Abstract
AIM: To report the incidence of epiretinal membrane (ERM) formation following panretinal photocoagulation (PRP) as the treatment for diabetic retinopathy (DR). METHODS: Retrospective cross-sectional study of patient charts between January 1(st), 2010 to January 1(st), 2017 with at least 1y follow-up data. All 809 patients treated with PRP for DR were evaluated for exclusion criteria and 73 eyes remained after exclusion for confounding variables related to ERM formation such as other procedures or diseases. Outcomes were determined through medical record review and masked review of optical coherence tomography (OCT) images. Cohen's kappa was completed to determine if there was an agreement between masked retinal specialists on OCT evaluations. Univariate logistic regression was used to determine the unadjusted odds ratio for patient and procedural characteristics on the formation of ERMs. Multiple logistic regression was then completed on select variables that met the cutoff of 0.25 for a statistically significant contribution by the Wald test with the sequential addition of clinical variables that contributed positively to the model. RESULTS: Among the 73 eyes studied, 9.6% formed an ERM with an average time to formation of 1.4y. The minimum power uses during PRP was found to be statistically significant between non-ERM formers and ERM formers (P=0.044). When adjusting for all selected variables aside from minimum power used, multiple logistic regression determined that for every 10 mW increase in minimum power used during PRP, there is an increase in log odds of 1.009 (SE: 0.003, P=0.014). CONCLUSION: The incidence of ERM formation in 2y following treatment of DR with PRP is roughly 1 in every 10 eyes treated.