Reproductive Exposures, Incident Age-RelatedCataracts, and Age-Related Maculopathy in

Women: The Beaver Dam Eye Study

BARBARA E. K. KLEIN, MD, RONALD KLEIN, MD, AND KRISTINE E. LEE, MS

● PURPOSE: To evaluate the relationship of reproductiveexposures and incident age-related cataract and macu-lopathy in women.● METHODS: This was a population-based cohort studyincluding all adults 43 to 84 years of age living in BeaverDam, Wisconsin (a representative midwestern community)who were identified during a census in 1987 to 1988. Theywere evaluated initially in 1988 to 1990 and at follow-up in1993 to 1995. Evaluations included medical histories andboth fundus and lens photography. All procedures weredone according to protocols that were the same at bothexaminations. All photographs were graded by trainedobservers using defined grading schemes. The severities ofage-related cataracts and maculopathy were determined bygrading of photographs. Information on hormone exposureswas ascertained from structured interviews.● RESULTS: After adjusting for age, the only significantfinding for lens end points was a trend indicating a possibleprotective effect of increasing number of live births andincident posterior subcapsular cataract. There were nosignificant associations of any reproductive exposure withlesions of early or late age-related maculopathy.● CONCLUSIONS: In these population-based data, thereis little evidence of association of hormone exposureswith incident age-related eye disease in women 5 yearslater. Longer follow-up of this population, whose meanage is approaching that of heightened incidence, maydisclose significant relationships. (Am J Ophthalmol2000;130:322–326. © 2000 by Elsevier Science Inc. Allrights reserved.)

W OMEN HAVE BEEN FOUND TO HAVE POORER

vision and to be more likely to have some typesof age-related cataracts1–7 and age-related

maculopathy2,8 than men of similar ages. Prevalence datafrom the Beaver Dam Eye Study suggested that hormonallyrelated exposures might influence the risk of some types ofcataract.9 For example, older age at menarche was associ-ated with increased severity of nuclear sclerosis, whereasolder age at menopause was associated with a decreasedrisk of cortical opacities, and years of use of postmeno-pausal estrogens was positively related to decreased severityof nuclear sclerosis. Some of these relationships have alsobeen found in the Blue Mountains Study.10

Investigation of hormonally related exposures and mac-ular degeneration in Beaver Dam suggested, at most, amodest protective effect of hormone replacement on earlylesions of age-related maculopathy.11 However, the EyeDisease Case-Control Study reported a significant protec-tive effect of hormone replacement therapy on exudativemacular degeneration.12 We sought to clarify some of therelationships of hormonal exposures to incidence of cata-ract and the incidence of lesions characteristic of age-related maculopathy 5 years after the prevalence survey ina community-based study.

METHODS

BEAVER DAM, WISCONSIN, IS A TOWN OF APPROXIMATELY

17,000 persons located northeast of Madison, Wisconsin.A private census of the population of Beaver Dam wasperformed from 1987 to 1988 to identify all persons 43years of age and older. Persons 43 to 86 years of age wereinvited for the study evaluation, which was performedduring a 2.5-year period beginning March 1, 1988. Theexamined group was 99% white. Tenets of the Declarationof Helsinki were followed, approval of the institutionalhuman subjects committee was granted, and a consentform was signed by each subject. During the study visit,standard measurements were made and questionnaireswere administered. All subjects identified at the initial

Accepted for publication Feb 29, 2000.From the Department of Ophthalmology and Visual Sciences, Univer-

sity of Wisconsin Medical School, Madison, Wisconsin.This research was supported by grant EYO6594 (Drs R Klein and BEK

Klein) from the National Institutes of Health, Bethesda, Maryland.Correspondence to Barbara E.K. Klein, Department of Ophthalmology

and Visual Sciences, University of Wisconsin—Madison, 610 N WalnutSt, 460 WARF, Madison, WI 53705-2397; fax: (608) 263-0279; e-mail:kleinb@epi.ophth.wisc.edu

Reprints not available.

