The Potential Association Between Postmenopausal Hormone Use and Primary Open-Angle Glaucoma

Importance  Retinal ganglion cells are known to express estrogen receptors and prior studies have suggested an association between postmenopausal hormone (PMH) use and decreased intraocular pressure, suggesting that PMH use may decrease the risk for primary open-angle glaucoma (POAG).

Objective  To determine whether the use of 3 different classes of PMH affects the risk for POAG.

Design, Setting, and Participants  Retrospective longitudinal cohort analysis of claims data from women 50 years or older enrolled in a US managed-care plan for at least 4 years in which enrollees had at least 2 visits to an eye care provider during the period 2001 through 2009.

Exposure  Postmenopausal hormone medications containing estrogen only, estrogen + progesterone, and estrogen + androgen, as captured from outpatient pharmacy claims over a 4-year period.

Main Outcomes and Measures  Hazard ratios (HRs) for developing incident POAG.

Results  Of 152 163 eligible enrollees, 2925 (1.9%) developed POAG. After adjustment for confounding factors, each additional month of use of PMH containing estrogen only was associated with a 0.4% reduced risk for POAG (HR, 0.996 [95% CI, 0.993-0.999]; P = .02). The risk for POAG did not differ with each additional month of use of estrogen + progesterone (HR, 0.994 [95% CI, 0.987-1.001]; P = .08) or estrogen + androgen (HR, 0.999 [95% CI, 0.988-1.011]; P = .89).

Conclusions and Relevance  Use of PMH preparations containing estrogen may help reduce the risk for POAG. If prospective studies confirm the findings of this analysis, novel treatments for this sight-threatening condition may follow.

Evidence suggests that sex hormones may play a role in the development of glaucoma.^1-7 Retinal ganglion cells (RGCs) express estrogen receptors,¹ and in a rat model of retinal ischemia, oral estrogen administration had a protective effect on RGCs that was mediated by these receptors.² Several clinic-based studies have shown that postmenopausal hormone (PMH) use, which usually consists of oral preparations containing estrogen, estrogen and progesterone, or estrogen along with an androgen, is associated with a modestly reduced intraocular pressure (IOP).^3-7 These data suggest that sex hormones may influence the development of open-angle glaucoma (OAG) through the lowering of IOP or protecting RGCs. Some population-based studies have found a statistically insignificant decreased risk of OAG with PMH use.^8,9 A secondary analysis from the Nurses’ Health Study (NHS), to our knowledge, has been the only report to find that women using estrogen and progesterone had a significantly lower risk of high-tension OAG, compared with nonusers of PMH.¹⁰

Benefits of PMH use include improvements in menopause-related vaginal and vasomotor symptoms and a decreased risk of fractures secondary to osteoporosis. However, in 2005, based largely on results from the Women’s Health Initiative (WHI),^11,12 the US Preventive Services Task Force recommended against routine and extended PMH use because of safety concerns including increased risks for heart disease, breast cancer, venous thromboembolism, stroke, and dementia.¹³ Interestingly, a 10-year follow-up of the WHI Estrogen-Alone Trial noted complete reversal of the adverse associations between conjugated equine estrogen use and systemic diseases among women with previous hysterectomy, whereas the protective effect against breast cancer persisted.¹⁴

Despite favorable results in the most recent WHI Estrogen-Alone Trial follow-up,¹⁴ PMH use has declined since the US Preventive Services Task Force recommendations were published.¹⁵ Additional evidence of an association between PMH use and primary open-angle glaucoma (POAG) would add support to the targeting of declining estrogen levels in the treatment of POAG in women. Using a large, nationwide health care claims database containing detailed medical records for more than 150 000 women 50 years or older, each in the plan for at least 4 years, we sought to evaluate the potential association between PMH use and development of POAG. The size of the sample allows us to investigate whether the risk for POAG differs according to the type of hormonal therapy prescribed: estrogen only (E), estrogen plus progesterone (E+P), or estrogen plus androgen (E+A), relative to nonusers of PMH.

