Racial Differences in Glaucoma Care:  The Longitudinal Pattern of Care FREE

Racial disparities in the provision of health care services have been a relatively unexplored topic in eye care services, whereas they have received prominent attention in other fields of medicine and health care. Within eye care, study results demonstrated that black persons have a rate of utilization of glaucoma services that is more than twice that of the white population.^1- 3 This result has been compared with estimates from population-based studies of glaucoma prevalence in which prevalence among blacks was 4 times higher than among whites, leading to the conclusion that glaucoma is undertreated in blacks.^1,4- 7 Growing evidence of undertreatment in blacks and other minorities in other health arenas,^8- 14 especially surgical services, is consistent with this concern of a potential disparity in access to and use of eye care services for glaucoma.

The authors of these prior studies used cross-sectional estimates of care for the Medicare population during 1 year’s course of care. In this study, we examined whether differences in care patterns exist on a longitudinal basis for individuals with glaucoma enrolled in the Medicare program. First, we determined whether the rate of diagnosis of glaucoma according to race comports with the most recent estimates by Prevent Blindness America and the National Eye Institute.¹⁵ Second, we investigated whether longitudinal care patterns, once glaucoma is diagnosed, vary according to race. In this way, we can begin to determine better the extent of racial disparities in access to and use of care, as well as in intensity of care after individuals gain access to the health care system.

The initial study population consisted of a nationally representative sample of 21 644 Medicare beneficiaries who were studied previously to estimate prevalence of glaucoma, diabetic retinopathy (DR), and age-related macular degeneration (ARMD).¹⁶ The sample was nationally representative of Medicare beneficiaries chosen for purposes of conducting the National Long-Term Care Survey (NLTCS), a longitudinal study of elderly persons, during 1982 and 1984 and every 5 years through 1999. The data are available on a restricted use basis; we obtained Duke University’s Institutional Review Board and US Centers for Medicare and Medicaid Services’ approval for use of the data for this study.

To obtain the analysis sample of 21 644 beneficiaries for this study, we performed the following steps. In brief, beginning with a sample of 41 931 beneficiaries who appeared in the data for at least part of the period from 1982 through 1999, we excluded persons who died before 1991. This exclusion resulted in a loss of 11 023 individuals. We next eliminated 8238 individuals who were younger than 65 years in 1991. We then removed 573 individuals who died between January 1 and June 30, 1991, for whom we had fewer than 6 months of claims data. Of the remaining sample, we were unable to match 98 individuals to 1991 through 1999 Medicare enrollment records for unknown reasons; 338 individuals also had incomplete enrollment data (missing data for 1 or more years) between 1991 and 1999. Finally, we removed 17 individuals with duplicate and conflicting enrollment records, leaving us with a net eligible sample of 21 644 Medicare beneficiaries.

The only loss in sample after the beginning of 1991 was because of death or health maintenance organization (HMO) enrollment. The sample of 21 644 persons was further reduced in those years in which the person was enrolled in a Medicare HMO for more than 6 months in a particular year. For example, if a beneficiary was enrolled in an HMO for more than 6 months during 1993 but was in an HMO for less time than this in the other years, the person would have been excluded from the analysis sample only in 1993. This adjustment was necessary because Medicare did not receive information about utilization of services of beneficiaries in HMOs until after 1999. Information about whether a beneficiary belonged to an HMO each month was obtained from separate Medicare enrollment files. The number of persons excluded from the analysis because of HMO enrollment varied each year and ranged from 1319 in 1991 to 2193 in 1999. We followed up persons through 2000, but no persons diagnosed as having glaucoma were added to the original sample after 1999 to allow for at least 1 year of follow-up data.

To identify persons with glaucoma, we searched for glaucoma-specific diagnosis codes in the Medicare claims data (Table 1, Panel A). Persons entered the sample in the year after diagnosis, but not before 1991, the first year in which diagnosis codes were available in Medicare physician carrier data. We identified medical records of eye examinations and surgical procedures performed for glaucoma treatment by using Current Procedural Terminology and International Classification of Diseases, Ninth Revision, Clinical Modification(ICD-9-CM) procedure codes (Table 1, Panel B). We identified the individual’s race from Medicare vital statistics information linked to our data set. We excluded 685 persons (3.2%) who were not explicitly classified as either black or white.

