Ellwein LB, Urato CJ. Use of Eye Care and Associated Charges Among the Medicare Population1991-1998. Arch Ophthalmol. 2002;120(6):804-811. doi:10.1001/archopht.120.6.804
Copyright 2002 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2002
To examine trends in the utilization and cost of eye care in the Medicare population.
Data were obtained from fee-for-service physician claims (Part B) from a 5% sample of Medicare beneficiaries 65 years and older. Use of eye care services and procedures, frequency of ocular diagnoses, and allowed charges were compared for each year from 1991 through 1998.
The proportion of beneficiaries receiving eye care increased from 41.4% to 48.1% during the 8-year period. Part B charges attributable to eye care decreased from 12.5% to 10.4%, with annual inflation-adjusted charges per beneficiary decreasing from $235 to $176 (1998 dollars). The proportion of beneficiaries with cataract-related claims increased from 23.4% to 27.3%, accounting for approximately 60% of eye care charges each year; beneficiaries with retinal disease claims increased from 7.8% to 11.4%, capturing 15.4% of eye care charges in 1998, up from 10.7% in 1991; and beneficiaries with glaucoma claims increased from 6.8% to 9.5%, accounting for nearly 10% of eye care charges each year.
The proportion of the Medicare population receiving eye care increased between 1991 and 1998. Nevertheless, eye care costs did not increase, primarily because of constraints in charges associated with the management of cataract.
THERE IS a continuing interest in determining the magnitude of ocular diseases and associated conditions within the US population, with estimates frequently based on prevalence data obtained in population-based surveys. While not a substitute for data from such surveys, the demand for ophthalmic services is an indicator of the extent to which ocular-related problems manifest themselves within a community. This demand-based perspective of disease is predicated, however, on the assumption that individuals with significant disease or conditions will eventually seek access to care. Administrative databases that record the provision of ophthalmic services within a geographic area are an efficient source for obtaining this information—with the recognition that asymptomatic conditions will be underrepresented, as will conditions for which screening and treatment costs are not reimbursed by insurance carriers.
Because of Medicare's nearly comprehensive coverage of the US population 65 years and older, Medicare claims data are a valuable source of information on the utilization and cost of eye care services. Using Medicare databases, we reviewed eye care services obtained through fee-for-service (FFS) providers, the allowed charges for these services, and the frequency of ocular diagnoses. Analyses were conducted on a cross-sectional, year-by-year basis for 1991 through 1998. Previous articles have addressed both specialized and comprehensive eye care utilization based on Medicare data1- 9 and, to a more limited extent, with data from the National Ambulatory Medical Care Survey.10- 12 To our knowledge, none have presented long-term year-by-year trends in eye care utilization or cost for the full spectrum of eye care services and procedures.
The study population for each calendar year was identified from Medicare Denominator Files for 1991 through 1998 obtained from the Health Care Financing Administration (HCFA). (This agency has recently been renamed the Centers for Medicare and Medicaid Services [CMS].) Beneficiaries included in the study were those at least 65 years of age before the end of the calendar year, and who had Medicare Medical Insurance for physician services (Part B) for at least part of the time after reaching age 65 years. Beneficiaries enrolled for the entire year in a Medicare health maintenance organization (HMO) or other Medicare managed care plan were excluded from the study population for that year. Full-time HMO enrollees were excluded because the use of services is not routinely reported to CMS, unlike the systematic claim-based reporting that occurs with FFS care. Railroad retirees, a subgroup of Medicare beneficiaries with a different benefit structure, were also excluded.
Eye care utilization data were obtained from a random 5% sample of FFS Medicare beneficiaries 65 years or older, as found in CMS's 5% Part B Physician/Supplier Files. These annual files contain claims submitted by physicians and limited-license practitioners for inpatient and outpatient services, as well as claims from free-standing ambulatory surgical centers (ASCs). (Part B claims for facility-related services are unique to ASCs. When such services are provided by hospitals for inpatients or for clinic outpatients, they are claimed under Part A of Medicare.) Claims for all services or procedures performed by an ophthalmologist or optometrist, and claims for ocular-related services or procedures performed by other providers, were used in identifying recipients of eye care. Each service or procedure in a Medicare claim (a claim line item) is coded by means of the HCFA Common Procedure Coding System (HCPCS), which is based on Physicians' Current Procedural Terminology codes.13 The HCPCS also includes CMS-specific codes for local (state-level) services and procedures. Eye care recipients were also identified through claims coded with an ocular-related diagnosis, even if the service or procedure was not necessarily ocular-related(eg, a claim for general evaluation and management services from a provider other than an ophthalmologist or optometrist). Claim diagnoses were coded by means of International Classification of Diseases, Ninth Revision, Clinical Modification, codes.14(The list of ocular-related HCPCS service or procedure codes and diagnosis codes that were used in identifying eye-related claims is available from the authors.)
