The Relationship Between D-dimer And Diabetic Retinopathy: A Meta-analysis

Objective In order to study the relationship between D-dimer and diabetic retinopathy,we performed a meta-analysis to assess the difference of the level of plasma D-dimer in patients with and without diabetic retinopathy of type 2 diabetes of different observational studies,included diffirent DR lesion degrees within the DR patients.This analysis was performed in order to provide basis for the DR patients’ clinical diagnosis and treatment.Datas and Methods We used the Internet browser to search the relevant references in different full-text database such as Pubmed,the Cochrane Library,PMC,Wanfang and VIP.The search keywords are diabetic retinopathy,D-dimer and dimer,retrieval time until March 2017.A total amount of 156 references were retrieved,included twelve from Pubmed,fortytwo from Wanfang database,eighty-three from PMC,nineteen from the VIP database and zero from the Cochrane Library.According to the advanced setting criterion of inclusion and exclusion,we used the Note Express software to manage all the references.We excluded nineteen references after duplicate checking.Then one hundred and three references were exclued after preliminary reading abstract of literature,included references considered as irrelative to the purpose of this study,the research object was type 1 diabetes,animal experiments,conference papers and medical review.The number of residual potential references reduced to thirty-four.After executed the full text retrieval of all potential references,three of which were not retrieved in full text,were excluded.The remaining amount of references was thirty-one.Twenty-five references were excluded,one for the wrong grouping of diabetic microvascular complications instead of retinopathy,one for the discrepancies between reference data and analysis conclusion,two for the incomparable skewed distribution of original data which was expressed in quartile,one for the possiblity of redundant publication,one for the lack of NDR control group,one for the missing of available data,one for the lack of group indicators and eighteen for the loss of elimination of medical history.After the aforementioned exclusion,six references were remained.Reference quality evaluation,evidence grading and data extraction were carried out within all six references.The extracted data included the sample size,the mean and the standard deviation of D-dimer level of each groups,which were served as combined statistics.The total sample size of case group(DR group)is three hundred and nightyseven cases.The control group(NDR)merged sample size of three hundred and 353 cases.The NPDR subgroup merged one hundred and eight-two cases while the PDR subgroup merged one hundred and forteen cases.Since the measure way of each study was inconsistent,standardized mean difference(SMD)was adopted to indicate the data.Heterogeneity test showed a high number of 96 percent of I squared after executed the comparison between combined effect sizes of DR and NDR group,while the p value was less than 0.05.Removed studies one by one,heterogeneity test stayed in a high level sostenuto,I squared continued showing a number of more than 50.Sensitivity analysis was negative.Due to the great heterogeneity exsisted in the inclued studies,random effects model is used to execute the following meta-analysis.Results NOS Study evaluation criterias were used to evaluate the literature quality of all references based on the study type.The quality is medium.Rev Man 5.3 software was used to performe the meta-analysis of combined statistics,analysis results showed that compare DR group to NDR group,average plasma D-dimer level of patients in DR group was 0.99 times that of the NDR group patients,the 95% CI was [0.12,1.85].As in the subgroup,average plasma D-dimer level of patients in NPDR group was 0.67 times that of NDR group,95% CI was [-0.06,1.39].As to the PDR group,the number of times increased to 1.82,the 95% CI was [0.24,3.40].While comparing the NPDR group to the PDR group,the number of times raised to 0.96,the 95%CI was [0.22,1.69].The combined effect and 95% CI of all compare group are statistically significant,except for the 95%CI of the NPDR subgroup.STATA 12.0 software was used to to draw the funnel figure,but asymmetry could be seen on the charted figure,indicated the possibility of publication bias.Begg and Egger’s test was performed as quantitative detection in order to measure the publication bias.Result shows p value was 0.326,which was more than 0.05,indicated that possibility of publication bias was low.The results of meta-analysis were credible.Conclusions 1.The average plasma D-dimer level is significantly higher in diabetic patients with PDR than NDR and NPDR.2.Monitoring the plasma D-dimer level helps to diagnose and screen patients with PDR in early stage,as well as the course of disease.3.Patients with diabetic retinopathy are more likely to have disorders of prethrombotic state in blood system and the inflammatory response.