Association between glaucoma and duration of statin use.
McGwin G, McNeal S, Owsley C, Girkin C, Epstein D, Lee PP. Statins and Other Cholesterol-Lowering Medications and the Presenceof Glaucoma. Arch Ophthalmol. 2004;122(6):822-826. doi:10.1001/archopht.122.6.822
To explore whether oral statin and other antihyperlipidemic medicationsare associated with open-angle glaucoma.
The administrative clinical databases maintained at the Veterans AffairsMedical Center, Birmingham, Ala, were used to conduct a matched case-controlstudy. Cases were all male patients aged 50 years and older with a new diagnosisof glaucoma on an outpatient or inpatient visit during the period January1, 1997, through December 31, 2001. Ten control subjects were matched to eachcase according to age (within 1 year). Prescription files were assessed forstatin use as well as additional medications to lower cholesterol levels.Information on comorbid medical conditions was also obtained. Conditionallogistic regression was used to calculate odds ratios (ORs) and 95% confidenceintervals (CIs).
Longer duration of statin use was associated with a lower risk of open-angleglaucoma (P for trend = .04) primarily among subjects with 24months or more of use (OR, 0.60; 95% CI, 0.39-0.92). When stratified by comorbidmedical condition, among those with cardiovascular disease (OR, 0.63; 95%CI, 0.42-0.97), lipid metabolism disorders (OR, 0.63; 95% CI, 0.41-0.99),and the absence of cerebrovascular disease (OR, 0.76; 95% CI, 0.58-0.99),statins demonstrated a protective effect on open-angle glaucoma. Finally,a protective association was also observed among those who used nonstatincholesterol-lowering agents (OR, 0.59; 95% CI, 0.37-0.97).
Initial examination of an administrative clinical database indicatesthe intriguing possibility that long-term use of oral statins may be associatedwith a reduced risk of open-angle glaucoma, particularly among those withcardiovascular and lipid diseases. Nonstatin cholesterol-lowering agents werealso associated with a reduced risk of having open-angle glaucoma. Additionalinvestigation is warranted as to whether these classes of agents may providean additional therapeutic addition for glaucoma.
The use of statins has been associated in some studies with a diminishedrisk of developing age-related macular degeneration as well as a potentialfor reducing risk for several medical conditions related to cardiovasculardisease.1- 4 Thepresumed direct causal mechanisms have centered on the effect of 3-hydroxy-3-methylglutarylcoenzyme A reductase inhibitors in reducing cholesterol production and enhancinglow-density lipoprotein cholesterol removal from plasma. To the extent thatexcess total cholesterol or low-density lipoprotein cholesterol is implicatedin these conditions, use of the statins would reduce the risk of developingthese conditions, or at least delay their onset. Another potential mechanismmay be their inhibition of nonsteroidal isoprenoid production, in which processesassociated with atherosclerotic progression may be likewise inhibited.5- 7 Many statins also inhibitrho-kinase activity, which has been shown to increase aqueous outflow.8 Finally, there may be as yet undiscovered or indirecteffects of these compounds to help explain their protective associations.
With a greater understanding of the multiple mechanisms that could potentiallyaffect the outflow system or the optic nerve in glaucoma, there are plausiblemeans by which statin use could be associated with a reduced risk of developingglaucoma. By reducing atherosclerotic processes and subsequent vascular diseases,statins may directly protect optic nerve head vasculature or may indirectlyimprove ocular blood flow. Since the trabecular meshwork has endothelial cellsthat share many attributes of vascular endothelial cells, statins may alsoexert a protective effect in enhancing trabecular endothelial cell function.Thus, in this article, we seek to address the issue of whether in a largecohort of patients the use of statins is associated with open-angle glaucoma.
The Birmingham (Alabama) Department of Veterans Affairs Medical Center(BVAMC) is a 134-bed acute tertiary care medical facility and serves as aVeterans Hospital Administration tertiary care referral center for Alabama.All patients who had at least one visit (inpatient or outpatient) to the BVAMCbetween January 1, 1997, and December 31, 2001, were eligible for study inclusion.Because the prevalence of glaucoma is low in individuals younger than 50 years,the study population was limited to patients aged 50 years and older. Womenwere also excluded because they represented such a small proportion of thepatient population (10.8%) that meaningful analyses were impossible.
