Medication Use and the 5-Year Incidence of Early Age-Related Maculopathy: The Beaver Dam Eye Study

Objective  To evaluate incident early age-related maculopathy (ARM) after a 5-year interval with respect to medication use.

Design  Population-based incidence study.

Setting  Participants were adults aged 43 to 86 years living in Beaver Dam, Wis, when first examined in 1988-1990 (n = 4926); they were reexamined in 1993-1995(n = 3684).

Methods  All participants were examined and interviewed and stereoscopic color fundus photographs were taken. All procedures were done by standard protocol at both examinations. Incidence of ARM was based on grading using the Wisconsin ARM Grading System. All prescribed and over-the-counter medications in current use were brought to the examination site and the names were recorded at that time.

Results  There were 678 drug preparations (active ingredients) being used at the baseline examination. No relations were found between most antihypertensive drugs, most central nervous system medications, aspirin and other nonsteroidal anti-inflammatory agents, estrogens, lipid-lowering agents, and incident early ARM over the 5-year period. Age- and sex-adjusted logistic regression analyses suggested possible associations (P<.10) between the use of phenothiazine antidopaminergics (odds ratio [OR], 2.83; 95% confidence interval [CI], 0.97-8.23; P = .06), desiccated thyroid hormones (OR, 2.32; 95% CI, 0.89-6.07; P = .09), and calcium channel blockers (OR, 1.70; 95% CI, 0.93-3.12; P = .08) with incident ARM. When additional information on past use was included in the regression model, the association remained for calcium channel blockers, but not for phenothiazines and desiccated thyroid hormones. A lower incidence of early ARM occurred in those who took antidepressants(OR, 0.34; 95% CI, 0.12-0.94; P = .04) at the baseline examination.

Conclusion  Although many different medications were being used at the baseline examination in the Beaver Dam Eye Study cohort, there were no striking associations with the 5-year incidence of early ARM.

AGE-RELATED maculopathy (ARM) is an important cause of visual loss in persons 65 years or older.^1,2 Its pathogenesis is not well understood.^3,4 Aside from family history, and to a lesser extent cigarette smoking, there are few strong or consistent associations with ARM.⁵ One particular class of exposure has not been widely studied with respect to ARM, namely, use of medications. Anecdotal cases of associations between the use of steroids and neovascular ARM have been reported.⁶ A study by el Baba et al⁷ described 15 patients with ARM complicated by massive subretinal and/or vitreous hemorrhages, of whom 8 were taking oral anticoagulants, salicylates, or salicylate-like medications. However, no relation between aspirin use and the presence⁸ or incidence of ARM⁹ was found in 2 other studies. In the Age-Related Eye Disease Study, use of hydrochlorothiazide diuretic was associated with the presence of large retinal drusen but not with geographic atrophy or exudative macular degeneration.¹⁰ In Oulu, Finland,⁸ and in Melton Mowbray, England,¹¹ there were no associations found between use of antihypertensive medications or sedatives and the presence of early or late ARM.

Although exposures to medications (eg, chloroquine, diiodohydroxyquin, thioridazine, chlorpromazine, and trifluoperazine) that are known to directly affect the retinal pigment epithelium (RPE) might also be expected to exacerbate the incidence and progression of ARM,¹² to our knowledge, there have been no population-based data examining these associations. The purpose of this report is to describe the relation between a wide range of commonly used medications and the incidence of early ARM and its component lesions in the Beaver Dam Eye Study.

Identification and description of the population appear in previous reports.^13-17 Briefly, a private census of the population of Beaver Dam, Wis (99% white), was performed from September 15, 1987, to May 4, 1988, to identify all residents in the city or township of Beaver Dam who were 43 to 84 years of age. Of the 5924 eligible individuals, 4926 participated in the baseline examination between March 1, 1988, and September 14, 1990. Nonparticipants consisted of 226 persons(3.8%) who had died before the examination, 18 (0.3%) who could not be located, 337 (5.7%) who permitted an interview only (of these, 61 had moved), and 417(7.0%) who refused to participate (of these, 39 had moved). Comparisons between participants and nonparticipants at the time of the baseline examination appear elsewhere.¹⁴

Prior to the start of the 5-year follow-up examination on March 1, 1993, 385 (7.8%) of the participants had died. Of the 4541 surviving participants in the baseline examination, 3684 (81.1%) participated in the follow-up examination between March 1, 1993, and June 14, 1995.¹⁵ Four (0.1%) participants could not be located, 259 (5.7%) permitted an interview only (of these, 48 had moved out of the area), 423 (9.3%) refused to participate(of these, 44 had moved out of the area), and 171 (3.8%) died during the examination period. Both the mean and median times between the baseline and 5-year follow-up examinations were 4.8 years and the SD was 0.4 years.

