Association of Retinal Vascular Occlusion With Women Filling a Prescription for Female Hormone Therapy

Key Points

Question  Is there an increased risk of retinal artery occlusion (RAO) or retinal vein occlusion (RVO) with women filling a prescription for female hormone therapy (FHT)?

Findings  In this cohort study of more than 205 000 women who filled a prescription for FHT, there was no evidence to suggest an increased risk of RAO or RVO.

Meaning  These findings suggest that filling a prescription for FHT, and presumably taking FHT, does not increase the risk of RAO or RVO; such history may not be relevant in the evaluation of an individual with an RAO or RVO.

Importance  Oral contraceptives have been associated with cardiovascular disease, ischemic stroke, venous thromboembolic disease, and breast cancer. Retinal vascular occlusions share the same risk factors as cardiovascular and cerebrovascular disease.

Objective  To determine whether filling a prescription of female hormone therapy (FHT) is associated with an increased risk of retinal artery occlusion (RAO) or retinal vein occlusions (RVO).

Design, Setting, and Participants  A multiple-cohort study was conducted using an administrative claims insurance database comparing women who filled a prescription for FHT with matched control individuals. Exclusion occurred for those enrolled for less than 2 years in the plan, with no prior ophthalmologic examination, with a history of a RAO/RVO, with systemic diseases/medications that affected estrogen levels, or a disease associated with an increased risk for thromboembolism.

Main Outcomes and Measures  The primary outcome was the incidence of a new diagnosis of RAO or RVO. Cox proportional hazard regression modeling with inverse probability of treatment weight was used to assess the hazard ratio (HR) for a new diagnosis of RAO or RVO relative to filling prescription for FHT. Subanalyses were conducted to stratify by age, race/ethnicity, diabetes, and hypertension.

Results  A total of 205 304 women who filled a prescription for FHT were matched to 755 462 control individuals. After inverse probability of treatment weight, the study cohort was a mean age of 47.2 years, 71% were White, 7% were Black, 6% were Hispanic, 3% were Asian, and 3% were unknown. There were 41 cases (0.01%) of RAO and 68 cases of RVO (0.02%) in the FHT cohort. In comparison, there were 373 cases of RAO (0.05%) and 617 cases of RVO (0.08%) in the control cohort. After inverse probability of treatment weight, Cox regression analysis showed no difference in hazard for RAO, RVO, or combined outcomes in the FHT cohort relative to the control cohort (RAO HR, 1.17; 95% CI, 0.83-1.65; P = .36; RVO HR, 1.07; 95% CI, 0.82-1.39; P = .65; combined HR, 1.10; 95% CI, 0.89-1.36; P = .37). Subanalyses that stratified by age, diabetes, and hypertension similarly showed no significant associations between the FHT prescription cohort and all outcomes.

Conclusions and Relevance  These findings suggest that filling a prescription for FHT, and presumably taking FHT, does not increase the risk of RAO or RVO. Such history may not be relevant in the evaluation of an individual with an RAO or RVO nor do our results support stopping FHT in an individual who develops an RAO or RVO.

Retinal vascular occlusion can lead to permanent vision loss and is the second most common cause of blindness from retinal vascular diseases.^1-4 Systemic risk factors, such as hypertension, hyperlipidemia, and type 2 diabetes, are associated with both retinal artery (RAO) or retinal vein occlusions (RVO).^3,4

Female hormone therapy (FHT) includes oral contraceptive pills (OCPs) in women of reproductive age and hormone therapy in postmenopausal women. Female hormone therapy has been shown to increase the risk of cardiovascular,^5-8 venous thromboembolism,^5,9-11 pulmonary embolism,^12,13 and stroke.^14-16 Given that retinal vascular occlusions share similar risk factors to those of cardiovascular and cerebrovascular diseases, it is therefore reasonable to suspect the FHT may predispose toward retinal vascular occlusion.