The Clinformatics Data Mart (formerly the i3 InVision Data Mart) database (Ingenix) contains detailed records of all beneficiaries in a managed-care network with enrollees throughout the United States. The data set that we used comprises all enrollees with at least 1 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code for an eye-related diagnosis (360-379.9); at least 1 Current Procedural Terminology code for any eye-related visits, or diagnostic or therapeutic procedures (65091-68899 or 92002-92499); or any other claim submitted by an ophthalmologist or optometrist from January 1, 2001, through December 31, 2009. We had information on all these enrollees’ medical claims for ocular and nonocular conditions; sociodemographic characteristics, including age, sex, race, education level, and household net worth; and outpatient pharmacy prescriptions that were filled. Beneficiaries in the medical plan were also enrolled in the pharmacy plan. This database was used previously to study patients with glaucoma.^16-18 Because the data are deidentified, the University of Michigan determined that this study was exempt from requiring institutional review board approval.

Patients were included in the analysis if they met the following criteria: female sex, age 50 years or older, continuous enrollment in the medical plan for at least 4 years, and at least 2 visits to an eye care provider (ophthalmologist or optometrist). The ICD-9-CM and Current Procedural Terminology codes used in the analyses are listed in eTable 1 (in Supplement). Individuals with nonincident POAG (≥1 diagnoses during a 4-year “look-back” period) were excluded. Also excluded were those assigned ICD-9-CM codes for non-POAG forms of glaucoma at any time during the study period. We also excluded women who had a recorded diagnosis or procedural code for breast cancer, hysterectomy, oophorectomy, or polycystic ovarian syndrome or 1 or more prescriptions for a selective estrogen receptor modulator (tamoxifen citrate, raloxifene hydrochloride), aromatase inhibitor (anastrozole, exemestane, toremifene citrate, fulvestrant, letrozole, testolactone), or progestational agent (norgestimate + estradiol, progesterone, medroxyprogesterone acetate) (eTable 2 in Supplement). All of these conditions or medications constitute relative contraindications to PMH use.

Individuals were classified as receiving PMH if they had at least 1 outpatient pharmacy prescription for any of the following medication classes in oral or topical forms: E, E+P, or E+A (eTable 2 in Supplement). Specific medications in each class were identified by using therapeutic drug class codes and American Hospital Formulary Service drug codes. The database contains information on the number of days for which an enrollee was supplied a given medication, thereby enabling us to quantify the amount each beneficiary was prescribed during her time in the plan.

The dependent variable for this analysis was the development of POAG. These diagnoses were identified by ICD-9-CM code 365.11. Beneficiaries were identified as developing incident POAG if they received no diagnosis of POAG during the 4-year look-back period but received a diagnosis of POAG during their subsequent time in the plan. Women with other forms of glaucoma were excluded, as were those who were suspected to have glaucoma and those with ocular hypertension who lacked codes for POAG. A recent study showed that billing codes are more than 90% accurate in identifying patients with POAG, as confirmed with review of medical records.¹⁹

Statistical analyses were performed by using SAS software, version 9.3 (SAS Institute). Participant characteristics were summarized for the sample by using means and standard deviations for continuous variables and frequencies and percentages for categorical variables.

