Incidence of Age-Related Cataract: The Beaver Dam Eye Study

Objective  To estimate the incidence of nuclear, cortical, and posterior subcapsular cataract in the Beaver Dam Eye Study cohort.

Design  A population-based study of the prevalence of age-related eye diseases was conducted between 1988 and 1990 (n=4926), and a follow-up study was conducted between 1993 and 1995 (n=3684). The evaluations included photographic documentation of the lens. Slit-lamp photographs were taken to assess nuclear opacity, and retroillumination photographs were taken to assess cortical and posterior subcapsular cataract. The grading of photographs was done in a masked manner by trained graders using the same protocols for baseline and follow-up photographs. The graders were the same for both evaluations.

Participants  Persons aged 43 to 86 years who were identified through a private census conducted from 1987 to 1988 of the population of Beaver Dam, Wis, were invited for a baseline examination held between 1988-1990 and again for a follow-up examination held between 1993-1995.

Results  Incident nuclear cataract occurred in 13.1%, cortical cataract in 8.0%, and posterior subcapsular cataract in 3.4% of right eyes. The cumulative incidence of nuclear cataract in right eyes increased from 2.9% in persons aged 43 to 54 years at baseline to 40.0% in those aged 75 years or older. For cortical and posterior subcapsular cataract, the corresponding values were 1.9% and 21.8% and 1.4% and 7.3%, respectively. Women were more likely than men to have nuclear cataract even after adjusting for age.

Conclusions  Incident cataracts are common age-related problems, and incidence increases with increasing age at baseline. These data will help in planning for future care (eg, cataract surgery and change in spectacle correction) and in investigating the importance of risk factors.

THE HIGH prevalence of cataract and the resulting need for care by sighted attendants, spectacles, or surgery is an age-related health problem in virtually all studies of this condition.^1-5 While cross-sectional data are important for estimating the burden associated with this condition, incidence data are necessary to develop insights regarding potential causative factors and, when long-term follow-up is possible, to estimate the interval between exposure and the development of disease.

Age-related cataracts are characterized by the 3 common lens lesions, nuclear sclerosis, cortical opacities, and posterior subcapsular opacities. While these 3 conditions are age related, they differ in their distributions in different groups, possibly because of different exposure levels.^6-11 For example, persons with greater exposure to UV light seem to be at greater risk for cortical opacities. Leske and Sperduto¹² and West and Valmadrid¹³ have summarized the published associations of various risk factors and specific cataract types. Therefore, it is important to evaluate the incidence of each of these lesions independently.

While data about the incidence of lens opacities have been described for select populations¹⁴ and for persons serving as controls in case-control studies,^15,16 there is a lack of population-based incidence data derived from an objective recording of disease status. Also, cataracts develop gradually; it is, therefore, necessary to permit a long enough interval between evaluations of the lens to be able to observe incident disease.¹⁵ Thus, this article describes the cumulative incidence of cataract in a well-described adult population who has been seen at baseline and after an interval of almost 5 years. The cohort was examined first between 1988 and 1990 and again between 1993 and 1995.

Beaver Dam, Wis, is a town of approximately 17000 persons, located northeast of Madison, Wis, the state capital. A private census of the population of Beaver Dam was performed from 1987 to 1988 to identify all persons aged 43 years and older. Persons aged 43 to 86 years were invited for the study evaluation, which was performed during a 2.5-year period beginning March 1, 1988. The examined group was 99% white. The tenets of the Declaration of Helsinki were followed, institutional human experimentation committee approval was granted, and informed consent was obtained from each subject. During the study visit, standard measurements were obtained and standard questionnaires were administered. All subjects identified at the initial census¹⁷ were invited for the second examination. Information pertinent to this article was supplied only by those who participated in the first and second examinations (n=3684). The differences between participants and nonparticipants at the baseline¹⁸ and follow-up¹⁹ examinations have been previously published. In brief, those who were alive but not participating at the second visit were older and had poorer visual acuity. Those who died were older, were more likely to be men, had a poorer visual acuity, were more likely to have diabetes,²⁰ and were more likely to have more severe nuclear sclerosis.²¹ There were 226 deaths before the first examination and an additional 556 deaths before the second examination. The average age of persons at the follow-up examination was 65 years.