© 2000 BY ELSEVIER SCIENCE INC. ALL RIGHTS RESERVED.322 0002-9394/00/$20.00PII S0002-9394(00)00474-8

census13 were invited for the second examination. Onlythose participating at the first and second examinations(n 5 3,684) supplied information pertinent to the currentanalysis. Differences between participants and nonpartici-pants at the baseline4 and follow-up14 examinations havebeen previously published. In brief, those who were alivebut not participating at the second visit were considerablyolder and had poorer visual acuity. Those who died weresignificantly older, were more often male, had poorer visualacuity, more often had diabetes,14 and had more severenuclear sclerosis.15 There were 226 deaths before the firstexamination and an additional 556 deaths before thesecond examination. The average age at follow-up was 65years.

The follow-up examinations occurred 4.8 years on av-erage after the first evaluations and were performed so thatparticipants were seen in approximately the same order asin the prevalence examinations. Procedures at the fol-low-up were the same as at the baseline examination.4

Blood pressures were measured according to the Hyper-tension Detection and Follow-up Program protocol.16 Pu-pils were pharmacologically dilated after examination ofthe anterior chamber.

Participants responded to an interview at the time of thestudy evaluation. They were asked whether they had everbeen told by their physician that they had diabetes and/orhypertension. Subjects were asked whether they had eversmoked (.100 cigarettes/lifetime); if so, they were askedthe number of cigarettes smoked per day and the numberof months they smoked, current smoking status, and thedate of stopping. Each participant was asked about con-sumption of beer, wine, and liquor in the last year and thenumber of drinks of each beverage type consumed perweek. The response was converted to grams of alcohol,according to the following estimates of ethanol per serving:12 ounces of beer contained 13 g, 4 ounces of winecontained 11.5 g, and 1.5 ounces of liquor contained 14 g.

Each participant was asked what the highest year ofschool they completed.

During the medical interview of women, historicalinformation that was pertinent to hormone status wasobtained, in addition to confirmation of date of birth. Thequestions asked of all women were as follows:

1. How old were you when you began to have periods(menstruate)?

2. Have you stopped having periods (menstruating)?3. How old were you when you stopped?4. Did you stop having menstrual periods because of

surgery such as hysterectomy?5. Have you ever had a hysterectomy?6. If yes (to question 5), how old were you when you

had your hysterectomy?7. If yes (to question 5), did you have both of your

ovaries removed?8. How many times, if ever, have you been pregnant?

9. (If the answer to question 8 was greater than 0) Ofthese pregnancies, how many resulted in live births?

10. Have you ever taken or used female hormones, suchas estrogens, for menopausal symptoms or aftermenopause?

11. If so, about how many years did you take thesefemale hormones?

12. Do you now use these hormones?13. What are the name of the hormones you are now

taking (or last took)?14. Have you ever taken birth control pills?15. For about how many years did you take birth control

pills?These questions permit a variety of possible strategies forevaluating hormone exposures. We present only those forwhich there was evidence in our data sets and or in publica-tions of others that suggested meaningful relationships.

Photographs of the lenses were taken with two cameras:a slit-lamp camera and a retroillumination camera.17 Grad-ing procedures for the lens were based on detailed codifieddecision rules.17 Graders were masked to subject identity,personal characteristics, and the presence and/or severityof lens lesions at the first examination. Scores for nuclearsclerosis were based on comparisons with standard photo-graphs. The scale has five steps of severity based on theopacity of the nucleus. Scores for cortical and posteriorsubcapsular cataracts were based on weighted estimates ofthe degree of opacity of lens area, as defined by a circulargrid divided into eight wedge-shaped peripheral areas anda central circular area overlaid on the photograph.

Any eye with evidence of direct trauma was excludedfrom all analyses. Any eye with a lesion that obscured thegrading of specific lens lesions was excluded from consid-eration for that lens lesion (for example, the presence ofpseudoexfoliation of the lens that interfered with photog-raphy/grading of cortical opacities). Eyes that had under-gone cataract surgery were excluded from analyses of riskfactors for specific cataract lesions, except where explicitlystated. Estimates of incidence were based on havingcorresponding gradable subfields at both visits.

