Longitudinal Rates of Cataract Surgery

Objective  To determine the cumulative probability of cataract surgery and factors accounting for such surgery.

Methods  Respondents to the Asset and Health Dynamics Among the Oldest Old survey, a national longitudinal panel, were interviewed in 1998, 2000, and 2002 to determine whether they had undergone cataract extraction since the previous interview (N = 8363 in 1998). Multivariate analysis was used to identify factors affecting cataract surgery rates.

Results  The annual incidence of cataract surgery from January 1, 1995, to December 31, 2002, was 7.4%. The prevalence of unilateral pseudophakia increased from 7.6% in 1998 to 9.8% in 2002; the prevalence of bilateral pseudophakia increased from 10.5% in 1998 to 22.3% in 2002. The self-reported vision of persons undergoing cataract surgery improved related to that of others (a difference of 0.4 on a 9-point scale; P<.001). Black individuals were less likely to undergo cataract surgery than white individuals (P<.01). The highest rates of surgery were for persons who were 65 years or older in 1998. However, persons with Medicare parts A and B coverage underwent more procedures than those with primary private employer-based coverage or the uninsured.

Conclusions  At 5.3%, the cataract surgery incidence is similar to that given in previous reports. Persons undergoing cataract surgery more often had low self-reported vision before surgery, and their vision improved on average relative to others after surgery.

Age-related cataract remains the leading cause of visual impairment among elderly persons, affecting more than 20.5 million Americans.¹ Americans 65 years or older compose 13% of the population,² and the prevalence of age-related cataract is projected to increase dramatically in the future.³ Cataract extraction is the most frequently performed surgical procedure in the Medicare-insured population.^4-6 Annual population rates in the United States vary between 2% and 6%, depending in part on the geographic region and whether the population is insured under fee-for-service or a health maintenance organization^4,6; incidence rates of 5.7% in populations 49 years or older (in 1997) and 9.1% in those 60 years or older (in 1998) have been reported in other countries.^7,8 Recent improvements in surgical techniques and new intraocular lens technologies will continue to contribute to growth in cataract surgical procedures.⁹

While several studies have generated estimates of cross-sectional or annual rates of cataract surgery, particularly in the Medicare-aged population,⁴ few data exist about the cumulative longitudinal probability of cataract surgery. This study uses national longitudinal data on cataract surgery use rates from January 1, 1995, to December 31, 2002, to provide initial information about this question. Furthermore, demographic factors related to variation in the probability of undergoing cataract surgery are examined in this nationally representative data set.

The Asset and Health Dynamics Among the Oldest Old (AHEAD) survey is a national panel survey of US households, drawn as an area probability sample with an oversampling of blacks, Latinos, and Floridians.¹⁰ When the first wave of interviews was conducted in 1993, the survey sampled noninstitutionalized persons 70 years or older and their spouses or partners (who could be any age). Follow-up interviews of the same persons were conducted in 1995, 1998, 2000, and 2002. Persons were followed up irrespective of the setting in which they resided; by 2002, 8.9% of respondents lived in nursing homes. Spouses or partners received identical interviews as sample persons. In 1998, AHEAD was merged with the Health and Retirement Study (HRS), a panel survey of persons primarily aged 51 to 61 years when first interviewed in 1992.

The AHEAD (and the AHEAD/HRS merged sample) is unique in collecting comprehensive, longitudinal, and national information on self-reported health services use, including cataract surgery, self-reported health, visual status, functional and cognitive status, demographic characteristics, and family structure. At each wave, additional respondents are added periodically to maintain sample size. Detailed and up-to-date information on survey design, sampling, data collection, and follow-up and institutional review board information are available from the AHEAD/HRS Web site.¹¹

Data from the 1993 AHEAD interviews were not used in this study because the survey only asked whether the person had ever undergone cataract surgery, but did not request information on the number of eyes undergoing such surgery. The cataract surgery question was a retrospective question covering the period since the last interview and was skipped for persons younger than 65 years in the survey year. We only included persons who were 65 years or older in 1998 and who had reported no prior cataract surgery in 1993 or 1995.