On average, the follow-up examinations occurred 4.8 years after the first examinations, and they were performed so that participants were seen in approximately the same order as they were seen in the prevalence examinations. The procedures used at the follow-up examination were the same as those used at the baseline examination.¹⁸

Photographs of the lenses were taken with 2 different cameras: a slit-lamp camera and a retroillumination camera.²⁰ The grading procedures for the lens were based on detailed, codified decision rules.²⁰ Graders were masked to subject identity, personal characteristics of the subjects, and presence and severity of lens lesions at the first examination. Scores for nuclear sclerosis were based on comparisons with standard photographs. The scale has 5 steps of severity based on opacity of the nucleus. Levels 4 and 5 were considered to be cases of nuclear cataract in publications of the prevalence data.¹ Scores for cortical and posterior subcapsular cataracts were based on weighted estimates of degree of opacity of lens area as defined by a circular grid, divided into 8 "pie-wedged" peripheral areas and a central circular area overlaid on the photograph. Prevalent cases of cortical cataract were those with an opacity of 5% or more of the lens "surface." Posterior subcapsular opacity was defined as 5% or more of a grid segment. The classification of prevalent cataract corresponded to a lens opacity of sufficient severity to meet the criteria that a clinical ophthalmologist would be comfortable labeling as a cataract.

Horizontal and vertical pupil diameters were measured from the photographs before grading cortical and posterior subcapsular opacities. The mean estimates were used in the analyses concerning pupil size.

Grading was subject to a quality review procedure such that a sample of 50 to 60 photographic sets was regraded biannually by graders who were masked to previous gradings. These photographs had been used during the baseline examinations. These gradings provided data for measures of intergrader and intragrader variability. In addition, a sample of photographs taken during the previous 6-month period was recirculated in a masked manner for regrading. The results of measures of variability of these gradings were reported to each grader in an effort to maintain consistency. In addition, 1 of the principal investigators (B.E.K.K.) met monthly with all graders to review gradings for a sample of eyes selected from the previous month's gradings. Measures of variability indicated no evidence of marked or consistent change in variability during the course of either the prevalence or the follow-up study or between the 2 study periods. The same graders evaluated photographs from the baseline and follow-up examinations. Data about the reproducibility of gradings using these systems were previously published prior to reporting the prevalence findings.¹⁶ In general, reproducibility was similar for intragrader and intergrader comparisons. Weighted κ values were between 0.65 and 0.75 for each cataract type.¹⁶ The estimates of incidence were based on all persons having corresponding gradable subfields at both visits.

We considered 2 different approaches to evaluate the lens end points during the interval between the 2 examinations: incidence and progression. In analyses for the incidence of a particular cataract type, the population at risk was all those eyes free of that cataract type at baseline. The incidence of "any" cataract was computed based on the presence of at least 1 of the cataract types. Progression was deemed as any increase in the severity or involvement of a lens opacity, including absence to presence. In analyses of progression for each type of opacity, nuclear sclerosis was defined as a change of 1 level. The progression of a cortical or posterior opacity was defined as a change of 0.75 U or more after square root transformation of the data. Examples of the amount of change in severity of gradings needed for a 0.75-U change on the square root scale are as follows:

This tabulation indicates that for an opacity involving 5% of the lens at baseline, 8.9% of the lens would need to be involved at follow-up for that cortical opacity to be considered to have progressed. For greater amounts of involvement at baseline, a greater increment in area involved is needed at follow-up to be considered as progression. The square root transformation was used in analyses of cortical and posterior subcapsular opacities to account for the increase in variability of estimated area of involvement as the base amount increased. The unit change of 0.75 on the square root transformed scale was chosen because it minimized the degree regression we view as spurious while also minimizing the amount of data that would have been disregarded were a larger interval of change chosen. The amount of change could be used in calculating progression to provide information on the magnitude of the change. However, interpretation of this variable is difficult, and results using any progression compared with amount of progression were almost identical. Therefore, progression in these analyses refers to any change greater than our prespecified amount. These amounts were determined based on our reproducibility results.