Cox regression with delayed entry was used to estimate the hazard for POAG associated with PMH use. We used the first 4 years of beneficiaries’ enrollment in the plan as their look-back period, after which the index date marked the beginning of the follow-up period. Individuals with at least 1 diagnosis of POAG during the look-back period were considered nonincident case patients with POAG and were excluded from the analysis. To ensure that each beneficiary had the opportunity to receive a diagnosis of POAG during the look-back period, we required all included beneficiaries to have at least 1 visit to an eye care provider during this period. We also required a second visit to an eye care provider during the follow-up period, to allow for an opportunity to receive a diagnosis of POAG during that time. Beneficiaries were followed in the model from the index date until they developed POAG or were censored. Censoring occurred at the last day of membership in the plan. The key predictor variable in our models was PMH use, which was treated as a time-dependent covariate. The number of days that each beneficiary was covered by a PMH prescription was totaled over a 4-year period, from the index date to POAG onset or censoring. The look-back period ensured that enrollees had a 4-year window available to measure prior PMH exposure. For the Cox regression, we used the best subset selection method to identify covariates to include in the models.²⁰ We evaluated for model inclusion the following covariates: age at index date, race, education level, household net worth, region of residence at medical plan enrollment, and urban or rural residence; the ocular comorbidities of cataract, pseudophakia or aphakia, macular degeneration, proliferative and nonproliferative diabetic retinopathy, and retinal vascular occlusion; the nonocular comorbidities of diabetes mellitus, systemic hypertension, dyslipidemia, obesity, systemic hypotension, sleep apnea, migraine headache, cardiovascular disease, cerebrovascular disease, osteoporosis, dementia, and depression; and Charlson comorbidity index (an overall health measure).²¹ Thirteen predictors were identified as the best-fitting explanatory variables for the outcome, including age at index date, race, household net worth, residential region, osteoporosis, retinal vascular occlusion, obesity, depression, diabetes mellitus, myocardial infarction, cataract, proliferative diabetic retinopathy, and pseudophakia or aphakia. The final multivariable model included the 13 predictors, along with use of each PMH class. For all analyses, P < .05 was considered statistically significant.

A total of 152 163 female enrollees 50 years or older met the study inclusion criteria (Figure). These beneficiaries were enrolled in the plan for a mean (SD) of 6.4 (1.5) years. Their mean age (SD) was 65.1 (8.9) years, and their racial distribution included 123 045 whites (80.9%), 6170 blacks (4.1%), 5763 Latinas (3.8%), 2502 Asian Americans (1.6%), and 998 individuals of other races (0.7%) (Table 1). The records of the remaining 13 685 enrollees (9.0%) were missing information on race/ethnicity. Nearly half the enrollees who met eligibility criteria (n = 70 292 [46.2%]) had received a prescription for at least 1 PMH medication of any class during their time in the plan. Among enrollees with PMH prescriptions, 59 847 (39.3%) had at least 1 prescription for E, 15 288 (10.1%) for E+P, and 4441 (2.9%) for E+A during their time in the plan. Those beneficiaries prescribed E medications had prescriptions filled for a mean (SD) of 753 (719) days. Beneficiaries prescribed E+P medications had prescriptions filled for a mean (SD) of 670 (594) days. Enrollees prescribed E+A medications had prescriptions filled for a mean (SD) of 708 (679) days. Over the course of the study, the median number of visits to eye care providers for women who used PMH and for those who did not use PMH was 5.

Overall, 2925 individuals (1.9%) had an incident POAG diagnosis during their time in the plan. The mean (SD) age of those with incident POAG was 66.3 (9.2) years, compared with 65.1 (8.9) years for those not developing POAG (Table 1). Relative to whites, blacks (adjusted hazard ratio [HR], 1.72 [95% CI, 1.48-2.01]; P < .001) and Latinas (adjusted HR, 1.41 [95% CI, 1.19-1.67]; P < .001) had an increased hazard for POAG. Relative to beneficiaries with a household net worth less than $25 000, beneficiaries with a household net worth of $150 000 to 500 000 or more than $500 000 had a decreased hazard for POAG (adjusted HR, 0.83 [95% CI, 0.70-0.98]; P = .03, and 0.79 [95% CI, 0.66-0.94]; P = .007, respectively) (Table 2).

The proportions of women taking E, E+P, and E+A who developed POAG were 1.7%, 1.9%, and 1.4%, respectively. By comparison, 2.1% of those using no PMH developed POAG. After adjustment for age alone, women using E or E+P had a slightly decreased hazard for POAG. Each additional month of use reduced the risk by 0.6% for E alone (adjusted HR, 0.994 [95% CI, 0.991-0.997]; P < .001) and by 1.2% for E+P (adjusted HR, 0.988 [95% CI, 0.982-0.995]; P < .001), relative to PMH nonusers. Use of E+A had no association with development of POAG (adjusted HR, 0.994 [95% CI, 0.983-1.006]; P = .33).

