Diabetic Retinopathy in Patients with Dyslipidemia: Development and Progression

PURPOSE:To investigate the association of dyslipidemia with the development of diabetic retinopathy (DR). DESIGN:A prospective cohort from the Longitudinal Health Insurance Database in Taiwan. PARTICIPANTS:Patients with diabetes mellitus (DM) aged ≥18 years from this cohort. METHODS:A logistic regression model considering age, sex, and adapted Diabetes Complication Severity Index (aDCSI) including diabetic retinopathy, diabetic neuropathy, diabetic nephropathy cardiovascular disease, cerebrovascular disease, peripheral arteriolar disease, and metabolic disease. We estimated the propensity score for disease assignment probability for each included patient with DM and dyslipidemia. For each patient, a comparison patient without dyslipidemia was matched with a propensity score using a greedy algorithm. The standardized mean differences method was used to measure the difference in means or proportions divided by the pooled standard deviation of a variable. We calculated the incidence densities of nonproliferative DR (NPDR), diabetic macular edema (DME), and proliferative DR (PDR) as total events divided by the sum of follow-up duration, and the incidence curves were measured using the Kaplan-Meier method. The log-rank test was applied to test the differences of incidence curves. MAIN OUTCOME MEASURES:Hazard ratios (HRs) for DR. RESULTS:Our results demonstrated that the cumulative incidence of NPDR, DME, and PDR significantly increased in patients with DM and dyslipidemia compared with those without before adjustment for covariates (P < 0.001). Adjusted HRs for NPDR, DME, and PDR in patients with dyslipidemia were 1.77 (95% confidence interval [CI] = 1.63-1.92), 2.34 (95% CI = 1.24-4.41), and 1.07 (95% CI = 0.91-1.27), respectively. The risks of NPDR, DME, and PDR increased in patients who had underlying complications according to the aDCSI. Only statin use had a protective effect against the development of NPDR (HR = 0.83; 95% CI = 0.76-0.90), but it had no effect on DME and PDR. The protective effect was not significantly different between patients with and without dyslipidemia. CONCLUSION:Dyslipidemia is involved in the development of DR at an earlier stage, but the role of lipid-modulating agents in DR requires additional study.