP districts (abs. + 0.37; rel. + 28%).

The contrast between the obvious lack of evidence for impact of the MDG-F JP on stunting and the observed positive trends regarding wasting-rates has many possible interpretations. First, stunting is considered a chronic form of undernutrition and is only associated with a child’s height, which cannot easily be addressed within a period of a few years, in contrast to wasting. Childhood stunting remains a strongly persistent problem in Timor-Leste compared to global trends, but the underlying causes are not yet understood [19]. It is suggested that maternal problems in Timor-Leste play a substantial role, confirming the importance of MDG-F JP's focus on mothers, but also highlighting the importance of further examination of underlying factors [19]. Although wasting and stunting are closely related, each represents a unique health threat. Wasting is associated with a greater risk of mortality, but there is also the possibility of treatment. Episodes of wasting seem to increase the risk of stunting in particular contexts, but evidence is mixed [40]. Therefore, in a sense, the decline in the incidence of wasting offers perspectives, but a broader range of socio-economic and maternal determinants plays a role in its occurrence and prevention.

Lastly, the lack of evidence for (un-)successful impact of the MDG-F JP on Z-scores can be largely explained by baseline differences and district-specific (unobserved) confounders. For instance, wealth appeared to be an important effect modifier when assessing underweight, and was also correlated with district of residence. The rationale behind the districts selected for specific program components remains largely unspecified. The uneven distribution of Timor Vita across MDG-F JP districts, for example, could have played a role [46]. Such unobserved contextual and program confounding factors challenge the demarcation of the MDG-F JP initiatives and their impact.

Determinants of child nutritional status

In order to develop relevant policy recommendations, we explored the context-specific determinants of food security. Regarding child age, the data showed that stunting and underweight were more severe as the child became older, corroborating previous studies [11, 31]. Next, in line with the literature, our findings suggest a major negative impact of various maternal health factors on children’s nutritional status, such as smoking [8, 62] and underweight [35]. Our findings show that both underweight and wasting appeared to be significantly more severe in children whose mothers use contraception. Previous studies have yielded mixed results on this subject [20, 30, 59]. Parents’ lack of education, occupation and wealth proved to be important determinants of childhood undernutrition both in current and previous studies [29, 31, 64]. Lastly, the literature commonly cites water quality and sanitation as important determinants of how food is prepared and children’s nutritional status [24, 29, 58]. It was therefore alarming to find that that fewer people had access to an improved source of drinking water in 2016 than in 2009–2010 (59% vs 26%), and access to improved sanitation facilities decreased from 48 to 33%. Yet, this study found mixed associations between the use of water and sanitation facilities and undernutrition, which accords with earlier studies by Fagbamigbe et al. [29] where complex structure of (confounding) factors blurred possible causal relationships [57].

Policy recommendations

The outcomes of current study reinforce the complex multi-factorial nature of undernutrition among under-fives in SIDS, and the pressing need for integrated and cross-sectoral programs. Integration is important, in order to effectively reach specific populations, namely mothers and children. Their situation is strongly interconnected with broader population health issues and far-reaching developmental consequences of undernutrition in early life; and an intergenerational cycle of poor health can potentially be avoided [4, 10, 39, 75]. According to this study, young women who are underweight and are particularly likely to have underweight children should receive nutritional support, for example during pregnancy [35]. The integrated focus on mothers and children aligns with the priorities set by local political authorities [19, 35]. Households’ dependence on agricultural assets and activities also seems to increase their vulnerability to undernutrition. Therefore, interventions should be adapted to occupational, biophysical and socio-economic context and prioritized based on prevalence of malnourished children.

