Contribution of home garden vegetables on reducing stunting among 6‐ to 23‐month‐old children in South Tigray, northern Ethiopia

Abstract The study was conducted to analyze the contribution of home garden vegetables in reducing stunting among 6‐ to 23‐month‐old children from South Tigray, Northern Ethiopia. The quasi‐experimental study design was used. Multistage sampling technique was used to select the districts and study communities. A total of 94 purposively selected vegetable producer (intervention) households and 260 randomly selected non‐producer (control) households were included in the study (1:3 ratio). The recumbent length of children was measured using horizontal wooden board to the nearest 0.1 cm. The length‐for‐age Z‐scores were computed using WHO‐Anthro 2006 software. Propensity score‐matching and difference‐in‐difference (DID) estimates were used to analyze data using STATA software version 12. Prevalence of child stunting was 19.8 (12.7–29.4) and 21.1 (16.4–26.7)% (baseline) and 43.5 (33.5–54.1) and 46.5 (45.7–47.2)% (end line) among intervention and control groups, respectively. Child stunting was higher for boys and older children from both intervention and control households. DID estimation revealed that there was no significant difference in child stunting between intervention and control households (DID = 1.7, p = .604). However, there was an intervention effect of −0.5, 2.5, and 1.7% in the prevalence of child stunting among females, males, and both sexes, respectively. Vegetable production as an intervention strategy reduced the prevalence of stunting in children aged 6–23 months. However, vegetable production needs to be well integrated with other nutrition‐sensitive interventions to realize the objective of reducing child stunting.


| INTRODUC TI ON
Stunting (low height for age) is the result of long-term chronic consumption of a low-quality and less diverse diet in combination with morbidity, infectious diseases, and environmental problems (Haile et al., 2016;WHO, 2014). Childhood stunting is associated with adverse functional consequences including child mortality, poor cognition, low educational performance, low adult wages, poor socioeconomic development, and poor reproductive outcomes (Haile et al., 2016;UNICEF, 2013). Moreover, stunting reflects nutritional deficiencies during the most critical periods of growth and development in early life (Moges et al., 2015). In countries where food availability and access to nutrient-rich food groups are severely restricted, the nutritional status of already vulnerable children is extremely concerning, with one in two children stunted (FSIN, 2020). In sub-Saharan Africa (SSA), 34% of children aged less than 5 years are stunted, which is significantly higher than the global average of 21.3% (FAO, IFAD, UNICEF, WFP, WHO, 2017). About 49% of children aged under 5 years in Tigray are stunted, which is much higher than the average prevalence of child stunting (37%) in Ethiopia (CSA (Central Statistical Agency of Ethiopia), 2019).
Dietary intake is one of the risk factors for child stunting (WHO, 2004(WHO, , 2011. Low fruit and vegetable consumption was listed as one of the top three diet-related risks accounting for global deaths in 2000 (Lock et al., 2005). Studies conducted in Ethiopia (Moges et al., 2011) and India (Chauhan, 2015) revealed that dietary diversity is negatively associated with child stunting. Vegetables are essential components of a well-balanced diet since they supply an abundant and inexpensive source of energy, body-building nutrients, vitamins, minerals, dietary fiber, and phytochemicals (Dias, 2012;FAO, 2020;Hunde, 2017). Vegetables are also excellent resources for overcoming micronutrient deficiencies as well as providing smallholder farmers with much higher incomes and more jobs per hectare than staple crops (Davey et al., 2009).
Despite the nutritional and health benefits and the comparative advantage of a favorable climate in Ethiopia, particularly in the Tigray region, production and consumption of fruits and vegetables are very limited (Hunde, 2017;WHO, 2015). In Ethiopia, less than 20% of 6-to 23-month-old children received the minimum dietary diversity (FSIN, 2020). There are also inconsistencies in the effect of vegetable home gardening on the nutritional status of children due to differences in social, economic, and cultural issues. Thus, within the local context, identifying important gaps and generating knowledge on the effect of vegetable production on reducing stunting are imperative, which will help the design of better approaches to address context-specific strategic interventions with home garden vegetables for improved nutritional status of children aged 6-23 months. The research was, therefore, conducted to determine the effect of home garden vegetables on reducing stunting in children aged 6-23 months.