Incident cataract refers to the classification of lensopacity in follow-up photographs of sufficient severity tomeet our standard definitions in an eye free of that cataracttype at baseline. The severity of each cataract was chosento comply with a severity of lens opacity that clinicianswould be comfortable labeling as a cataract if they saw it intheir patients. These categories were the same as thoseused to define prevalent cases in previous publications.7The specific definitions used to classify such opacities wereas follows: nuclear sclerosis level of four or five in thegrading scheme, which specifies five levels of increasingseverity; cortical opacities involving 5% or more of thelens area, assuming the lens to be a plane surface; andposterior subcapsular opacity involving 5% or more of anyof nine segments defined by a circular grid of the lens“surface.” Data are presented for the worse eye at follow-

AGE-RELATED EYE DISEASE IN WOMENVOL. 130, NO. 3 323

up, where worse indicates the development of an incidentlesion. Cataract as an all-inclusive category refers to anyeye meeting criteria for any of the three types describedabove. For estimating the association of risk factors withany age-related cataract, eyes at risk were those that hadnone of the three cataract types at baseline. This dimin-ishes the number of eyes at risk compared with analyses forspecific cataract types. However, the incidence rate wasnot materially affected.

Photographs were taken of the retina through a dilatedpupil with a 30-degree fundus camera. Three standardphotographic fields including the disk, macula, and a fieldtemporal to the macula were taken by protocol.8 Therewere 3,583 people with at least one eye free of confoundingretinal lesions at both examinations. Grading for age-related maculopathy was performed in a masked fashion bymeans of a standardized protocol, the Wisconsin Age-Related Maculopathy Grading System.18 Grading proce-dures, lesion descriptions, and detailed definitions of thepresence and severity of specific lesions have appearedelsewhere. For each subfield, the presence and severity ofeach of the following characteristics were evaluated: thelargest drusen size (in order of increasing size: 0, ,63 mm,63 to 124 mm, 125 to 249 mm, $250 mm in diameter); themost severe drusen type (in order of increasing severity:none or questionable [hard indistinct], hard distinct, softdistinct, and soft indistinct or reticular drusen); areainvolved by drusen; drusen confluence, increased retinalpigment; retinal pigment epithelial depigmentation; reti-nal pigment epithelial detachment or serous detachment ofthe sensory retina; subretinal or subretinal pigment epithe-lial hemorrhages; subretinal fibrous scars; and geographicatrophy. When present, the size, appearance, and areainvolved are used to determine the severity of drusen. Theareas with increased retinal pigment or retinal pigmentepithelial depigmentation involvement are used to deter-mine the severity of retinal pigment epithelial abnormal-ities within a subfield.

Incidence of a specific lesion of maculopathy for an eyewas defined by the development of such a lesion when itwas not present at baseline in any of the subfields thatcould be graded at both examinations. For example, ifnone of the subfields had soft indistinct drusen at baselineand soft indistinct drusen were present in one or moresubfields at follow-up, the eye would be considered to haveincident soft indistinct drusen. Data are presented for theworse eye, as was done for the cataract types. Incident earlyage-related maculopathy refers to the new appearance ofany soft drusen and/or pigmentary abnormalities. Incidentlate age-related maculopathy refers to the new appearanceof retinal pigment epithelial detachment or serous detach-ment of the sensory retina, subretinal or subretinal pig-ment epithelial hemorrhage, or subretinal fibrous scar orgeographic atrophy.

Diabetes was defined as a history of diabetes treated withinsulin, oral hypoglycemic agents, or diet. Newly diag-

nosed diabetes (at baseline) was defined as no previousmedical history of diabetes in the presence of an elevatedglycated hemoglobin level at the baseline examination.19

All women meeting these criteria were excluded fromfurther analyses to avoid the possibility of confounding ofrelationships between hormonal status and the ocularlesions by diabetes.

Smoking was used as a continuous variable (pack-years).Alcohol consumption was considered a continuous vari-able (grams per week).

Hypertension was defined as a history of hypertensionand current use of medication for it or a mean systolicblood pressure of 160 mm Hg or diastolic blood pressure of95 mm Hg at the time of the baseline examination(definition used at the time the baseline examinationswere designed).