Comparisons between participants and nonparticipants at follow-up are presented elsewhere.¹⁵ Persons who were alive and did not participate in the follow-up eye examination (n = 686) were older at baseline than those who did (62.7 vs 60.4 years; P<.001). After adjusting for age, those who were alive during the study period and did not participate were more likely to have fewer years of education, lower income, poorer visual acuity, a history of cardiovascular disease, a history of never drinking alcohol, smoked more (pack-years), higher serum cholesterol levels, and higher systolic and diastolic blood pressures, and to be retired at baseline than persons who participated.

Similar procedures were used at both baseline and follow-up examinations and are described in detail elsewhere.^14-21 Informed consent was obtained from each participant at the beginning of the examination. Medication use was assessed using a standardized questionnaire administered by the examiners at each examination.¹⁶ Participants were asked to bring to the examination all medications (prescription and over-the-counter) that they were regularly taking. The examiner asked whether there were other medications that were being taken but were not brought. If there were, the participant was asked to call the study examiner with the medication's name. In addition, at the baseline examination they were asked if they had used specific classes of drugs in the past. Information on duration of use was not obtained. Participants were asked to list the pharmacy or pharmacies where they usuallyobtained their medications. Participants, their physicians, and their pharmacies were called, when necessary, to verify medication and reason for use.

The name of the drug entered into the drug database is structured as a code table in which the code is assigned a record number reflecting that of the American Hospital Formulary Service for each active ingredient.²² In addition, subclassification information was included(eg, type of diuretic: thiazide, potassium sparing, loop).

Stereoscopic 30° color fundus photographs centered on the disc (Diabetic Retinopathy Study [DRS] standard field 1)²³ and macula (DRS standard field 2), and a nonstereoscopic color fundus photograph temporal to but including the fovea of each eye were taken. For purposes of this study, the 3583 people with at least 1 eye free of confounding lesions(eg, chorioretinal scars due to trauma, total retinal detachment) at both examinations (right eye, n = 3497; left eye, n = 3519; both eyes, n = 3435) are included in the analyses.

Grading for ARM was performed in a masked fashion using a standardized protocol, the Wisconsin ARM Grading System.^18-20 Grading procedures, lesion descriptions, and detailed definitions for the presence and severity of specific lesions appear elsewhere.^19-21

The incidence of early ARM was defined by the presence of either soft indistinct drusen or the presence of any type of drusen associated with RPE depigmentation or increased retinal pigment at follow-up when none of these lesions was present at baseline. Current age was defined as the age at the time of the baseline examination. For cigarette smoking, a participant was classified as having never smoked if he/she had smoked fewer than 100 cigarettes in his/her lifetime, as being an ex-smoker if he/she had smoked at least 100 cigarettes in his/her lifetime but had stopped smoking before the baseline examination, or as a current smoker if he/she had not stopped.

The SAS statistical software package was used for all statistical analyses.²⁴ The 5-year incidence of specific lesions of early ARM (soft indistinct drusen, increased retinal pigment, and RPE depigmentation) and early ARM were used as end points for all analyses. We did not examine associations of medication use at baseline with incident exudative macular degeneration (0.5%), geographic atrophy (0.3%), or late ARM (0.9%) because of the low incidence of these conditions in the population.²¹ The relations between current use of medications at baseline, categorized by indication for use and for some specific medications, and the incidence of specific maculopathy lesions were examined using logistic regression models controlling for age and sex. When associations of P<.10 were found, further models were created to include past use when available(ever used vs never used). Final models including age, sex, and other risk factors for ARM lesions, including smoking status at baseline (never, past, current), history of beer consumption at baseline (no, yes), and history of heavy alcohol drinking status at baseline (never, past, current) were created.²⁵

Persons who were alive and did not participate in the follow-up examination were less likely to be taking thyroid medications at the baseline examination(7.9 vs 10.0%; P = .05). There were no statistically significant differences in the age-adjusted frequency of use of oral steroids, gout medications, quinine, cholesterol- and blood pressure–lowering medications, diuretics, or aspirin at baseline between participants and nonparticipants. Nonparticipants in the 5-year follow-up had a slightly higher average number of medications taken at baseline compared with participants (3.6 vs 3.4; P = .08). After adjusting for age and sex, participants with early ARM at baseline were more likely to participate in the 5-year follow-up than those in whom ARM was absent (Cochran-Mantel-Haenszel test for general association, P = .003).

