Menu
Menu
Menu
Menu
Background: Armed conflicts greatly affect the health, nutrition, and food security of conflict affected settings particularly children. However, no empirical data exist regarding context specific factors contributing towards acute malnutrition in the war-torn Tigray, Ethiopia. Thus, this study aimed to identify individual and community level factors associated with acute malnutrition among children aged 6–59 months from armed conflict affected settings of Tigray, Ethiopia. Methods: A community based cross-sectional study was conducted among 3,614 children aged 6–59 months in Tigray, from July 15 to Aug 15, 2021. Study participants were selected using a two-stage random sampling method. A structured questionnaire was used to collect data by interviewing mothers/caregivers. Mid upper arm circumference (MUAC) measurements were taken from upper left arm of the children using MUAC tapes. Multivariable multilevel logistic regression analysis was used to determine factors associated with acute malnutrition. Adjusted Odds ratio (AOR) with 95% CI were estimated to describe the strength of associations at p < 0.05. Results: More than half (52.5%) of the sampled children were males in sex. Immediately after the first nine months into the conflict, the prevalence of severe, moderate, and global acute malnutrition was very high (5.1%, 21.8%, and 26.9%, respectively) in Tigray. The lowest and highest burden of child acute malnutrition was reported from Mekelle zone (13.3%) and Southeastern zone (36.7%), respectively. Individual-level factors such as older child age (AOR = 0.13, 95% CI: 0.10, 0.18), female child sex (AOR = 1.24, 95% CI 1.05, 1.480.95), Vitamin-A supplementation (AOR = 1.3, 95% CI: 1.05, 1.65), and history of diarrhea (AOR = 1.22, 95%CI: 1.02, 1.53) and community-level factors like unimproved drinking water source (AOR = 1.31, 95%CI: 1.08, 1.58), unimproved toilet facility (AOR = 1.24, 95% CI: 1.01, 1.52), and severe food insecurity (AOR = 1.55, 95% CI: 1.16. 2.07) were significantly associated with childhood acute malnutrition. Conclusions: The burden of acute malnutrition is a severe public health problem in Tigray. To prevent the untimely suffering and death of children, regular nutrition screening, speedy, and appropriate referral of all malnourished children to nutritional services and large-scale humanitarian assistance including access to food; nutrition supplies; water, sanitation and hygiene supplies; and health care in a timely manner are required. In the prevailing armed conflict, these have been very difficult to achieve. Thus, immediate international intervention is needed.
The study was conducted in Tigray, the northern most part of Ethiopia. Tigray is bordered with Eritrea from the north, Sudan from the west, Afar region from the east and Amhara region from the south. Administratively, Tigray is divided in to seven zones namely Central, Eastern, Mekelle, North western, Southern, South eastern and Western zones and 93 districts. In this study, 52 districts in all the zones except the western zone were included. The western zone was excluded for security concerns. The study was conducted between July 15 and Aug 15, 2021. A community based cross-sectional study was conducted in six zones and 52 randomly selected districts of Tigray. Only mothers or caretakers of children 6–59 months of age were included in the study. Inclusion criteria: all households with under one years of age children were included. Then, all children under five in the selected household were measured for nutritional status. Exclusion criteria: Serious illness in children under five. The minimum sample size was estimated using proportional allocation of the minimum sample size for a Tabia (smallest administrative unit) from each of the districts. The sample size was determined based on the burden of acute malnutrition. According to the EDHS 2016, the burden of acute malnutrition was 11% in Tigray [15]. For situations where power and prevalence are known, effective sample size can easily be estimated. For 11% prevalence, 20 subjects are sufficient to reach a power value greater than 80 [16]. Accordingly, 20 households were included from each Tabia and four Tabias were selected from each district. Thus, adding 5% of non-response, the total sample size was calculated to be 4368 [20 subjects per Tabia*4 Tabias per district*52 districts) + (5%*4160)]. However, 754 subjects were excluded from the final analysis for the following reasons; 1) 444 were under the age of 6 months; 2) 308 had no MUAC measurements, and 3) 2 were outliers. Therefore, the findings of this study were based on data from 3614 subjects. A multistage sampling technique was employed. All the zones except the western zone were included in the study. At the first stage, a total of 52 districts were randomly selected from the 93 districts. In the second stage, four Tabias were randomly selected from each of the 52 districts. Then, 20 households with under one year of age