| Study area
The study was conducted in Raya-Azebo and Emba-Alaje districts of the Southern zone of the Tigray region, Ethiopia. Five villages, namely Genete and Tsigea from Raya-Azebo district and Ayba, Atsela, and Tek'a from Emba-Alaje district, representing lowland and highland agroecological conditions, respectively, were selected and included in the study. The study areas were purposefully selected with the help of local experts based on better experiences in home garden vegetable production.

| Study design
The study was conducted using a quasi-experimental design to analyze the effect of home garden vegetable production in reducing child stunting. The schematic representation of the study design is shown in Figure 1.

| Sample size and sampling technique
A total of 96 vegetable-producer households and 288 non-producer households were included in the study using the formula provided in the Food and Nutrition Technical Assistance III Sampling Guide. n = required minimum sample size per survey round or comparison group; P1 = stunting rate at baseline, 39.3% = 0.39 (EDHS, 2016); P2 = the expected level of stunting at end line for the program area such that the quantity (P2 − P1) is the size of the magnitude of change it is desired to be able to detect, considering target reduction of 10 percentage points, 29% = 0.29; Zα = the Z-score corresponding to the degree of confidence with which it is desired to be able to conclude that an observed change of magnitude (P2 − P1) would not have occurred by chance (α-the level of statistical significance for one-tailed test), 95% = 1.645; Zβ = the Z-score corresponding to the degree of confidence with which it is desired to be certain of detecting a change of magnitude (P2 − P1) if one actually occurred (β-statistical power), 80% = 0.840; D = design effect for stunting = 1.0; Nf = Non-response factor (assuming a 5% non-response rate) = 1.05. The intervention households were limited to 96 because it was not possible to get more than 96 households with access to home gardening and children 6 months of age. Hence, all households in the intervention group were included. Therefore, a total sample size of 384 (288 control and 96 intervention households) was used for the study.
Multistage sampling method was used to identify study districts and villages. Vegetable producer households were purposefully selected, while vegetable non-producer households were randomly selected from the list of the total number of eligible households identified during the household listing exercises. Respondents were mothers or caretakers of the target households.

| Description of the intervention
Households with a child 6 months of age and access to home gardening were included as intervention groups in the study.
The selected households were supported by the development agents (DAs) and the researcher to grow a variety of vegetables.
Participants in the intervention groups received training on vegetable production management. They were also provided training focusing on child nutrition using the nutrition education guidelines

| Data collection
Data on demographic and socioeconomic variables and anthropometrics were collected during the study period to evaluate the effect of home garden vegetables on child stunting. Except for the anthropometric measurements, all data were collected through face-to-face interviews using a structured questionnaire, which was administered in two phases: at the beginning and end of the intervention.

| Anthropometric measurements
Recumbent length was collected three times during the study period (6-, 15-, and 21-month-old children). The length of a child was measured using a horizontal wooden length board in the recumbent position and read to the nearest 0.1 cm. The equipment needed to measure the length was placed on a flat surface, and the measurement was conducted by trained HEW as per the standard techniques of anthropometric measurements.
F I G U R E 1 Schematic representation of the study design.

| Data processing and analysis
Length for age of a child was expressed as a standard deviation unit from the median for the reference group and was compared to the 2006 WHO growth standards. Children who fall below −2 SD from the median of the standard population were regarded as moderately stunted, while those who fall below −3 SD from the median of the standard population were considered severely stunted. Children were, therefore, categorized by their length-forage Z-score (LAZ).
Descriptive statistics was used to determine the frequencies and percentages of the sociodemographic characteristics of the study participants. The length-for-age Z-scores were computed using the WHO Anthro 2006 software. A propensity score matching (PSM) t-test was conducted to check the significance level of the variables at baseline between the control and intervention groups. Difference-in-difference (DID) estimate was used to assess the effect of home garden vegetables on child stunting. An independent t-test was performed using the gain scores of children. Analysis was done using tools in the STATA software version 12 at a statistical significance level of 5%.