SAS/STAT (SAS Institute, Cary, North Carolina) wasused to calculate chi-square statistics and perform logisticregression.20,21 For all analyses, independent variables con-sidered for inclusion were age, hypertension status, pack-years of cigarettes smoked, total grams of alcohol currentlyconsumed, and years of education.

RESULTS

WE EVALUATED THE ASSOCIATION OF SEVERAL HORMONE-

related exposures to age-related ocular end points. Theexposures were number of years of menses, menopausalstatus, number of live births, and years of use of hormonereplacement. The association of hormone-related expo-sures with each end point were considered to be indepen-dent of other lesions in that eye or in the fellow eye.Because of the powerful influence of age, significance levelswere age adjusted. We also examined the relationship ofthe following reproductive exposures: age at menarche, ageat menopause, and years of use of oral contraceptives. Wefound no significant relationship of any of these exposuresto the incidence of any specific cataract type or to anycataract. For economy of space, we present data in Table 1only for the exposures that are currently considered to bemost revealing and only those for the more inclusivecategories of the age-related eye lesions. We give oddsratios and 95% confidence intervals, because these may beof interest even in the absence of statistical significance.There was a significant trend of a lower incidence ofposterior subcapsular cataract with a greater number of livebirths (data not displayed in the table).

In multiple variable analyses, we considered the follow-ing as potentially important influences on the relationshipbetween the lens end points and the reproductive expo-sures: age, hypertension, pack-years smoked, alcohol in-take, and years of education (data not shown). Only agewas significant in explaining the rates of nuclear, cortical,or posterior subcapsular or any cataract. These character-

AMERICAN JOURNAL OF OPHTHALMOLOGY324 SEPTEMBER 2000

istics did not alter the relationship between the reproduc-tive exposures and the cataract end points.

Reproductive exposures and lesions characteristic of earlyage-related maculopathy were unrelated. Pure geographicatrophy and exudative macular degeneration occurred tooinfrequently in this population to afford meaningful analyses(eight and nine cases, respectively). Considering the inci-dence of any of these severe lesions of age-related maculopa-thy, we found no significant associations.

Multivariable analyses were performed considering thesame characteristics as described for cataracts. In thesemodels, there were no consistently significant patterns ofassociation for early or for severe age-related maculopathy.

DISCUSSION

IN OUR ANALYSES OF THE PREVALENCE DATA, WE FOUND

limited evidence to support the possibility of the protectiveeffects of postmenopausal estrogen use on nuclear cataract9

and pigmentary abnormalities.11 The incidence data didnot lend support to such an effect. We noted that in the

Eye Disease Case-Control Study, a protective effect ofpostmenopausal estrogen use was found for exudativemacular degeneration.12 We found no evidence of such aneffect, but our number of incident cases of exudativemacular degeneration was small.

Cumming and Mitchell10 reported that a history ofpostmenopausal estrogen use was more common in womenwith posterior subcapsular cataract, and use of estrogen andprogesterone in combination was associated with a greaterrisk than estrogen use alone. We found no evidence in theprevalence data,9 nor did we find any in the incidencedata, of any trend in that direction. We found a protectiveeffect of number of live births, an endogenous exposure, onthe incidence of posterior subcapsular cataract that we didnot find in the prevalence data.9 However, prevalence datafrom Cumming and Mitchell give suggest that an increasednumber of children was associated with increased rates ofposterior subcapsular cataracts.10 It is unclear whether thedifferences between the findings in the Beaver Dam andthe Blue Mountains studies regarding cataract end pointsresult from differences in the age distributions, other

TABLE 1. Reproductive Exposures and 5-Year Incidences of Age-Related Cataract and Maculopathy, Worse Eye, Beaver DamEye Study

Exposure

Any Age-Related Cataract Early Age-Related Maculopathy Late Age-Related Maculopathy