We evaluated relations for right and left eyes for each opacity. We provide incidence findings for right eyes and for first or both eyes. For progression, we provide findings for right eyes and for either eye. When either eye was considered, the individual had to have had data on both eyes available at both visits to be included.

There were specific characteristics causing an eye to be excluded from the calculation of incidence of an age-related cataract. These included lens surgery, trauma, confounding lesions in the photographs, absence of a photograph, or an ungradable photograph at the baseline or follow-up examination.

Exclusions from analyses of these longitudinal data were approached by a hierarchical scheme: baseline exclusions were for confounding ocular lesions, past cataract surgery, ungradable or missing photographs, and prevalent disease; follow-up exclusions were for confounding ocular lesions, past cataract surgery, and ungradable or missing photographs.

A software system (Wisconsin Storage and Retrieval, University of Wisconsin Clinical Cancer Center, Madison) was used to process subject files. Another system (SAS, SAS Institute, Cary, NC) was used to calculate χ² statistics and logistic regression. Tests of trend were calculated by the Mantel-Haenszel method.

After excluding the 785 right eyes with missing information, direct trauma, or prevalent disease, 2899 eyes remained for further analyses of nuclear sclerosis. The incidence of nuclear cataract was 13.1% for right eyes and 17.3% for first or both eyes (Table 1). The distribution of incident nuclear cataract by age and sex indicated an age effect in both sexes. The incidence for women was higher than for men.

There were 791 right eyes excluded from consideration for analyses of cortical opacities, leaving 2893 eyes for analysis. The incidence was 8.0% for right eyes and 10.2% for first or both eyes (Table 2). An age trend was apparent for both sexes.

There were 656 right eyes excluded from consideration for analyses of posterior subcapsular opacities, leaving 3028 eyes for analysis. The incidence was 3.4% for the right eyes and 4.3% for first or both eyes (Table 3).

Progression of nuclear sclerosis occurred in about 48% of right eyes (Table 1). There was little evidence of a consistent effect of age on progression, although the lowest rates occurred in the oldest group. The differences between men and women were not significant. Progression of cortical opacities occurred in about 20% of right eyes (Table 2). There was a significant trend by age in analyses for right eyes and for either eye. There was no difference by sex. Progression of posterior subcapsular opacities occurred in about 2.4% of right eyes (Table 3). There was a significant effect of age in analyses for right eyes and for either eye (Table 3). There was no difference by sex.

For some public health purposes, it may be useful to estimate the incidence of any of the age-related opacities that we are considering (Table 4). Incidence increased significantly (P<.001) with age. Women were slightly more likely to have incident cataract than men because of their higher incidence of nuclear cataract.

Of the 245 cases of nuclear cataract in the right eye as classified at baseline, 24 (9.8%) were graded as less severe at follow-up. There were 6 (2.7%) of 224 such cases for cortical cataract and 8 (20.0%) of 40 such cases for posterior subcapsular cataract. We consider these changes to be due to misclassification. Although misclassification was likely to occur in both directions, if we assume that the same percentage of incident cases was similarly misclassified and modify the incidence accordingly, the rate would be 11.3% for nuclear, 7.6% for cortical, and 2.6% for posterior subcapsular cataract.

In this study, we made a conscious effort to achieve the same pupil size at the baseline and follow-up examinations. The baseline mean pupil diameter for right eyes was 7.31 mm, and the follow-up mean pupil diameter was 7.33 mm. There were 454 right eyes for which the pupil diameter differed by 10% or more between the visits. These differences in pupil size were not systematically associated with higher or lower frequencies of incidence of any cataract type when comparing with those eyes with no change in pupil size.

For this study, we have defined cataract as it refers to the specific lens lesion. Visual acuity was not a criterion included in the definition. The most common visual acuity in eyes without any cataract was 20/20; those with a single cataract type were most likely to have a visual acuity of 20/25. There were few eyes with 2 or more cataract types; the most common visual acuity in this group was 20/40. Of eyes with a cataract, the visual acuity was worst in those with posterior subcapsular cataract and best in those with cortical cataract.