The BVAMC provided data files containing demographic information (age,sex, race) and clinical and medication information for each patient. The clinicalfile contained a description of each diagnosis made at the BVAMC during inpatientand outpatient visits and the diagnosis date. All diagnoses were coded usingthe International Classification of Diseases, Ninth Revision,Clinical Modification (ICD-9-CM).9 The medication file contained information on eachmedication prescribed during each patient visit. This file also containedthe prescription date and the date the prescription was filled. For both theclinical and medication files, the information provided pertained to all diagnosesand medications during the course of each patient's history with the BVAMCand not just 1997 through 2001. All data received from the BVAMC containedno information that would allow patients to be identified. The institutionalreview board of the BVAMC approved the protocol.
Within the study population, a nested case-control study was conducted.Cases of glaucoma were defined using the ICD-9-CM codes365.1 (open-angle glaucoma), 365.8 (other specified forms of glaucoma), and365.9 (unspecified glaucoma). Information on the glaucoma diagnosis date wasprocured and will heretofore be referred to as the index date. Because thisstudy addressed the association between statin use and the new diagnosis ofglaucoma, patients who had a glaucoma diagnosis prior to the observation periodof the study (prevalent cases) were excluded.
Control subjects were randomly selected from the study population whodid not have a glaucoma diagnosis by the end of the observation period. Tobe considered an eligible control subject for a given case, the control subjectmust have had an encounter with the BVAMC (inpatient or outpatient) on orbefore the index date of the matched case. Ten control subjects were selectedfor each case and matched on age (±1 year). Each control subject wasassigned the index date associated with their matched case.
The prescription file was queried for the presence of filled statin(atorvastatin calcium, cerivastatin sodium, fluvastatin sodium, pravastatinsodium, simvastatin, lovastatin) prescriptions. Nonstatin lipid-lowering agents(eg, fibrates, nicotinic acid) were also extracted from the prescription file.Only those prescriptions that were filled prior to the index date for eachmatched set of cases and control subjects were considered. Time since firststatin use was calculated as the time between the first statin prescriptionand the index date. Statin users were also classified as being current orpast users with the former being those who had a statin prescription filledwithin 6 months before the index date and the latter being those whose lastprescription fill date was greater than 6 months before the index date. Ananalogous set of variables was created for the nonstatin lipid-lowering agents.
Information on the presence of the following conditions was extractedfrom the clinical data file: ischemic heart disease (ICD-9-CM codes 410-414); cerebrovascular disease (ICD-9-CM codes 430-438); lipid metabolism disorders (ICD-9-CM code 272); hypertension (ICD-9-CM codes 401-405);diseases of the arteries, arterioles, and capillaries (ICD-9-CM codes 440-448); and diabetes (ICD-9-CM code250). For the purposes of analysis, only those diagnoses that were recordedprior to the index date were considered.
Conditional logistic regression was used to calculate an odds ratio(OR) and 95% confidence interval (CI) for the association between any statinuse and the risk of developing glaucoma. The ORs and 95% CIs were also estimatedfor current and past statin users relative to nonusers and according to timesince first prescription. A similar set of analyses was conducted for nonstatinlipid-lowering agents. Stratified analyses were conducted to determine ifdiabetes, lipid metabolism disorders, hypertension, cardiovascular disease,cerebrovascular disease, and arterial disease modified the association betweenstatin use and glaucoma. There was an insufficient number of patients usingnonstatin lipid-lowering agents to conduct a similar set of stratified analyses.
We selected 667 cases; 170 had a diagnosis of open-angle glaucoma (ICD-9-CM code 365.1), 31 had other specified forms of glaucoma(ICD-9-CM code 365.8), and 466 had unspecified glaucoma(ICD-9-CM code 365.9).
Table 1 presents the demographicand medical characteristics among the glaucoma cases and control subjects.By design, the mean age of both cases and control subjects was equivalent.There were twice as many African American individuals among the cases comparedwith control subjects. Those with glaucoma were more likely to also have diabetes,lipid metabolism disorders, and hypertension.
Table 2 demonstrates thestatin and nonstatin medication use characteristics among glaucoma cases andcontrol subjects, as well as the unadjusted and adjusted ORs. While caseswere more likely to have filled a statin prescription (OR, 1.23; 95% CI, 0.99-1.51),following adjustment for diabetes, lipid metabolism disorders, hypertension,cardiovascular disease, cerebrovascular disease, and arterial disease, a protectiveassociation was observed (OR, 0.85; 95% CI, 0.66-1.09), albeit not a statisticallysignificant one. This association, although not significant, was also observedfor past (OR, 0.74; 95% CI, 0.53-1.04) but not current (OR, 0.94; 95% CI,0.70-1.27) statin use. There was a significant trend toward a reduced riskof glaucoma with longer-term statin use (P = .04)(Figure 1). Indeed, use of statinsfor greater than 23 months was associated with a statistically significantreduction in the risk of glaucoma (OR, 0.60; 95% CI, 0.39-0.92).
