Study participants. BCN indicates Blue Care Network.
Multivariable analysis of the hazard of death associated with the indicated factors. ART indicates arterial disease; BB, β-blocker; CAN, cancer (any type); CHF, congestive heart failure; CKD, chronic kidney disease; CLD, chronic liver disease; CVD, cerebrovascular disease; DM, diabetes mellitus; IHD, ischemic heart disease; NAG, narrow-angle glaucoma; NVG, neovascular glaucoma; and OAG, open-angle glaucoma. Error bars represent 95% confidence intervals.
Graph of glaucoma medication use in beneficiaries taking glaucoma medications on January 1, 2006. AA indicates α-agonist; BB, β-blocker; other, topical and oral anhydrase inhibitors, miotics, and epinephrine compounds; and PGA, prostaglandin analogue.
Stein JD, Newman-Casey PA, Niziol LM, Gillespie BW, Lichter PR, Musch DC. Association Between the Use of Glaucoma Medications and Mortality. Arch Ophthalmol. 2010;128(2):235-240. doi:10.1001/archophthalmol.2009.378
PAUL P.LEEMDAuthor Affiliations: Department of Ophthalmology (Drs Stein, Newman-Casey, Lichter, and Musch and Ms Niziol) and School of Public Health (Drs Gillespie and Musch), University of Michigan, Ann Arbor.
Copyright 2010 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2010
To evaluate the relationship between glaucoma medication use and death.
This study uses longitudinal data from 2003 to 2007 on persons 40 years and older with glaucoma or suspected glaucoma enrolled in a large managed care network. Cox regression analysis was performed to estimate the hazard of death associated with the use of various glaucoma medication classes and combinations thereof. Multivariable models were adjusted for demographic characteristics and comorbid medical conditions.
Of 21 506 participants with glaucoma or suspected glaucoma, 237 (1.1%) died during the study period. The use of any class of glaucoma medication was associated with a 74% reduced hazard of death (adjusted hazard ratio [HR], 0.26; 95% confidence interval [CI], 0.16-0.40) compared with no glaucoma medication use. This association was observed for use of a single agent alone, such as a topical β-antagonist (0.44; 0.24-0.83) or a prostaglandin analogue (0.31; 0.18-0.54), and for use of different combinations of drug classes.
After adjustment for potential confounding variables, the use of glaucoma medications was associated with a reduced likelihood of death in this large sample of US adults with glaucoma. Future investigations should explore this association further because these findings may have important clinical implications.
In recent years, numerous studies1- 17 have assessed whether glaucoma is associated with mortality. Some studies have found a decreased risk of survival in patients with glaucoma, whereas others have found no such relationship. Because most glaucomas affect older adults in particular, persons with glaucoma tend to be at elevated risk for comorbid medical conditions thatcan adversely affect patient survival; thus, efforts to evaluate the relationship between glaucoma and death must account for these and other factors. Few studies,3,15,18,19 however, have considered whether the medications commonly used to treat glaucoma may affect the association between glaucoma and death. Given that certain topically applied glaucoma medications are absorbed systemically, affecting cardiac and pulmonary functioning and systemic blood pressure,20 systemic effects of glaucoma medication use may confound the relationship between glaucoma and survival.
Studies that have assessed the relationship between use of glaucoma medications, specifically, topical β-blockers, and mortality include 2 clinical trials and 2 population-based studies. In 2 of these studies, the Ocular Hypertension Treatment Study19 and the Rotterdam Study,15 no association was found between topical β-blocker use and mortality. In the Early Manifest Glaucoma Trial,18 the rate of death was nonsignificantly lower in untreated persons with open-angle glaucoma than in patients with open-angle glaucoma receiving topical betaxolol hydrochloride, a difference the investigators attributed to chance. Finally, the Blue Mountains Eye Study3 demonstrated an increased hazard of death in participants taking topical β-blockers. An editorial21 that accompanied the Blue Mountains Eye Study report questioned why topical β-blockers would be associated with an elevated risk of death given the voluminous evidence in the medical literature supporting the survival benefits of β-blockers. The editorialist recommended that additional studies be performed to better understand the relationship between glaucoma medication use and death.