Nuclear sclerosis is a process that is ongoing throughout adulthood if not throughout the entire life span. Thus, the diagnosis of nuclear cataract depends on an arbitrary definition of "amount" of sclerosis. We have arbitrarily defined this by a 35-mm color transparency of a lens characterized by some loss of distinction of lamellae and of the central sulcus in the slit-lamp photograph. The level we chose corresponds to a severity of sclerosis that we believe is a useful definition of clinical nuclear cataract. Progression of nuclear sclerosis, although common, is not clearly age related, while incidence is. This may be related to differences in (1) rates of progression, dependent on the initial severity of sclerosis; (2) incident cataract surgery; or (3) differences in the "size" of intervals of the measurement scale. We believe that all 3 components may contribute to this finding.

We consider that all subjects may progress for this lesion regardless of whether they were considered to be cases at baseline because of the continuous nature of sclerosis of the nucleus throughout life. It is unlikely that determinants of progression, after reaching the arbitrary level of 3 in our scheme, would differ from those affecting the lens that had lower levels of sclerosis at baseline. In addition, the number of cases of nuclear sclerosis progressing after being classified as prevalent disease is too small for meaningful analyses.

Any cortical opacity is usually considered to be a pathologic change in the lens. However, in a clinical situation, physicians often will not diagnose or label a cortical opacity as cataract until obvious characteristic changes are seen in a retroilluminated examination. We chose greater than 5% of the lens (considered for grading to be a plane circle) as the definition of cortical cataract because this amount is comparable with practice patterns. The estimation of the frequency of cortical cataract may be influenced by pupillary dilation and by other pathologic features that may be found in the same anatomical region (eg, pseudoexfoliation). One concern may be about underestimating the severity of a cortical opacity because of pupil size, especially when it occurs in the peripheral cortex in a circumferential or arcuate pattern. This distribution of opacity is not commonly found as the sole type of cortical opacity. While its extent may be underestimated, as will the basilar involvement by the more common cuneiform opacities, this effect would have to obscure enough to cause a lens with more than 5% involvement to be graded as less than this. While we cannot be certain how much of the lens area we do not observe because of less than complete pupil dilation, we can estimate the effect of different pupil sizes at given diameters on observable total area. In our data, the average pupil areas were 41.97 mm² and 42.20 mm² at the baseline and follow-up examinations, respectively. We may have missed some true prevalent cases at the baseline examination if the pupils were not maximally dilated and if the unseen area constituted at least 5% of the area and it had a substantial cortical opacity that was revealed at the follow-up examination. While we observed an effect of pupil size on the estimated area of involvement, its influence is apparently not great when the variation in pupil size between evaluations is small.

Some persons with cortical opacities that were of insufficient severity to be classified as cortical cataract at baseline and those whose cortical opacity was obscured by the iris would have been ineligible to contribute to analyses of risk factors for progressive disease if we included only those with prevalent cataract to be subject to progression. It is also true that the number of persons with prevalent cortical cataract in whom we could search for causes of progression was relatively small when considering the need to adjust for confounders. Thus, because we are not convinced that factors for progression in "cases" vs "noncases" are different or that we could distinguish them in our analyses, we have included all subjects as possible candidates for progression.

Any posterior subcapsular opacity is considered a pathologic change. However, imaging these opacities is difficult. Photographing this opacity requires a clear focus of the most posterior part of the lens. Thus, any other opacity that blocks the illuminating beam (and retroillumination) may obscure the pathologic features and influence the grading. Lesions that we have specifically identified as confounding pathologic features are central corneal scars, dense vitreous hemorrhage, extensive anterior or posterior cortical cataract "retro dots" (white anterior cortical opacities that may obscure the pathologic feature of interest), and dense nuclear cataract. Mittendorf dots may be difficult to distinguish from a posterior subcapsular opacity. However, because of their characteristic location, morphological features, and size, we believe that we have correctly identified them and have not misclassified them as posterior subcapsular cataract.

To estimate the rates of progression and the risk factors for progression of posterior subcapsular cataract, only those who had this lesion at baseline could be considered. However, the number of such cases was small, and cataract surgery has further depleted this number. Thus, for posterior subcapsular cataract, we include progressed with incident disease, although we understand that these analyses are "weighted" heavily by those with incident disease.

