Level of Overweight, Obesity and Associated Factors Among Preschool Children in Dire Dawa City, Eastern Ethiopia, 2016

Background : Obesity is accumulation of excess fat that have negative consequences on health. In 21 st century overweight and obesity of children is becoming burning issue. About 42 million under five children are overweight and obese in 2013. Close to 31 million of these are living in developing countries. The objective of this study was to assess the level of overweight and obesity and associated factors among preschool children aged 3-5years in Dire Dawa city, Ethiopia, 2016. Methodology: A community based cross sectional study design was used from March 1-15 2016. Simple random sampling followed by systematic random sampling technique was used to select study participants. A pretested structured questionnaire was used to collect data from total samples of 627 preschool age children with their respective mothers or caretakers. Weight and height were measured using standardized and calibrated equipment. Finally Binary logistic regression analysis was done to identify factors associated with overweight and obesity. Level of statistical significance was declared at p- value less than 0.05. Results : The magnitude of overweight and obesity in the study area were 11.4% and 5.5% respectively. Female sex (AOR = 2.67, 95% CI: 1.13 - 6.32), Mother BMI (AOR = 16.5, 95% CI: 6.80 - 40.29, p value < 0.001) and Father Education (AOR =7.179, 95% CI: 1.06-48.6, p value <0.05) were among factors associated with preschool children overweight and obesity. is 16.9%. So as Government of Ethiopia continues in its ﬁght towards reducing under nutrition, it should also expand or modify these efforts to include ways that may help prevent and reduce prevalence of overweight within the community. multivariable logistic regression analysis. Level of statistical significance was declared at P-value < 0.05.

Other issue for obesity of preschool age children attention is issue of adiposity rebound. BMI or adiposity rebound refers to a period, usually between 4 and 7 years of age, when BMI reaches a nadir and then begins to increase throughout the rest of childhood, adolescence, and young adulthood. Even once obesity is established in children (as in adults) it is hard to reverse (De Onis, Blo¨ssner et al. 2010). Hence obese child is more likely to be obese adult as compared to normal child; intervention during early life stage is beneficiary (Daniels, Arnett et al. 2005).
In Ethiopia, most studies were dealing with under nutrition than over nutrition due to the major problem of the country was under nutrition. But now days due to nutrition transition over nutrition is becoming public problem in urban areas. So that over burden the country both with over nutrition and under nutrition.

Study Design, Sample
Population and sample size determination: community based cross-sectional study design was used. Randomly selected children aged 3-5years and their mothers or care takers in randomly selected kebele's of Dire Dawa City Administration were study subjects. Mothers or care takers who have been living permanently at least for six months in the study area having Preschool children aged 3-5years were included in this study whereas children with chronic illness, visible physical deformity and edema were excluded from this study as such factors may distort anthropometric measurements.
Sample size for the first objective (prevalence of overweight and obesity) was estimated based on single population proportion formula (n = (Z /2) 2 pq/d2) considering confidence level at 95% =1.96, margin of error (d) = 0.05 and p=0.073 from previous study done in Hawassa city in 2014, (Wolde and Belachew 2014). Sample size for the second objective (associated factors) was calculated by using Epi-info 7 statistical software considering different factors associated with children overweight and obesity in previous study. Finally when we compare sample size calculated for first objective (120) and for second objective (418) the second one is the largest and considering 1.5 design effects on the largest sample size, the final sample 627 was studied.

Sampling Procedure and Data Collection:
First three kebeles (kebele refers to the lowest administrative unit) in Dire Dawa city were selected randomly by lottery method. Then study subjects were allocated to selected kebeles proportion to population size based on number of preschool age children found in each kebeles. Finally systematic random sampling was used to reach at households containing eligible study subjects using house numbers. The first house containing eligible subject was obtained by Lottery method. Data was collected by using face-to-face interviewer administered questionnaire. The data collectors were trained 3 Diploma Nurses and 3 Health extension workers mean while supervision was done by 3 Health Officers. Anthropometric measurements were taken by standardized and calibrated equipments. Height was measured making the child barefoot, in light clothing and remove any pins and braids from the hair that could affect the measurement. Height then recorded to the nearest 0.1cm and positioning the subject at the Frankfurt plane using a stadiometer seca (Germany). Weight measured children with light clothing and recorded to the nearest 0.1 kg using UNICEF seca digital weighing scale (Germany) (Gibson, 2005). Age of the child was asked from both the child's date of birth and age on the day measured, and then the mothers or care takers were probed to report the correct age. Finally the age calculated by subtracting date of birth from date of data collection (WHO 2009). Finally WHO 2006 growth reference standards, was used to transform children's weight and height measurements into sex and age specific Z-score (i.e. BMI-forage Z-score BAZ)