Establishing evidence of a relationship between glaucoma medication use and death would be beneficial for clinical decision making. For example, an eye care professional's decisions regarding which medications to prescribe, whether to treat medically or simply to observe someone with early or suspected glaucoma, and whether to consider surgical intervention relatively early in the disease course could be informed by the knowledge that the use of certain glaucoma drug classes increases or decreases the risk of death. Moreover, if glaucoma medication use were known to be associated with adverse systemic outcomes, individual patients, in consultation with their physicians, would need to weigh the potential risk of a shortened lifespan against the quality-of-life–related benefits associated with preserved sight.
The data source for this study is a large de-identified health care claims database of beneficiaries enrolled in the Blue Care Network (BCN), a managed care company in southeastern Michigan. The database contains demographic information on each beneficiary, along with data on all health care services received at inpatient, outpatient, and skilled nursing facilities during the study period. For beneficiaries who participated in the BCN pharmacy plan, all prescription medications purchased using the plan were captured. Data were linked, making available longitudinal, person-specific data from January 1, 2003, through December 31, 2007.
All beneficiaries with 1 or more diagnoses of glaucoma or suspected glaucoma during the study period were identified by using International Classification of Diseases, Ninth Revision, Clinical Modification22 billing codes 365.0 to 365.9. Beneficiaries were excluded if they were younger than 40 years. Other exclusion criteria included enrollment in the BCN for fewer than 180 days, total time enrolled in the medical plan of less than 90% of the time from the enrollment date to the date of disenrollment or death, and beneficiary enrollment in the pharmacy plan of less than 90% of the individual's time in the database (Figure 1).
National Drug Code23 numbers were used to identify glaucoma medications in the BCN database. Medications were grouped in the following classes: topical β-antagonists, α-agonists, topical carbonic anhydrase inhibitors (CAIs), oral CAIs, miotics, epinephrine compounds, and prostaglandin analogues (PGAs) (eTable 1http://www.archophthalmol.com). The numbers of beneficiaries taking oral CAIs, epinephrine compounds, and miotics were insufficient to include in the univariable and multivariable analyses. Glaucoma medication use was defined as filling 1 or more prescriptions for a 30-day-or-more supply of the drug during the study period. The combination agent dorzolamide hydrochloride–timolol maleate (Cosopt; Merck & Co Inc, Whitehouse Station, New Jersey) was recorded as a β-antagonist and as a topical CAI.
The main outcome variable was death during the study period. The date of death was typically reported to the BCN by family members, employers, or health care professionals. To assess potential confounding variables, we captured data on age, sex, self-reported race, insurance type, glaucoma type (open angle, narrow angle, suspected, neovascular, other, or >1 type), glaucoma surgery (laser trabeculoplasty, iridotomy, trabeculectomy, glaucoma drainage device, and cyclophotocoagulation), oral β-blocker use, and the following comorbid conditions: diabetes mellitus, congestive heart failure, cancer (any type), chronic liver disease, chronic kidney disease, asthma, bradycardia, hyperlipidemia, arterial disease, ischemic heart disease, cerebrovascular disease, osteoporosis, hypotension, atrioventricular block, and depression. eTable 2 provides the International Classification of Diseases, Ninth Revision, Clinical Modification, and Current Procedural Terminology424 billing codes used to identify the type of glaucoma diagnosis, surgical procedures, and comorbid medical conditions. Participant characteristics were summarized for the entire sample using means and standard deviations for continuous variables and frequencies and percentages for categorical variables. Cox regression was used to estimate the hazard of death associated with various glaucoma medications. Using age as the time axis, the Cox model was left truncated at the age of first recording of glaucoma in the database. Participants were followed up until death or were censored at the age of disenrollment or at the end of the study period (December 31, 2007). Initially, univariable models were run to test potential predictors individually. Multivariable models were adjusted for sex, glaucoma surgery, glaucoma type, and chronic medical conditions. Because glaucoma surgery, glaucoma type, chronic medical conditions, and prescription medications purchased could change across time, these variables were entered into the Cox models as time-dependent covariates. These time-dependent covariates were indicator variables, for example, taking the value of 1 when the patient was prescribed a particular medication and the value of 0 when the medication was not prescribed. Multiple medication use at a given time was accommodated using this method because each medication was entered as a separate indicator. Statistical analyses were performed by using a commercially available software program (SAS version 9.1; SAS Institute Inc, Cary, North Carolina). This study was approved by the institutional review boards of the University of Michigan, Ann Arbor, and the BCN.
