Socio-demographic factors associated with underweight and stunting among adolescents in Ethiopia

Introduction Nutrition during adolescence plays an important role in the individual's life. There are different factors that affect nutritional status of adolescents. Socio Economic Status, age, sex and mothers’ educational level are among the important determinants factors of nutritional status of adolescents. Younger adolescents tend to be more undernourished than older adolescents, and, contrary to expectations that boys are almost twice as undernourished as girls. In this study, we test the competing hypothesis about the correlates of nutritional status among Ethiopian adolescents. Methods We report a total of 2084 adolescents from the second round of a 5-year longitudinal study in Jimma zone, southwest Ethiopia. Univariate and Multivariable linear regression were used to assess socio-demographic factors associated with Underweight and Stunting among Adolescents in Jimma zone. Results Age, highest grade completed, job and last attended in community school were positively associated with BMI for Age z-score and highest grade completed, household income and job were positively associated with Height for Age z-score. However, male gender was negatively associated with BMI for Age z-score and male gender, last attended in community school, abdominal pain and household size were negatively associated with Height for Age z-score. Conclusion Age of the adolescents, gender, educational status, employment status and type of last school attended are associated with underweight; and gender, place of residence, household size, household income, educational status, employment status, type of last school attended and abdominal pain are associated with stunted. The findings reflect socio-demographic characteristics are associated with underweight and stunting.


Introduction
The world population is believed to have reached over 6.6 billion [1]. Adolescence is a particularly unique period in life because it is a time of intense physical, psychosocial, and cognitive development.
Nutrition during adolescence plays an important role in the individual's life. Increased nutritional needs to adolescents gain up to 50% of their adult weight, more than 20% of their adult height, and 50% of their adult skeletal mass [2].
There are different factors that affect nutritional status of adolescents. Socio Economic Status, age, sex and mothers' educational level are among the important determinants factors of nutritional status of adolescents [3][4][5][6]. Studies have shown that adolescent women from low economic status households were most affected by malnutrition and the higher the level of education, the lower the proportion of undernourished adolescents women and rural adolescent women are more likely to suffer from chronic energy deficiency than adolescents women in urban areas [7,8].
There are only a handful of studies on adolescent malnutrition in developing countries-there is indicates that younger adolescents tend to be more undernourished than older adolescents, and, contrary to expectations that boys are almost twice as undernourished as girls. In addition, these few studies suggest that there are more undernourished adolescents in Sub-Saharan Africa, and a higher prevalence in rural than in urban areas [9].
In Ethiopia, undernourishment among preschool aged children has been well documented [10], but studies on factors affecting the nutritional status of adolescents have not been studied in detail [11]. The available studies focused on factors affecting pregnancy outcomes rather than on problems associated with normal growth and development of adolescents [12,13]. A recent study on the nutritional status of adolescent girls from rural communities of

Study sample
A community-based cross-sectional study was conducted from 2010 adolescents enrolled in the second round of the five year longitudinal study of adolescents in Jimma zone, Southwest Ethiopia. A census was done to generate the list of all households which gave a sampling frame for random selection of 3,700 households from the total of 5,795 households in the list. A twostage sampling plan was used to select the target sample of adolescents. Households were classified into urban (Jimma City), semi-urban (Serbo, Dedo and Yebbu Towns) and six rural communities in the vicinity of the small towns. At the first stage, households were randomly sampled with the sample size in each kebele determined by the relative proportion of the study population in the kebele and the overall target sample size. In the second stage, one adolescent (a boy or a girl) was randomly selected from each household using a Kish Table [ 14]. Using this sampling strategy a total of 1059 boys and 1025 girls were interviewed in round one. This paper reports on all the 2084 adolescents in the second round of the five year longitudinal family survey of youth.