A "visit" variable was created to quantify beneficiary eye care visits. A claim was taken to represent more than 1 visit if it covered services on multiple days (claim expense dates) by 1 or more providers. For example, if services were received from 1 provider on 2 different days, and if a second provider specialty also billed for services on 1 of these days, 3 visits took place. Each claim represented at least 1 visit, even if 2 claims were submitted(by 2 providers) for the same beneficiary for services on the same date.
The first, or principal, claim diagnosis was used in the classification of visits according to 16 previously defined categories and 97 subgroupings of eye diseases and disorders.15 Claims with a general, nonspecific diagnosis (eg, diabetes) that could encompass the provision of ocular-related services were categorized as "other ocular" (one of the 16 categories) when the service was provided by an ophthalmologist or optometrist, but as "nonocular" (another category) when some other specialty was involved.
For each calendar year, case incidence within diagnostic categories and subgroupings was determined. A beneficiary with 1 or more visits during the calendar year, all with principal diagnoses represented by a single diagnostic category (or subgrouping), was an incidence case for that particular category (subgrouping) in that year. Beneficiaries with visits corresponding to multiple diagnostic categories within a specific year represented an incidence case for each of the categories.
Costs of services and procedures were examined by means of allowed charges, the amount Medicare will reimburse for the service or procedure on the basis of a periodically updated fee schedule. (The claim file also includes the actual payment amount, which is an adjustment of the allowed charge taking deductibles and beneficiary coinsurance into account.) Allowed charges were reported on a per-beneficiary or per-recipient basis to accommodate the changing number of beneficiaries in each year.
Calculation of service or procedure utilization rates and allowed charges per beneficiary and eye care recipient was based on denominators adjusted to reflect the duration of active FFS Part B enrollment, thus representing person-year rates. Month-to-month changes in beneficiary status because of reaching 65 years of age, Part B enrollment, HMO enrollment, or death were taken into account. Allowed charges were adjusted to account for inflation by means of the medical expense component of the Consumer Price Index.
Because of its large size, the 5% Medicare sample is an unusually precise representation of what actually takes place within the entire Medicare population. To illustrate, with an estimated case incidence or service or provider utilization rate of 25 per 10 000 beneficiaries, the 95% confidence interval would be 24/10 000 to 26/10 000 (±4% of the point estimate) when calculated on a sample of 1 million beneficiaries. The confidence interval as a percentage of the point estimate would be even tighter for more elevated incidence or utilization rates. Considering that our estimates were derived from a sample approaching 1.5 million beneficiaries, confidence intervals around these estimates were ignored.
The number of Medicare Part B enrollees aged 65 years or older in the 5% sample for 1991 through 1998 is shown in Table 1. Although US demographics are reflected in the increasing number of enrollees each year, because of the accelerating popularity of HMOs the FFS beneficiary population actually declined during the latter part of the 8-year period—dropping from 92.3% of enrollees in 1991 to 81.1% in 1998. (Railroad retirees, who are not included in the study population, represented approximately 2% of enrollees.) The combined effect of HMO enrollment, which was particularly popular among newly eligible beneficiaries, and the aging of the US population produced a "graying" of the study population. (In 1991, 5.4% of all Part B enrollees were excluded from our sample because of HMO enrollment; by 1998, this percentage had increased to 16.7%. Among beneficiaries who were aged 65 years, 19.5% were excluded because of HMO enrollment in 1991 and nearly 50% were excluded in 1998.) The sex and race composition of the study population remained comparatively stable.
As shown in Table 2, the percentage of FFS beneficiaries receiving eye care increased steadily, from 41.4% in 1991 to 48.1% in 1998 (using adjusted numerators and denominators), representing an average increase of 2.4% per year. Despite a declining number of study beneficiaries beginning in 1994 (Table 1), the number of eye care recipients continued to increase until 1997. This increase was reflected in the average number of eye care visits per FFS beneficiary, which increased from 1.16 to 1.36 during the 8-year period. The number of eye care visits per recipient was reasonably constant, however, averaging between 2.67 and 2.83.
