Severe stunting and its associated factors among children aged 6–59 months in Ethiopia; multilevel ordinal logistic regression model

Background: In Ethiopia, stunting is the most common form of undernutriton. Identifying the determinants of severe stunting among children is crucial for public health interventions to improve child health. Therefore, this study aimed to identify the determinants of severe stunting among under-five children in Ethiopia. Methods: A community-based cross-sectional study design was employed. A two stage stratified cluster sampling technique was used. A multilevel ordinal logistic regression model was fitted to identify independent determinants. Adjusted odds ratio (AOR) and median odds ratio (MOR) with its 95% confidence interval at p-value< 0.05 were used to declare statistical significance. Results: The result of this study showed that about 18% of the children were severely stunted. Being male increased the severity of stunting in children by 26% adjusted odds ratio (AOR): 1.26 (95% CI: 1.09–1.46), compared to female sex; over-weight mothers increased the severity of stunting in their children AOR: 3.43 (95% CI: 2.21–5.33) compared to normal BMI mothers; and children from middle, poorer, and poorest wealth index households were 1.84 (95% CI:1.27–2.67), 2.13 (95% CI, CI:1.45–3.14) and 2.52 (95% CI,1.72–3.68). In contrast, severe stunting was reduced by 62% (AOR: 0.38, 95% CI: 0.20–0.74) and 48% (AOR = 0.52, 95% CI: 0.37–0.72) in children of educated mothers compared to children of uneducated mothers and children of underweight mothers compared with those children of normal BMI mothers respectively. For each one-unit increase in maternal height, there is a 5% significant reduction in the child’s odds of being severely stunted. After controlling for other factors, the effect of predictors on the likelihood of stunting in high risk clusters increased by a median odds ratio (MOR) of 1.83 (95% CI: 1.69–2.00). Conclusions: The magnitude of severe childhood stunting was still high with regional variation in Ethiopia. Child age, sex, maternal height, age, education and household wealth index as well as administrative regions were significantly associated factors with severe stunting. Significant interventions shall be implemented at the individual, household and community levels in order to reduce the problem.

