A community-based cross-sectional study was conducted from 25 August to 20 September 2020, among children aged 6–59 months, 6 months after discharge from an outpatient therapeutic program in Habro Woreda, Oromia, Eastern Ethiopia. Habro Woreda is located about 320 km east of Addis Ababa, the capital city of Ethiopia. Habro Woreda has 1 general hospital, 7 type B functional health centers, and 32 functional health posts. In 2020, there were approximately 44,240 under-five children and 55,765 households in the region, with a total population size of 269,272. According to a projection by the Ethiopian Central Statistical Agency in 2011, the population was 190,455 (17). With a mostly rural population (88%), the area is characterized by a high population density (242/km²), fragmented farmland ownership, and limited income-generating opportunities. However, the major source of income was cash crops, mainly focused on chat cultivating and chat trading. We used the strengthening the reporting of observational studies in epidemiology (STROBE) cross-sectional checklist for this study (25).

All children aged 6–59 months who lived in the Habro district served as the study population. Mothers who had resided for at least 6 months in the study area and had a child aged 6–59 months were included. However, children with evidence of physical impairments, serious illness or mental impairments, children whose names differed from those present on the outpatient therapeutic program registration logbook, children who had left the selected study area, and mothers or caregivers who were unable to communicate were excluded from the study.

A questionnaire was developed after reviewing related studies in different literature, with some questions adopted from the related studies and contextualized to align with the study objectives and data extraction form accordingly. Data were collected using a pretested and structured interviewer-administered questionnaire to collect data on socio-demographic characteristics, environmental hygiene and sanitation characteristics, and childcare practice and health characteristics, which comprised the three survey components. The wording and sequence of questions were designed in such a way that the sequence of ideas (from general to specific and from easy to difficult questions) was maintained. Standard anthropometric measurements were taken from all subjects during data collection. We use the Waterlow height-for-age (%) method of classification (26). Length was measured for children who were < 87 cm or those too weak to stand in supine (lying down). However, if the children were more than 87 cm in height, their height was recorded using a stadiometer to the nearest 1 cm. To ensure the quality of data, seven skilled health professionals who had completed BSc Nursing and one supervisor who was proficient in both Afaan Oromo and Amharic collected the data, and the task was closely supervised by the investigators. Data collectors and supervisors received 4 days of training in the data-gathering processes. Initially, we translated the questions while keeping the purpose of the questionnaire and the intent of the questions in mind. It was translated by group members who were bilingual experts. To ensure the accuracy of the translation, the questionnaire was translated back into English by someone who had not seen the original version and was unfamiliar with the questionnaire's context. The back-translated version was then compared with the original, and any differences in meaning were corrected. Then, to ensure cross-validity, we attempted to interview a set of respondents in English and another set in the local language, such as Afaan Oromo and Amharic; their responses were then compared to identify differences in understanding. Finally, pretesting was conducted to identify questions that were poorly understood, ambiguous, or elicited hostile or other undesirable responses. We attempted to conduct a pretest using the already translated questionnaire. We tried to implement all the steps in pretesting, such as obtaining an evaluation of a questionnaire and testing the revised questionnaire through its paces with friends and colleagues. We statistically calculated Cronbach's alpha, which is a measure used to assess the quality of the instruments we used. The outcome was 0.87, which was within acceptable ranges.

Wasting: Wasting is defined as the child's weight-for-height Z-score (WHZ) < -2 SD from the median WHO reference values; WHZ ≥-3SD and < -2SD indicates moderate wasting; and WHZ < -3SD indicates severe wasting (19, 27).

Undernutrition: A mid-upper arm circumference (MUAC) below 12.5 cm indicates acute undernutrition; MUAC ≥ 11.5 cm and < 12.5 cm indicates moderate acute undernutrition; and MUAC < 11.5 cm indicates severe acute undernutrition (19, 27).

Household food security: It is measured by whether the respondent worries that the household is lacking enough food for the past 4 weeks (4, 19).

Food secure: If the respondent does not worry about the household lacking enough food for the past few weeks (4, 19).

Food insecurity: The lack of regular access to enough safe and nutritious food for normal growth and development and active and healthy life. This may be due to the unavailability of food and/or a lack of resources to obtain food (4, 19, 28).

Mildly food insecure: It is when a respondent rarely worries about food (once or twice in the past 4 weeks) (4, 19, 28).

Moderately food insecure: It is when a respondent sometimes worries about food (3 to 10 times in the past 4 weeks) (4, 19, 28).

Severely food insecure: It is when a respondent often worries about food (more than 10 times in the past 4 weeks) (4, 19, 28).

Household dietary diversity: Food intake that includes all of the diversified dietary needs of the organism in the correct proportion (29).