There were 678 drug preparations (678 unique active ingredients) currently used at the baseline examination, most (n = 610) were currently used by fewer than 50 persons who were at risk for incidence of early ARM. Three hundred fifty-seven of the 678 drug preparations were included in the present analyses. We limit the data presented to medication categories for which there has been recent interest in possible relations with ARM or for when there was some reason to suggest a possible association (P<.10). In some cases we present results for categories of drugs that may contain preparations with different active ingredients (eg, antihistamines), and in other cases we present the data for a specific active ingredient (eg, aspirin).

After adjusting for age and sex, we found one association (antidepressants) with incident early ARM of P<.05 (Table 1, Table 2, and Table 3). There were no associations between antihypertensive medications as a group or specific types of antihypertensive medications except for calcium channel blockers (age- and sex-adjusted odds ratio [OR], 1.70; 95% confidence interval [CI], 0.93-3.12; P = .08) (Table 1) and early ARM. For the 3 specific calcium channel blockers, diltiazem (n = 17), nifedipine (n = 30), and verapamil (n = 42), used at baseline, only nifedipine was associated with the incidence of early ARM (OR, 2.29; 95% CI, 0.89-5.87; P = .08) and this was largely due to the incidence of soft indistinct drusen (OR, 2.50; 95% CI, 0.97-6.41; P= .06). This association of current use of nifedipine and incident early ARM remained after controlling for smoking (OR, 2.30; 95% CI, 0.90-5.91; P = .08) or beer drinking (OR, 2.30; 95% CI, 0.93-6.14; P = .07) history status at baseline and when past users were included with current users of calcium channel blockers (OR, 1.84; 95% CI, 1.02-3.34; P = .04). We found no association of calcium channel blocker use and early ARM in those with no history of hypertension(data not shown).

The use of antidepressants at baseline was associated with a 66% reduction in the incidence of early ARM (OR, 0.34; 95% CI, 0.12-0.94) (Table 2). In the 68 persons using tricyclic tertiary amine compounds(eg, amitriptyline, clomipramine, doxepin, imipramine, triamipramine) the age-and sex-adjusted OR was 0.25 for the development of incident early ARM(95% CI, 0.06-1.03; P = .06) compared with those not using these types of antidepressants. This association remained after controlling for smoking, beer drinking history, and hypertension status at baseline (data not shown). The incidence of soft drusen (OR, 0.31; 95% CI, 0.08-1.29; P = .11) and RPE abnormalities (OR, 0.62; 95% CI, 0.19-2.05; P = .44) was lower in users of tertiary amine tricyclic compounds; however, the associations were not statistically significant. Although the incidence of early ARM was lower for those using amitriptylines, the most frequently used of the antidepressants, compared with persons not using this drug, the association was not statistically significant(Table 2). There were too few persons using secondary amines (n = 11), tetracyclics (n = 3), or serotonin reuptake inhibitors (n = 4) at baseline to examine whether they were associated with incident early ARM. When past use was included with current users of tertiary amine tricyclic compounds, an association with lower incidence of early ARM (OR, 0.35; 95% CI, 0.13-0.98; P = .05) was found.

Association between the use of antipsychotic agents and incident early ARM are shown by class/active agent in Table 2. Persons taking phenothiazines at baseline had a higher 5-year incidence of early ARM (OR, 2.83; 95% CI, 0.97-8.23; P = .06). Use of phenothiazines was not statistically significantly associated with incident increased retinal pigment (OR, 2.72; 95% CI, 0.74-9.93; P = .13) or soft drusen (OR, 1.54; 95% CI, 0.44-5.43; P = .50). When past use of antipsychotics was combined with current use (OR, 1.44; 95% CI, 0.77-2.68; P= .26) or when past use of phenothiazines was combined with current use (OR, 1.49; 95% CI, 0.60-3.69; P = .39), there was no association with incident early ARM. Other antipsychotic drugs (eg, phenylbutylpiperadines, thioxanthenes, and dibenzepins) were too infrequently used in the population to examine their relation with early incident ARM. Barbiturate and nonbarbiturate sedatives and antianxiety drugs were not associated with incident early ARM(Table 2).