Operational Definitions
Non Overweight and non obese was defined as proportion of preschool children with values Body mass Index for age Z score (BAZ) < 2 Standard Deviation (SDs). Over weight and obese was defined as proportion of preschool children with values BAZ ≥2SDs. Maternal BMI can be classified based on international cut-off point as follows. Underweight was defined as BMI<18.5 kg/m2 normal body weight was defined as BMI =18.5-24.9 kg/m2, overweight as defined as BMI =25.0-29.9 kg/m2 and obese was defined as BMI ≥30.0 kg/m (WHO 2006). Physical activity in children was defined as playing out door games ≥ 1hour per day. Screen time was defined as watching TV, DVD or playing computer game ≥ 2hours per a day (Barlow, 2007).
Dietary diversity score (DDS) was categorized as high, medium and low for score of >6 DDS, 3-5 DDS and < 3 DDS respectively (here giving score of 1 for Yes and 0 for No to each question of DDS) (Wolde and Belachew 2014).

Data Quality Control
: the questionnaire was translated into the local languages. Two days of training was provided to the data collectors and supervisors. The questionnaire was pretested on 5% of the sample size. Data collectors were closely supervised. Completeness of each questionnaire was checked. All anthropometric measurements were measured by three trained Diploma nurses and Relative TEM calculated to check intra observer variability. Weight scale was calibrated to zero level with no object on it and placed in level surface before measurement performed. The scales of measurements were checked continuously for their reliability. Finally, multivariate analysis was conducted to control the confounding factors.

Data Processing and Analysis:
The data were checked for completeness and consistency before entered to computer. Then it was coded and entered in the computer using EPI-DATA 3.1 software and then sex, age, height and weight transferred with ID number to WHO Anthros version 3.1.0 software to convert nutritional data into Zscores of the indices; BAZ using new WHO growth standard reference. Then entered data were transformed to SPSS version 20 for further analysis and descriptive summary like frequency, proportion, cross tabs and graphical presentation.
Initially the analysis of data was done by using bivariate logistic Regression to determine association between dependent variable and predictors. Here the association was checked by crude odds ratio and 95% CI. Then significant variable (p value ≤ 0.3) obtained by bivariate analysis included in multivariable analysis. Hosmerlemeshow test (>0.05) and multicollinearity checked (VIF<3) before proceeding multivariate analysis. Adjusted Odd ratios, along with 95%CI, were estimated to identify factors associated with overweight and obesity using multivariable logistic regression analysis. Level of statistical significance was declared at P-value < 0.05.

Ethical considerations
The study was reviewed and approved by the Institutional Health Research Ethics Review Committee (IHRERC) of Haramaya University of Ethiopia. Informed written consent was obtained from parents or caregivers. Child assent was taken for anthropometric measurements. Confidentiality of information collected from each study participant was not disclosed. Health education and medical advice was given for family as well as obese and overweight child.

RESULTS
In this study data were collected from 590 children aged 3-5years and their respective mothers/care givers with response rate of 94%. Thus the collected data were used to determine the prevalence and factors associated with overweight and obesity in the study area.

Socio-economic and demographic characteristics of study participants
Among study participants 40.2% were Muslims, 28% were orthodox and 14.2 % were protestant by their religion. Majorities of mothers were house wife (70%) followed by merchant (13.4%) and government employ (11.2%) whereas the majorities of fathers were merchant (34.9%) followed by government employ (30.9%) and farmers (19.2%) respectively. About 48% fathers and 31.4% mothers joined collage/ University and above whereas about 5.6% mothers and 3.9% mothers were unable to read and write (see Table 1).