Reduction of undernutrition cannot be solved merely by health and nutrition interventions [9], but also requires better linking with the agricultural sector [75]. The MDG-F JP was mainly focused on nutrition security, and so took an integrated approach only to a limited extent [44]. The agricultural challenge, however, is not only to increase food production, but also to sustainably produce better-quality and health-enhancing foods [38]. Therefore, in the optimal case, programs should combine agricultural components with complementary non-agricultural components, like education, on maternal health-seeking, caring and nutritional practices [2]. Better education is needed in the broad sense, as well as being focused on agriculture. Both the local experts interviewed for this study and previous studies indicate that there is a lack of alignment of research, government budgeting, farmers’ practices, local wisdom and extension workers’ ideas in Timor-Leste [19]. Therefore, participatory research and education on (the adoption of) high-yield and highly nutritious crop varieties need specific attention and should combine external resources with farmers’ existing livelihood systems and environmental context, and address improving farmers’ conservation and seed-preservation techniques, as well as flexible recommendations and long-term commitment of aid agencies [24, 56, 67, 76].

Programs encouraging participatory selection of seed varieties have proved effective in giving farmers ownership [42]. Examples of improved agricultural practices are intercropping, replacing traditional slash and burn (swidden) agriculture and seasonal bush burning [47]. These developments would contribute to achieving the goals of the Timor-Leste’s Strategic Development Plan, stating that by 2030 every citizen has the opportunity to acquire new skills based on 21st-century technologies, such as high-yield agriculture [33]. Crop selection was reported as part of the MDG-F JP, albeit to a limited and barely described extent [71]. For the MDG-F JP, constraints for sustainable outcomes were also found in institutional cooperation [44]. As part of the MDG-F JP, public–private partnerships (PPPs) were set up for production of fortified food supplements at the local level [71]. However, those who attended were often unfamiliar with the prescribed methods and formats, which constrained efficient cooperation and action [71]. The next linkage is therefore to increase cooperation with, and capacity of, local community-level organizations, to be able to explore and access new technologies, as well as to set up community-based seed businesses. This contributes to the sustainability of efforts often made by aid-supported programs [12, 23, 38].

The impact of being either born or measured in Timor-Leste’s hungry season on nutritional status was striking. Despite the fact that globally, acute hunger and undernutrition are more prevalent in annual hunger seasons than in conflicts and natural disasters, it receives far less public attention and underlines the importance of intelligent and context-specific selection of crop varieties [7, 72]. There is also a need for food price regulations and improved storage techniques to prevent households from falling into seasonal hunger. In addition, this would reduce the country’s increasing dependence on imported food, seeds and fertilizers. The promotion and protection of regional markets and smallholders’ businesses is needed to prevent great transport distances and unfair prices not covering full production costs [34]. To that end, despite the lack of details, the MDG-F JP was reported to have improved access to markets, which in turn has led to a more varied diet and greater food sovereignty [34, 71]. After all, in Timor-Leste as on other SIDS, developments should entail a self-defined food and agricultural system, putting those who produce, distribute and consume the food at the core of the system and policies, while also being healthy, culturally appropriate, ecologically sound and sustainable [34]. In addition, this includes promoting the preservation of local seeds, fish, and livestock varieties, dietary diversity through traditional foods, and the support of local innovations and the value-adding development of local agri-processing and marketing [34].

Conclusions

We found that undernutrition is seasonal and locational, and that in addition, sex, age, size at birth, mother’s weight and wealth are major contributing factors to undernutrition in this group of under-five children at the household and individual level in Timor-Leste. There has been a drop in the severity of wasting over the years in districts in which the MDG-F JP was implemented, whereas stunting severity has unfortunately increased in these districts. Both run counter to trends seen in the overall population of under-fives in Timor-Leste. In theory, the MDG-F JP (partially) included many components that are necessary for a successful program, such as integration, cooperation, prioritization of children and young mothers, linking of nutrition and agriculture, and support of local community-level organizations and innovations. At the same time, one of the (potential) flaws in the MDG-F JP and other food-security programs, is the lack of policy and cooperation embracing different sectors, as well as the application of participatory research and education. Climate change, population growth and persistent failure to generate sufficient food and revenues on at the macro and micro level also reinforce the need to strengthen the focus on the selection and adoption of high-yield and highly nutritious crop varieties, on improved agricultural practices, and on facilitation of agri-processing industries and markets.