A total of 1 350 518 beneficiaries were enrolled in the BCN during the study period, and 28 695 beneficiaries (2.1%) had 1 or more diagnoses of glaucoma or suspected glaucoma. Of the individuals with glaucoma or suspected glaucoma, 7189 (25.1%) did not meet the inclusion criteria. After exclusions, data on 21 506 persons were analyzed. During the study period, 237 beneficiaries with glaucoma or suspected glaucoma (1.1%) died.
The mean (SD) age of the beneficiaries in the analysis was 60.0 (10.9) years; 55.0% were female, and 82.3% were white (Table 1). The beneficiaries had the following types of glaucoma: 50.5% had suspected glaucoma, 21.5% had open-angle glaucoma, 20.3% had multiple types, 4.3% had some other type of glaucoma, 3.3% had narrow-angle glaucoma, and 0.1% had neovascular glaucoma. Common comorbid systemic medical conditions included hypertension (52.7%), diabetes mellitus (41.3%), hyperlipidemia (36.0%), and ischemic heart disease (30.7%) (Table 2).
During the study period, 6049 beneficiaries (28.1%) filled 1 or more prescriptions for a glaucoma medication; 2021 individuals (9.4%) underwent glaucoma surgery. The most commonly used glaucoma medications were PGAs (20.8%) and β-antagonists (12.8%) (Table 3). Rates of laser trabeculoplasty, trabeculectomy, and glaucoma drainage device implantation were 5.5%, 1.5%, and 0.3%, respectively. A total of 6032 beneficiaries (28.0%) were prescribed oral β-blockers.
In the univariable analysis of time to death, the following glaucoma medication classes were associated with a decreased hazard of death: PGAs (unadjusted HR, 0.27; 95% confidence interval [CI], 0.16-0.46), α-agonists (0.14; 0.04-0.57), and topical β-blockers (0.29; 0.16-0.53) (eTable 3). No glaucoma medication class was associated with an increased hazard of death. In the univariable analysis, arterial disease, cancer, diabetes mellitus, congestive heart failure, ischemic heart disease, chronic kidney and liver disease, and cerebrovascular disease were associated with an increased hazard of death. The likelihood of death was elevated for men (unadjusted HR, 1.56; 95% CI, 1.21-2.02).
In the multivariable Cox regression analysis, the HR for each variable was adjusted for all other variables, including use of any class of glaucoma medication, age, sex, race, insurance type, glaucoma type, glaucoma surgery, oral β-blocker use, and comorbid conditions (arterial disease, cancer, congestive heart failure, ischemic heart disease, chronic kidney and liver disease, cerebrovascular disease, and diabetes mellitus) (Figure 2). After these adjustments, use of any class of glaucoma medication was associated with a 74% reduced hazard of death (adjusted HR, 0.26; 95% CI, 0.16-0.40; P < .001) compared with no use of glaucoma medications (eTable 4).
