Abstract: Background：The diagnosis of true obesity was recommended to be based on body fat quantity and distribution by body composition measurement. However, the riskbased overfat cutoffs were scarce for pediatric population. Objective：To develop cutoffs and the optimal combination for body fat indices for screening persistent hyperglycemia and dyslipidemia among the pediatric population. Design：Prospective cohort study. Methods：Subjects who participated in the 2017 baseline and 2019 followup survey of Schoolbased Cardiovascular and Bone Health (SCVBH) Promotion Program with complete data of body composition and blood test, were selected as the study population. The gold standard was persistent hyperglycemia and dyslipidemia in both baseline and followup surveys, including persistent impaired fasting glucose (IFG), persistent high total cholesterol (TC), persistent high TG, persistent high low density lipoprotein cholesterol (LDLC), persistent low high density lipoprotein cholesterol (HDLC) and persistent high NonHDLC. The predictors included body mass index (BMI) and body fat indices derived from bioelectrical impedance analysis, including fat mass index (FMI), fat mass percentage (FMP), trunk to leg fat ratio (TLR).The area under the receiver operating characteristic curve was used to determine the best combination and optimal cutoffs of body fat indices for detecting persistent hyperglycemia and dyslipidemia. Main outcome measuresThe best combination and optimal cutoffs of body fat indices for detecting persistent hyperglycemia and dyslipidemia. Results：A total of 10 603 (mean age at baseline: 10.9 ± 3.3 years, 49.4% males) children and adolescents aged 618years were included for analysis. Among, 371 (3.5%) were diagnosed as persistent lFG,131 (1.2%) as persistent high TC, 128 (1.2%) as persistent high TG, 118 (1.1%) as persistent high LDLC, 448 (4.2%) as persistent low HDLC, and 212 (2.0%) as persistent high nonHDLC. According to the results for ROC analyses and Delong tests, the capability of FMI+TLR combination for detecting persistent IFG, persistent high TC and persistent high LDLC were statistically higher than BMI in both sexes (All P<0.05). Moreover, the combined use of FMI+TLR presented higher capability for detecting persistent high NonHDLC ［AUCFMI+TLR：0.664 (0.6150.713) vs AUCBMI：0.617 (0.5570.677), P<0.001］ than BMI in girls. According to the ROC analysis, the optimal overfat cutoffs of FMI were determined at the range of 75th percentile to 95th percentile, and TLR were determined at the range of 75th percentile to 90th percentile, varied by indicators for persistent hyperglycemia and dyslipidemia. Conclusion：The FMI + TLR combination presented higher predictability for discriminating persistent hyperglycemia and dyslipidemia among children and adolescents. We suggest the 75th percentile of FMI to be the cutoff for mild general overfat, the 90th percentile of FMI to be the cutoff for severe general overfat, the75th percentile of TLR to be the cutoff for mild central overfat, the 90th percentile of TLR to be the cutoff for severe central overfat.

Key words: Bioelectrical impedance analysis, Body fat quantity, Body fat distribution, Persistent hyperglycemia and dyslipidemia, Overfat cutoffs