Strengths and limitations

The observational nature of this study did not allow for unambiguous determination of causality and increased the probability of residual confounding effects. Moreover, the use of anthropometric characteristics as indicators for food insecurity aligned with the literature, but might not necessarily have been a flawless classification strategy, as it does not completely recognize the complex interaction between food security and health status [28, 68]. As a rule, an increase of mean Z-scores was indicated as positive in this study, without taking account of where on the growth chart individual increases occurred. Likewise, validity issues occurred in other variables, for example in the asset-based index of household wealth and the minimum dietary diversity, resulting in weak or non-detectable associations. Due to a lack of integrated data sources, biophysical variables were aggregates from districts (MVHI) or clusters (altitude, slope, population density, soil suitability), of which the timing data collection was not exactly similar to the survey data collection. Similarly, the timing of DHS-data collection did not exactly conform to the baseline and end line of the MDG-F JP. Determination of the MDG-F JP’s actual impact remained a challenge, despite the use of representative DHS surveys, a comprehensive set of spatially tailored covariates, and the quasi-experimental design partially decreasing residual confounding and increasing precision [18].

A strength of this study is its careful selection of variables based on bivariate analyses of scientifically plausibly related variables, in order to prevent model-selection issues like data dredging and overfitting [3, 51]. Also, the thorough sensitivity analyses supported the primary findings and thus confirm reliability. Our multiple linear regression models had low adjusted R-square values of between 0.11 and 0.19. This is, however, not unusual in social research. Moreover, the model was used to identify explanatory variables, but not to provide a full predictive model of undernutrition among under-fives [1, 49]. Due to the use of context-specific variables and the specific program assessed in the current study, generalizability of outcomes to other countries and programs is limited, although the methodology may inspire future program evaluations. Moreover, the spatially embedded findings ideally meet the need for tailored evidence on health and socio-demographic indicators, applicable to local, national and international purposes. Furthermore, many challenges that generally face SIDS are reflected in this sketch of the situation in Timor-Leste. SIDS are on the frontlines of climate change, and many climate-related challenges described in this paper merit increasing attention [60].

Future research

The current study has given an exploratory outline of the way in which spatial, temporal and interpersonal variation plays a role in both the risk of undernutrition and the impact of interventions. There is a need for more in-depth geo-spatial, socio-economic and institutional research to facilitate a precise alignment of future policies and programs for specific contexts. For example, an improved empirical base could explain how agricultural assets and production of individual households affect undernutrition in Timor-Leste. Especially in SIDS, there is a need for a rapid increase of nutrient-rich food production to tackle the problem of persistent food-insufficiency, in addition to population growth and climate change. Therefore, there is a need for participatory agricultural enhancement studies to facilitate self-supply based on sustainable food production systems. Next, studies in which households actively participate through surveys and focus group discussions (FGDs) would offer insights into sustainable and effective livelihood strategies on child-feeding practices. Furthermore, to increase food self-sufficiency in Timor-Leste as a nation, as well as for individual farmers, it would be useful to explore farmers’ market mechanisms and (potential) revenue models, as the local food markets seem unable to supply food that is both affordable and profitable. Technology and agri-processing industries have the potential to improve sustainable production and conservation, but should again be affordable, profitable and adoptable, and should therefore be properly considered and facilitated through participatory research and policymaking. Given that stunting appears to be a persistent problem over the years, especially in Timor-Leste and despite the MDG-F JP and decline in wasting, there is a clear need for further exploration of interventions that might curb the intergenerational cycle of growth failure. Lastly, the way in which seasonality and maternal (pre-) pregnancy health status determine the perfect timing of interventions should be further explored.

Availability of data and materials

The datasets analyzed during the current study are available in the DHS repository, https://www.dhsprogram.com/data/dataset/Timor-Leste_Standard-DHS_2009.cfm and https://www.dhsprogram.com/data/dataset/Timor-Leste_Standard-DHS_2016.cfm.

Notes

  1. Sampling of DHS participants was based on the 2004 Timor-Leste Population and Housing Census, which has yielded independent samples from rural and urban strata in all 13 districts [55].

  2. For each DHS data point the distance to the attribute grid was calculated. A descending sort, next, identified the nearest neighbor of the DHS points in the attribute map.