Use of nonstatin lipid-lowering medications was also associated witha significantly reduced risk of glaucoma (OR, 0.59; 95% CI, 0.37-0.97) thatwas also apparent among both current and past users, although neither associationwas statistically significant. However, this association was limited to thosewith less than 12 months of use (OR, 0.38; 95% CI, 0.18-0.79). When consideringthe joint effect of statin and nonstatin medications, the largest risk reductionwas associated with use of both types of medications (OR, 0.52), followedby nonstatin use only (OR, 0.60), and statin use only (OR, 0.86). However,none of these associations were statistically significant.
Table 3 presents ORs and95% CIs for the association between statin use and glaucoma stratified accordingto the presence of comorbidities. Once the other medical characteristics werecontrolled for, significant associations between statin use and glaucoma wereobserved among those with lipid metabolism disorders, cardiovascular disease,and those without cerebrovascular disease.
The results of this analysis demonstrate a significant and meaningfulassociation between glaucoma and long-term statin use. Evaluating larger datasetswith longer follow-up periods, such as those with managed care companies orwith insurance company claims files where both disease and pharmacy data aremaintained, would allow a more definitive evaluation of the possibility ofstatins being an additional therapy for glaucoma or its prevention.
Such an endeavor is clearly worthwhile, based not only on the resultsof the current study but also on some plausible potential mechanisms wherebysuch a protective effect might occur. First, many statins inhibit the activityof rho kinase; such inhibition has been shown to enhance aqueous outflow andthereby presumably lower intraocular pressure.8 Sincewe did not have access to medical record data in this study, it would be interestingfor other investigators to examine the effect of statin use on the level ofintraocular pressure, controlling for the status of glaucoma and the intensityof treatment. Second, the ability of statins to reduce cardiovascular diseasemay directly or indirectly protect the vascular supply to the optic nerveor eye. Interestingly, statin use is associated with a higher, albeit notstatistically meaningful, elevation of risk of glaucoma among those with cerebrovasculardisease compared with a lower risk among those with cardiovascular disease.This may indicate that the direct and indirect mechanisms have opposing effectsor that there are additional factors at work that have yet to be identified.
If the trend and the magnitude of effect seen with statin use of greaterthan 23 months is upheld with larger sample sizes in future studies, the protectiveeffect will rival that of lowering intraocular pressure through the use ofmedications discussed in the Ocular Hypertension Treatment Study10 andof other treatments discussed in other studies.11 Thiswould also imply that a new therapeutic class of agents might be effectivefor the care and treatment of patients with glaucoma.
The adverse effects of statins on ocular structures have been studiedto a limited degree. While animal models might indicate a higher risk of cataractdevelopment because of the rho-kinase inhibition of statins, human studieshave demonstrated no elevated risk of cataracts among those taking statins.12,13 Obviously, longer-term follow-upacross many years will be needed to be able to conclude that such long-termuse is completely safe for the eye. In addition, the potential systemic adverseeffects of statins need to be carefully considered, as do the costs of themedications. Lastly, the ability to topically administer statins or statin-likecompounds has yet to be investigated.
The intriguing finding of some protective association with nonstatinuse also raises the possibility that lipid diseases as a whole may be associatedwith the presence of glaucoma, as seen in the higher proportion of cases withlipid disorders compared with control subjects. There have been no studiesto suggest this is the case in population-based evaluations to date. Nevertheless,the associations found here, both in the protective association with cholesterol-loweringagents and the higher rate of lipid disorders seen in those with glaucomacompared with control subjects, suggest that such inquiry is warranted. Thus,additional work is also needed to clarify the nature of the association betweenthe presence of lipid disorders and the use of medications to lower lipids.The evidence in favor of an independent role for treatment, however, is foundin the significant OR of 0.63 among those with lipid disorders who have receivedstatin treatment compared with those who have not, even after adjusting forcomorbid conditions.
