Epiretinal membranes (ERMs) are fibro-cellular sheets on the inner retinal surface that can distort the macula and reduce vision. In the United States, their overall prevalence is approximately 11% in adults aged 40 and older, with notably higher rates in Chinese Americans and non-white Hispanics. ^, Although some ERMs remain asymptomatic, approximately 60% to 70% of patients can experience metamorphopsia (vision distortion) and aniseikonia (image distortion) along with reduced contrast sensitivity. ^, Over time, up to 40% of ERMs enlarge enough to require surgical peeling. ERMs are thought to develop after disruption of the internal limiting membrane (ILM), triggering Müller cell activation, cellular migration, and collagen deposition on the vitreoretinal interface, creating a scar-like contractile membrane on the retinal surface that may distort vision. ^,,

ERM formation can be idiopathic and is typically associated with aging, however it can also occur in response to various retinal insults. Previously identified risk factors include a history of intraocular surgery, retinal vascular occlusion, diabetic retinopathy (DR), retinal detachment, vitreous degeneration, and ocular trauma. ^,,,, Retinal vein occlusion (RVO) may contribute to a secondary ERM formation by stimulating the release of inflammatory cytokines such as VEGF (vascular endothelial growth factor). These cytokines can drive vascular leakage and fibrosis, leading to macular swelling and visual impairment. ^, Several studies have documented ERMs in approximately 15% of eyes following central (CRVO) or branch (BRVO) retinal vein occlusion. ^,,, This structural change due to ERMs may also limit the delivery of standard treatments for RVO, such as intravitreal anti-VEGF therapy, by potentially acting as a physical barrier thereby limiting visual recovery.

Prior studies have primarily examined ERM formation as a binary outcome in eyes with prior RVO and have not evaluated whether treatment with anti-VEGF agents may influence its development. ^,, By leveraging a national multicenter database, we aim to examine the time period when patients with RVO are at greatest risk for ERM development and assess whether anti-VEGF therapy can modify outcomes.

Retrospective cohort study using a multi-institutional U.S. clinical database (TriNetX), comprised of 70 U.S. healthcare organizations with over 110 million patients. The database follows Health Insurance Portability and Accountability Act compliance protocols and has an institution review board waiver. The study design and reporting adhere to all tenets of the Declaration of Helsinki. This data was extracted on August 1st, 2025, and included demographics, comorbidities, procedures, and laboratory records dating back to August 1st, 2005. Sex (as defined at birth) and race are self-reported. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines were applied in this study.

In this study, patients aged ≥18 years who underwent ophthalmic examination were identified using current procedural terminology (CPT) for ophthalmology services and procedures (CPT 1 012 793). Patients were then stratified using International Classification of Diseases, 10th Revision (ICD-10) by the presence of BRVO only (BRVO cohort; H34.83), CRVO only (CRVO cohort; H34.81), and neither BRVO nor CRVO (controls). We hypothesized a priori that retinal vein occlusion (either BRVO or CRVO) may have an elevated risk for ERM development and an increased incidence for subsequent ERM peel. Exclusion criteria included patients diagnosed with proliferative diabetic retinopathy (PDR) (ICD-10: E10.35/E11.35) and nonproliferative diabetic retinopathy (NPDR) (E10.32/E11.32, E10.311/E11.311, E11.331, E11.341, E11.351) to reduce the possibility that surgical intervention was confounded by DR sequalae. Patients were also excluded if they had a history of surgical retinal detachment repair (current procedural terminology (CPT): 67108, 67113), panretinal photocoagulation (CPT: 67228) or a history of outcome measures. Patients were required to have at least 6 months of follow-up.

Outcomes were evaluated at 3 months, 6 months, 1, 3, and 5 years after the index date, defined as the time from initial diagnosis of either a BRVO or CRVO with an encounter for ophthalmic services or procedures. In the control group, the index date was considered at the time of an encounter for ophthalmic services and procedures without a history of vascular occlusion. Primary outcomes included the development of ERM (ICD-10: H35.37) and the incidence of subsequent ERM (including ILM) peel (CPT: 67041, 67042).

A sensitivity analysis stratified the primary cohorts by eye laterality (right [OD] or left [OS]) to account for potential asymmetry in disease presentation and outcomes. Patients were categorized as BRVO-right (BRVO-R; ICD-10: H34.831), BRVO-left (BRVO-L; H34.832), CRVO-right (CRVO-R; H34.811), and CRVO-left (CRVO-L; H34.812) and compared with their respective controls. Exclusion criteria included patients diagnosed with proliferative diabetic retinopathy (PDR) (ICD-10: E10.35/E11.35) and had a history of surgical retinal detachment repair (current procedural terminology (CPT): 67108, 67113), panretinal photocoagulation (CPT: 67228) or a history of outcome measures. Patients were required to have at least 6 months of follow-up. Main measure outcomes were assessed at 1 year follow-up.

