Application of multilevel models in analyzing on province-level social determinants of obesity in China

Objective To explore province-level social determinants of obesity in Chinese population, using multilevel analysis. Methods Information of individual outcomes and individual-level variables was acquired from data of Chronic Disease Risk Factor Surveillance in China, 2007, while province-level factors were extracted from 7 social indicators, which were obtained from data of National Bureau of Statistics, using factor analysis method. Then, multilevel models were built to calculate the correlations of province-level factors with body mass index (BMI), obesity and central obesity. Results Among the nationwide population aged 18-69 years in 2007, mean BMI level was (23.27±3.37) kg/m², rate of obesity and central obesity was 8.49% and 30.92%, respectively. Two province-level factors were extracted from 7 social indicators, and factor 2, which represented the provincial social and economic comprehensive development level, was found positively correlated with BMI (OR=1.09, 95% CI:1.04-1.10), obesity (OR=1.17, 95% CI:1.07-1.28) and central obesity (OR=1.19, 95% CI:1.10-1.30), while factor 1, which represented the provincial level of residents' consumption and health service, wasn't found statistically correlated with these three outcomes(all P>0.05). Conclusions In China, higher provincial social-economic comprehensive development level might be a risk factor for individual obesity, especially central obesity, which indicated that limited resources for obesity prevention and control may need to be allocated to provinces or regions with higher economic development level. In conclusion, in order to provide more evidence on improving the efficiency of non-communicable diseases (NCDs) prevention and control and help the Chinese policy makers to allocate health resources more reasonably, multilevel models could be used to perform researches on area-level social-economic influencers of NCDs.