Measurements
Structured adolescent level questionnaires were used to collect data. The questionnaires were interviewer-administered and translated in to Amharic and Oromifa languages and checked for consistency by another person who speaks both Oromifa and English. The questionnaire focused on issues related to adolescents' experiences of Nutritional status, socio-demographic, health, food insecurity and anthropometric measurements of the adolescent. The interview was conducted in a private place by an interviewer of the same sex.
The interviewers received one week of intensive training prior to the pretest and an additional week of training was given with the final version of the questionnaire before the start of the actual interviews. Supervisors checked the data collection process and filled questionnaires daily to ensure accuracy of the data. The research team supervised the data collection team every week through meetings and checking of the filled questionnaires.
We measured height to the nearest 0.1 cm using a stadiometer (SECA, Hannover Germany) and weight to the nearest 0.1 kg using digital scales (SECA). The level of stunting (height for age z-scores), which is an indicator of chronic malnutrition and wasting (weight for age z-scores), which is another indicator of malnutrition were calculated using WHO Athro-Plus software [15]. Thus, those below -2 standard deviations of the NCHS median reference for height-forage and weight-for-height were defined as stunting and wasting, respectively. recorded. The questionnaire was tested on 200 adolescents (not included in the sample) selected from a community in Jimma city.

Data analysis
The data were double entered, checked for missing values and outliers using SPSS (SPSS Inc. version 16.1, Chicago, Illinois); and analyzed using SAS version 9.2. First, univariate linear regression analyses were conducted. To identify the predictors of Underweight (BMI for Age z-score) and Stunting (Height for Age z-score), multivariable linear regression analysis with stepwise variable selection and AIC criteria procedure was employed and variables that showed a statistically significant (P < 0.05) in the univariate analyses were entered to the multivariable linear regression.
Goodness of fit of the models was assessed using Akaike's information criterion (AIC) and Adjusted-R 2 and Partial regression residual plots showed that all had linear relationship. Normality of the data was assessed using a Q-Q plot and there was no need of transformation; and Outliers and Influential observations were not as such influential and then retained in the final model. Co-linearity between predictor variables were checked using variance inflation factor (VIF); and Interaction and confounding variables were checked. All tests were two-sided and a P < 0.05 was considered statistically significant. We present the results of the linear regression as parameter estimates (ß), P-values and 95% confidence intervals.

Ethical consideration
Informed verbal consent was obtained both from the parents and each adolescent before the interview or measurement. The study obtained ethical clearance from the Ethical Review Board of Jimma University (Ethiopia). The head of the household was interviewed by using the household questionnaire.
We were not obtained written verbal consent, because most of the respondents did not write and read well. We first read the consent paper and explain the objective of the study in front of each respondent and after they agree on the objective of the study and they signed. The ethics committees also approved the consent procedure after we explain the procedures.
In bivariate linear regression models (Table 3), residence in semirural, age, highest grade completed, last attended in community school, household income, adolescent mothers with secondary education and adolescent fathers with secondary education were positively associated with BMI for Age z-score, while residence in rural, male gender and Household size were negatively associated with BMI for Age z-score.
After adjusting for all other variables in the multivariable linear regression model (Table 3), age, highest grade completed, job and last attended in community school were positively associated with BMI for Age z-score. However, male gender was negatively associated with BMI for Age z-score. The effect of residence in semi-urban, residence in rural, household income, household size, adolescent mothers with secondary education and adolescent fathers with secondary education disappeared in the multivariable linear regression model.

Determinants of nutrition status of adolescents in terms of height for age z-scores
In bivariate linear regression models (Table 4), age, highest grade completed, household income, job and adolescent fathers with secondary education were positively associated with BMI for Age zscore, while residence in rural, male gender, last school attended in community school, abdominal pain and household size were negatively associated with the Height for Age z-score.
After adjusting for all other variables in the multivariable linear regression model (Table 4), highest grade completed, household income and job were positively associated with Height for Age zscore, while male gender, last attended in community school, abdominal pain and household size were negatively associated with Height for Age z-score. The effect of adolescents' residence in rural, age of the adolescent, adolescent mothers with secondary education and adolescents fathers with secondary education disappeared in the multivariable linear regression model.

