21 ISAAC Phase Three used the ISAAC Phase One standardised core questionnaire on symptoms of asthma, rhinoconjunctivitis and eczema. Phase Three provided an additional opportunity to explore the relationship between lifestyle factors such as vitamin d fast-food consumption and BMI, as heights, weights and information on the frequency of fast-food consumption of participants

were gathered in many centres through an optional environmental questionnaire that was answered by the parents of the children and by the adolescents themselves. The ISAAC Phase One standardised core questionnaire and ISAAC Phase Three environmental questionnaire are on the ISAAC website: isaac.auckland.ac.nz Main outcome variable: BMI Height and weight were reported by the parents of the children, and were self-reported by adolescents. In some centres, each participant’s height and weight were measured objectively; there were no standardised or specific instructions for doing this. BMI was calculated as weight (kg)/height (m)2. Explanatory variables Fast-food consumption was established by asking participants to answer the following question: “In the past 12 months, how often, on

average, did you [your child] eat the following?” ‘Fast-food/Burgers’ were listed as one option along with 14 other foodstuffs including meat, seafood, fruit and vegetables. The participants were asked to categorise their intake of each listed food as “Never or only occasionally”; “once or twice per week”; or “Three or more times a week.” These responses were categorised as ‘infrequent’, ‘frequent’ and ‘very frequent’. Each variable was examined separately for both age groups. Country Gross National Index (GNI) was based on the 2006 World Bank of Gross National Income by country. The World Bank categories of high-income, high middle income, low middle income and low-income countries were dichotomised into ‘high-income’ (high and high middle income) and ‘low-income’ (low middle and low income) categories. Participants For Anacetrapib children

aged 6–7 years, data were submitted from 73 centres in 32 countries (214 706 participants). For adolescents aged 13–14 years, data were submitted from 122 centres in 53 countries (362 019 participants). Centres that provided data on height, weight and fast-food consumption for at least 70% of participants were included in our analyses. Individuals without complete age, sex, fast-food consumption, height or weight data were excluded. Data cleaning To eliminate likely erroneous BMI data, we applied the following thresholds: For children in each centre, those in the top and bottom 0.5% of weights and heights, and those with heights less than 1 m were excluded. Children with a BMI less than 9 kg/m2 and greater than 40 kg/m2 were excluded.