Abbreviations

DHS:

Demographic and Health Survey

HAZ:

Height-for-age Z-scores

JP:

Joint Program

MDG-F:

Millennium Development Goals Achievement Fund

MDG-F JP:

Millennium Development Goals Achievement Fund Joint Program

MDGs:

Millennium Development Goals

MVHI:

Mean Vegetation Health Index

SIDS:

Small Island Developing State

WAZ:

Weight-for-age Z-scores

WHZ:

Weight-for-height Z-scores

References

  1. Akossou AYJ, Palm R. Impact of data structure on the estimators R-square and adjusted R-square in linear regression. Int J Math Comput. 2013;20:84–93.

    Google Scholar 

  2. Alderman HH, Elder LK, Goyal A, Herforth AW, Hoberg YT, Marini A, Zaman H. Improving nutrition through multisectoral approaches (No. 75102). World Bank; 2013. p. 1.

    Google Scholar 

  3. Anderson D, Burnham K. Model selection and multi-model inference. 2nd ed. New York: Springer; 2004.

    Google Scholar 

  4. Ashiabi GS, O’Neal KK. A framework for understanding the association between food insecurity and children’s developmental outcomes. Child Dev Perspect. 2008;2(2):71–7.

    Article  Google Scholar 

  5. Barnett J, Dessai S, Jones RN. Vulnerability to climate variability and change in East Timor. AMBIO; 2007. p. 372–8.

    Google Scholar 

  6. Batterbury SP, Palmer LR, Reuter TR, do Amaral de Carvalho D, Kehi B, Cullen A. Land access and livelihoods in post-conflict Timor-Leste: no magic bullets. Int J Commons. 2015;9(2):619–47.

    Article  Google Scholar 

  7. Baye K, Hirvonen K. Seasonality: a missing link in preventing undernutrition. Lancet Child Adolescent Health. 2020;4(1): e3.

    Article  PubMed  Google Scholar 

  8. Best CM, Sun K, De Pee S, Sari M, Bloem MW, Semba RD. Paternal smoking and increased risk of child malnutrition among families in rural Indonesia. Tob Control. 2008;17(1):38–45.

    Article  CAS  PubMed  Google Scholar 

  9. Bhutta ZA, Akseer N, Keats EC, Vaivada T, Baker S, Horton SE, Katz J, Menon P, Piwoz E, Shekar M, Victora C, Black R. How countries can reduce child stunting at scale: lessons from exemplar countries. Am J Clin Nutr. 2020;112(Suppl 2):894S-904S.

    Article  PubMed  PubMed Central  Google Scholar 

  10. Bhutta ZA, Cabral S, Chan CW, Keenan WJ. Reducing maternal, newborn, and infant mortality globally: an integrated action agenda. Int J Gynecol Obstet. 2012;119:S13–7.

    Article  Google Scholar 

  11. Boah M, Azupogo F, Amporfro DA, Abada LA. The epidemiology of undernutrition and its determinants in children under five years in Ghana. PLoS ONE. 2019;14(7): e0219665.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  12. Borges LF, do Rosario Ferreira A, Da Silva D, Williams R, Andersen R, Dalley A, Erskine W. Improving food security through agricultural research and development in Timor-Leste: a country emerging from conflict. Food Secur. 2009;1(4):403–12.

    Article  Google Scholar 

  13. Breiman L. The little bootstrap and other methods for dimensionality selection in regression: X-fixed prediction error. J Am Stat Assoc. 1992;87:738–54.

    Article  Google Scholar 

  14. Buchhorn M, Smets B, Bertels L, Lesiv M, Tsendbazar NE, Masiliunas D, Fritz S. Copernicus global land service: land cover 100m: collection 3: epoch 2019: globe. Version V3. 0.1; 2020.

  15. Burnham KP, Anderson DR. Model selection and multimodel inference. 2nd ed. New York: Springer; 2002.

    Google Scholar 

  16. Butsic V, Lewis DJ, Radeloff VC, Baumann M, Kuemmerle T. Quasi-experimental methods enable stronger inferences from observational data in ecology. Basic Appl Ecol. 2017;19:1–10.