Among those without cerebrovascular disease, there was a protectiveassociation between statin use and glaucoma (OR, 0.76; 95% CI, 0.58-0.99),whereas those with cerebrovascular disease demonstrated a positive associationbetween statin use and glaucoma (OR, 2.01; 95% CI, 0.99-4.10). This resultis difficult to interpret. This could be a spurious association or an anomalyin the data or results. Alternatively, it may reflect the more general stateof vascular supply to the central nervous system. Those without a historyof cerebrovascular disease have a statin association similar to the othersubpopulations and the population as a whole in this study. Those with a historymay reflect a strong difficulty in vascular flow that overrides the effectsof statins because even those who do not use statins have an elevated, albeitnot significant, risk of glaucoma (OR, 1.65). Further, those who have suchconditions and who have been placed on statins may have cerebrovascular diseaseof greater severity than those who do not.
The study has several limitations that should be kept in mind. First,the study population consisted entirely of men because it was an older veteranpopulation. Additional research on the association between statin use andglaucoma among women is necessary. Second, the diagnoses of glaucoma weremade by individual physicians without the use of standardized criteria, whichcould introduce significant differences relative to other studies and studypopulations. However, there is no reason to expect the diagnosis of glaucomato have been biased by the use of statins. Third, the diagnoses were subjectto miscoding into ICD-9-CM codes; again, however,there is no reason to suspect that bias would result from this. Also, we includednonspecified forms of glaucoma assuming these were likely to reflect open-angleglaucoma that was imprecisely coded. To the extent that they truly representother forms of glaucoma that are not associated with statin use, the biasso introduced would be toward the null. It should also be noted that whilethe majority of cases had visited the optometry and/or ophthalmology clinicsat the BVAMC, only 40% of the control subjects had made such a visit, thusintroducing the potential for misclassification of glaucoma. Fourth, no clinicaldata was available, so we could not comment on the severity of glaucoma. Fifth,statin use was defined on the basis of a filled prescription within the BVAMCpharmacy service. This suggests that a patient with a statin prescriptionrecord but no matching fill record would be classified as a nonstatin usereven though he did indeed use statins by filling the prescription outsidethe BVAMC system. Such misclassification, however, would only bias to thenull. Additionally, since more than 90% of statin prescriptions were filledat the BVAMC, this is unlikely to have produced a significant effect. We alsodid not have information on statin use outside the veteran affairs system.Thus, we may have underestimated some subjects' duration of use. As long assuch misclassification is not differential according to case status, thenthe effect on the study results is likely to be minimal. Finally, race wasunknown for a large proportion of our study population (both cases and controlsubjects). However, the race distribution among those with known data wassimilar to what would be expected given population-based studies and thusunlikely to introduce spurious results in the analyses. When stratified accordingto race, the protective association of statin use was apparent among whiteindividuals (OR, 0.56) and African American individuals (OR, 0.76) as wellas those with unknown race (OR, 0.58). Given the consistency of the associations,even if race had been known for all subjects, adjustment would likely havehad little effect on the observed results. Information on additional potentiallyconfounding characteristics (eg, smoking) was similarly not available.
An important methodological issue is the possibility of left-censoreddata in that many patients with preexisting glaucoma would have been capturedin the first or second year of the study, because case identification wasmade on the basis of the first visit at which an open-angle or unspecifiedtype of glaucoma was diagnosed. Thus, we fully expect that many of the newdiagnosis cases of glaucoma were actually prevalent cases that were firstseen at the BVAMC and recorded into the database and captured by our selectionalgorithm as a new diagnosis case because of the enrollment intervals. Toaddress this issue, the analyses were also conducted including the prevalentcases, and the results were highly consistent with those reported herein.
In summary, the results of the current study thus suggest that the associationbetween statin use and the risk of glaucoma deserves further investigation.Any such study should address the difficulties faced when using administrativedata sources such as in the present study. If confirmation is forthcoming,a randomized clinical trial of the use of statins or other antilipid/cholesterol-loweringagents may lead to the application of a new class of medications for the treatmentof glaucoma.
Corresponding author: Paul P. Lee, MD, JD, Duke Eye Center, Box 3802,Erwin Road, Durham, NC 27710 (e-mail: firstname.lastname@example.org).
Submitted for publication December 31, 2002; final revision receivedAugust 12, 2003; accepted November 6, 2003.
This study was supported by grants R21-EY14071 and R01-AG04212 fromthe National Institutes of Health, Bethesda, Md; a Lew Wasserman Merit Awardfrom Research to Prevent Blindness, New York, NY (Dr Lee); and the EyeSightFoundation of Alabama, Birmingham.
Dr Owsley is a senior scientific investigator for Research to PreventBlindness.