Community-based cross-sectional study design was conducted among children aged 6–59 months. The 2016 Ethiopian Demographic and Health Survey (EDHS) is the fourth Demographic and Health Survey conducted in Ethiopia, which was conducted from January 18, 2016 to June 27, 2016 [3]. Administratively, Ethiopia is divided into nine geographical regions and two administrative cities. The sampling frame used for the 2016 EDHS is the Ethiopia Population and Housing Census (PHC), which was conducted in 2007 by the Ethiopian Central Statistical Agency. The census frame is a complete list of 84,915 enumeration areas (EAs) created for the 2007 PHC. An EA is a geographical area covering on average 181 households [3]. The source populations are children aged 6–59 months who were residing in the households in Ethiopia. The study population is children aged 6–59 months who were living in the selected households in Ethiopia. All children aged 6–59 months in the selected households. Children who fulfilled the inclusion criteria had severe medical conditions at the time of survey. In two phases, the 2016 EDHS sample was stratified and picked. There were 21 sampling strata in each region, which were divided into urban and rural areas. In two phases, EA samples were selected independently in each stratum. Implicit stratification and proportional allocation were achieved at each of the lower administrative levels by sorting the sampling frame within each sampling stratum before sample selection, according to administrative units in different units in different levels, and by using a probability to size selection at the first stage of sampling [3]. In the first stage, a total of 645 (202 in urban areas and 443 in rural areas) were selected with probabilities proportional to EA size (based on the 2007 PHC) and with independent selection in each sampling stratum. A household listing operation was carried out in all of the selected EAs from September to December 2015. The resulting lists of households served as a sampling frame for the selection of households in the second stage. Some of the selected EAs were large, consisting of more than 300 households. To minimize the task of household listing, each large EA selected for the 2016 EDHS was segmented. Only one segment was selected for the survey with probability proportional to segment size. Household listing was conducted only in the selected segment; that is, a 2016 EDHS cluster is either an EA or a segment of an EA of 2007 [3]. In the second stage of selection, a fixed number of 28 households per cluster were selected with an equal probability systematic selection from the newly created household listing. All women age 15–49 and all men age 15–59 who were either permanent residents of the selected households or visitors who stayed in the household the night before the survey were eligible to be interviewed. In all of the selected households, height and weight measurements were collected from children age 0–59 months. Anemia testing was performed on children age 6–59 months whose parent/guardian consented to the testing [3]. Dependent variable – Stunting (ordinal variable which was categorized as severely stunted if a child’s HAZ score less than − 3 SD, moderately stunted (− 3 ≤ HAZ < − 2), not stunted (HAZ ≥ − 2 SD) [4, 27]. Covariates – community level factors: region, enumeration areas (EAs/clusters), residence, community education and community wealth index. Household factors: wealth index, sex of household head, family size, number of under five children in the household, source of drinking water and type of toilet facility. Maternal characteristics: educational level, marital status, body mass index (BMI), maternal height, birth interval and paternal education. Child characteristics: age, sex, type of birth, birth order and anemia level. Improved drinking water source: include piped water, public tap, standpipes, tube wells, boreholes, protected dug wells and spring, rain water, and bottled water [28]. Unimproved drinking water source: include unprotected well, unprotected spring, surface water (river/dam/lake/pond/stream/canal/irrigation channel), tanker truck, cart with small tank and other [28]. Improved toilet facility: include any non-shared toilet of those types: flush/pour flush toilets to piped water systems, septic tanks and, and pit latrines; ventilated improved pit (VIP) latrines, pit latrines with slabs, and composting toilets [28]. Unimproved toilet facility: includes flush to somewhere else, flush do not know where, pit latrine without slab/open pit, no facility/bush/field, bucket toilet, hanging toilet/latrine, others [28]. Anemia level: hemoglobin levels are adjusted for altitude in enumeration areas that are above 1000 m. Not anemic (≥11.0), mild anemic (10.0–10.9), moderate anemic (7.0–9.9), severe anemic (< 7.0) [3]. Body mass index: BMI is calculated by divided weight in kilograms by height in meters square (kg/m2). Women aged 15–49 years who are not pregnant and who have not had a birth in the last 2 months before the survey. Underweight (BMI < 18.5 kg/m2), normal (BMI 18.5–24.9 kg/m2) and over weight (≥25.0 kg/m2) [3]. The length of children aged < 24 months was measured during the survey in a recumbent position to the nearest 0.1 cm using a locally made measuring board (Shorr Board®) with an upright wooden base and moveable headpieces. Children ≥24 months were measured while standing upright. The length/height-for-age Z-score, an indicator of nutritional status, was compared with reference data from the WHO Multicenter Growth Reference Study Group, 2006 [29]. Children whose height-for-age Z-score is < − 2 SD from the median of the WHO reference population are considered stunted (short for their age). The EDHS dataset accessed available at the DHS website; https://www.dhsprogram.com/data/available-datasets.cfm after registering for the dataset access permission the 2016 EDHS data set was accessed through the DHS website; https://dhsprogram.com/data/dataset_admin/login_main.cfm . The kid recode (KR) data set in STATA file was the data set containing the outcome and predictor variables of our study. The data was explored, cleaned, coded, re-categorized and recoded to be suitable for analysis. This study was based on secondary data analysis of 2016 EDHS by adjusting sampling weights. Categorical characteristics and outcome of the study was described in terms of percentage and frequencies. Tables, bar graph and pie chart were used to present the data some selected variables which has significant association with stunting. A bi-variable multi-level ordinal logistic regression analysis was carried out to see the crude effect of each independent variable on stunting, and then variables with p. value of < 0.25 were entered to the multivariable multi-level ordinal logistic regression analysis. The deviance information criterion (DIC) [30] statistic was calculated for the different models (individual level, community level and both individual and community level) fitted with logit, probit and clog log link functions. The DIC was used to evaluate and compare model performance of the full model and the reduced model. A model with lower DIC was considered as one with a better fit. Variance partition coefficient (VPC) [31] median odds ratio (MOR) [31] and proportional change in variance (PCV) statistic were calculated to measure the variation between clusters (the random effect variable). VPC represents the percentage variance explained by higher level (clusters). Hence, it was calculated as below. VPC=σu2σe2+σu2, where σu2 is the between cluster (Enumeration area) variance, σe2=3.29 [31, 32]. Median odds ratio is the median value of the odds ratio between the highest risk and the area at lowest risk when randomly picking out two areas and it was calculated using the formula; MOR=exp2∗σu2×0.75≈exp0.95σu2 [31, 33, 34]. The proportional change in variance (PCV) measures the total variation attributed by the individual level factors and area level in the multilevel model. The PCV is calculated as: PCV=VA−VBVA×100, Where VA = variance of the initial Model, VB = variance of the model with more terms [31, 33, 34]. The data was downloaded after the purpose of the analysis was sent and access permission received confirmation letter that was approved by MEASURE DHS. The original data was collected in confirmation with international and national ethical guidelines. Ethical clearance for the survey was provided by the Ethiopian Public Health Institute (EPHI) Review Board, the National Research Ethics Review Committee (NRERC) at the Ministry of Science and Technology, the Institutional Review Board of ICF Macro International, and the United States Center for Disease Control and Prevention (CDC). The Ethiopian Demographic and Health Survey ensured the principle of respondent’s protection and prevention from unnecessary risk. Verbal informed consent was obtained from participants before data collection began. Participants were informed of their anthropometric measurements (weight, height and edema screening).

N/A

• Dima Health