We began with 8670 respondents, interviewed in 1998, who were 62 years or older and observed these persons through 2002 unless respondents were not reinterviewed because of death or for other reasons (Figure 1). All persons in the analysis sample had not undergone cataract surgery on 2 eyes at the previous interview. A total of 8363 respondents were still available in 1998, 6572 in 2000, and 5218 in 2002. Slightly different frequencies were used in regression analysis because we excluded an observation if there was missing information on any of the covariates. Deaths accounted for about half of total losses of the sample from 1995 to 2002. The data set did not include other reasons for nonresponse. Proxy responses for persons with substantial physical or cognitive impairments and those willing to have someone else answer on their behalf were included in the analysis, and were identified by an indicator variable in the regression analysis. Persons with missing data were excluded from the regression analyses. Persons entered the sample repeatedly until death or loss to follow-up.

The AHEAD/HRS survey asked respondents whether they had ever undergone cataract surgery. Those who gave affirmative responses were asked whether they had undergone such surgery since the previous interview. If yes, they were asked how many eyes were operated on. To compute number of cataract surgical procedures since the previous interview, we took the difference between the number of eyes operated on at the current and previous interviews. In all 3 years combined (1998, 2000, and 2002), there were discrepant values for the cataract surgery variable in 1.6% of cases. Because the rate is low, the effect on our estimates of prevalence and incidence is minor.

We used fixed-effects regression analyses to assess the likelihood of having undergone cataract surgery between successive interviews and among individuals. The primary dependent variable was number of cataract surgical procedures since the last interview. We also assessed whether having undergone cataract surgery improved self-reported visual functioning, taking advantage of a survey question that asked respondents to rate their eyesight as excellent, very good, good, fair, poor, or legally blind. In particular, transitions from vision ranging from legally blind to fair to vision ranging from good to excellent were assessed.

Explanatory health variables, defined for the beginning of the time interval (eg, for 1995, when the dependent variable was the number of cataract procedures the person underwent between 1995 and 1998), included self-reported measures of overall health, a hearing impairment, and mental health. Self-reported health was measured by responses to a question that asked individuals to rate their health as excellent, very good, good, fair, or poor. Of the 5 mutually exclusive groups, we created 1 indicator variable, fair or poor health compared with excellent to good health, the omitted reference group. Previous research¹² has indicated that this subjective measure of self-reported health is systematically related to objective measures of health, including mortality. The mental health measure was a binary variable based on the following question, “Have you had or has a doctor told you that you have any emotional, nervous, or psychiatric problems?”

Analyses were adjusted for depression, using an abridged version of the 20-item Center for Epidemiological Studies Depression Scale.¹³ The abridged version asked whether, during the week before the interview, the respondent: (1) felt depressed, (2) felt everything he or she did was an effort, (3) experienced restless sleep, (4) could not get going, (5) felt lonely, (6) felt sad much of the time, (7) enjoyed life, and (8) was happy. Each affirmative response to the first 6 and each negative response to the last 2 questions received a score of 1. The sum was the depressive symptoms score used in our analysis. The validity and reliability of the Center for Epidemiological Studies Depression Scale has been assessed.¹³

The AHEAD survey also included a battery of identical questions in each wave to elicit the cognitive functioning of respondents.¹⁴ These questions included immediate and delayed word recall, a serial 7s subtraction test, counting backward, providing the date, and naming objects, yielding a maximum score of 35 points. Questions were also included to measure knowledge, language, and orientation. Total cognitive scores per individual ranged from zero (lowest functioning) to 35 (highest functioning). The validity and reliability of AHEAD's cognitive test have been evaluated.^15,16

Cognitive tasks represented a range of difficulty levels, with naming tasks being the easiest and the recall and the serial 7s tasks being the most difficult.¹⁵ Most of the cognitive measures were adapted from the Telephone Interview for Cognitive Status.¹⁶ Follow-up surveys were conducted by telephone for individuals younger than 80 years and in person for others. Herzog and Rodgers¹⁵ found no difference in measured performance depending on whether the cognitive test was conducted over the telephone or in person. We defined a variable for whether the person had a total cognitive score of 10 or less.