To date, numerous case reports have associated oral estrogen use with developing RVO^17-19 or RAO.^20-22 However, besides a single study from 1998,²³ no large cohort series have been published on this topic. Very early studies (published prior to 1980) were based on the use of second-generation oral contraceptive pills that contained 80 or 100 μg of estrogen.^14,24 Third-generation OCPs are most frequently used in the United States and contain only 30 or 35 μg of estrogen.¹⁴ Low-dose OCPs appear to be less associated with an increased risk of cerebrovascular diseases.¹⁴

In this study, we investigated the incidence of RVO and RAO in patients who filled the prescription of FHT compared with age and race-matched control patients who were not prescribed with estrogen products. Also, we evaluated whether filling of FHT further increased the risk of RVO and RAO in patients with other known risk factors.

All data were abstracted from the Clinformatics Data Mart Database (OptumInsight), which includes the deidentified medical claims of all beneficiaries in a commercial and Medicare Advantage database obtained by a large nationwide US insurance provider. Included within the database are all outpatient medical claims for office visit and their associated diagnoses, outpatient pharmaceutical prescriptions, and demographic data for each beneficiary during their enrollment in the insurance plan from January 1, 2000, to June 30, 2019. Because of the deidentified nature of the data set and minimal risk to patients within the data set, the University of Pennsylvania institutional review board deemed this project exempt from review and waived the requirement for informed consent.

Quiz Ref IDMultiple cohorts were created based on whether a prescription was filled for FHT for all women 18 years and older. The index date was considered the earliest date a prescription was filled. Exclusion occurred for patients with less than 2 years in the insurance plan or who did not have a visit to an eye-care clinician prior to the index date. Patients were also excluded if, prior to the index date, they had a prescription for an estrogen-modifying drug or a diagnosis of an estrogen-modifying disease. Furthermore, patients were excluded if they had systemic thrombotic diseases including collagen vascular disease, coagulopathy, sickle cell disease, homocystinuria, platelet abnormalities, or hyperviscosity. Additionally, to avoid confounders and to decrease the chance of a misdiagnosis, patients with the following diagnoses were also excluded: history of intraocular surgery, diabetic retinopathy, glaucoma, optic disc drusen, Susac syndrome, RAO, RVO, giant-cell arteritis, or nonarteritic anterior ischemic optic neuropathy (see the eTable in the Supplement for a full list of the International Classification of Diseases, Ninth Revision and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes used within this study). Each patient with a prescription of FHT was then matched with up to 5 female patients who had not been prescribed with FHT and who met all of the inclusion and exclusion criteria. Patients were matched on age (SD, 3 years), race/ethnicity, and insurance eligibility start date (SD, 3 months). Control patients were assigned the same index date as their corresponding match. All unmatched FHT-prescribed patients were excluded (a comparison of baseline demographic and systemic differences between unmatched and matched women who filled a FHT prescription was summarized in the eTable in the Supplement).

In an effort to better balance the covariates between the FHT prescription-filling group and control groups, a propensity score for having been prescribed an FHT was created using multivariable logistic regression. Covariates that were included in the propensity score were continuous age (to account for residual differences in age after matching within 3 years), education level, household income, geographic region of the country, index year, the diabetes complications severity index, previous diagnosis of hypertension, hypercholesterolemia, chronic kidney disease, and smoking status. The diabetes complications severity index is a diabetes disease severity scale derived from outpatient diagnosis codes that are divided into 6 diabetes-related complication categories, each scored 0 to 2. This scale has been shown to better predict important clinical end points than hemoglobin A_1c and duration of disease.²⁵ While smoking is a difficult variable to obtain in an administrative data set, we applied a previously used approach consisting of smoking diagnosis codes, use of antismoking drugs and Current Procedural Terminology (CPT) codes for smoking cessation counseling, which previously found smoking rates in administrative databases as high as 10% to 11%, similar to the general population.²⁶ Inverse probability of treatment weighting (IPTW) with the propensity scores was then applied to address residual confounding by baseline covariates between the FHT prescription and control cohorts after matching.