Child feeding practices
In this study only about 84% of the children were exclusively breastfed among whom 57.5%, 40.9% and 1.6% fed for the first four months, for the first six months and greater than six months respectively. Altogether 52.5% continued BF for <12months, 28% for 12-18 months and 3.6% for 19-24 months. Regarding infant formula feeding about 63.2% fed formula in which 25.2% of them started during the early infancy 0-3 months while 40.2% and 34.6% started during 4-6 months and >6 months respectively.(

Dietary diversity score by preschool children
The Dietary Diversity Score of preschool children in the study area were reported by mothers/care takers in the last 24 hours. Most Fruits and vegetables were available in the market in the area from March to April and October to February respectively. The minimum and maximum DDS were 1 and 8 respectively with mean (±SD) score of 3.57±1.63. Foods from grains, tubers and roots were consumed by majorities (67.6%) whereas meat, poultry, fish and sea foods were least consumed (32.7%) food groups in this study (Table 3)

Sedentary behavior and physical activities of preschool children
As described in the table below, sedentary behaviors of preschool children were assessed by asking screen time (TV, DVD and computer use). In this study majorities (66.9 %) view television among whom only 28.1% view greater than or equal to 2hours per day. Children who play computer and mobile game constitutes 39% of total sample, among whom only 23% played greater than or equal to 2 hours per day. Regarding physical activities, the majorities about 91% play outdoor games, among whom 36.9% play less than 1 hour per day. (Table 5)