An association was found between desiccated thyroid hormone use and incident early ARM (Table 3). This was largely due to the incidence of soft indistinct drusen (OR, 2.74; 95% CI, 1.06-7.08; P = .04). There was no association of synthetic thyroid hormone use at baseline and incident early ARM. Aspirin, other nonsteroidal anti-inflammatory agents, acetaminophens, oral and nasal steroids, estrogens and progesterones, gout medications, lipid-lowering agents, cardiac glycosides, chloroquine and quinine, and antihistamines used at baseline were not associated with incident early ARM (Table 2 and Table 3).

Data from the Beaver Dam Eye Study cohort offered an opportunity to examine the associations of medication use with the incidence of early ARM and its specific lesions. The cohort was large, the currently used medications were brought to the examination site, and trained examiners recorded the names of the medications. The ARM lesions were determined using standardized photographic and grading protocols at both the baseline and follow-up examinations.

There were few significant associations between incident early ARM and medications used in the population. The borderline association of antipsychotic drugs with incident early ARM (incident large drusen and pigmentary abnormalities) was not unexpected, as antidopaminergic phenothiazines (eg, chlorpromazine, prochlorperazine, and perphenazine) have long been known to be associated with pigmentary abnormalities of the RPE.¹² However, the finding of a protective effect of antidepressants with incident early ARM was unexpected. The antidepressants most frequently used were tricyclic tertiary amines (eg, amitriptyline, desipramine, and imipramine). The biologic reason why these drugs, which have complex pharmacologic actions on neurologic pathways, have a protective effect for ARM in this study is not known. This may represent a chance or spurious finding.

Any conclusions or explanations regarding associations, or lack of them, described herein must be made with caution for a number of reasons. First, because this was not a study primarily aimed at identifying effects of commonly used drugs, we have virtually no information about dosage, duration of use, or longest exposure. The lack of such information might be expected to decrease the ability to estimate the strength of an association as persons taking low doses of a drug for short periods would be similarly classified as persons taking high doses for long periods. Also, age may be an important consideration when evaluating drug and drug dosage effects (ie, an age of "vulnerability" may exist). Second, the concomitant low frequency of medication use and of early ARM incidence limits our ability to detect (or reject) meaningful relations. For example, to achieve a power of 0.80 to detect an OR of a given drug of equal to 2.0 or greater with an α = .05, with an overall incidence of early ARM of 8.2% and a total sample size of 2832 (as in the Beaver Dam cohort), we estimated that 175 persons had to be currently taking the drug of interest. Although there may be some potentially meaningful relations among these that are not statistically significant, the associations are likely to be small if "real." However, as suggested by Rothman,²⁶ it may be that some of these relations are important and warrant further study. Third, some findings (eg, those between calcium channel blockers, desiccated thyroid hormones, and antidepressants), which may be of potential biological significance, may be entirely due to chance.

In summary, we find no striking associations between medication use and incident early ARM. A few of these associations may be interesting to look at in examining other cohorts. Also, studies in larger cohorts with information on dosage and duration of medication use as well as late ARM will be necessary to find weaker associations with ARM lesions.

Accepted for publication March 8, 2001.

This research is supported by National Institutes of Health grant EY06594(Drs R. Klein and B. E. K. Klein) and, in part, by Research to Prevent Blindness, Inc, New York, NY (Dr R. Klein, Senior Scientific Investigator Award).

We acknowledge the work of Lee Vermeulen, who helped in the development of the drug database; Rick Chappell, PhD, who provided statistical advice; and the guidance and input of the Beaver Dam Scientific Advisory Board (Mary Frances Cotch, PhD, Mae Gordon, PhD, Lee Jampol, MD, Dan Seigel, PhD, and Robert Wallace, MD).

Corresponding author: Ronald Klein, MD, MPH, Department of Ophthalmology and Visual Sciences, University of Wisconsin–Madison, 610 N Walnut St, 460 WARF, Madison, WI 53705-2397.