| BackgroundType 2 diabetes mellitus(T2DM)continues to spread throughout the world,and number of T2DM in China has even exploded in these twenty years,which made it to be a country with the most diabetic patients.Meanwhile,more than one third of the population in China is now at high risk of diabetes.The epidemic of diabetes and prediabetes both put efforts of diabetes prevention and treatment in extremely tough situation.Evidence from numerous follow-up studies has shown that intervention of relevant risk factors in high-risk groups can effectively prevent or postpone the onset of diabetes.Therefore,early identification of risk factors that influence the occurrence and development of prediabetes and diabetes is the basis for effective screening of high-risk populations and implementing comprehensive prevention and treatment programs,and also a key step to work against uncontrolled diabetes.However,not much data based on prospective cohort study is available regarding the profiles of prediabetes and diabetes incidence and diabetes incidence in people with prediabetes,and their related risk factors in Chinese population.Objectives1.To estimate the incidences of prediabetes and diabetes in the Chinese population2.To explore factors that influence the incidences of prediabetes,diabetes and diabetes in people with prediabetes,so as to provide scientific basis for prevention and control of diabetes3.To establish a diabetes incidence prediction model that was suitable for Chinese population,which could serve as core technologies for diabetes screening and follow-up management.Methods and contentsThe data we used came from the Chinese Major Chronic Disease Risk Assessment Cohort Project(here in after referred to as "Chronic Disease Risk Assessment Cohort Project").This project was a prospective cohort study design,baseline information was obtained from the 2010 Chinese Chronic Disease and Risk Factor Surveillance(CCDRFS,here in after referred to as "Chronic Disease and Risk Factor Surveillance")which was a nationwide cross-sectional survey recruited adults aged 18 years or older as participants by adopting a multistage stratified cluster sampling strategy from all 31 provinces,autonomous regions,and municipalities in mainland China.The contents of the survey included face-to-face questionnaire interviews,physical measurements,and laboratory tests.Then the Chronic Disease Risk Assessment Cohort Project selected ten provinces from the CCDRFS study,within each province(autonomous regions or municipalities),two surveillance spots(one in urban and one in rural area)were further sampled for follow up investigation in 2016-2017.Finally,10 874 subjects who were free of diabetes in baseline were recruited.The contents and methods of follow-up survey was the same as baseline study,and involved information gathering of outcome events additionally.Our present study aimed to use the baseline and follow-up data of the Chronic Disease Risk Assessment Cohort Project to firstly examine the incidence of prediabetes,diabetes and diabetes in individuals with prediabetes,and related risk factors influencing prediabetes and diabetes incidence.Then we went a step further to explore whether there existed interactions between two factors and which factor's dynamic changes contributed most to prediabetes and diabetes.Finally we would establish a diabetes incidence prediction model.Mean and standard deviation were used for statistical description of quantitative data,and t test was applied for comparison between groups.Rate or constituent ratio were adopted for statistical description of numerical data,and ?2 test was used for group comparison.Person-years was used for the calculation of incidence rate,which would be standardized by the direct method based on the Year 2010 Population Census of China.A method recommended in the WHO mortality analysis manual was applied for statistics analysis of standardized incidence rate comparison.Univariate and multivariate cox proportional hazards analysis were fitted to explore related risk factors that influence the incidence of prediabetes and diabetes.The interactions between two factors were analyzed by cox multiplication and additive model.Adjusted HR(95%CI)was estimated to evaluate associations between risk factors and prediabetes and diabetes incidence.We applied non-conditional Logistic regression to establish a T2DM incident prediction model,then we established rules to characterize risk levels of the study population based on the cut-off points of the probability,after being transformed into normal distribution by log-transformation.The receiver operating characteristic curve(ROC)and the Hosmer lemeshow x2 were used to appraise