Meta-analysis On Risk Factors Related To Retinopathy In Type 2 Diabetes

In recent years with the changes of lifestyle and the acceleration of aging, patients with diabetes have already reached to 40,000,000 in our country. About 90 percent of patients belong to type 2 diabetes mellitus. It has been common disease in adult. Diabetic retinopathy is the most severe complication in eyes, and its incidence and growing blind rate have been increasing year by year. Diabetic retinopathy has been one of four ophthalmocaces which leads to be blind, which drew the attention of many researchers to the pathogenesis of diabetic retinopathy. They performed a lot of research work and found out many risk factors involving in the incidence of diabetic retinopathy. As the results were not consistent, further investigation is needed to determine the relationship between theses factors and diabetic retinopathy.Objective: To evaluate the relationship between diabetic retinopathy and the following risk factors including angiotensin-converting enzyme gene polymorphism, duration of diabetes, glycosylated haemoglobin A1c, and fasting plasma glucose, postprandial plasma glucose,triglycerides, cholesterol, uric acid,body mass index after referring to the published papers on risk factors associated with retinopathy in type 2 diabetes in China by means of Meta-analysis.Method: After collecting the epidemiology literature published domestic from 1994 to 2006 on risk factors of diabetic retinopathy were searched, which were angiotensin-converting enzyme gene polymorphism ,duration of diabetes, glycosylated haemoglobin A1c, and fasting plasma glucose, postprandial plasma glucose, triglycerides, cholesterol, uric acid, body mass index, Meta-analysis was carried out using Review manager 4.2. Fixed effort model or random effort model were applied for data processing to calculate the pooled odds ratio (OR) and its corresponding 95% confidence interval (95%CI) based on heterogeneity test. Sensitivity analysis and cumulation Meta-analysis were performed and the publication bias was evaluate by funnel plot, liner regression and fail-safe number.Results: 26 literatures were selected for Meta-analysis based on our including and excluding standards. Factors which had the significant difference were as followings: angiotensin- converting enzyme gene polymorphism, duration of diabetes, glycosylated haemoglobin A1c, fasting plasma glucose. The combined OR and 95% confidence interval were respectively as followings: 1.84 (1.47,2.31), 1.19(1.11,1.27), 1.45(1.20,1.75), 1.25(1.00,1.55). There were no significant differences between postprandial plasma glucose, cholesterol, uric acid, body mass index and diabetic retinopathy.According to the characters of the literatures, sensitivity analysis was carried out to different factors. Except for triglycerides, other sensitivity analysis results were similar to the original results, which proved that the conclusion of this research work was stable and reliable. In the Meta-analysis to triglycerides, the result became more significant difference after kicking out two small samples, so that sample size needed to be augment to study.Funnel plots for risk factors were all symmetrical by the standard ofα=0.1 and that the fail-safe number was more than (5k+1) showing that the publication bias was under control. But it didn't mean that the publication bias was thoroughly eliminated to some research without enough including literatures even if funnel plots were proved to be symmetrical by linear regression test or the fail-safe number was large.The cumulative Meta-analysis results showed significant differences in each point between diabetic retinopathy and the following factors Including angiotensin-converting enzyme gene polymorphism, duration of diabetes, point estimate of OR and 95% confidence interval (95%CI) went stable as the increasing of sample size. The cumulative Meta-analysis result between glycosylated haemoglobin A1c and diabetic retinopathy showed that the outcome had no statistical differences at the time point of"2000 and 2002"after that, with the increasing of sample size, the point estimate of OR and the 95%CI became stable and the 95% confidence interval also became narrower. The cumulative Meta-analysis between fasting plasma glucose and diabetic retinopathy showed that there were no statistical differences except for the last time point what we had searched at present, but when the sample size was added, the point estimate of OR and the 95%CI became stable and the 95% confidence interval also became narrower and P value tended to less than 0.05.Conclusion: Factors associated with retinopathy were angiotensin-converting enzyme gene polymorphism duration of diabetes, glycosylated haemoglobin A1c. Cumulative Meta- analysis outcomes showed the dynamic changing of the correlation between the risk factors and diabetic retinopathy, suggesting a stable relationship. The sensitivity analysis and publication bias estimation(excluding triglycerides) showed that the outcomes of this research work was stable and reliable which could authentically reflect the epidemiology research status on risk factors to diabetic retinopathy in the latest decade. Because the sensitivity analysis of triglycerides had significant difference,the sample size is needed to add.