In the multivariable analysis, after adjustment for age plus other sociodemographic factors and ocular and systemic comorbidities, the association between E use and development of POAG was statistically significant. Every additional month of E use reduced the hazard for POAG by 0.4% (adjusted HR, 0.996 [95% CI, 0.993-0.999]; P = .02). We found no association between E+P (adjusted HR, 0.994 [95% CI, 0.987-1.001]; P = .08) or E+A (adjusted HR, 0.999 [95% CI, 0.988-1.011]; P = .89) use and POAG (Table 2).

In our analysis of more than 150 000 US women, a lower proportion of PMH users than nonusers developed POAG. After adjustment for age and other possible confounding factors, women prescribed E had a 0.4% per month reduction in POAG risk and those prescribed E+P had a 0.6% per month reduction.

Several other studies have assessed the relationship between PMH use and risk for glaucoma. The Rotterdam Study⁸ and the Blue Mountain Eye Study⁹ each showed up to a 50% reduced risk of OAG among users of PMH; neither of these study findings was determined to be statistically significant, which may have been due to their relatively small sample sizes of those experiencing the outcome. A secondary analysis of data from the NHS¹⁰ found that the use of medications containing E+P was associated with a reduction in POAG risk in a subset of women with elevated IOP, a finding that was statistically significant. One population-based study that did not demonstrate an association between PMH use and risk of POAG was the Los Angeles Latino Eye Study.²² Direct comparison of these various studies with one another and with ours can be challenging because the sample populations differ with respect to sociodemographic characteristics, such as age and race/ethnicity, both of which are key factors that themselves are known to affect the risk of developing POAG. Although in our study only PMH containing E alone significantly reduced the risk for POAG, other population-based studies did not collect data on the type of PMH used, as they simply asked women whether they had used PMH previously. Differences in study design and ways of identifying OAG development may also explain why some, but not all, of these studies found a reduced risk for glaucoma among PMH users. Collectively, the findings of our analyses, when viewed along with some others, offer evidence that sex hormone use may indeed affect the likelihood of developing glaucoma in a subset of women.

Longer exposure to estrogen has been postulated to protect the optic nerve from OAG in women compared with men until menopause, when endogenous estrogen exposure becomes similar between the sexes. Women, on average, live longer than men and thus have a longer opportunity to experience glaucoma-related visual impairment after estrogen protection of the optic nerve ceases. The Rotterdam study²³ found that early menopause was associated with an increased risk for OAG, and Pasquale and colleagues¹⁰ showed that among NHS participants older than 65 years, the incidence of OAG was decreased with older age at menopause. Unfortunately, no consistent billing code exists for menopause status and thus we could not assess the potential effect of this variable on OAG risk; conceivably, the inability to account for menopause status may have affected the results.

We also explored the possibility of a dose-response effect and found that each additional month of E or E+P use further reduced the risk for POAG. Assuming a linear relationship between PMH exposure and POAG, we found that use of E or E+P continuously for 4 years yielded a hazard reduction of 18% and 26%, respectively (data not shown). In the NHS—which captured participants’ medication use for as long as 20 years and, to our knowledge, is the only study that quantified PMH use for longer than our study—women’s risk for POAG was reduced by 42% with use of E+P, although more than 5 years of use was not associated with an additional protective effect.¹⁰

Others have demonstrated that the impact of PMH use on glaucoma risk may be genotype specific or may depend on gene-environment interactions. For example, Kang and colleagues²⁴ evaluated the relationship between certain nitric oxide synthase (NOS3) single-nucleotide polymorphisms, sex, PMH use, and POAG. They found that, overall, the different NOS3 polymorphisms were not associated with POAG. However, for women with certain NOS3 genotypes, PMH therapy reduced the risk of the high-tension variant of POAG.

Although the inverse association between E use and POAG risk could be due to chance, the mounting biological data linking declining sex hormones to POAG—including estrogen protection of RGCs from death in an animal model of retinal ischemia² and clinic-based studies in which PMH use is associated with IOP reduction—suggest that the association is valid, although additional confirmation is needed. Evidence that estrogen protects against other neuronal diseases in animal models and humans is compelling. For example, when estrogen therapy has been initiated close to menopause onset, denying neurons a critical period without estrogen, women’s risks for Alzheimer disease, Parkinson disease, and cognitive decline decrease.^25-28 Our subsample of E users (n = 59 847) was considerably larger than the E+P (n = 15 288) and E+A (n = 4441) groups, and this may partially explain our finding of statistical significance for the E-alone users only, not the other PMH-using groups.