N

Incidence

(%) OR* 95% CI N

Incidence

(%) OR* 95% CI N

Incidence

(%) OR* 95% CI

No. of years of menses

#29 261 22.2 1.00 335 9.0 1.00 410 2.0 1.00

30–32 114 28.1 0.93 0.53–1.64 151 8.6 0.79 0.39–1.61 189 0.5 0.18 0.02–1.58

33–35 147 25.9 0.79 0.47–1.33 192 9.9 0.83 0.44–1.57 247 0.4 0.14 0.02–1.19

36–38 168 34.5 1.37 0.84–2.21 232 6.5 0.59 0.30–1.15 294 1.7 0.66 0.20–2.19

391 190 32.1 0.81 0.50–1.31 265 12.1 0.94 0.54–1.64 341 1.8 0.54 0.18–1.68

Menopause

No 248 9.7 1.00 261 2.7 1.00 283 0.0

Partial hysterectomy 210 18.1 0.75 0.40–1.38 262 6.5 0.86 0.32–2.27 314 1.3

Total hysterectomy 152 25.7 0.80 0.42–1.53 208 13.0 1.40 0.54–3.63 254 0.8

Natural 506 32.4 0.94 0.54–1.64 683 9.4 0.87 0.35–2.17 887 1.5

No. of live births

None 108 31.5 1.00 164 8.5 1.00 222 2.7 1.00

1 100 33.0 1.01 0.53–1.95 141 8.5 1.02 0.44–2.35 185 1.1 0.55 0.10–2.97

2 261 21.1 0.70 0.40–1.23 334 10.5 1.73 0.88–3.44 408 1.2 0.84 0.24–3.02

3 273 22.3 0.71 0.41–1.24 326 8.3 1.32 0.65–2.68 386 1.0 0.84 0.22–3.25

4 202 20.3 0.63 0.35–1.14 241 7.5 1.15 0.54–2.46 294 1.4 1.08 0.27–4.26

51 195 24.1 0.67 0.38–1.20 249 5.2 0.76 0.34–1.71 295 0.3 0.31 0.03–2.75

Hormone replacement (yrs)

Never 799 23.9 1.00 991 7.9 1.00 1223 1.1 1.00

,1 83 21.7 0.74 0.41–1.34 109 10.1 1.11 0.54–2.27 134 1.5 1.40 0.29–6.66

1 61 13.1 0.51 0.23–1.12 87 5.8 0.84 0.32–2.18 106 0.9 1.14 0.14–9.66

2 56 26.8 0.78 0.40–1.52 66 9.1 1.15 0.47–2.82 80 2.5 3.67 0.72–18.6

31 133 27.1 0.79 0.50–1.25 190 9.0 0.98 0.56–1.73 231 1.3 1.38 0.36–5.21

OR 5 odds ratio; CI 5 confidence interval.

*ORs are adjusted for age.

AGE-RELATED EYE DISEASE IN WOMENVOL. 130, NO. 3 325

environmental exposures, genetics, indications for cataractsurgery, or chance. Also, we are evaluating our longitudi-nal experience, whereas Blue Mountains researchers havepublished only prevalence data. Regarding our findings,the 5-year interval between the baseline and follow-up forthese slowly developing chronic conditions may be tooshort to demonstrate any consistent effect on incidentdisease 5 years later. Clearly, further data are needed toattempt to sort out these disparate findings.

Estrogen exposures have been found to have salutaryeffects on the nervous system22–24 and have antioxidantproperties.25 Nevertheless, none of the reproductive expo-sures that we investigated were related to the maculopathyend points. There were relatively few cases of late macu-lopathy. Further longitudinal follow-up would be neededto test the hypotheses of significant protective or deleteri-ous influences if such effects are mild to moderate.

REFERENCES

1. Kahn HA, Moorhead HB. Statistics on blindness in themodel reporting area, 1969–1970. Bethesda: US Dept ofHealth, Education, and Welfare, 1973. DHEW Pub No.(NIH) 73-427.

2. Leibowitz HM, Krueger DE, Maunder LR, et al. The Fram-ingham Eye Study monograph: an ophthalmological andepidemiological study of cataract, glaucoma, diabetic reti-nopathy, macular degeneration, and visual acuity in a generalpopulation of 2631 adults, 1973–1975. Surv Ophthalmol1980;24(suppl):335–610.

3. Jonasson F, Thordarson K. Prevalence of ocular disease andblindness in a rural area in the eastern region of Iceland during1980 through 1984. Acta Ophthalmol Suppl 1987;182:40–43.

