The Association Between Body Composition Parameters And Chronic Disease And Its Predictive Value

ObjectiveBased on the analysis of community population data,this study compared the parameters of human body composition with the conventional obesity related indexes such as BMI.In addition,this study explored the feasibility of human body composition as a predictor of chronic diseases and established the diabetes screening model based on these body composition parameters.This research can add research evidence about the relationship between human body composition and chronic diseases,explain some disease causes from the perspective of human components and provide an important basis for prevention and treatment of chronic diseases.Materials and MethodFrom May to September in 2015,a clustering sampling method was used to select the residents of 18?80 years old in Lanxi City,Zhejiang Province.The data were collected by questionnaire,physical examination,body composition measurement and blood test data.Variance analysis and chi-square test were used to analyze the differences between male and female,and multivariate logistic regression models were used to examine the association between body composition parameters and obesity-related chronic diseases.Univariate regression analysis was used to analyze the effects of different body composition and traditional BMI index on disease prediction.U-test was used to compare the predictive effects.Multivariate logistic regression was used to establish the prediction model of diabetes risk.The ROC curve was used to analyze the predicted cut-off value and its predictive effect.ResultsA total of 3268 subjects were included in this study,including 1234 men and 2034 women and the average age of the study population was 52.7 years.In men,Total fat mass percent,Arms fat mass percent,Android fat mass percent.Android and Gynoid area adipose tissue ratio and Android and Gynoid area lean tissue ratio were risk factors for various diseases;Legs fat mass percent and Gynoid fat mass percent were protective factors for various diseases.In women,Trunk fat mass percent.Arms fat mass percent,Android fat mass percent,Total fat mass percent,Trunk fat mass percent,Arms fat mass percent,Android and Gynoid area adipose tissue ratio were risk factors for various diseases;Legs fat mass percent,Gynoid fat mass percent and Legs lean mass percent were protective factors for various diseases.The AUC of all the body composition and other traditional indicators of obesity like BMI were all larger than 0.5,most of female AUC values were larger than men.In the prediction of diabetes mellitus,the best indicator is Android and Gynoid area lean tissue ratio for man and Android and Gynoid area adipose tissue ratio for female.On hypertension prediction,the best index for male is the waist-to-hip ratio and for female is Android and Gynoid area adipose tissue ratio In the prediction of dyslipidemia,the best indicator is Android and Gynoid area adipose tissue ratio for both man and woman;For the prediction of hyperuricemia,the best indicator for males was Android fat mass percent and the waist-to-height ratio for females;In the prediction of fatty liver,the best indicator formales was BMI and the waist circumference for females.Eventually,The established diabetes risk prediction model include following indicators:Android fat mass percent,Android lean mass percent,Arms fat.mass percent,Arms lean mass percent,Android and Gynoid area adipose tissue ratio,age,family history of diabetes and.hypertension.Model AUC reached 0.85,95%interval:0.83?0.87.The AUC in the validation data was 0.86(0.82-0.89),The prediction effect was significantly better than the United States NHANES II model AUC:0.77(0.75-0.80)and the Chinese Qd_DPP model AUC:0.74(0.71-0.77)in this population.ConclusionExcept for diabetes,in both men and women,The percentage of fat tissue in upper body was positively correlated with disease:Legs fat mass percent and Gynoid fat mass percent were negatively correlated with all diseases;The other muscle tissue or fattissue percentage showed different results with different diseases.The prediction effect of most human body composition parameters on the disease in women were better than in men.In both men and women,single body composition indicators had good predictive effect for diabetes,hypertension,dyslipidemia and hyperuricemia and fatty liver,and most of them were better than the traditional BMI indicators.Especially in diabetes,the index of human body composition had a greater effect on the prediction of diabetes than the BMI index.The model on the risk of the prevalence of diabetes mellitus was reasonable andstable,and could predict the prevalence of diabetes without diagnosis.It could be usedas a screening tool for diagnosing diabetes mellitus.