The primary outcome of interest was an incident diagnosis of RAO, RVO, and a combined outcome of either RAO or RVO. Women in the FHT prescription group who had greater than a 30-day gap in FHT prescription coverage were censored at 30 days after end of coverage but were allowed to reenter the cohort as a new patient, assuming all inclusion and exclusion criteria were still met as of the date of a new FHT prescription. Partly conditional Cox proportional hazards regression was performed for all analyses to account for reentry of patients with a new FHT-prescription after a 30-day gap in prescription coverage, assuming they still met eligibility criteria. Patients were also censored at the end of their insurance eligibility for the occurrence of an exclusion criterion. Demographic data at baseline were evaluated at the time of index date by descriptive statistics. Means and ranges were used for continuous variables, and percentages were used for categorical variables weighted by IPTW. All statistical analysis was performed via SAS, version 9.4 (SAS Institute Inc) software. Given that increased age, diabetes, and hypertension have each been associated with increased hazard of the outcomes,²⁷ stratified analysis was performed by age, diabetes, and hypertension.

A total of 205 304 patients who filled prescription of FHT and 755 462 matched control patients (Figure 1) met criteria to be studied. The IPTW weighted mean (SD) age was 47.2 (16.3) years in the FHT prescription and 47.1 (16.7) years in the control cohorts. The baseline characteristics of both groups weighted by IPTW are demonstrated in Table 1. After accounting for IPTW, no differences in baseline covariates were seen (defined as a standard mean difference of <0.1). The median time in plan after the index date for the FHT prescription cohort was 190 days (interquartile range [IQR], 58-211 days) and 719 days (IQR, 202-994 days) for the control patients. The FHT prescription cohort was much shorter owing to the requirement that the patient continue to have an active prescription for the medication, a requirement not made of the control patients.

Quiz Ref IDOverall in the FHT prescription group, 41 RAO cases (0.01%), 68 RVO cases (0.02%), and 111 combined outcomes (0.03%) were found. In comparison, there were 373 RAO cases (0.05%), 617 RVO cases (0.08%), and 1003 combined cases (0.13%) in the control cohort (Table 2). The IPTW Kaplan-Meier (KM) event rates at 1 year for patients who filled prescription of FHT were 0.00015 and 0.00019 for the control groups with regard to RAOs. The RVO KM event rate was 0.00038 and 0.00033 for FHT prescription group and control group, respectively. The combined KM rate was 0.00053 for the FHT-prescription cohort and 0.00054 in the control group. The KM curves of all RAO, RVO, and combined outcomes are shown in Figure 2.

Quiz Ref IDAfter weighted by the IPTW, no difference was seen in the hazard of developing RAO (hazard ratio [HR], 1.17; 95% CI, 0.83-1.65; P = .36), RVO (HR, 1.07; 95% CI, 0.82-1.39; P = .65), or the combined outcome (HR, 1.10; 95% CI, 0.89-1.36; P = .37) when comparing patients who filled a prescription of FHT with control patients. After stratification by age, diabetes, and hypertension no significant difference was found between FHT prescription cohort vs control cohort for all 3 outcomes (Table 3).

Our study assessed the risk of RAO and RVO or a combined outcome in a cohort of patients who filled a prescription for FHT. In this study, we found no evidence that filling a prescription for FHT confers any increased risk of RAO or RVO. This was true across all analyses, including analyses stratified by age, diabetes, and hypertension. These findings suggest that filling a prescription for FHT, and presumably taking that female hormone therapy, does not increase the risk of RAO or RVO. As such, a history of FHT use may not be relevant in the evaluation of an individual with a retinal artery or retinal vein occlusion. Nor do our results support considering cessation of FHT in an individual who develops an RAO or RVO. Despite many forms of oral contraceptives and female hormonal therapy being recognized as a risk factor for thromboembolisms,^9,12,14,28 these hormones do not appear to affect the eye as they do in other areas of the body.