    Article  Google Scholar 

  17. Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale: Lawrence Erlbaum Associates Inc.; 1988. p. 13.

    Google Scholar 

  18. Cummins JR. On the use and misuse of child height-for-age Z-score in the demographic and health surveys. Unpublished working paper; 2013. University of California Riverside.

  19. Deen J, da Conceicao Matos L, Temple B, Su JY, da Silva J, Liberato S, Mulholland K. Identifying national health research priorities in Timor-Leste through a scoping review of existing health data. Health Res Policy Syst. 2013;11(1):1–13.

    Article  Google Scholar 

  20. Dewey KG, Cohen RJ. Does birth spacing affect maternal or child nutritional status? A systematic literature review. Matern Child Nutr. 2007;3(3):151–73.

    Article  PubMed  PubMed Central  Google Scholar 

  21. Do Rosario Pacheco C, Picauly I, Sinaga M. Health, food consumption, social economy, and stunting incidency in Timor Leste. KEMAS Jurnal Kesehatan Masyarakat. 2017;13(2):261–9.

    Google Scholar 

  22. Ecker O, Breisinger C. The food security system: a new conceptual framework (No. 1166); 2012. Washington, DC: International Food Policy Research Institute (IFPRI).

  23. Erskine W, Nesbitt H. How can agriculture research make a difference in countries emerging from conflict? Exp Agric. 2009;45(3):313.

    Article  Google Scholar 

  24. FAO, IFAD & WFP. The state of food insecurity in the world, 2013: the multiple dimensions of food security. Rome: Food and Agricultural Organization of the United Nations; 2013.

    Google Scholar 

  25. FAO. Food security and nutrition in small island developing states. Rome: Food and Agriculture Organization of the United Nations; 2014. Retrieved from: https://sustainabledevelopment.un.org/content/documents/2231Food%20Security%20and%20Nutrition%20in%20SIDS.pdf.

  26. FAO. FAO STAT. Agri-environmental indicators—land use domain. Food and Agriculture Organization of the United Nations. Last Update: September 10, 2020; 2020. Accessed on 2021 March 6. Retrieved from http://www.fao.org/faostat/en/#data/EL.

  27. FAO, IFAD, UNICEF, WFP & WHO. The state of food security and nutrition in the world 2017. Building resilience for peace and food security; 2017. Rome: Food and Agriculture Organization of the United Nations.

  28. FAO, IFAD, UNICEF, WFP & WHO. The state of food security and nutrition in the world 2018. Building climate resilience for food security and nutrition; 2018. Rome: Food and Agriculture Organization of the United Nations.

  29. Fagbamigbe AF, Kandala NB, Uthman OA. Severe acute malnutrition among under-5 children in low-and middle-income countries: a hierarchical analysis of associated risk factors. Nutrition. 2020;75: 110768.

    Article  PubMed  Google Scholar 

  30. Flood D, Petersen A, Martinez B, Chary A, Austad K, Rohloff P. Associations between contraception and stunting in Guatemala: secondary analysis of the 2014–2015 Demographic and Health Survey. BMJ Paediatrics Open. 2019;3(1):35.

    Article  Google Scholar 

  31. GDS, MoF & ICF Macro. Timor-Leste demographic and health survey 2016. Dili, Timor-Leste and Rockville, MD: General Directorate of Statistics [Timor-Leste], Ministry of Finance [Timor-Leste] and ICF Macro. Accessed on 2021 Feb 23; 2018a. Retrieved from: https://dhsprogram.com/pubs/pdf/FR329/FR329.pdf.

  32. GDS, MoF & ICF Macro. Timor-Leste demographic and health survey 2016. Dili, Timor-Leste and Rockville, MD General Directorate of Statistics [Timor-Leste], Ministry of Finance [Timor-Leste] and ICF Macro. . TLIR71SD.sas7bdat; 2018b. Rockville, Maryland: GDS, MoF, and ICF [Producers]. ICF [Distributor], 2018.