Our study has several limitations. First, we relied on claims data—a source that lacks information on clinical parameters (eg, IOP, extent of visual field loss, central corneal thickness, age at menopause, duration of endogenous estrogen exposure, and parity). By using billing data instead of clinical data, we may have included some patients who were erroneously assigned POAG-related codes or excluded others whose true diagnosis of POAG was not coded. However, such coding errors would be problematic only if a differential misclassification of glaucoma status existed between PMH users and nonusers, which we consider unlikely. Second, because all patients had health insurance, our findings may not be generalizable to uninsured persons, who tend to disproportionately comprise racial minorities and socioeconomically disadvantaged persons. Third, we did not consider patients’ possible lack of adherence to prescribed PMH therapies. However, if many recipients of PMH prescriptions were nonadherent, the results would be biased toward the null hypothesis. If all enrollees prescribed PMH were fully adherent, it is likely that the risk of POAG would be reduced even more than what we report. Finally, PMH users may be more health conscious than other enrollees are and more prone to seek eye care—although this would bias the results toward an increased risk for POAG diagnosis with PMH use, which is not what we found.

To our knowledge, this is the largest study to assess a potential relationship between PMH use and glaucoma. Claims data allowed us to more accurately capture PMH exposure, by using pharmacy records rather than relying on patients’ self-report. Our study also benefits from a patient sample that is more diverse than those of population-based studies limited to particular geographic regions and the available supply of willing participants. In addition, our numbers were ample to explore the relationship between several classes of PMH and to account for several key potential confounding variables.

Over all, these findings, along with results of other population-based studies, suggest that PMH use may generally affect women’s risk for glaucoma. More research is needed to better elucidate the complex relationship between PMH use and glaucoma. Additional work should also further explore whether the risk for POAG is affected only by use of E alone, or by use of any PMH class. Ongoing studies exploring how estrogen may affect apoptosis of the RGC layer may ultimately lead to the identification of novel mechanisms by which sex hormones can affect the development of POAG or novel, nonoral routes of PMH administration that may further decrease the risk for POAG.

Submitted for Publication: March 13, 2013; final revision received August 10, 2013; accepted August 27, 2013.

Corresponding Author: Joshua D. Stein, MD, MS, University of Michigan, Department of Ophthalmology and Visual Sciences, W. K. Kellogg Eye Center, 1000 Wall St, Ann Arbor, MI 48105 (jdstein@med.umich.edu).

Published Online: January 30, 2014. doi:10.1001/jamaophthalmol.2013.7618.

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

Study concept and design: Newman-Casey, Stein.

Acquisition of data: Stein.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: Newman-Casey.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Talwar, Nan.

Obtained funding: Stein.

Administrative, technical, and material support: Newman-Casey, Stein.

Study supervision: Pasquale.

Funding/Support: Grant support: National Eye Institute K23 Mentored Clinician Scientist Award (1K23EY019511-01 to J.D.S.); American Glaucoma Society Clinician Scientist Grant (to J.D.S.); Blue Cross Blue Shield of Michigan Foundation (to J.D.S. and P.A.N.-C.); Research to Prevent Blindness (to D.C.M. and J.D.S.); Heed Foundation Fellowship (P.A.N.-C.); National Eye Institute Core Grant EY07003; National Eye Institute RO1 EY015473 (to L.R.P.); Harvard Glaucoma Center of Excellence, the Margolis Fund (to L.R.P.); and the Arthur Ashley Foundation (to L.R.P.).

Role of the Sponsors: The sponsors had no role in the design and conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Previous Presentations: Presented in part at the American Glaucoma Society Annual Meeting; March 3, 2012; New York, New York; and the American Academy of Ophthalmology Annual Meeting; November 12, 2012; Chicago, Illinois.