A list of all the students attending the 8th and 9th grades in 2004-2005 was provided by the Palestinian Ministry of Education and Higher Education (which included number of students per classroom). The school enrollment rate for Palestinian children in the West Bank in 2007/2008 was 84.5% in the basic education stage (1st to the 9th grade) . The total number of 8th and 9th grade students was 13011 in Ramallah and 26681 in Hebron. The classes were divided into nine strata using school gender category (single sex-boy, single sex-girl and co-educational) and school ownership type (public, United Nation Relief and Works Agency (UNRWA), or private). The sample was selected using single stage probability proportional-to-size sampling procedure from each of the nine strata, using the class as the primary sampling unit.
A sample size of 1000 students in each governorate was set to have a maximum margin of error of 2.9% in the prevalence estimates of overweight/obesity using the following formula:
where p is the proportion of overweight/obesity (in this case 18%, based on a similar study from the North Gaza strip) ; DE is the design effect that is estimated at 1.5 based on previous studies in schools; the value 1.96, assures a confidence level of 95%; and E is the expected maximum margin of error.
As the average class size was about 31 students in Ramallah and 34 students in Hebron, 65 classes were selected: 34 in Ramallah and 31 in Hebron. All students in the chosen classes were invited to participate in the study.
The cross-sectional survey was conducted between March and May 2005. Self administered questionnaires were used to collect information from students and parents. The researcher and trained field workers gave students standardized instructions on filling in the questionnaires, which were completed in the classroom in 1.5-2 hours. Field workers were present during administering the questionnaires to clarify questions when needed. The students' questionnaire included questions about age, residence, household amenities and onset of puberty. Seven core questions in the students' questionnaire were adapted from the World Health Organization (WHO)-questionnaire on Health Behaviour in School-aged children (HBSC) ; these were: month and year of birth, onset of puberty, ownership of car, computer and Internet connection at home. The parents' questionnaire included household information such as family size, parents' education, and indicators of economic situation such as the ability to pay bills, cover family needs and having any debts. The parents' questionnaire was taken home by students and returned to school the following day. The students' and parents' questionnaires were piloted and adjusted before the survey. The students' questionnaire was tested for reliability (one week test-retest) on a different sample of 115 students in the same age group.
Students' weights were measured in light clothes without shoes to the nearest 0.1 kg using a portable scale (Seca 780/783, Hamburg, Germany). The scales were calibrated daily at the study site with a known weight. Heights were measured to the nearest 0.1 cm using a portable stadiometer (Seca 220) mounted on the scale. Students stood up straight in bare feet, with heels, buttocks and back touching the stadiometer. Measurements were taken once and recorded by one trained field worker during the early morning classes before the school recess for all students. Intensive training was conducted before the field work to ensure reliability of the measurements and decrease inter- and intra field workers' errors. BMI was calculated as weight in kilograms divided by height in meters squared.
For the purpose of regional comparisons, BMI was categorized based on age-and sex-specific cut-off values of the 2000 Centers for Disease Control and Prevention (CDC) growth charts. The categories were underweight (< 5th percentile), normal weight (5th to 85th percentile), overweight (85th to 95th percentile), and obese (> 95th percentile) .
For the purpose of international comparisons, BMI was also categorized using the IOTF criteria. Overweight and obesity were defined as BMI cut offs corresponding to adult cut-offs of overweight and obesity (BMI of 25 and 30 kg/m2, respectively) .
Stunting was defined as height-for-age below the 3rd percentile .
Age of the students was obtained from the date of birth as reported by the students. Three age groups were constructed: age group 13: 13 years 0 months to 13 years 12 months, age group 14: 14 years 0 months to 14 years 12 months, and age group 15: 15 years 0 months to 15 years 12 months.
Urban/rural classification was based on the Palestinian Central Bureau of Statistics (PCBS) classification which is based on services and population size .
Father's educational level was divided into 3 categories as follows: "Low": illiteracy, less than secondary school education, "Medium": secondary school education" and "High": college, university or higher.
Mother's educational level was constructed in the same way as the father's, but because of few subjects in the high educational category, the last two categories were collapsed to one, that is, medium/high.
The household standard of living (STL) index was based on household possessions. Ownership of central heating, family car, family mobile phone, personal mobile phone, indoor bathroom, water pipes, refrigerator, full automatic washing machine, color TV, satellite, video, computer, dish washer, microwave, vacuum cleaner and internet connection were summed; each item was given a value of 1. Three categories were constructed: 'low': 0-6, 'medium': 7-10, 'high':11-16. The 1 week test-retest of the ownership of these amenities showed consistent answers ranging between 86.1% and 99.1%.
The present household economic situation index was constructed based on 4 questions from the parents' questionnaire using a 5-point Likert Scale where 1 represented the best situation and 5 the worst. These questions were: "Do you have enough money to cover family needs now?", "At the present time, do you have any debts?", "At the present time, to what extent do you have to postpone paying bills?", and "During the past school year, did you have to borrow money to cover family needs?" A total index was formulated by adding up these values. Three categories were constructed: 'worse off': 14-20; 'medium': 7-13; 'better off': < 7. Spearman correlation between this index and STL was 0.42 (p < 0.001).
Food availability was assessed based on the following question from the parents' questionnaire: "Is there enough food for all family members on daily basis now?" Food was considered available if the family had enough food for all members "most of the time" or "always."
Onset of puberty
Onset of puberty was assessed by the question "have you reached puberty i.e. "have you had your first period?" for girls and "have you noticed deepening of your voice?" for boys. The 1 week test-retest for the onset of puberty showed consist answers in 92.3% in boys and 95.2% in girls.
EPI 2002 (Center for Disease Control and Prevention, Atlanta, Georgia) was used to calculate BMI, BMI and height percentiles. All other statistical analyses were done using Stata 10.0 (Stata Corporation, College Station, TX) and were adjusted for cluster sampling design. The sample was weighed according to sample and population size (inverse of sampling probability) in each grade and in each governorate. χ2 tests were used to compare frequencies and t-tests were used to compare means. A significance level of 0.05 was used for all statistical analyses. Separate analyses were run for boys and girls in each governorate. Multivariate logistic regression analysis was used to model the association between the outcome variables (overweight/obesity), (underweight) and (stunting) and selected sociodemographic characteristics. As no interactions were found between region (Ramallah/Hebron) and the rest of the independent variables in any of the models, regression analysis was done for both regions together.
Age, region and variables with p < 0.05 in the bivariate analysis tables for each gender in any of the two governorates or in both of them combined were included in the regression models. Colinearity and plausible interactions between variables in the model were investigated.
Informed consent was obtained from parents, students and school principals with standard assurances of confidentiality. They were informed that answering the anonymous questionnaire was voluntary. Moreover, the class teacher was not present while administering the questionnaires to assure confidentiality. The study was a collaboration between the Institute of General Practice and Community Medicine at the University of Oslo and the Institute of Community and Public Health at Birzeit University. The protocol was approved by both institutes. The study was also approved by the Palestinian Ministry of Education and Higher Education, by UNRWA Office of Education and by Regional Ethical Committee of Norway.