Diarrhea: It is defined as having three or more loose, watery stools in 24 h in the 2 weeks before the survey (30).

Improved drinking water: Drinking water that is obtained from piped water into dwellings, public taps, protected dug wells, protected springs, and rainwater (29).

Wasting: A nutritional deficiency condition of recent onset related to sudden food deprivation or malabsorption of nutrients, resulting in weight loss and weight-for-height below −2 SD, but resolved upon treatment (31).

Moderate wasting: Weight-for-height below −2 Z-score of children under 5 years of age (32).

Severe wasting: Weight-for-height or length < -3 Z-score of children under 5 years of age (31).

There were a number of biases involved while conducting this research, and the researcher took explicit measures to avoid them. One of the biases was social desirability and recall bias, and to avoid it, the researcher paraphrased the questions in a way that was not socially desirable and made an attempt to provide a clue for the intended question that needed to be answered.

The sample size was determined by using the single population proportion formula, considering the following assumptions:

where n = minimum sample size required for the study, Z = standard normal distribution (Z = 1.96), with a confidence interval of 95% and α= 0.05, p = proportion of wasting taken from proportion of wasting taken from the Ethiopian Demographic Health Survey, which is 10% (33), d = maximum acceptable margin of sampling error (d) = 5% = 0.05, and design effect = 1.5. Therefore, the final total sample size, including the contingency (10%) of the non-response rate, was 188. The sample size for the second objective (factors) was also determined by using the double population proportion formula for a cross-sectional study by considering the following assumptions: power = 80%, CI = 95%, and ratio = 1:1. The largest sample size from those samples was taken and the final sample size was calculated to be 306 children paired with their mothers or caregivers.

In this study, a three-stage sampling strategy was followed. In the first stage, four health centers and three health posts were randomly selected and included. The number of study units to be sampled from each health facility was determined using proportional allocation to size based on the number of client flows in the current year (2019), ensuring equal weighting despite varying numbers of health centers and posts (34). Children aged 6–59 months who had recovered from the OTP program (November 2019–February 2020) were recruited from the OTP registration logbook. Then, the systematic random sampling method was employed to select each study subject. The sampling fraction of the Kth interval was calculated as (N/n = 640/306 = 2.1~.2). Based on the client's registration logbook number, the starting participant was determined by using the lottery method, and every second discharged child was included until the required sample was achieved (33). The eligible participants who did not make themselves available during the data collection period in the selected Kebele were revisited up to three times, and if they did not make themselves available after three visits, the data collector skipped that household and interviewed the next household as a substitute. Finally, a total of 306 study samples were selected using a systematic sampling procedure. Then, each patient was given a unique code.

A multilingual specialist (Afaan Oromo and Amharic) created the questionnaire first in English and then translated it into the local languages. To maintain uniformity, it was then translated back into English. Data collectors and supervisor were trained on how to use the data-gathering instrument and processes. A pretest was conducted with 5% of the research participants under comparable conditions prior to the real data collection. Regular supervision was provided by investigators and competent research supervisors.

First, the obtained data were reviewed for completeness and consistency. They were then cleaned, coded, and entered into EpiData version 3.1 for further analysis. The entered data was transferred to SPSS version 25 for analysis. Descriptive and summary statistics were collected and reported using frequency tables and figures. A binary logistic regression model was constructed to test for a connection between independent factors and the outcome variable. Hosmer–Lemeshow statistics and Omnibus tests were used to evaluate the model's fitness. To determine the genuine determinants of the outcome variables, a multivariable analysis was performed. Using the standard error and co-linearity statistics, a multi-collinearity test was performed to assess for the presence of correlation between independent variables, and no collinearity effects were observed. As a result, the variance inflation factor (VIF) was 0.951. The odds ratio (OR) and 95% confidence interval (CI) were used to determine the direction and degree of the statistical connection. In all bivariable and multivariable analyses, a p-value of 0.05 was considered statistically significant.

In this study, a total of 306 children aged 6–59 months were included. Out of the total study participants, 156 (51%) were boys, and the mean age of the study children and their mothers was 30 (SD+6.5) months and 29.4 (SD+10.8) years, respectively. Regarding religion, most of the caregivers 278 (90.8%) were Muslim. Of the participants, 138 (45.1%) mothers and caregivers of children belong to an age group of 21–29 years. With regard to educational status, most of the mothers or caregivers 217 (70.9%) were illiterate. The majority 288 (94.1) of the caregivers were housewives (Table 1).

Among the total households, 195 (63.7%) had a latrine, and of those households that had a latrine, only 19 (9.7%) of them had a handwashing facility. A total of 215 (70.3%) households had access to improved sources of drinking water, 209 (68.3%) households lacked proper waste disposal facilities, and 39 (12.7%) households had no separate kitchens (Table 2).