Stata 12.1 version

The STATA 12.1 version software was used for data analysis. Participants who had a healthy response to all questions were classified as having a good eating habit. We further analysed and presented some details about eating habits of the adolescents as to whether they were eaten breakfast, snack after breakfast, lunch, snack after lunch, supper, snack after supper, eating at the right time, and frequency of the intake of various food groups and products. Again, twenty questions were used to evaluate adolescents' nutritional knowledge, where each correct response was rated 1 point and a wrong response was rated zero. Participants' overall nutritional knowledge was categorized using modified Bloom's cut-off point, as high if the score was between 80 and 100% (16–20 points), moderate if the score was between 50 and 79% (10–15 points), and poor if the score was less than 50% (<10 points). Therefore, sixteen or more points were classified as high knowledge, between ten and fifteen points were rated as moderate knowledge, and fewer than 10 points were regarded as low knowledge. Descriptive and inferential statistics comprising frequency, percentage, chi-square, and logistic regression were employed in analysing the data. All statistical analyses were considered significant at p value <0.05.

Bivariate and multivariate techniques were used to analyze the association between childhood overweight/obesity and sociodemographic, household, and dietary characteristics. The data were also examined using descriptive statistics. We first calculated the proportion of overweight/obese children and that of normal children. The frequency and percentage of the study variable were determined using descriptive statistics as the next step. Pearson’s chi-square tests were used in bivariate analysis to determine the sociodemographic and household characteristics associated with the prevalence of overweight/obesity and the significant level across the independent variables. Binary logistic regression models were used to assess the unadjusted risk ratio (URR) and adjusted risk ratio (ARR) with 95% confidence intervals (C.I.s) between childhood overweight/obesity and sociodemographic and household characteristics. The ARR was controlled for the sex of the child, the child’s age, currently breastfeeding, birth rank, mother’s educational level, age at marriage, mother’s BMI, place of residence, region, social group, religious beliefs, wealth quintile, and dietary diversity score. Data analyses were executed with STATA 12.1 version (StataCorp L.P., Lakeway Drive, College Station, TX, USA).