  33. Government of Timor-Leste. Timor-Leste strategic development plan 2011–2030; 2010. p. 213. Retrieved from: http://www.tls.searo.who.int/LinkFiles/Home_NATIONAL_STRATEGIC_DEVELOPMENT_PLAN_2011-2030.pdf.

  34. Guttal S. New and old faces of hunger: Cambodia, Timor Leste, and food crises. Review (Fernand Braudel Center); 2009. p. 61–89.

  35. Hall A, Hyder Z, Karyadi E. Malnutrition in young children and their mothers in Timor-Leste. Health, nutrition and population discussion paper. Washington, DC: World Bank; 2020.

    Book  Google Scholar 

  36. Hamill PV, Drizd TA, Johnson CL, Reed RB, Roche AF. NCHS growth curves for children birth-18 years. Washington, DC: Department of Health Education and Welfare; 1977.

    Google Scholar 

  37. Hauke J, Kossowski T. Comparison of values of Pearson’s and Spearman’s correlation coefficients on the same sets of data. Quaestiones Geographicae. 2011;30(2):87.

    Article  Google Scholar 

  38. Hawkes C, Ruel MT, editors. Understanding the links between agriculture and health. Washington, DC: International Food Policy Research Institute; 2006.

    Google Scholar 

  39. Hoddinott J, Alderman H, Behrman JR, Haddad L, Horton S. The economic rationale for investing in stunting reduction. Matern Child Nutr. 2013;9:69–82.

    Article  PubMed  PubMed Central  Google Scholar 

  40. Khara T, Dolan C. Technical briefing paper. The relationship between wasting and stunting: policy, programming and research implications. Oxford: Emergency Nutrition Network; 2014.

    Google Scholar 

  41. Kutner MH, Nachtsheim CJ, Neter J, Li W. Applied linear statistical models, vol. 5. Boston: McGraw-Hill Irwin; 2005.

    Google Scholar 

  42. Lacoste M, Williams R, Erskine W, Nesbitt H, Pereira L, Marçal A. Varietal diffusion in marginal seed systems: Participatory trials initiate change in East Timor. J Crop Improv. 2012;26(4):468–88.

    Article  Google Scholar 

  43. Lawlor DA, Tilling K, Smith GD. Triangulation in aetiological epidemiology. Int J Epidemiol. 2016;45(6):1866–86.

    PubMed  Google Scholar 

  44. Lenci S. MDG‐F JP sustainable food and nutrition security in Timor-Leste final evaluation report. MDG Fund; 2012. Retrieved from: http://www.mdgfund.org/sites/default/files/Timor%20Leste%20-%20Nutrition%20-%20Final%20Evaluation%20Report.pdf.

  45. Leroy JL, Ruel M, Frongillo EA, Harris J, Ballard TJ. Measuring the food access dimension of food security: a critical review and mapping of indicators. Food Nutr Bull. 2015;36(2):167–95.

    Article  PubMed  Google Scholar 

  46. Lister S, Keylock J, Silkin T. Country portfolio evaluation. Timor Leste: an evaluation of WFP’s Portfolio (2008–2012); 2013. Retrieved from: http://documents.wfp.org/stellent/groups/public/documents/reports/wfp257249.pdf.

  47. Lopes M, Nesbitt H. Improving food security in Timor-Leste with higher yield crop varieties (No. 423-2016-27050); 2012.

  48. MDG-F. MDG-F: Joint program promoting sustainable food and nutrition security. In: Timor-Leste. Signed document. Accessed on 9 March 2021; 2009c. Retrieved from: http://mdgfund.org/sites/default/files/Signed_JP_Timor_Leste_Children_Sept09.pdf.

  49. Moksony F, Heged R. Small is beautiful. The use and interpretation of R2 in social research. Szociológiai Szemle. 1990;8:130–8.

    Google Scholar 

  50. Molyneux N, Da Cruz GR, Williams RL, Andersen R, Turner NC. Climate change and population growth in Timor Leste: implications for food security. Ambio. 2012;41(8):823–40.

