A Meta-analysis Of Risk Factors And Establishment Of Risk Assessment Model In Type 2 Diabetic Retinopathy

BackgroundThe prevalence of diabetes is increasing worldwide in recent years.It has been the third major chronic disease after cardiovascular disease and cancer.Diabetic retinopathy(DR)is the leading causes of vision loss in in working age people and it's a serious microvascular complication of diabetes.Blindness caused by DR is completely irreversible.All diabetic patients are at risk of diabetic retinopathy.Early diagnosis and treatment of DR is of great significance.The occurrence and development of diabetic retinopathy is a complex process.Many molecular and biochemical mechanisms are involved,but its specific mechanism is not clear.There are many researches about risk factors of DR.Considering the factors such as geography,race,research methods and sample size,whether hyperlipemia,age,body mass index and some other factors are the risk factors of diabetic retinopathy are not for sure.In order to combine more recent evidences and get more intuitive and reliable multiple DR risk factors,we used meta-analysis to analyze the relationship between DR and multiple risk factors.It is essential to make regular fundus examination for patients with diabetes.However,the rate of ophthalmology in developing countries is relatively low,because of the reasons for the backward economy,the lack of medical resources,and the lack of patients' understanding of the DR.In China,due to the lack of medical resources and other reasons,most of the patients did not get early diagnosis and treatment,which reduced the quality of life of patients and brought heavy financial burden to social health care.Whether clinicians or patients,especially in remote areas lack of medical resources are hoping to find a scientific and effective method,can be used to predict the risk of diabetes in patients with DR,in order to make an accurate judgment to take reasonable measures for clinical.Method1.Relevant studies were identified through EMbase,Web of Science,PubMed,CNKI,Wanfang database and VIP database using a search strategy.Relevant data were extracted from eligible studies.Fixed and random effects models were used,and outcomes are presented as pooled odds ratios(ORs)and 95%confidence intervals(CIs).2.Select the type 2 diabetes inpatient in The Third Affiliated Hospital of Southern Medical University from January 2016 to December 2017.General information and laboratory examination were collected.Single factor analysis,multivariate logistic regression analysis and point system score were used to establish DR risk assessment model.The receiver's characteristic working curve(ROC curve)and Hosmer-Lemeshow goodness of fit test were used to test the predictive efficacy and fitting efficiency of the model.Results1.A total of 110 studies were identified,with 56,667 type 2 diabetes mellitus(T2DM)individuals from China.The main risk factors significantly associated with DR in Chinese patients with T2DM were diabetes duration(OR=1.106;95%CI 1.093-1.120),family history of diabetes mellitus(OR=3.048;95%CI 1.704-4.392),systolic blood pressure(OR=1.130;95%CI 1.052-1.208),BMI(OR=1.082;95%CI 1.012-1.152),fasting blood glucose concentration(OR=1.130;95%CI 1.052-1.208),glycosylated haemoglobin(OR=1.122;95%CI 1.096-1.148,total cholesterol(OR=1.306;95%CI 1.191-1.421),triglycerides(OR=1.334;95%CI 1.078-1.591),low-density lipoprotein cholesterol(OR=1.435;95%CI 1.234-1.636),high-sensitivity C-reactive protein(OR=1.105;95%CI 1.062-1.148),vascular endothelial growth factor(OR=1.046;95%CI 1.010-1.081),fibrinogen(OR=1.59;95%CI 1.081-2.099),homeostasis model of assessment for insulin resistence index(OR=1.334;95%CI,1.209-1.458),hypertension(OR=1.383;95%CI1.291-1.475),diabetic neuropathy(OR=9.202;95%CI 3.173-15.232),and diabetic nephropathy(OR=2.022;95%CI 1.598-2.445).Interestingly,fasting serum C-peptide(OR=0.528;95%CI 0.130-0.925),total bilirubin(OR=0.85;95%CI,0.764-0.935)and 25-hydroxyvitamin D3(OR=0.815;95%CI 0.653-0.977)may be protective against DR in Chinese patients with T2DM2.Multiple logistic analysis display,duration of diabetes?4.5 years,Glycated Hemoglobin Alc?7.75%,high density lipoprotein cholesterol<1.38mmol/L,Apolipoprotein-A ?1.18g/L,blood urea?6.46mmol/L,total bilirubin<6.65?mol/L,fibrinogen?2.94g/L,diabetic nephropathy is an independent risk factor of DR.3.DR risk assessment model:(1)duration of diabetes?4.5 years,got 4 points;(2)total bilirubin<6.65 mol/L,got 2 points;(3)Apolipoprotein-A?1.18g/L,got 2 points;(4)blood urea?6.46mmol/L,got 1 points;(5)Glycated Hemoglobin Alc?7.75%,got 2 points;(6)HDL-c<1.38mmol/L,got 2 points;(7)diabetic nephropathy,got 3 points;(8)fibrinogen,got 1 points.4.The area under ROC curve of DR risk assessment model is 0.787,which has diagnostic value for diabetic retinopathy.Hosmer-Lemeshow goodness of fit test(?2=10.125,df=8,P=0.256)shows that the fitting effect is good.5.when the DR risk score was 0?8 points,it was defined as low risk.The actual incidence of DR in patients with T2DM was 13.79%.When the DR risk score was 9?11 points,it was defined as a medium risk,and the actual incidence of DR in T2DM patients was 37.43%.When the DR risk score in 12?17 points,defined as high risk,real DR in patients with T2DM incidence rate was 70.29%,three different risk stratification of DR the difference between the rates was statistically significant(P<0.001),the incidence of low risk group and risk group DR was statistically significant(P<0.001),the incidence of low risk the rate of DR group and high risk group with statistical difference(P<0.001),disease risk group and high-risk group of DR rate was statistically significant(P<0.001).6.The DR model was used to verify that the group demonstrated good discrimination and fitting correction.The area under the ROC curve is 0.869.The Hosmer-Lemeshow goodness of fit test results(x2=5.345,df=7,P=0.618).Conclusion1.This meta-analysis presented a robust and intuitionistic results with considering geographical and racial differences.This meta-analysis comprehensively investigates the risk factors of DR and highlights the role of these risk factors in DR risk development.2.the DR risk factors model is simple and convenient to be operated and can accurately predict the occurrence of DR and provide scientific theoretical basis for the early prevention and treatment of DR.