Each person entered the database repeatedly every year after the year of diagnosis until 2000 or death, whichever occurred first. We calculated annual prevalence as the number of persons with glaucoma diagnosed before year end, divided by the total number of persons, excluding those alive for fewer than 6 months or those covered by Medicare HMOs for more than 6 months during the year.

Our sample yielded 353 blacks (8.9%) and 3598 whites (91.1%) with glaucoma diagnosed, for at least 1 year from 1992 through 2000, including those persons with codes for suspected glaucoma status (Table 2). On average, we observed black persons for 5.5 years and white persons for 5.7 years (P = .11). Annual rates of eye examinations for persons diagnosed with glaucoma were nearly identical for blacks (1.85 per year) and for whites (1.89 per year) (P = .51). However, blacks were more likely than whites to have no eye examination during the year (42.5%, blacks; 33.6%, whites [P<.001]). More blacks (22.1%) than whites (15.1%) with a prior diagnosis of some form of glaucoma had no eye examination during the entire observational period from 1991 through 2000 (P<.001). Of persons without an eye examination, 23.1% of blacks and 27.4% of whites were suspected of having glaucoma (P = .42, data not shown).

Rates of surgical procedures for glaucoma were almost twice as high for blacks as for whites (0.15 surgical procedures per person per year for blacks compared with 0.08 surgical procedures per person per year for whites P< .001). Furthermore, blacks with a diagnosis of glaucoma were more likely to have surgery during the entire observational period, as well as in any given year.

Most of the explanatory variables were based on data from Medicare claims and enrollment files. Information about some demographic characteristics, functional status, insurance, and income came from the 1989, 1994, and 1999 waves of the NLTCS, which was merged with the Medicare claims file. We used information from the NLTCS interview nearest to the observational year. For example, for 1991, we used 1989 data from NLTCS; for 1993, we used 1994 NLTCS data.

We used logistic regression to predict whether a person underwent eye examination or eye surgery during the year and negative binomial regression to predict the number of eye examinations and surgical procedures for glaucoma per year (STATA version 7.0; StataCorp LP, College Station, Tex).¹⁷ Covariates and their distribution according to race are shown in Table 3. P<.05 was considered statistically significant.

Black race was the main covariate of interest. Other covariates were as follows: diagnosis of glaucoma (suspected glaucoma—omitted reference group, primary open-angle glaucoma [POAG], narrow-angle glaucoma, other), ability to read newsprint with or without glasses, other ocular comorbidities—DR (background, proliferative, unspecified), cataract, ARMD (dry, wet, unspecified), death during the year, death during the next year, number of complications from diabetes (count of ICD-9-CM 250.xx codes), diagnosed dementia, DxCG score (measure of case mix severity based on all diagnoses/procedures recorded in claims in the year before the observational year),¹⁸ age, male sex, years of schooling, family income, number of limitations in activities of daily living, marital status, children living in the elderly person’s home, children living within an hour of the elderly person but not in the home, Medigap coverage, number of months during the year the person was enrolled in Medicaid, number of months during the year the person was enrolled in an HMO, number of ophthalmologists in the primary sampling unit per capita, standard metropolitan statistical area or county for persons not living in a standard metropolitan statistical area, number of optometrists per capita in the primary sampling unit, nonresidence in a standard metropolitan statistical area, living in nursing home, and binary variables for each year. We included interactions between glaucoma and diagnoses of other major eye diseases—cataract, ARMD, and diabetes mellitus. In some specifications, we also included interaction terms for black race and years since year of first diagnosis of glaucoma but excluded these covariates from the results presented because they did not alter the results. When information for a variable from NLTCS was not available, we set the variable to zero and included a binary variable to identify the value as missing. Standard errors were adjusted for repeated observations in the same individuals.

There were statistically significant differences between white and black persons in the prevalence of eye diseases (Table 3). A greater percentage of blacks had POAG or narrow-angle glaucoma, diabetes, DR, and complications from diabetes and reported having difficulty reading newsprint. Blacks were less likely to have other forms of glaucoma (suspected glaucoma, the omitted reference group, and other glaucoma), cataracts, and ARMD.

Compared with white persons in the sample, black persons were slightly older (<1 year on average) and were more likely to be female, live in urban areas, be less educated, have a lower income, be less likely to be married, be more likely to have children living within 1 hour distance, and have more comorbidities. Black persons were also more likely to have a diagnosis of dementia, reported more limitations in activities of daily living, were less likely to be in a nursing home, and were more likely to die during any given year.

