Study On The Relationships Between Obesity Index And Hypertension-diabete Comorbidity In Occupational Group And Development Of Comorbidity Prediction Model

Background and purposeHypertension progresses in tandem with diabetes,which is a significant predictor of hypertension,and hypertension is also a significant predictor of diabetes.The prevalence of hypertension-diabetes comorbidity(HDC)is high worldwide and continues to increase.Insulin resistance is a common feature of prehypertension and prediabetes,an early stage that can develop into two different disease states.Of the many causes of insulin resistance,obesity is the most important.The prevalence of hypertension and diabetes among obese people is high.Obesity in occupational group has become an important public health problem both at home and abroad.HDC is very dangerous and are closely related to the progression of cardiovascular disease,stroke,kidney disease and diabetic retinopathy.The HDC and other related diseases caused by it severely consumes medical and social resources and imposes a heavy economic burden on families and countries.As the backbone of social development,the health level of occupational group directly affects the social development process.In order to reduce the prevalence of HDC in the occupational group and reduce the economic loss caused by HDC to the family and society,it is necessary to trace the source and conduct in-depth research on the risk factors related to HDC.Therefore,it is particularly urgent and important to carry out in-depth research on the relationship between obesity and HDC.At present,most studies are limited to the relationship between obesity indexes and hypertension or diabetes,while there are few reports on obesity indexes and HDC.Moreover,there are still some limitations and research gaps in these few studies.First of all,when discussing the relationship between obesity indexes and HDC,the indexes included are not comprehensive enough.Secondly,in current studies at home and abroad,continuous data such as body mass index(BMI)and waist circumference(WC)are mostly converted into classification data.Third,the traditional research and analysis methods have some limitations.For example,they caould not present a multifactorial relationship in a model;could not construct latent variables;The measurement errors of independent variables and the relationship between them were not considered;could not handle multiple dependent variables at the same time.Fourth,studies on the relationship between obesity indexes and HDC have not been reported specifically for the occupational group in China.Finally,no studies have reported to construct a risk prediction model for HDC and compare the predictive efficacy of obesity indexes.The purposes of this study are as follows:(1)To analysis the relationship between obesity indexes and HDC in occupational group by controlling the influence of confounding factors;(2)To establish a model for predicting HDC by obesity indexes and other factors and to explore the predictive efficacy of different obesity indexes.(3)To described the dose-response relationship between obesity indexes and HDC in different gender and age groups;(4)To analysis the direct effect of obesity indexes on HDC,and to explore its mediating effect in the relationship between sociodemographic factors,lifestyle,health knowledge awareness and HDC.MethodA multi-stage stratified cluster sampling method was used to randomly select occupational group from 11 cities in Hubei province for a cross-sectional survey.Self-designed questionnaire was used to understand the factors related to HDC,including demographic and sociological data,behavioral lifestyle,health knowledge awareness,obesity indexes,etc.The height and weight of the occupational group were measured by using the height and weight scale,and the BMI value was calculated.A tape measure was used to measure the WC of the occupational group and calculate the WHt R value accordingly.Blood pressure and blood glucose were measured by mercury sphygmomanometer and blood glucose meter respectively,and the HDC was evaluated;SPSS21.0 was used to analyze the relationship between BMI,WC,waist-to-height ratio(WHt R)and HDC of occupational group through univariate and multivariate logistic regression.p<0.05 means the difference is statistically significant;Logistic regression analysis and decision tree classification were used to establish HDC prediction models by SPSS21.0.Med Calc19.0.7 was used to draw the receiver operating characteristic curve and calculate the sensitivity,specificity and sensitivity,as well as the area under curve.Compare the advantages and disadvantages of the prediction model and the prediction efficiency of different obesity indexes.p<0.05 means the difference is statistically significant;Stata14.0 and restrictive cubic spline method were used to explore the dose-response relationship between obesity indexes and HDC in occupational group,and stratified by gender and age group.p<0.05 means the difference is statistically significant;Social and demographic factors,lifestyle,health knowledge awareness and obesity level were taken as latent variables,and HDC as endogenous observed variable.Mplus 8.4 was used to construct a structural equation model to analysis the direct effect of obesity level on HDC reflected by indexes such as BMI,WC and WHt R,and its mediating effect on the relationship between socio-demographic factors,lifestyle,health knowledge awareness and HDC.ResultA total of 25,356 subjects were included as samples in this study,of which 713(2.8%)were HDC.Multivariate Logistic analysis showed that there was no statistical correlation between the higher WC and HDC;the risk of HDC with 24kg/m²?BMI?26.9kg/m²and BMI?27kg/m²was 5.66 times and 7.96 times higher than normal BMI,respectively;the HDC risk of 25th percentil(P₂₅)?WHt R<50th percentile(P₅₀),P₅₀?WHt R<75th percentile(P₇₅)and WHt R?P₇₅is 