We included explanatory variables for sex, race/ethnicity (black, Latino, and other, with white being the omitted reference group), years of schooling, marital status, annual income, and age. We defined separate binary variables for age in 5-year intervals, but all persons older than 90 years were represented by a single binary variable, with those aged 62 to 64 years being the omitted reference group. A binary variable was included for whether the survey respondent was a proxy for the sample person. Finally, we included binary variables for the type of insurance coverage that individuals had. Categories were as follows: (1) employer provided as the primary source of insurance, (2) Medicare parts A and B with other supplemental coverage (eg, Medigap, Civilian Health and Medical Program of the Uniformed Services, Civilian Health and Medical Program of the Department of Veterans Affairs, and health maintenance organization enrollment), (3) Medicare parts A and B with Medicaid, (4) Medicare part A but not part B, and (5) uninsured. The omitted reference group was Medicare parts A and B only.

The primary dependent variable was the number of cataract surgical procedures the person reported having between 2 adjacent interviews. We used linear fixed-effects regression with year and individual fixed effects. The fixed-effects approach uses within-individual changes over time to identify effects of changes in the explanatory variables on changes in the outcome variable. Therefore, in the estimation of each model, only information on persons experiencing at least 1 change in the dependent variable is used. For comparison, we used ordinary least squares regression. We used Stata, version 8.2 software.¹⁷

The fixed-effects methods exploit the repeated cross-sectional nature of panel data using changes in explanatory variables for each person over time (each wave of the panel) to predict changes in the dependent variables. Variables that did not change over time for each person (eg, race and schooling) had to be excluded. Use of the fixed-effects method eliminates the possibility of biased variable estimates resulting from a correlation between covariates and time-invariant factors specific to the individual and not measured by the survey.¹⁸ Results from a fixed-effects specification show effects of the changes in the explanatory variables on the number of cataract surgical procedures between 2 adjacent interviews.

The prevalence of fair or poor self-reported vision increased from 27% to 31% during the 7-year period over which sample persons were observed (Table 1). Persons were between the ages of 62 and 106 years. The mean age increased from 76.32 to 80.32 years. Household annual income decreased from $25 000 in 1998 to $18 000 in 2002 (to be expected, given retirement onset and patterns of spending relative to annual income after retirement). Persons in the cohort whose primary coverage was employer based declined from 9% to 5% from 1998 to 2000. The uninsured were 13% of the sample in 1998, but were 2% of the sample in 2000.

Between January 1, 1995, and December 31, 1998, 7.1% of respondents underwent surgery on 1 eye and an additional 5.6% underwent surgery on 2 eyes (Table 2). Between January 1, 1998, and December 31, 2000, 7.0% underwent surgery on 1 eye and 4.9% underwent surgery on 2 eyes. Between January 1, 2000, and December 31, 2002, the corresponding values were 7.4% and 4.8%, respectively. Taken together, these estimates imply an annual rate of cataract surgery of 5.3%.

As of 1995, 88.4% of subjects had not undergone cataract surgery (Table 3). Seven years later, 67.9% had not undergone such surgery. The prevalence of individuals who underwent unilateral cataract surgery was 7.6% by 1998, 8.7% by 2000, and 9.8% by 2002. The prevalence of 2 operated-on eyes increased from 5.5% to 22.3%.

Persons who underwent surgery had lower self-reported vision before surgery than did the others (χ²₄ = 357.5) (Figure 2). Persons who underwent surgery on average experienced 0.4 greater improvement in vision than those who did not (P<.001). Of those persons who underwent surgery during the past 2 to 3 years and who reported having poor vision at the beginning of the period, 36.0% reported continued poor vision, with the remaining 64.0% transitioning to vision that was excellent to fair. Among those who did not undergo surgery and who reported poor vision at the beginning of the period, 64.2% still had poor vision at the end of the period, with 35.8% having improved vision. Among those with excellent self-reported vision in the prior period, 34.2% reported excellent vision after surgery. For those who did not undergo surgery, only 27.8% still had vision.