the performance of the prediction model.Results1.General information of the sampleA total of 10 874 individuals who were free of diabetes in baseline were recruited in the follow-up queue,and 7 289 participants were followed up.After data cleaning,we excluded 241 individuals who were not considered to be the same person at baseline and follow up survey,finally a total of 7 048 participants received completely re-investigation,which generated a thirty-five percent of the loss ratio of follow up.Of the 7 048 participants,57.2%were female,42.8%were male.The proportions of age group 18?45,45?60,and ?60 in the sample were 44.2%,37.8%,and 18.0%,respectively.2.The incidence of prediabetes and diabetes Crude incidence of prediabetes in the Chinese population aged 18 years and above was 24.4 per 1000 person-years(24.9 pre 1000 person-years for men and 23.9 per 1000 person-years for women),age-standard incidence of prediabetes was 20.2 per 1000 person-years,there was no difference in prediabetes incidence between gender(20.3 per 1000 person-years for men and 19.4 per 1000 person-years for women,P>0.05).Crude incidence of diabetes was 13.4 per 1000 person-years(14.8 per 1000 person-years for men and 12.3 per 1000 person-years),age-standard incidence of diabetes was 10.3 per 1000 person-years with male(11.1 per 1000 person-years)higher than female(10.0 per 1000 person-years)(P<0.05),while crude incidence of diabetes in people with prediabetes was 26.9 per 1000 person-years(26.4 per 1000 person-years for men and 27.2 per 1000 person-years for women),age-standard incidence of diabetes in people with prediabetes was 17.5/1000 person-years,and there was no difference in diabetes incidence in participants with prediabetes between gender(19.2 per 1000 person-years for men and 15.4 per 1000 person-years for women,P>0.05).Incidence rate of prediabetes and diabetes increased with elevated levels of age,BMI,blood pressure,TG,and LDL-C,and decreased with rising HDL-C3.Risk factors related to prediabetes and diabetes and interactions between two factorsCox multivariate analysis showed that overweight,obesity,hypertriglyceridemia,hypertension,and low HDL-C hyperlipidemia were risk factors for prediabetes,the prediabetes risk increased by 28%(overweight:HR=1.28,95%CI:1.05?1.56),67%(obesity:HR=1.67,95%CI:1.25?2.24),32%(hypertriglyceridemia:HR=1.32,95%CI:1.04?1.67),27%(hypertension:HR=1.27,95%CI:1.07?1.53),and 19%(low HDL-C hyperlipidemia:HR=1.19,95%CI:1.01?1.42),respectively than their normal counterparts.Cox multiplication model failed to show interactions between two factors.Additive model showed there existed a positive synergistic effect between overweight/obesity and hypertriglyceridemia,the value of RERI was 1.29(95%CI:0.10?2.58)and of AP was 0.41(95%CI:0.03?1.24).Age group 45?59,obesity,family history of diabetes,hypertriglyceridemia,and high LDL-C hyperlipidemia were risk factors for diabetes,and the diabetes risk increased by 33%(age group 45?59:HR=1.33,95%Cl:1.06?1.68),91%(obesity:HR=1.91,95%CI:1.36?2.68),44%(family history of diabetes:HR=1.44,95%Cl:1.02?2.07),64%(hypertriglyceridemia:HR=1.64,95%CI:1.28?2.12),and 67%(high LDL-C hyperlipidemia:HR=1.67,95%CI:1.04?2.76),respectively than their normal counterparts.Cox multiplication failed to show interactions between two factors.Additive model showed there existed a positive interaction between high LDL-C hyperlipidemia and hypertriglyceridemia,the value of RERI was 1.53(95%CI:0.1 1?5.02)and of AP was 0.40(95%CI:0.09?1.66).Family history of diabetes,abdominal obesity,and hypertriglyceridemia were risk factors for progression to diabetes in individuals with prediabetes,and the diabetes risk increased by 96%(HR=1.96,95%CI:1.18?3.25),52%(HR=1.52,95%CI:1.02?2.28),and 45%(HR=1.45,95%CI:1.02?2.06),respectively than their normal counterparts.Cox multiplication failed to show interactions between two factors.Additive model showed there existed a positive interaction between family history of diabetes and obesity,the value of RERI was 1.18(95%CI:0.25?3.94)?and of AP was 0.37(95%CI:0.10?1.26).4.Relationship between dynamic change of risk factors from baseline to follow-up and prediabetes and diabetes risksCox multivariate analysis showed that after adjustment for potential confounding factors,in the baseline normal BMI,blood pressure,TG,and HDL-C group,prediabetes risk remarkably increased when baseline normal BMI,blood pressure,TG or HDL-C developed abnormality during follow-up,and the risk increased by 37%(BMI:HR=1.37,95%CI:1.08?1.75),74%(blood pressure:HR=1.74,95%CI:1.39?2.17),87%(TG:HR=1.87,95%CI:1.54?2,25)or 32%(LDL-C:HR=1.32,95%CI:1.02?1.71),respectively than those who still remained normal.In the baseline