Additional models of time to death were evaluated for different combinations of glaucoma medications as predictors of death, with adjustment for demographic variables and comorbid medical conditions (Table 4). The use of PGAs, alone (adjusted HR, 0.27; 95% CI, 0.41-0.52; P < .001) or combined with other glaucoma medication classes (0.21; 0.09-0.52; P < .001), was associated with a decrease in mortality compared with individuals who did not receive PGAs. Similarly, the use of topical β-blockers alone (adjusted HR, 0.40; 95% CI, 0.18-0.86; P = .02) or combined with other glaucoma drug classes (0.19; 0.07-0.52; P = .001) also was associated with a reduced likelihood of death compared with those not prescribed these agents. Use of other combinations of glaucoma medication classes was associated with a reduction in mortality as well. As the number of prescribed glaucoma medication classes increased, the patient's likelihood of death decreased (adjusted HR, 0.37; 95% CI, 0.26-0.53; P < .001). Figure 3 shows a snapshot of glaucoma medication use in beneficiaries on January 1, 2006. Of persons receiving 1 glaucoma medication on this date, 65.4% were using a PGA, 28.2% a topical β-blocker, 5.3% an α-agonist, and 1.2% another type. Of those receiving 2 glaucoma drugs, 85.6% were using a PGA, 75.9% a topical β-blocker, 25.3% an α-agonist, and 13.2% another medication type. Compared with beneficiaries taking no medications, the adjusted HR for those receiving 1 glaucoma medication of any type was 0.29 (95% CI, 0.18-0.48; P < .001); 2 medications of any type, 0.19 (0.07-0.52; P = .001); and 3 or more, 0.12 (0.02-0.90; P = .04) (eTable 5). Finally, additional Cox regression models were performed to assess the mortality HR associated with glaucoma medication use in persons with suspected glaucoma and separately in those with open-angle glaucoma. For beneficiaries with suspected glaucoma, there was no association between use of any glaucoma medication type and death (adjusted HR, 1.19; 95% CI, 0.43-3.27), whereas glaucoma medication use was associated with a reduction in mortality in beneficiaries with open-angle glaucoma (0.23; 0.14-0.37).
In this large cohort of beneficiaries in a Midwestern US managed care network who had glaucoma or suspected glaucoma and were followed up in the community setting, glaucoma drug use was associated with a reduced mortality rate. This association remained statistically significant for various glaucoma medication classes and for different combinations of glaucoma medications after adjustment for various demographic and clinical variables and comorbid medical conditions. Although it is unclear why glaucoma medication use might confer a reduced likelihood of death in beneficiaries with glaucoma, the observed findings may be attributable to effects of the medications themselves, the types of beneficiaries receiving these medications, or providers' prescribing patterns.
Research supports the benefits of oral β-blocker use in reducing mortality rates in patients after myocardial infarction, in patients with advanced stages of cardiac failure, and in patients undergoing general anesthesia for major surgery.25- 27 Because some portion of topically administered glaucoma medications get absorbed into the systemic circulation, persons using these topical agents might experience additional benefits of the drug besides intraocular pressure reduction, some of which may be associated with improved overall health. For example, β-antagonist therapy may reduce systemic blood pressure,28- 30 a known risk factor for myocardial infarction, cerebrovascular accidents, and other potentially life-threatening conditions. β-Blockers can also help patients with arrhythmias.31,32 The use of CAIs has proved to be beneficial for persons with congestive heart failure.33 There may be other beneficial systemic effects of topical glaucoma medications that have yet to be appreciated.
Another possible explanation for this finding is that in beneficiaries with glaucoma or suspected glaucoma, those who received medical treatment for the condition may be healthier than those who did not. Studies34 have demonstrated that individuals undergoing treatment for life-threatening medical illnesses have a reduced likelihood of receiving concurrent treatment for their less-serious comorbid conditions. In some cases, for example, because certain glaucoma drug classes can exacerbate preexisting systemic conditions, the risks of prescribing these agents in persons with illnesses such as asthma, chronic obstructive pulmonary disease, and chronic renal insufficiency may outweigh the potential pressure-lowering benefits. Furthermore, enrollees may opt to receive medications for their symptomatic conditions only, yet most types of glaucoma are relatively asymptomatic until late in the disease course. Among nonwealthy beneficiaries with multiple medical conditions who may be unable to afford the prescription drug co-payments for all their illnesses, the common choice may be to buy the drugs for their most acutely debilitating conditions only. Finally, individuals who regularly visit their eye care professional and receive treatment to prevent disease progression may be more health conscious generally than are others. Although this analysis adjusted for several prevalent comorbid conditions, persons at relatively high risk for death may have been overrepresented in the subgroup undergoing no medical treatment for their glaucoma.