An additional secondary analysis was performed in patients with either BRVO or CRVO treated with or without intravitreal anti-VEGF therapy (CPT: 67028) to further evaluate the effects of anti-VEGF treatment on ERM formation and the incidence of subsequent ERM peel. This analysis was conducted post hoc, following initial identification of the association between retinal vein occlusions and ERM formation. In this analysis, patients with a diagnosis of BRVO or CRVO ICD-10 codes were stratified by the presence or absence of anti-VEGF treatment (BRVO-T: branch retinal vein occlusion anti-VEGF–treated; CRVO-T: central retinal vein occlusion anti-VEGF–treated) in each respective cohort. Exclusion criteria included patients diagnosed with proliferative diabetic retinopathy (PDR) (ICD-10: E10.35/E11.35) and had a history of surgical retinal detachment repair (current procedural terminology (CPT): 67108, 67113), panretinal photocoagulation (CPT: 67228) or a history of outcome measures. Patients were required to have at least 6 months of follow-up. Main measure outcomes were evaluated at 3 months, 6 months, 1, 3, and 5 years after the index date.

Further sub-analyses were performed to evaluate the role of posterior vitreous detachment (PVD) on outcomes. Initially, the risk of PVD (ICD-10: H43.81) development was analyzed among patients with BRVO and CRVO cohorts compared with their respective matched controls without RVO at 1 year to determine whether RVO was associated with an increased risk of PVD. Eyes with baseline PVD were excluded. Similarly, the anti-VEGF–treated cohorts (BRVO-T and CRVO-T) were analyzed separately and compared with matched RVO controls who did not receive anti-VEGF therapy to assess whether anti-VEGF treatment influenced the risk of PVD development at 1 year. A separate analysis then evaluated the risk of ERM development at 1 year among anti-VEGF–treated RVO eyes with PVD compared to anti-VEGF–treated RVO eyes without PVD to determine whether PVD in treated RVO eyes was associated with an increased risk of ERM development.

A post hoc sub-analysis was conducted using a multivariate Cox proportional hazards regression model to assess the relative contribution of covariates on the risk of ERM formation (ICD-10: H35.37) and ERM peel (CPT: 67041, 67042) at 5 years in the primary BRVO and CRVO cohorts, respectively. Examined variables included age, sex, anti-VEGF exposure (aflibercept: RxNorm 1232150; bevacizumab: 253337; ranibizumab: 595060) and intravitreal steroid exposure (10759). Additional agents, for example, faricimab could not be included into the Cox model due to low sample size.

Before PSM, 20,385 individuals with BRVO, 15,823 individuals with CRVO, and 2670,657 controls aged ≥ 18 years were identified. After PSM, a total of 19,172 patients with BRVO and 14,4974 patients with CRVO were compared with their respective 1:1 matched control cohort. The baseline characteristics before and after PSM in each cohort are shown in Tables 1–2 .

Prior to PSM, there was an increased prevalence across all demographics and ocular comorbidities in both BRVO and CRVO cohorts compared with their respective unmatched controls. The mean (± SD) age was 72.6 (10.3) years in the BRVO cohort compared with 66.9 (11.2) years in the unmatched control cohort, and 71.2 (11.0) years in the CRVO cohort compared with 66.9 (11.2) years in the unmatched control cohort. The BRVO cohort included 8429 males (43.5%) compared with 1080,014 males (42.2%) in the unmatched control cohort, while the CRVO cohort included 7158 males (47.4%) compared with 1080,014 males (42.2%) in the unmatched control cohort. Interestingly, vitreous degeneration was present in 12.7% of the BRVO cohort compared with 2.8% of controls, and in 10.4% of the CRVO cohort compared with 2.8% of controls prior to matching.

After PSM, the mean (± SD) age at index was 72.5 (910.3) years in the BRVO cohort compared with 72.3 (10.3) years in the matched control cohort, and 72.6 (11) years in the CRVO cohort compared with 72.5 (10.8) years in the matched control cohort. The BRVO cohort included 8348 males (43.5%) compared with 8454 males (44.1%) in the control cohort, while the CRVO cohort contained 7097 males (47.4%) compared with 7135 males (47.7%) in the control cohort. After PSM, these cohorts were statistically balanced, with no significant differences specifically in risk factors such as vitreous degeneration, cataract surgery, nicotine dependence, vascular disease, or eye injury.

Table 3 demonstrates the incidence of ERM formation and ERM peel in patients with BRVO compared with matched controls without BRVO. The mean (± SD) follow-up time was 1084 (682) days for the BRVO cohort and 1156 (689) days for matched controls. The BRVO cohort had a higher risk of ERM formation than matched controls at 3 months (1.33% vs. 0.29%; RR, 4.54; 95% CI, 3.40-6.05; p <.0001), 6 months (1.87% vs. 0.46%; RR, 4.07; 95% CI, 3.23-5.13; p <.0001), and 1 year (2.74% vs. 0.65%; RR, 4.24; 95% CI, 3.49-5.15; p <.0001). Figure 1 depicts the Kaplan–Meier curves for the probability of ERM formation over time in BRVO cohorts compared with matched controls. The BRVO cohort also had a higher incidence of ERM peel compared with matched controls at 3 months 0.05% vs. 0.05%; RR, 3.30; 95% CI, 1.63-6.69; p <.0001), 6 months (0.18% vs. 0.05%; RR, 3.50; 95% CI, 1.73-7.07; p =.0002), and 1 year (0.24% vs. 0.06%; RR, 3.83; 95% CI, 2.03-7.23; p <.0001). Figure 2 depicts the Kaplan–Meier curves for the probability of ERM peel incidence over time in BRVO cohorts compared with matched controls.

Probability of ERM formation over time in BRVO cohorts compared with matched controls. BRVO = branch retinal vein occlusion; ERM = epiretinal membrane.

Stay updated, free articles. Join our Telegram channel

Join