One possible explanation specific to RVOs could be owing to their pathophysiologic nature. Retinal vein occlusions are thought to occur most frequently when the supple venular wall is impinged from a sclerosing artery,¹ which would be independent of a clotting effect induced by hormone-related clotting factor changes. Furthermore, third-generation oral contraceptives, also called low-dose oral contraceptives (30 to 35 μg of estrogen and 150 μg levonorgestrel or 1 mg of norethindrone), account for more than 85% of pharmacy purchases of oral contraceptives in developed countries since the late 1980s.²⁹ This lower-dose formulation may likely also confer less risk. This is also true of hormone in postmenopausal women that typically uses lower or “more physiologic” doses of hormones as well.³⁰

Few epidemiologic studies have investigated FHT use and retinal vascular diseases. Two large British cohort studies²³ of the benefits and risks of oral contraceptives analyzed the issue. They reported a 2-fold increase in the risk of retinal vascular lesions in the FHT users in both studies.²³ However, participants in both studies were recruited from 1968 to 1974 and therefore also likely assessed the riskier second-generation formulations of oral contraceptives.

Quiz Ref IDOne limitation of our study is that while we know that a prescription of FHT was filled by a patient, we are unable to confirm whether the medication was actually taken by the patient. Additionally, a limitation of claims data is the inability to link a pharmacy prescription with a specific diagnosis. This limited our study’s ability to specifically assess oral contraceptives vs postmenopausal hormone therapy. To address this, we performed an age-stratified analysis presuming those women older than 45 years received hormone therapy and those younger than 45 years received oral contraceptives. Regardless, no differences were seen in our age-stratified analyses for any of our outcomes. Another limitation is the inability to verify diagnosis codes with medical record data. Next, it is possible that the risk for RAO/RVO is dose dependent and the follow-up time within the database was not long enough to confer an identifiable risk to FHT users. Another issue that may have affected the study was a lack of mandated eye examinations after the index date. Without mandated eye examinations, it is possible that asymptomatic RAO and RVOs went undetected, and we cannot rule out this possibility. Lastly, the data come from a single large US insurer and may or may not reflect patients from other insurers or from the uninsured population.

In summary, we investigated the risk of RAO and RVO in women using FHT and found that the incidence of RAO and RVO in young women is very low and in our study was not associated with their status of filling prescription for FHT. Our results offer evidence suggesting that filling a prescription for FHT does not increase the risk of RAO or RAO in women.

Corresponding Author: Brian L. VanderBeek, MD, MPH, MSCE, Scheie Eye Institute, Department of Ophthalmology, Perelman School of Medicine at the University of Pennsylvania, 51 N 39th St, Philadelphia, PA 19104 (brian.vanderbeek@pennmedicine.upenn.edu).

Accepted for Publication: September 27, 2020.

Published Online: November 12, 2020. doi:10.1001/jamaophthalmol.2020.4884

Author Contributions: Dr VanderBeek had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Song, VanderBeek.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Song, Nadelmann, VanderBeek.

Critical revision of the manuscript for important intellectual content: Song, Nadelmann, Yu.

Statistical analysis: Song, Yu, VanderBeek.

Obtained funding: VanderBeek.

Administrative, technical, or material support: Nadelmann, VanderBeek.

Supervision: VanderBeek.

Conflict of Interest Disclosures: None reported.

Funding/Support: National Institutes of Health K23 Award (1K23EY025729-01) and University of Pennsylvania Core Grant for Vision Research (2P30EY001583). Additional funding was provided by Research to Prevent Blindness, Karen & Herbert Lotman Fund for Macular Vision Research Foundation, and the Paul and Evanina Mackall Foundation.

Role of the Funder/Sponsor: None of the funding organizations had any role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.