LAUSR

OBJECTIVES The objective of this study to determine whether body mass index (BMI) in different genders and age groups need different thresholds when predicting chronic non-communicable diseases (CNCDs). STUDY DESIGN This is a cross-sectional study. METHODS Data were obtained from the China Health and Nutrition Survey conducted in 2009. Sequential sample cluster analysis was used to group age according to BMI. Propensity score matching was used to eliminate the influence of age. Receiver operating characteristic curve based on gender and age group was used to evaluate the cut-off values and efficiency of BMI in each group. RESULTS A total of 8469 individuals were enrolled in this study. Results of sequential sample cluster analyses showed age was divided into three groups: 18-39, 40-59 and 60-99 years. There were significant differences in the distribution of BMI among the three groups for both males and females (P < 0.001). Statistical differences were observed in the distribution of BMI between genders in the 18-39- and 60-99-year-old age groups (P < 0.001). For men, the cut-off values of BMI were ≥25 kg/m2, ≥24 kg/m2 and ≥23 kg/m2 in the 18-39, 40-59 and 60-99 years old groups, respectively; for women, the corresponding cut-off points were ≥25 kg/m2, ≥23 kg/m2 and ≥25 kg/m2 in groups. CONCLUSIONS The thresholds for BMI might be different between gender and age group. In addition, it might not be suitable to determine cut-off values of BMI to predict CNCDs for people aged ≥60 years.