Rates of diagnosed glaucoma among those who survived the year increased over time for both blacks and whites (Table 4). Rates for blacks increased from 11.2% in 1991 to 32.9% in 1999, whereas those for whites increased from 8.8% to 26.0%. Prevalence was significantly higher among blacks in all 3 years (P< .01 in 1991 and P< .001 in 1995 and 1999). Prevalence of POAG in blacks relative to whites was 1.6 in 1991 (7.5% vs 4.7%, P< .001). As the cohort aged, the ratios of black to white glaucoma prevalence increased to 1.44 in 1995 and 1.46 in 1999. Rates of narrow-angle glaucoma were higher among blacks than whites (P< .01 in 1995 and 1999), but there were no statistically significant differences in prevalence of other and suspected glaucoma.

Having other major eye diseases in combination with glaucoma may not only complicate therapy but also lead to poorer visual outcomes on average than having glaucoma alone. Joint prevalence of glaucoma and DR was the same for blacks as for whites in 1991, but the joint prevalence of glaucoma and DR increased substantially thereafter (Table 5). By 1999, 11.1% of blacks with glaucoma also had DR diagnosed, compared with 5.4% for whites. Joint prevalence of glaucoma and ARMD and joint prevalence of glaucoma, DR, and ARMD were generally lower for blacks than for whites, but the differences were not statistically significant at conventional levels.

In the second phase of the study, we examined use of eye care services conditional on a diagnosis of glaucoma. The purpose of limiting this analysis to persons with diagnosed glaucoma was to obtain a better sense of whether racial disparities existed in the provision of services after initial access barriers were overcome and blacks entered the health care system with a diagnosis of some form of glaucoma.

Blacks with glaucoma (including suspected glaucoma) diagnosed were 32% (P< .001) less likely to undergo eye examination during the year, not accounting for other factors likely to affect use (Table 6). Controlling for the other covariates reduced this difference in half to 16% (P = .09). Thus, it appears that more blacks than whites were likely to be sporadic in their care patterns, and the nearly identical mean annual utilization rates (Table 2) suggest that blacks who received care did so in a more intense fashion.

Blacks were 88% more likely to have eye surgery for glaucoma during a year than were whites (P< .001). When we controlled for other covariates, the difference remained high—64% (P< .001). Rates of eye examinations were not significantly different between blacks and whites, whereas rates of surgery were 83% higher for blacks than for whites diagnosed as having glaucoma without other covariates included in the analysis and 62% higher with other covariates included (P< .001).

The relationship between race and eye examinations and surgical procedures was similar when we restricted the analysis to persons with POAG diagnosed (Table 7). When we did not control for covariates, blacks with POAG were 36% less likely to undergo any eye examination during a given year but 75% more likely to undergo surgery (P< .001). Controlling for the other factors changed the differences to –17% (P = .20) and 78% (P< .001). As with glaucoma patients overall, there was no difference between blacks and whites in the mean number of annual examinations, but surgery rates were 71% to 76% higher for blacks than whites (P< .001).

Blacks were more likely than whites to have a diagnosis of glaucoma. However, our study estimates of prevalence according to race demonstrated a lower ratio than expected of diagnosed glaucoma in blacks than in whites among those 65 years and older.^1,5 The rate of suspected glaucoma was lower for blacks than whites at all intervals as the cohort of survivors grew older. If patients with a definite POAG diagnosis are combined with those suspected of having glaucoma, the ratio of rates for blacks to whites decreases further below the 1.8 ratio estimated by Prevent Blindness America; the race-specific estimates and ratios from the Baltimore Eye Survey, which exceed 4 in the Medicare age group; and ratios from other studies.^5,15,19- 21

The longitudinal ratio of glaucoma diagnosis rates among blacks compared with that among whites was less than that estimated in cross-sectional studies. Unlike the authors of studies in which cross-sectional data were used to demonstrate differences in rates of diagnosis relative to population-based estimates,^1,5,19 we longitudinally followed up a nationally representative cohort of persons 65 years and older to estimate these rates of glaucoma (particularly POAG) diagnosis. Our analysis revealed lower glaucoma diagnosis rates among blacks in general, and of POAG in particular, than did analyses in earlier studies.