Graders and grading schemes were identical for the baseline and follow-up examinations. However, misclassification has occurred, as it has in other studies.¹² In our recalculations of the possible influence of misclassification on rates, we used generous assumptions about the direction and amount of misclassification. Because the rates were close to the unmodified rates, misclassification does not seem to be a major problem.

The reasons for cataract surgery are unknown to us. Therefore, we are not able to attribute surgery to the presence of any of the cataract types we identified at baseline, to others developing anew in the interval that we did not observe, to other ocular pathologic features, or to socioeconomic reasons. One of us (B.E.K.K.) contacted the private offices of ophthalmologists who participants identified as the surgeon performing the cataract surgery. Medical records were reviewed by one of us (B.E.K.K.) or through telephone interviews of the surgeons or their office staff after they retrieved the medical record of the patient. Of the 274 right eyes that underwent surgery, there were 104 cases of nuclear cataract, 10 cases of cortical cataract, 42 cases of posterior subcapsular cataract, 72 cases of mixed cataracts, and 46 cases for which no specific cataract type was recorded in the medical record preceding the relevant surgery or for which the medical record was unobtainable. The numbers are similar for left eyes. Including these data as though they were equivalent to our gradings, the rates of incident nuclear, cortical, and posterior subcapsular cataract for persons would have been 18.6%, 10.6%, and 6.7%, respectively. Because we cannot ascertain the similarity of the clinicians' grading conventions compared with each other and with ours, we have based our analyses solely on the grading data. From our own data, we do know that having at least 1 of the age-related cataract types increased the risk of surgery in the interval and that the risk was greatest for posterior subcapsular cataract, but this does not entirely explain the frequency of incident surgery (21 surgical procedures were performed in right eyes that do not meet our definition of cataract at baseline).¹⁵

Cortical cataract was more prevalent in women than in men in this population, but there was no sex difference for incidence. Cataract surgery occurring during the interval between the baseline and follow-up examinations was more common in women. Cataract surgery was performed more frequently in persons with nuclear and posterior subcapsular cataracts than in those with cortical cataracts.¹⁹ Were surgery not performed, many of these persons may have progressed to incident cortical cataract. Also, there may have been cohort differences in exposure levels that were causally related to cortical cataract in our population that would be reflected in a higher prevalence than incidence in women. It is not possible to determine whether either or neither of the previously mentioned possibilities is related to the discrepancy.

As in any incidence study, it is not possible to ascertain the effect of nonparticipants on the estimated rates. The live nonparticipants were older and had a poorer visual acuity than those returning for the second visit; therefore, they were more likely to have incident disease. Thus, we may be more likely to have underestimated incidence in the older age groups. In those who died, incident disease may have slightly less importance in projecting needs for long-term care (especially cataract surgery) as many of these people would not have undergone surgery before death. However, incident cataract may have influenced the quality of their lives before death. In analyses using the severity of nuclear sclerosis at baseline, there was a significant independent relationship of this lens lesion to mortality.²¹ Thus, there may be selective effects of mortality on estimates of incidence and progression of specific lens opacities.

The extrapolation of our findings to other communities, or to the population of the United States in general, requires many assumptions; the validity of some of these assumptions may be unknown. For example, we do not have an ethnically diverse population. Therefore, we cannot determine whether the experience of persons in Beaver Dam, who are largely of northern European background, is likely to be similar to that of persons of other backgrounds. Also, if urban and rural differences affect lens opacities, we are unable to estimate such effects. Nevertheless, Beaver Dam is a community that is similar to small towns in the midwest, and we are unaware of any unique exposure levels that would cause us to suspect that the population is at particular risk. We have found that incident cataract and progressive lens opacities are common and that, while persons in the oldest age groups are generally at highest risk, there is substantial incidence and progression, even in our youngest age group. These data suggest that the search for risk and modifying factors should begin during the young adult years and possibly during childhood.

Accepted for publication September 25, 1997.

Corresponding author: Barbara E. K. Klein, MD, MPH, Department of Ophthalmology and Visual Sciences, University of Wisconsin Medical School, 610 N Walnut St, 460 WARF, Madison, WI 53705-2397.