With individual and time fixed effects (Table 4), persons with fair or poor self-reported vision were more likely to undergo surgery than those with excellent to good vision at the beginning of the period. On average, the effect of fair or poor visual status was to increase the number of eyes operated on by 0.11 within the 2- to 3-year interval. Persons in the groups aged 65 to 69, 70 to 74, and 75 to 79 years at the beginning of the period had fewer procedures during the 2- to 3-year periods than did those in the omitted 62 to 64 years age group. Men were not as likely as women to undergo cataract surgery, but the difference was not statistically significant. Although the 1995 to 1998 period was 3 years, and subsequent periods were 2 years, individuals underwent more procedures in the latter 2 periods, holding the other factors constant.

When we dropped time-invariant binary variables for interview year in the regression with individual fixed effects, because they are perfectly correlated with the fixed effects, married persons underwent fewer procedures and persons 70 years and older were more likely to have undergone surgery. Persons with private employer-based coverage underwent fewer procedures than those in the omitted reference group (persons with Medicare parts A and B coverage). The uninsured also underwent fewer procedures. In ordinary least squares regression, black individuals underwent fewer procedures than did white individuals (the omitted reference group).

Nonsignificant factors pertaining to the likelihood of cataract surgery were depression (determined by the Center for Epidemiological Studies Depression Scale score), overall health, mental status, hearing problems, sex, years of education, memory status, annual income, and Medicaid insurance (Table 4). Patients with supplemental insurance (62%) were more likely to undergo cataract surgery (mean probability, 0.27; P=.05) than the uninsured or persons with only Medicare parts A and B. Black individuals underwent slightly fewer cataract surgical procedures (−0.04 per 2 to 3 years; P<.01) than did white individuals.

By the end of the observational period, nearly a third of persons in the United States 69 years or older underwent a cataract procedure in at least 1 eye, more than double the number of such persons in the cohort who had undergone such surgery 7 years previously. The incidence of such surgery from 1995 to 2002 was 7.4%. This is slightly higher than previously reported.^4,6

The regression analysis identified factors that lead to such surgery and factors that do not. Important in the former group are race, age structure, and source of payment. Psychological problems, including depression, hearing impairments, low general health status, and poor cognition are not determinants of undergoing such surgery. Other studies^19,20 have reported changing trends in the frequency of cataract surgery, but have done this for limited geographic populations.

The prevalence of cataract extraction in 1 eye increased to 8.7% in 2000, and then to 9.8% by 2002. Yet, the overall probability of having 1 or 2 eyes operated on increased to 30% during the observational period, mostly attributable to cohort aging. These estimates are higher than the 5.1% previously reported; however, this discrepancy may be expected because the lower estimates were for persons 40 years or older.³ Estimates in that study were obtained by applying prevalence rates from several population-based studies³ and were cross-sectional. Equivalent estimates from our study were 11.6% in 1995 in the initial survey of those who had undergone cataract surgery. The difference most likely reflects the nature of the 2 studies and their populations. While individuals may overstate the rate at which they undergo cataract surgery by self-report, such bias may not be sufficiently high as to account for the differences noted. Furthermore, 28% of the Medicare-insured population never sees an eye care provider in a 5-year period, suggesting other variances in estimates.²¹

Self-reported vision was elicited on a 5-point scale, ranging from poor to excellent. Individuals with self-reported fair or poor vision were more likely to undergo cataract surgery than others. Changes in vision were assigned numerical values from −4 to 4, depending on the number and direction of changes in responses between 2 adjacent interviews (Figure 2). By using the 9-point range as the denominator and changes on the scale as a measure of change in vision, relative to those who did not undergo surgery, those who underwent surgery experienced a nearly 4.5% improvement in self-reported vision. This relative change is probably a lower bound on the true estimate because we measured changes in self-reported vision over a 2- to 3-year interval. In fact, vision may have declined after an interview but before surgery.