    Article  PubMed  PubMed Central  Google Scholar 

  51. Nansen C, Geremias LD, Xue Y, Huang F, Parra JR. Agricultural case studies of classification accuracy, spectral resolution, and model over-fitting. Appl Spectrosc. 2013;67(11):1332–8.

    Article  CAS  PubMed  Google Scholar 

  52. Noij F. Joint program promoting sustainable food and nutrition security in Timor-Leste: mid-term evaluation—final report; 2011.

  53. Nørgaard M, Ehrenstein V, Vandenbroucke JP. Confounding in observational studies based on large health care databases: problems and potential solutions–a primer for the clinician. Clin Epidemiol. 2017;9:185.

    Article  PubMed  PubMed Central  Google Scholar 

  54. NSD, MoF and ICF Macro. Timor-Leste demographic and health survey 2009–10. Dili. National Statistics Directorate [Timor-Leste], Ministry of Finance [Timor-Leste] and ICF Macro; 2010a. Accessed on 2021 Feb 5. Retrieved from: https://dhsprogram.com/pubs/pdf/FR235/FR235.pdf.

  55. NSD, MoF and ICF Macro. Timor-Leste demographic and health survey 2009–10. Dili. National Statistics Directorate [Timor-Leste], Ministry of Finance [Timor-Leste] and ICF Macro. TLIR61SD.sas7bdat. Rockville, MD: NSD, MoF, and ICF [Producers]. ICF [Distributor]; 2010b.

  56. Oxfam Australia. Timor-Leste food security baseline survey report. Oxfam Australia; 2008.

  57. Rah JH, Sukotjo S, Badgaiyan N, Cronin AA, Torlesse H. Improved sanitation is associated with reduced child stunting amongst Indonesian children under 3 years of age. Matern Child Nutr. 2020;16: e12741.

    Article  PubMed  PubMed Central  Google Scholar 

  58. Raihan MJ, Farzana FD, Sultana S, Haque MA, Rahman AS, Waid JL, Ahmed T. Examining the relationship between socio-economic status, WASH practices and wasting. PLoS ONE. 2017;12(3): e0172134.

    Article  PubMed  PubMed Central  Google Scholar 

  59. Rana MJ, Goli S. The returns of family planning: macro-level assessment of the effect of contraceptive use on women’s anaemia and childhood undernutrition. J Biosoc Sci. 2017;49(6):773.

    Article  PubMed  Google Scholar 

  60. Robinson SA. Climate change adaptation in SIDS: a systematic review of the literature pre and post the IPCC fifth assessment report. Wiley Interdiscip Rev Clim Change. 2020;11(4): e653.

    Article  Google Scholar 

  61. Schober P, Boer C, Schwarte LA. Correlation coefficients: appropriate use and interpretation. Anesth Analg. 2018;126(5):1763–8.

    Article  PubMed  Google Scholar 

  62. Semba RD, Kalm LM, De Pee S, Ricks MO, Sari M, Bloem MW. Paternal smoking is associated with increased risk of child malnutrition among poor urban families in Indonesia. Public Health Nutr. 2007;10(1):7–15.

    Article  PubMed  Google Scholar 

  63. Smith LC, Haddad L. Reducing child undernutrition: past drivers and priorities for the post-MDG era. World Dev. 2015;68:180–204.

    Article  Google Scholar 

  64. Smith LC, Haddad LJ. Explaining child malnutrition in developing countries: a cross-country analysis, vol. 111. Washington, DC: International Food Policy Research Institute (IFPRI); 2000.

    Google Scholar 

  65. Sonneveld B, Alfarra A, Merbis M, Atzori D. Sustainable management of freshwater resources for food and nutrition security in Small Island Developing States. Rome: FAO; 2021.

    Google Scholar 

  66. Sullivan KM, Gorstein J. ANTHRO: software for calculating pediatric anthropometry. version 1.02; 1999. Software and documentation available at https://stacks.cdc.gov/view/cdc/44746.

  67. Tiwari TP, Virk DS, Sinclair FL. Rapid gains in yield and adoption of new maize varieties for complex hillside environments through farmer participation: I. Improving options through participatory varietal selection (PVS). Field crops Res. 2009;111(1–2):137–43.