abnormal BMI or TG group,prediabetes risk obviously decreased when baseline abnormal BMI or TG developed normality during follow-up,and the risk decreased by 34%(BMI:HR=0.66,95%CI:0.49?0.90)or 38%(TG:HR=0.62,95%CI:0.42?0.91),respectively than those with uncontrolled BMI or TG.However,we failed to observe a negative relationship between controlled abnormal blood pressure or HDL-C with decreased prediabetes risk(P>0.05)In the baseline normal BMI,TG or LDL-C group,diabetes risk remarkably increased when baseline normal BMI,TG or LDL-C developed abnormality during follow-up,and the risk increased by 84%(BMI:HR=1.84,95%CI:1.36?2.49),163%(TG:HR=2.63,95%CI:2.10?3.28),or 78%(LDL-C:HR=1.78,95%CI:1.32?2.28),respectively than those who still remained normal.In the baseline abnormal TG group,diabetes risk was lower for those whose TG had been under control comparing with those with uncontrolled TG,and the risk decreased by 48%(HR=0.52,95%CI:0.34?0.79),P<0.05.However,there did not exist a negative relationship between controlled abnormal BMI or LDL-C with decreased diabetes risk(P>0.05).Comparing with those who maintained normal WC or TG both at baseline and follow-up,diabetes risk in people with prediabetes increased for individuals who had the normal WC or TG at baseline but developed abnormality during follow-up,and the risk increased by 90%(WC:HR=1.90,95%CI:1.12?3.24)or 113%(TG HR=2.13,95%CI:1.36?3.34),respectively.In the baseline central obesity group,diabetes risk was lower for those whose WC had been under control during follow-up comparing with subjects with uncontrolled WC,and the risk decreased by 57%(HR=0.43,95%CI:0.23?0.80).However,significance difference was not observed between controlled TG in subjects with abnormal TG and decreased diabetes risk5.Establishment and evaluation of risk prediction model of T2DMThe 6-year T2DM incidence prediction model established using non-condition Logistic regression model was Logit(P)=-7.940+0.015×age-0.203×gender(female=1)+0.416×family history of T2DM(positive=1)+0.056×BMI+0.029×SBP(mmHg)+0.397xfasting blood glucose(mmol/L)+0.892×high LDL-C hyperlipidemia(positive=1).The area under the ROC curve is 0.709(0.701?0.722),and the corresponding sensitivity is 65.4%,the specificity is 66.6%when the youden index reaches its maximum(0.320),The goodness of fit test ?2=7.966,P=0.437.Risk probabilities of diabetes incidence in people with normal blood glucose were divided into four levels.After labeling the degrees the four risks,it was found that 0?3%of the incidence probability was in the low-risk group,3%?13%of the incidence probability was in the population under generate-risk,13%?27%of the incidence probability was in the moderate-risk population,while over 27%was in the high risk population.Conclusion1.Age-standard incidence of prediabetes in the Chinese population aged 18 years and above was 20.2 per 1000 person-years,of diabetes in the general population was 10.3 per 1000 person-years,of diabetes in people with prediabetes was 17.5 per 1000 person-years.New cases of prediabetes in Chinese population might occure at a low rate compared with studies at home and abroad,while diabetes incidence was at a relatively high level compared with domestic documents.2.Overweight,obesity,hypertension,hypertriglyceridemia,and low HDL-C hyperlipidemia were risk factors for prediabetes in the Chinese population,there did exist a positive additive interaction between overweight/obesity and hypertriglyceridemia;Age group 45?59,obesity,family history of diabetes,hypertriglyceridemia,and high LDL-C hyperlipidemia were risk factors for diabetes,there did exist a positive additive interaction between hypertriglyceridemia and high LDL-C hyperlipidemia;Family history of diabetes,hypertriglyceridemia,and abdominal obesity were the main predictiors for the development of diabetes in people with prediabetes,there did exist a positive additive interaction between family history of diabetes and central obesity.3.For those with normal BMI,blood pressure,TG,LDL-C,HDL-C,or WC,the development of abnormalities in all indicators will lead to a significant increase in prediabetes or diabetes risk.For subjects with abnormal TG,WC or BMI,controlling TG,WC or BMI at normal levels can significantly generate reduced risk of prediabetes or diabetes.While the dynamic change of TG has the most significant effect on prediabetes and diabetes.4.We established a 6-year T2DM incidence prediction model characterized with relative high discrimination and good calibration in Chinese adults population based on nationwide data of prospective cohort study,which can provide technology support for diabetes screening and follow-up management. |
PDF Full Text Request