Prior studies²² have documented improvements in visual function following cataract surgery. Visual impairment correlates with overall perception of well-being. Prior studies³ have consistently shown a higher incidence of cataract surgery among women and with increasing age. In our study, women also had higher rates, but this result was not statistically significant at conventional levels. Also consistent with prior studies,³ persons 70 years or older in our study were more likely to undergo cataract surgery compared with persons younger than 70 years. This pattern is likely explained by the slowly progressive and age-related nature of cataract. It is possible that this trend does not continue into the 90 years and older demographic category because the individuals have already undergone prior cataract surgery. Also, other health concerns may cause individuals 90 years or older to be deemed poor surgical candidates.²³

Not surprisingly, black individuals were less likely to undergo cataract surgery than white individuals in our study. An oversampling of black individuals was used to ensure representative inclusion of such persons. Historically, there has been underuse of many health care services among the black population. It is unlikely that lower rates of cataract extraction could reflect lower rates of lens opacification among black individuals because it has been reported that unoperated-on cataract accounts for 27% of all blindness among black residents of Baltimore, Md; also, blindness from cataract is much more likely among black than among white individuals.²⁴ More plausible explanations to understand the slightly lower rates of cataract surgery among black persons include lack of access to cataract surgery, resignation to decreased vision with the aging process, and lack of understanding of the possible benefits of cataract surgery.²⁵

We acknowledge several study limitations. First, almost 30% of the original sample was lost to follow-up. Although sample attrition may cause bias, it is not clear what the direction of bias would be. Second, the self-report data relied on the respondent's subjective assessment of his or her own visual acuity and the accuracy of reporting cataract surgery. While this is likely to correlate with objective clinical measures, there may be some variation due to subjective interpretation, lack of recall, or misunderstanding of a surgical procedure. Some respondents, for example, may confuse laser capsulotomy with posterior capsular opacification resulting from previous cataract surgery as having undergone an additional cataract surgery. But particular confusion is unlikely given that only 1.6% of respondents had discrepant results across interviews. Third, we did not have clinical information from the physician's perspective. Type of lens opacity present within the individual was not assessed. The likelihood of undergoing cataract surgery has been linked to the presence of specific types of cataract; some studies²⁵ cite mixed opacities as more likely to be associated with visual impairment, while others²⁶ cite nuclear sclerotic cataracts as the subtype present before cataract surgery. Posterior subcapsular cataract has been linked to increased odds of cataract surgery in other studies.^27,28 Subtype of lens opacity is an important determinant of likelihood of surgery.²⁶

Along with age-related cataract, age-related macular degeneration and glaucoma are the leading causes of vision loss in the United States, and advancing age is among the risk factors for the development of both. The AHEAD/HRS survey did not ask individuals about these conditions, and visual loss may have resulted from these entities rather than from age-related cataract. Indeed, the fact that the condition of many of those with fair or poor vision did not improve after cataract surgery could be some evidence of the effect of these conditions.

Understanding longitudinal cataract surgery patterns and factors influencing the likelihood of undergoing surgery can have a significant effect on Medicare and health care delivery. Information from such analysis will better allow us to develop policies and procedures to ensure access to appropriate care.

Correspondence: Frank A. Sloan, PhD, Center for Health Policy, Law, and Management, Duke University, 114 Rubenstein Hall, Campus Box 90253, Durham, NC 27708 (fsloan@hpolicy.duke.edu).

Submitted for Publication: June 30, 2005; final revision received February 21, 2006; accepted February 23, 2006.

Financial Disclosure: None reported.

Funding/Support: This study was supported in part by grant RO1-AG-17473 from the National Institute on Aging and by Research to Prevent Blindness. Dr Lee is a recipient of the Lou Wasserman Merit Award.

Role of the Sponsor: The funding bodies had no role in data extraction and analyses, in the writing of the manuscript, or in the decision to submit the manuscript for publication.