children were randomly selected from the selected tabias using the registration book of the Health Extension Workers (HEWs) as a sampling frame. When the registration book of the HEWs was not available, a new list of the households with under one year of age children was generated and used as a sampling frame to randomly select the study households. However, it must be noted that the districts from the Western zone of Tigray were not included in the sampling frame. Data were collected using a pretested and interviewer-administered structured questionnaire. The tool contained items regarding socio-demographic, health and obstetric, childhood illness and vaccination characteristics; water, hygiene and sanitation conditions; and household food insecurity, and anthropometric measurement (MUAC). Experienced HEWs were the data collectors and health and nutrition researchers/experts from Mekelle University and Tigray Health Bureau worked as supervisors of the data collection. MUAC measurements were taken from the upper left arm of the children using MUAC tapes; household food insecurity was measured using the Household Food Insecurity Access Scale (HFIAS), which was answered by the mothers/caregivers of the children. The questionnaire was initially prepared in English and then translated into the local language (Tigrigna) and was then translated back to English for consistency check. Three days training was given for the data collectors and supervisors. Moreover, fieldwork was accompanied by strict follow up and supervision. These included child age, child sex, Vitamin-A supplementation status, measles vaccine status, deworming, maternal education, paternal education, child had diarrhea in the last nine months, child had fever in the last nine months, child had cough in the last nine months, place of delivery, and antenatal care (ANC) visits. Residence, toilet facility, drinking water source, handwashing facility close to toilet, solid waste disposal, liquid waste disposal, family size, and household food insecurity were considered as community level factors. The dependent variable for this study was child acute malnutrition as measured by MUAC with two categories (“Yes” if MUAC < 12.5 cm and “No” if MUAC ≥ 12.5 cm). Data (Additional file 1) were cleaned and analyzed using Statistical package for Social Sciences (SPSS) software version 25. Categorical variables were summarized using percentages. Multilevel binary logistic regression analysis was employed to identify factors significantly associated with the outcome variable. Within the multilevel multivariable logistical regression analysis, Adjusted odds ratios (AOR) with their 95% confidence intervals were computed to measure the fixed effects of individual-level and community-levels factors on the prevalence of child acute malnutrition. During bi-variable logistic regression analysis, we used p-value of ≤ 0.2 to screen factors for multivariable logistic regression analysis. In the final multivariable logistic model, four models including an intercept-only model containing no explanatory variables, an individual-only model, a community-only model, and a combined model containing both individual and community-level variables were fitted. This helped to come up with a model where the effect of clustering is controlled and to determine the independent effect of each individual and community level factors on the dependent variable. Associations were declared statistically significant at a p-value of < 0.05. Prior to running the multivariable logistic analysis, multicollinearity among individual and community level variables was checked using Variance Inflation Factor (VIF) cutoff value of 10. From the four fitted models, the one with the lowest value of Akaike information criterion (AIC) and/or Bayesian information criterion (BIC) was selected as the best model to our data. Both AIC and BIC consist of a part that represents model fitness and a part that represent the size and dimensionality of the model as shown in the equation: IC= − 2logf(y│Ӫ)+λd, where IC stands for information criterion, f(y│Ӫ) is the likelihood of the data y evaluated using the model parameters, λ denotes the penalty weight that differs for AIC and BIC, and d represents the size or dimensionality of the model [17]. Variables that showed significant association at p ≤ 0.20 in the bivariate analysis were entered to multivariable logistic regression analysis. These included individual level variables like child age, child sex, Vitamin A supplementation, diarrhea in the last nine months, treatment sought for diarrhea, and fever. Community level variables that met this criteria were drinking water source, toilet facility, and HFIAS. Our analysis showed multicollinearity between presence of diarrhea and treatment sought for diarrhea. From the bivariate logistic regression, the p-value for treatment sought for diarrhea was higher than the p-value for presence of diarrhea, thus, we removed the variable treatment sought for diarrhea.