Factors associated with overweight and obesity
In this study, variables that are linked to overweight and obesity the first step analysis were child sex, monthly income, maternal BMI, father education, mother occupation, DDS, child age in months, duration of EBF, screen time and maternal estimation of her child weight status. In the second step analysis, all variables that show association in bivariate analysis were included in multivariable analysis. (Table 6) There was a significant association between overweight & obesity and child sex. Female preschool children were 2.67 times more likely to be overweight and obese as compared to male (AOR = 2.67, 95% CI: 1.13 -6.32, p value < 0.05). Similarly strong association was seen between preschool children overweight & obesity and maternal body mass index. Children from overweight and obese mothers were 16.5 times more likely to be overweight and obese as compared to children from non overweight and non obese mothers (AOR = 16.5, 95% CI: 6.80 -40.29, p value < 0.001) ( Table 6).
Father educational status was also associated with child overweight and obesity. The odds of being overweight and obese was 7 times higher for children whose father was less than or equal to grade four compared to children whose father have completed collage/University (AOR =7.179, 95% CI: 1.06-48.6, p value <0.05). Duration of exclusive breast feeding was also another factor linked to preschool children overweight and obesity. The odds of being overweight and obese was 2.4 times higher for children who fed only for the first 4 months compared to children who fed for the first 6 months (AOR =2.4, CI: 1.02-5.74, p <0.05) ( Table 6). The prevalence of overweight and obesity in Basrah (Iraq) were 7.6% and 3.6% respectively which is lower than the finding of this study (Musa and Hassan 2010). One reason may be time gap between this study and the previous study and there may be good prevention strategies against childhood overweight and obesity in middle income countries than SSA country like Ethiopia.
When compared to the national data of Ethiopia, the current prevalence of overweight and obesity was much higher than EDHS 2011 report. Overall, 2 percent of children below age five years were overweight or obese. But for Addis Ababa, 6 percent of children under five, the highest percentage in all regions, were overweight or obese. Generally the prevalence variation may be attributed to fast socioeconomic development in the country.
According to this study the prevalence of overweight and obesity were 8.5% &4.4% for boys and 13.9% & 6.5% for girls respectively. And this disparity was statistically significant where girls were 2.67 times more likely to be overweight and obese as compared to boys (AOR = 2.67, 95% CI: 1.13 -6.32, p value < 0.05). Similar results have been reported from Ghana, girls were more likely to be overweight and obese as compared to boys (p value < 0.001) (Mohammed and Vuvor 2012), additionally from Egypt about 6.1% girls and 4.7% boys were obese and this was also shown statistically significant (AOR = 1.32, p value < 0.001) (Safiya, Shaker et al. 2014).
In contrast, studies from Iran showed that the prevalence of overweight was higher in boys (23.6%) than girls (19.3%) and this difference was statistically significant (p=0.001) (Tabesh, Hosseiny et al. 2014). In addition to this a pocket study in Hawassa city, South Ethiopia also reported that the prevalence of overweight and obesity was higher among boys 9% and 3.4% while in girls 5.5% and 3.3%,respectively (Wolde and Belachew 2014). But when compared to the present study prevalence of overweight and obesity of both sex reported in Hawassa city was smaller than this result. The variation seen in these studies may be attributed to difference in screen time (TV, DVD) of participants. In this study about 39% girls and 19% boys had screen time greater than 2hrs per day. Other possible explanation may be time gap between this study and the previous.
In this study maternal BMI was associated with preschools' children overweight and obesity. The odds of being overweight and obese was 16 times more likely for children from overweight and obese mothers compared to children from non overweight and non obese mothers (AOR = 16.5, 95% CI: 6.80 -40.29, p value < 0.001). Similar result have been reported from China which found that maternal BMI was significantly associated with offspring elevated BMI (β = 0.134, P = .002) (McCarthy, Ye et al. 2015). In addition study from America also stated that there was strong association between maternal BMI and child elevated weight status than father BMI (Whitaker, Jarvis et al. 2010).
In the present study father educational status was significantly associated with preschool children overweight and obesity (p < 0.05). About 27% overweight and obese children belongs to father who learned up to grade four, while only about 7.2% overweight and obese children belongs to Collage/University completed fathers. Similar result have been reported from Ghana, increase in parental educational level was associated with decrease in overweight/obesity of children (B=-0.242, p-value=0.043) (Mohammed and Vuvor 2012). The possible explanation may be increased paternal educational status has been associated with enhanced acquisition and use of health knowledge and health services, affects preference to child health and family size as well as healthy life styles.
Conversely, study from Western Algeria stated that higher parental education is predictor of childhood overweight (p<0.001) (Saker, Merzouk et al. 2011) and from Egypt More than half of obese children belongs to university educated fathers (54.9%) while about 60.5% normal children belongs to pre University educated fathers (Safiya, Shaker et al. 2014). This might be attributed to cultural influence posed on fathers in these countries to make decision on child health issues.
Regarding breast feeding, duration of Exclusive breastfeeding was one of the factors associated with preschool children overweight and obesity. Children who exclusively breastfed only for the first 4 months were more likely to be overweight and obese, than those who exclusively breastfed for the first 6 months (p <0.05). In support to this finding, a review of about 9000 articles in England reports, breastfeeding had a protective effect against obesity (Ip, Chung et al. 2007). The reason might be the smaller the duration of EBF the earlier the children could start either formula or complementary feeding which may be then a risk for childhood overweight and obesity.
This study is limited by the fact that Body Mass Index was used to measure body fat in these children. The waist circumference or Waist/Hip ratios are better tools to measure the health impact of excess fat. Food frequency questionnaires rely on memory of the respondents and were not validated. Furthermore this study had also strength like taking anthropometric measurement twice, using TEM to identify intra observer variability of measurement errors and probing the respondents to get valid information.
In conclusion the prevalence of overweight and obesity was 11.4% & 5.5% respectively. The factors that have been identified as predictors of preschool children overweight and obesity were child sex, mother BMI, father education and duration of exclusive breast feeding. So that the government should expand or modify efforts to include ways that may help prevent and reduce prevalence of overweight within the community.

LIST OF ABBREVIATION 6. Declaration
Ethics approval and consent to participate: The study was reviewed and approved by the Health Research Ethics Review Committee of Haramaya University of Ethiopia. Informed written consent was obtained from parents or caregivers. Confidentiality of information collected from each study participant was not disclosed. Consent to publish: All authors have read the content of the manuscript and had common understanding regarding the finding of the study. They also have agreement to publish it under this journal. Availability of data and materials: all data regarding this study is available and it can be send up on request. Competing interests: The authors also declare that there are no any other financial or non-financial competing interests. Funding: The author did not receive payments, funding, or salary from any organization in relation to the work and publication of this paper. There is no any organization affected positively or negatively by the publication of this paper. Author's contribution: BE and KT guide proposal preparation, data collection and assists final report writing. TT conceptualized the research question, analyzed and interpreted the findings. All Authors have read the manuscript and had common understanding regarding the study.

Acknowledgement
We are grateful to Dire Dawa city Administration along with dwellers of Addis Ketema, Sabian and Laga Harre kebele's and all our study participants.