1.73 times,2.51 times and 3.22times of WHt R<P₂₅,respectively.For male,there was no statistical association between higher WC and HDC;the risk of HDC with 24kg/m²?BMI?26.9kg/m²and BMI?27kg/m²is 7.39 times and 12.19 times higher than the normal BMI range,respectively;the risk of HDC with WHt R?P₇₅was 2.27 times greater than WHt R<P₂₅.For females,there was no statistical correlation between the higher WC and HDC;the risk of HDC with 24kg/m²?BMI?26.9kg/m²and BMI?27kg/m²were4.69 and 6.08 times than the normal BMI,respectively;the HDC risk of P₂₅?WHt R<P₅₀,P₅₀?WHt R<P₇₅,and WHt R?P₇₅were 3.27,3,and 4.53 times than WHt R<P₂₅,respectively.For young people,There was no statistical correlation between higher WC,WHt R and HDC;the risk of HDC with BMI?27kg/m²was380.57 times than the normal BMI.For middle-aged adults,the risk of HDC with24kg/m²?BMI?26.9kg/m²and BMI?27kg/m²was 9.74 times and 10.90 times than the normal BMI,respectively;the HDC risk of P₅₀?WHt R<P₇₅and WHt R?P₇₅was9.57 times and 12.25 times than WHt R<P₂₅,respectively.In the elderly,the risk of HDC with 24kg/m²?BMI?26.9kg/m²and BMI?27kg/m²was 8.42 times and 12.23times than normal BMI,respectively;there was 5.24 times more likely to have HDC with WHt R?P₇₅than WHt R<P₂₅.The hosmer-lemeshow test showed that?²=12.617,degrees of freedom=8,p=0.126.The sensitivity,specificity and sensitivity of the Logistic regression model were 92.1%,90.8%and 92.15%,respectively.The AUC was 0.968,and the 95%confidence interval was(0.962,0.975).In the model of decision tree classification,a total of 23 nodes were obtained,including 14 leaf nodes with a depth of 4.The independent influencing factors of HDC were:work intensity,smoking,marriage,gender,BMI,WHt R,age,etc.The sensitivity,specificity and sensitivity of the decision tree model were 89.2%,86.7%and 89.20%,respectively.The AUC was0.947,with a 95%confidence interval(0.944,0.950).The AUC of Logistic regression prediction model is higher than that of decision tree classification prediction model,and the difference between them is statistically significant(z=5.859,p<0.0001).The sensitivity of BMI,WC and WHt R to predict HDC separately is.90.74%,89.06%and89.20%.The sensitivity of BMI and WHt R to predict HDC was 92.15%.The results of the restricted cubic spline model analysis of the whole sample population showed that the strength of the association between BMI and HDC showed a significant nonlinear dose-response relationship(nonlinear test,?²=87.42,p<0.001).The correlation strength between WC and HDC showed a nonlinear dose-response relationship(nonlinear test,?²=6.87,p<0.05).The correlation strength between WHt R and HDC showed a nonlinear dose-response relationship(nonlinear test,?²=34.30,p<0.001).The stratified analysis based on gender showed that the strength of the association between BMI and HDC presented a significant nonlinear dose-response relationship(nonlinear test,?²_male=68.56,p<0.001;?²_female=8.31,p<0.05);The strength of the association between WC and HDC in males presented a nonlinear dose-response relationship,while that between WC and HDC in females presented no nonlinear dose-response relationship(nonlinear test,?²_male=6.95,p<0.05;?²_female=0.29,p>0.05);The intensity of WHt R and HDC in both men and women presented a nonlinear dose-response relationship(nonlinear test,?²_male=9.09,p<0.05;?²_female=18.42,p<0.001).Stratified analysis based on age showed that except for the young occupational group(nonlinear test,?²_{young adulthood}=2.16,p>0.05),the intensity of the association between BMI and HDC in the middle-aged and the elderly occupational group showed a significant nonlinear dose-response relationship(nonlinear test,?²_middle-aged=47.22,p<0.001;?²_elderly=46.22,p<0.001);Except for the elderly occupational group(nonlinear test,?²_elderly=0.08,p>0.05),the intensity of the association between WC and HDC in the young and middle-aged occupational group showed an obvious nonlinear dose-response relationship(nonlinear test,?²_{young adulthood}=10.52,p<0.01;?²_middle-aged=7.47,p<0.05);Except for the young occupational group(nonlinear test,?²_{young adulthood}=2.15,p>0.05),the correlation strength of WHt R and HDC in the middle-aged and the elderly occupational group showed obvious nonlinear dose-response relationship(nonlinear test,?²_middle-aged=18.83,p<0.001;?²_elderly=14.61,p<0.001).The results of the structural equation model showed that obesity have direct effect on HDC,and the influence path coefficient was 0.354.In the influence of demographic and sociological characteristics on HDC,the mediating effect of obesity was 0.102,accounting for 32.4%of the total effect.In the influence of health knowledge awareness on HDC,the mediating effect of obesity was-0.032,accounting for 36.0%of the total effect.In the influence of lifestyle on HDC,the mediating effect of obesity was 0.045,accounting for 5.7%of the total effect.The total mediating effect of obesity was 0.115.ConclusionsCompared with the reported prevalence of HDC at home and abroad,the prevalence of HDC in the occupational group in this study was at a medium level.Obesity is an important risk factor associated with the HDC.The related indexes of obesity are closely related to the HDC.Both men and women,young adults,middle-aged adults,and older adults,higher BMI and WHt R were associated with an increased likelihood of HDC in the occupational group,as compared to those with normal BMI and WHt R,and were independent,dose-dependent correlative factors for HDC.The Logistic regression prediction model is more accurate than the decision tree classification prediction model in predicting the HDC.Combined BMI and WHt R were more effective than a single index of obesity in predicting HDC in occupational group.The effect of obesity on the HDC in occupational group is both direct and mediating.This study suggests that it is necessary to prevent HDC in the occupational group focusing on obesity control.