4. Klein R, Klein BEK, Linton KLP, De Mets DL. The BeaverDam Eye Study: visual acuity. Ophthalmology 1991;98:1310–1315.

5. Sorsby A. The incidence and causes of blindness in Englandand Wales, 1948–1962. London: Her Majesty’s StationeryOffice, 1966. Report on public health and medical subjectsno. 114.

6. Mitchell P, Cumming RG, Attebo K, Panchapakesan J.Prevalence of cataract in Australia: the Blue Mountains EyeStudy. Ophthalmology 1997;104:581–588.

7. Klein BEK, Klein R, Linton KLP. Prevalence of age-relatedlens opacities in a population: the Beaver Dam Eye Study.Ophthalmology 1992;99:546–552.

8. Klein R, Klein BEK, Linton KLP. Prevalence of age-relatedmaculopathy: the Beaver Dam Eye Study. Ophthalmology1992;99:933–943.

9. Klein BEK. Lens opacities in women in Beaver Dam,Wisconsin: is there evidence of an effect of sex hormones?Trans Am Ophthalmol Soc 1993;91:517–544.

10. Cumming RG, Mitchell P. Hormone replacement therapy,reproductive factors, and cataract: the Blue Mountains EyeStudy. Am J Epidemiol 1997;145:242–249.

11. Klein BEK, Klein R, Jensen SC, Ritter LL. Are sex hormonesassociated with age-related maculopathy in women? TransAm Ophthalmol Soc 1994;92:289–297.

12. Eye Disease Case-Control Study Group. Risk factors forneovascular age-related macular degeneration. Arch Oph-thalmol 1992;110:1701–1708.

13. Campbell JA, Palit CD. Total digit dialing for a small areacensus by phone. In the 1988 proceedings of the section onsurvey research methods. Alexandria, Virginia. AmericanStatistical Association, 1988:549–551.

14. Klein BEK, Klein R, Moss SE. Incident cataract surgery: theBeaver Dam Eye Study. Ophthalmology 1997;104:573–580.

15. Klein R, Klein BEK, Moss SE. Age-related eye disease andsurvival: the Beaver Dam Eye Study. Arch Ophthalmol1995;113:333–339.

16. Hypertension Detection and Follow-up Program Coopera-tive Group. The Hypertension Detection and Follow-upProgram. Prev Med 1976;5:207–215.

17. Klein BEK, Klein R, Linton KLP, et al. Assessment ofcataracts from photographs in the Beaver Dam Eye Study.Ophthalmology 1990;97:1428–1433.

18. Klein R, Davis MD, Magli YL, et al. The WisconsinAge-Related Maculopathy Grading System. Ophthalmology1991;98:1128–1134.

19. Klein R, Klein BEK, Moss SE, Linton KLP. The Beaver DamEye Study: retinopathy in adults with newly discovered andpreviously diagnosed diabetes mellitus. Ophthalmology1992;99:58–62.

20. SAS/STAT User’s Guide. Version 6, 4th ed. Cary, NorthCarolina: SAS Institute, 1990.

21. Mantel N. Chi-square tests with one degree of freedom:extensions of the Mantel-Haenszel procedure. J Am StatAssoc 1963;58:690–700.

22. Brinton RD, Tran J, Proffitt P, Montoya M. 17 Beta-estradiolenhances the outgrowth and survival of neocortical neuronsin culture. Neurochem Res 1997;22:1339–1351.

23. Singh M, Meyer EM, Simpkins JW. The effect of ovariec-tomy and estradiol replacement on brain-derived neurotro-phic factor messenger ribonucleic acid expression in corticaland hippocampal brain regions of female Sprague-Dawleyrats. Endocrinology 1995;136:2320–2324.

24. Paganini-Hill A, Henderson VW. Estrogen replacementtherapy and risk of Alzheimer disease. Arch Intern Med1996;156:2213–2217.

25. Gerhard M, Ganz P. How do we explain the clinical benefits ofestrogen? From bedside to bench. Circulation 1995;92:5–8.

The full-text of AJO is now available online at www.ajo.com. AuthorsInteractivet, currently available in limited form, is undergoing an upgrade.

AMERICAN JOURNAL OF OPHTHALMOLOGY326 SEPTEMBER 2000