The observed findings may also reflect differences in beneficiaries' access to care. Although everyone in this cohort had at least some form of insurance and had visited an eye care professional at least once (to receive their glaucoma diagnosis), some beneficiaries may have been limited in their ability to receive continued glaucoma care and, similarly, to receive timely treatment or preventive care for potentially life-threatening conditions. Examples of barriers to care include lack of affordable or otherwise accessible transportation, inflexible work hours at a low-wage–earning job, and a dearth of eye care professionals in the community. Most administrative billing databases contain inadequate information on these types of variables to analyze their potential influence. Findings from the present analysis differ considerably from analyses of data from the Early Manifest Glaucoma Trial18 and the Blue Mountains Eye Study,3 in which the rate of death was elevated in topical β-blocker users. The present analysis also differs from results of the Ocular Hypertension Treatment Study19 and the Rotterdam Study15 that found no association between β-blocker use and mortality. Directly comparing findings from population-based observational studies and randomized controlled trials with results of the present study is difficult because of differences in participants' demographic profiles, type of diagnosed glaucoma, and comorbidities accounted for in the multivariable analyses.
In a multivariable analysis of claims data on hospitalized and nonhospitalized patients enrolled in Medicare, Glynn et al35 found that patients who were prescribed glaucoma medications experienced a 17% reduced odds of death. Compared with the present study, the study by Glynn et al involved an older, Medicare-recipient population and used data from a different period (1991-1994), when β-antagonists and parasympathomimetics were the mainstays of glaucoma treatment, and other glaucoma drug classes, such as PGAs, α-agonists, and topical CAIs, were not yet available or were just becoming available. Nevertheless, these studies share a common conclusion: the data suggest that use of glaucoma medications may protect against death in persons with diagnosed glaucoma.
An advantage of using health care claims databases to study the potential association between glaucoma medication use and mortality is the considerably larger size of this data source compared with population-based observational studies or randomized controlled trials. Moreover, claims databases may be more representative of patients in the community receiving treatment for glaucoma. In addition, the databases capture information on the presence or absence of glaucoma and comorbid medical conditions using billing codes rather than relying on participant self-report.
This study has several important limitations. First, the data came from a claims database and not from patient records. As Coleman and Morgenstern36 and other researchers have noted, because the purpose of billing records is to obtain reimbursement for health care services provided and not to document or study patterns of care or health outcomes, these resources may not fully capture all the beneficiaries' medical conditions, treatments, and outcomes. In addition, claims records include data on only those variables with billing codes. Therefore, other potential confounding variables, such as blood pressure, body mass index, and smoking status, could not be evaluated. Important clinical variables, such as visual acuity and visual field loss severity, similarly could not be considered in this analysis. Second, these findings are not necessarily generalizable beyond this particular cohort of insured beneficiaries in a single US managed care plan. Third, because the data set was completely de-identified, we could not verify the death status of beneficiaries in the study or the specific cause of death using the National Death Index. Additional deaths could have occurred during the study period but gone unreported to the BCN, or, alternatively, some beneficiaries' classification as deceased may have been recorded erroneously. Fourth, although this database captures all filled prescriptions for glaucoma medications, we cannot know whether the beneficiaries were actually taking the prescribed medications. Fifth, owing to the limited number of beneficiaries with glaucoma taking oral CAIs or epinephrine compounds in the database, we could not adequately study the relationship between these agents and death. Because it is well appreciated that these medication classes can cause serious systemic adverse effects, additional studies are needed to determine the impact of these particular medications on mortality.37
In conclusion, in this large, community-based study of beneficiaries with glaucoma or suspected glaucoma in a US managed care network, glaucoma medication use was associated with decreased mortality after adjustment for important confounding variables. Additional studies are needed to determine whether this result is best explained by a protective effect of the medications themselves or by other confounding factors, such as access to care or providers' prescribing patterns.
Correspondence: Joshua D. Stein, MD, MS, Department of Ophthalmology, University of Michigan, 1000 Wall St, Ann Arbor, MI 48105 (firstname.lastname@example.org).
Submitted for Publication: February 18, 2009; final revision received June 29, 2009; accepted July 9, 2009.
Financial Disclosure: None reported.
Previous Presentation: This study was presented in part at the 19th Annual Meeting of the American Glaucoma Society; March 7, 2009; San Diego, California.