| Sociodemographic characteristics
The mean ages of the women respondents from vegetable producer and

| Prevalence of stunting among 6-to 23-month-old children
The prevalence of stunting was 19.8 (12.7-29.4) and 21.1 (16.4-26.7)% at baseline and 43.5 (33.5-54.1) and 46.5 (45.7-47.2)% at the end line among children from vegetable producer and non-producer households, respectively. The prevalence of child stunting was higher in male compared to female children from both vegetableproducer and non-producer households (Table 2). Similarly, child stunting was found to be higher for older children in both vegetableproducer and non-producer households. The trend of child growth at the age of 6-23 months was, however, good in both groups in terms of their length-for-age Z-score ( Figure 2

| Difference-in-differences estimation results
Kernel PSM t-test of the control and intervention groups at the baseline showed that there was no significant difference between the groups for the outcome variable with covariates (Table 3). As the PSM t-test of the variables was insignificant at the baseline, the effect of the intervention was determined using DIDs estimate. Table 4 shows that there was a difference of 1.3 and 3.0% in the prevalence of child stunting between the control and intervention groups at the baseline and endline surveys, respectively, with a higher prevalence of child stunting for the control groups (vegetable non-producer households). DIDs estimation results revealed that there was no significant difference between female (p = .579) and male children (p = .775) in control and intervention groups based on the prevalence of stunting (Table 4). The difference observed in the prevalence of child stunting due to the intervention effect was only −0.5% and 2.5% for female and male children, respectively (Table 4).
Similarly, there was no significant difference for all children (both sexes; p = .604) by which a difference of 1.7% in the prevalence of child stunting was observed between the intervention and control groups over the study period.

| DISCUSS ION
In this study, the sex and age of the children were identified as The trend of child growth based on the length-for-age Z-score was similar in both the control and intervention groups during the study. Although significant difference was not observed, DID estimation results revealed that the prevalence of child stunting was reduced by 1.7% due to vegetable production compared to vegetable non-producers. This indicates that vegetable production could improve the chance of reducing child stunting and, thus, more effort is needed to significantly and positively affect the length-for-age Z-score of children. According to Baliki et al. (2019), the effect of vegetable consumption at the beginning was not statistically different from consumption 1 year later, which demonstrates that impact was observed in the long term. The most important determinants of children's vegetable intake are age of the child, social and cultural norms, economic status, preferences, parental intake, and home availability/accessibility (Jeong & Lee, 2021;Kehoe et al., 2019;Raggio & Gámbaro, 2018). Parents, especially mothers, influence children's vegetable preferences and the majority of them mentioned the sensory characteristics of vegetables as the main determinant of children's vegetable consumption (Kral & Faith, 2009;Laureati, 2022;Mahmood et al., 2021). Sensory attributes such as bitter taste, green color, texture, and appearance were reported as drivers for the rejection of vegetable consumption in children (Estay et al., 2019;Mustonen et al., 2012). Hence, the lower impact of vegetables on the prevalence of child stunting in the study areas could be due to the It is believed that home gardening can directly enhance household food security and then nutritional status by providing access to a diversity of nutritionally-rich foods, increased purchasing power from savings on food bills and income from sales of garden products, and fallback food provision during seasonal lean periods TA B L E 3 Kernel propensity score matching (PSM) t-test at baseline.  (Beyene et al., 2019). However, in this study, it is only due to the vegetable intervention that the prevalence of child stunting was reduced by 1.7% depicting that more reduction in the prevalence of child stunting could be achieved if vegetable intervention is integrated with other nutrition-sensitive interventions (e.g., animal source foods).

| CON CLUS ION
There is a high prevalence of child stunting in the study areas but the status of child stunting is different for different sex and age groups of children, where it was highest for boys and older children. Vegetable production as an intervention strategy reduced the prevalence of stunting in children aged 6-23 months. The availability of vegetable production is not enough for improving nutritional status of children. Hence, home garden vegetable production needs to be well integrated with the existing demographic and socioeconomic factors such as feeding practices to improve consumption of diverse and nutritious foods by the households, particularly children so as to realize the objective of reducing child stunting. Further study is also needed to investigate better design of strategic interventions of vegetables to address the problem of child stunting.

ACK N OWLED G M ENTS
The authors acknowledge the contributions of ACCAI (African Climate Change Adaptation Initiative) project, Mekelle University, for the financial support to conduct the study. We would also like to thank staff of the Institute of Climate and Society, Mekelle University, for their support in facilitating the necessary materials to conduct the study and our team of enumerators and respondents for their time and kind cooperation.

This work was supported by the African Climate Change Adaptation
Initiative project (ACCAI) of the Open Society Institute, Mekelle University, with grant No. OR2014-18350 to conduct the research.
The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in figshare at http://doi.org/10.6084/m9.figsh are.22730066. University. Moreover, the study was approved by the local offices at the district and sub-district levels. After the approval from local officials, each participant consented orally to participate in the study and was informed of the objectives of the study, and their rights to refuse participation, stop participating at any time during the interview, and skip specific questions or topics they do not want to answer.