The present study, moreover, also provides some insights into why this shortfall relative to the expected ratio might or might not be an important problem. The prevalence rates among blacks in this cohort, whose average age was in the middle to high 80s by 2000, increased to more than 30%, an expected level based on findings in the Barbados Eye Study population.²² In contrast, relative to findings in the Baltimore Eye Survey,¹⁹ rates were nearly 3 times higher among blacks, and they were almost 10 times higher among whites in our study relative to the Barbados Eye Study²²; the resulting ratio is less than might be expected from results of either previous study. Whether this finding represents overdiagnosis among whites or underdiagnosis among blacks, or both, cannot be answered from our data. In fact, it may be neither of these but a manifestation of the longitudinal nature of our study—as opposed to findings with cross-sectional estimates—and the larger sample sizes involved in this study relative to those in the population-based studies. (The Baltimore Eye Survey¹⁹ included only 11 black and 4 white patients with definite POAG cases among persons 80 years and older.) At a minimum, our estimates provide an alternative estimate of prevalence rates.

Finally, as noted in an earlier study,¹⁶ the influence of using physician-based diagnostic criteria as opposed to specific study criteria may affect our reported rates, so it is less certain whether there is a significant degree of underdiagnosis of glaucoma among blacks. The study results suggest that it would be worth examining actual rates of glaucoma diagnosis to determine if access is being garnered in the health care system so that diagnoses can be made.

Our study results also indicate that there are 2 distinct groups once blacks have received a diagnosis of glaucoma. First, there is a group that does not obtain access to regular care, as is evident by the larger share of black individuals without annual examinations. There is, however, also a second group for whom the rate of visits to the Medicare physician and service use (especially surgical) is at least as great as, if not greater than, that of whites. The rates of surgery are much higher among blacks on per person and subpopulation bases, confirming the direction of findings in earlier cross-sectional studies.

In our study, blacks were much more likely to undergo glaucoma surgery once they had glaucoma diagnosed. Compared with authors of previous cross-sectional studies, we were able to monitor surgical use per person across time. Thus, in contrast to conclusions of prior studies such as that of Wang et al,⁵ our conclusions do not indicate a lower rate than expected of glaucoma surgery among blacks with glaucoma; instead, we found a higher rate among blacks, so we cannot draw any significant conclusions about underservice.

Higher rates of glaucoma surgery reported for blacks in our study most likely reflect the greater clinical severity of disease among blacks. Although, to our knowledge, the recommendation of surgery according to race has not been studied within ophthalmology, research in other fields of medicine suggests that blacks are less likely than whites to be offered invasive or surgical therapies.^23- 24 To the extent that this result generalizes to ophthalmology, there is the implication that the higher rates of surgery for blacks in our study reflect greater clinical severity of glaucoma for blacks than for whites at diagnosis. Unfortunately, we did not have direct information about the clinical severity of glaucoma among black patients.

The results of this study demonstrate the value of longitudinal data for learning about the roles of race and other demographic factors in use of health services. Because of the relatively larger sample sizes of individuals in the older age cohorts compared with those in population-based studies, rate estimates may be robust in different ways in longitudinal than in population-based studies. Counterbalanced against this gain is the loss of precision of disease definition. However, prevalence and incidence estimates from this study population provide estimates similar to those in population-based studies for glaucoma,¹⁶ suggesting that this case identification issue may not be overly problematic.

In conclusion, the present study results demonstrate that, to the extent possible with administrative data, no clear pattern of significant undertreatment of glaucoma among blacks exists. Although the ratio of diagnosed cases in blacks compared with that in whites is lower than expected from the Baltimore Eye Survey,¹⁹ there is a somewhat surprising finding in that the rate of glaucoma among the white population is relatively much higher, which reduces the ratio. Additional investigation as to the clinical severity of disease and the resulting influences on patterns of care are the next step. We hope that other investigators also will pursue these studies.

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

Submitted for Publication: December 4, 2003; final revision received June 20, 2005; accepted June 20, 2005.

Financial Disclosure: None.

Funding/Support: This study was supported in part by grant RO1-AG-17473 from the National Institute on Aging, Bethesda, Md, and the Lew Wasserman Merit Award from Research to Prevent Blindness, New York, NY (Dr Lee).