    Article  Google Scholar 

  68. Townsend MS, Peerson J, Love B, Achterberg C, Murphy SP. Food insecurity is positively related to overweight in women. J Nutr. 2001;131(6):1738–45.

    Article  CAS  PubMed  Google Scholar 

  69. UNICEF, WHO & World Bank Group. Levels and trends in child malnutrition: key findings of the 2020 edition of the joint child malnutrition estimates. Geneva: World Health Organization; 2020a.

    Google Scholar 

  70. UNICEF, WHO & World Bank Group. Joint child malnutrition estimates (JME). Prevalence of stunting, height for age (% of children under 5); 2020b. Retrieved from: data.worldbank.org on 2021 Feb 23.

  71. United Nations. Joint Programmes for food security and nutrition. A qualitative review of agricultural programming for nutrition among the Millennium Development Goals Achievement Fund joint programmes; 2013. Accessed on 12 Feb. 2021. Retrieved from: https://www.sdgfund.org/joint-programmes-food-and-nutrition-security.

  72. Vaitla B, Devereux S, Swan SH. Seasonal hunger: a neglected problem with proven solutions. PLoS Med. 2009;6(6): e1000101.

    Article  PubMed  PubMed Central  Google Scholar 

  73. WHO. Fact sheet on millennium development goals (MDGs). Published on 19 February 2018. The World Health Organization [online]. Geneva: World Health Organization; 2018a. Accessed on March 11 2021. Retrieved from: https://www.who.int/news-room/fact-sheets/detail/millennium-development-goals-(mdgs).

  74. WHO. Child growth standards. In: The World Health Organization [online]. Geneva: World Health Organization; 2018b. Retrieved from: www.who.int/childgrowth.

  75. Wiegers ES, van Dorp M, Torgerson S. Improving nutrition through agriculture: viewing agriculture-nutrition linkages along the smallholder value chain. Centre for Development Innovation, Wageningen University & Research centre/REACH; 2011.

  76. Williams R, Borges LF, Lacoste M, Andersen R, Nesbitt H, Johansen C. On-farm evaluation of introduced maize varieties and their yield determining factors in East Timor. Field Crop Res. 2012;137:170–7.

    Article  Google Scholar 

  77. World Bank Group & International Monetary Fund. Global Monitoring Report 2014/2015: Ending Poverty and Sharing Prosperity. Washington, DC: World Bank Group; 2015.

Download references

Acknowledgements

Not applicable.

Funding

The authors did not receive funds, grants, or other support from any organization for the submitted work.

Author information

Authors and Affiliations

Authors

Contributions

Both authors contributed to the study conception and design. Data analysis and preparation of the first manuscript draft were performed by LS, supervised by BS. Both authors collaborated on and approved the final manuscript.

Corresponding author

Correspondence to L. Van der Spek.

Ethics declarations

Ethics approval and consent to participate

Regarding compliance with Ethical Standards, participants of the standard DHS surveys gave informed consent in advance. The surveys comply with the U.S. Department of Health and Human Services regulations for the protection of human subjects (45 CFR 46) and ethical approval from the Timor-Leste Ministry of Health was obtained. This study complies with the Code of Ethics for Research involving Human Participants of the VU University’s Faculty of Science.

Consent for publication

Not applicable.

Competing interests

The authors declared that they have no conflict of interest.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1

. Overview of variables.

Additional file 2

. Background information on the MDG-F JP.

Additional file 3

. Flow chart of the study population.

Additional file 4

. Characteristics of children in the sample.

Additional file 5

. Bivariate analyses.

Additional file 6

. Interaction terms.

Additional file 7

. Multiple linear regression models.

Additional file 8

. Model selection.

Additional file 9

. MDG-F JP impact.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Van der Spek, L., Sonneveld, B.G.J.S. Analyzing the impact of an MDG-Fund program on childhood malnutrition in Timor-Leste. J Health Popul Nutr 43, 46 (2024). https://doi.org/10.1186/s41043-024-00539-x

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s41043-024-00539-x

Keywords