Discussion
Our results showed that, age was positively associated with BMI for Age z-score. An association of age with BMI for Age z-score has previously been reported [5,8]. The risk of underweight was, on average, significantly higher for younger adolescents than older adolescents.
Studies have suggested a positive association between BMI for Age z-score and educational status and employment [7,8,12,16]. Our study found that highest grade completed was positively associated with BMI for Age z-score and Height for Age z-score. They indicated that adolescents who receive even a minimal education are generally more aware than those who have no education of how to utilize available resources for the improvement of their own nutritional status and that of their families. Education may enable to make independent decisions, to be accepted by other household members, and to have greater access to household resources that are important to nutritional status [16].
Our study found that job was positively associated with BMI for Age z-score and Height for Age z-score. The risk of underweight and stunted was significantly high, on average, for unemployed adolescents than employed adolescents in Jimma zone. This finding is consistent with other studies [17]. They indicated that unemployment is a significant factor for chronic energy deficiency in these adolescents as compared with employed adolescents. Our study also found that adolescents attended their last school in community school was positively associated with BMI for Age zscore. Studies have demonstrated the connection between Height for Age z-score and household income [6-8]. Our study also indicated Household income was positively associated with Height for Age z-score.
We found male gender was negatively associated with BMI for Age z-score and Height for Age z-score, as previously reported by others [4]. The risk of underweight was, on average, significantly higher for adolescent males than adolescent females. This may be because of the fact that biological, behavioral, and socio-cultural mechanisms have been proposed for the gender differences in morbidity and mortality. Biologically, female subjects have an advantage for better health and longer survival because of the role of sex hormones in modulating lipid levels and increasing immune response. In addition, the difference in morbidity and mortality between boys and girls is further related to individual lifestyle, the use of health care, and health and illness behaviors and practices. For example, adolescent boys are more likely to smoke and have higher propensities of taking greater risks that expose them to injury.
The result also showed that residence in rural was negatively associated with the Height for Age z-score which is consistent with reports of studies in Ethiopia and Sub-Saharan Africa [8,9,17,18].
The observed urban-rural difference could be an indication of low access and use of health services in the rural areas as compared with urban areas. In general, people living in cities have better health and lower death rates than rural residents, even though the urban poor often live in unsanitary and crowded conditions. Compared with rural residents, urban residents have better access to medical services and are more easily reached by immunization and educational campaign.
We found also household size was negatively associated with the Height for Age z-score, as previously reported by others [3,6]. In addition, abdominal pain and last school attended in community were negatively associated with the Height for Age z-score.
We acknowledge a number of limitations in our study. As the study involved adolescents who are at the different stages of academic status, we used the highest grade completed as a measure of educational attainment that can serve across all age groups.
However, there are other measures of educational attainment that were not captured. The fact that we did not have data from the school records regarding the academic performances is also a limitation of our analysis.

Conclusion
In conclusion, the study revealed age of the adolescents, gender, educational status, employment status and type of last school attended were associated with underweight; and gender, place of residence, household size, household income, educational status, employment status, type of last school attended and abdominal pain were associated with stunted. Thus, underweight and stunting is the reflection of socioeconomic development demanding combination of different types of policies and programs for its solution. To reduce and prevent underweight and stunting Strategies need to involve and more targeted interventions. Promoting gender equality through effective behavior change communications needs to be considered.
Further research will help to understand to identify the most effective strategies for reducing adolescent malnutrition in the study area.

Competing interests
The authors declare no competing interests.

Authors' contributions
The authors' responsibilities were as follows: HA, TB, & LN: Designed and supervised the study and assisted in the analysis and interpretation of the data. All authors critically reviewed the manuscript. HA, the corresponding author did the analysis & drafted the manuscript and had the responsibility to submit the manuscript for publication. All authors read and agreed to the final version of this manuscript.

Acknowledgments
The Jimma Longitudinal Family Survey of Youth was funded by the Packard Foundation, Campton Foundation, National Institute of Health and National Science Foundation. We are extremely grateful to research team members and Prof. Tefera Belachew.