Average allowed charges for eye care services per beneficiary followed a generally decreasing trend during the 8-year period, decreasing from $235 to $176, a 25% decline. Average overall Part B charges per beneficiary decreased 10%, from an average of $1877 per beneficiary in 1991 to $1689 in 1998. Accordingly, eye care represented 12.5% of Part B charges in 1991, but only 10.4% by 1998. On a recipient basis, average eye care charges decreased from $567 to $366 during the 8-year period.
Table 3 shows case incidence, per 100 Medicare beneficiaries, within diagnostic categories for each calendar year. For broad diagnostic categories, specific subgroupings are also shown when the subgroup incidence was 0.4/100 or more in any 1 year. Case incidence increased for almost all of the diagnostic categories and subgroupings; exceptions were refractive conditions (refractive error, myopia, astigmatism), which decreased, and strabismus (amblyopia) and uveitis, which remained constant. The reduction in cases with a missing or nonocular diagnosis, from 7.09/100 to 4.38/100, contributed to the increase in diagnosis-specific incidence.(The diagnosis coding improved over time.) Most of the cases with missing or nonocular diagnoses, especially by 1998, were those treated by providers other than ophthalmologists or optometrists.
Cataract-related cases were particularly common, with an 18.0% increase in case incidence between 1991 (23.44/100) and 1997 (27.65/100) before dropping 1.3% in 1998 (to 27.29/100). Cataract-related cases represented approximately 55% of eye care recipients in each year. Glaucoma cases had the next highest incidence, 6.81/100 in 1991 and 9.51/100 in 1998, a 39.6% increase during the 8-year period. Particularly large percentage increases in case incidence were also seen with retinal diseases, 46.6% (including an 87.5% increase in diabetic retinopathy); neurologic disorders, 45.8%; orbital disorders, 44.4%; plastics, 37.6%; and external diseases, 33.0% (including a 57.4% increase in dry eye). Because Medicare reimburses for correction of refractive error in only very limited circumstances, the incidence of refraction cases was low and nowhere close to representing the true magnitude of refractive error within the Medicare population.
Table 4 shows the distribution of allowed charges for eye care across the 16 diagnosis categories for each of 3 representative years. Visits with a cataract-related principal diagnosis accounted for approximately 60% of all eye care charges, clearly dominating all other disease categories. Retinal diseases and glaucoma were in a distant second and third place, respectively. Retinal disease–related claims represented 10.7% of eye care allowed charges in 1991, increasing to 15.4% in 1998. Glaucoma accounted for slightly less than 10% across the entire study period. Changes in the relative distribution of allowed charges across time, as exemplified by the relative increase in charges for retinal diseases, paralleled changes in case incidence (Table 3). As noted earlier, the reduction in cases with missing or nonocular diagnoses contributed to the increases seen in diagnosis-specific charges.
Table 5 presents the use of eye care procedure and service codes during the 1991 to 1998 period; those with a frequency in any year of at least 25 per 10 000 beneficiaries are itemized. (Fitting of spectacles [92340-92371], which had a frequency between 32/10 000 and 41/10 000 during 1991 to 1995 before being discontinued from 1996 onward, is not shown separately but is included in"other procedures." Claims representing the purchase of frames, lenses, and coatings [HCPCS codes V2020-V2799] were not included in the analysis.) Substantial increases were seen during the 8-year period for procedures associated with cataract (extracapsular cataract extraction and intraocular lense implantation[ECCE/IOL] increased at an average annual rate of 6.9%; ophthalmic biometry at 6.5%; lens surgery anesthesia at 4.5%; and laser capsulotomy at 2.4%), retinal diseases (fundus photography increased at an average annual rate of 5.7%; fluorescein angiography at 5.2%; and laser photocoagulation at 4.2%), glaucoma (trabeculectomy increased at an average annual rate of 3.0% and visual field study at 2.7%), and dry eye (lacrimal punctum closure increased at an average annual rate of 33.2%). Substantial increases were also seen with eyelid epilation (9.4%) and external ocular photography (5.7%). Serial tonometry and anterior segment photography experienced particularly dramatic decreases(average annual decreases of 22.0% and 15.7%, respectively). With regard to services, ophthalmic examination and office evaluations (taken together) increased at an average annual rate of 1.8% for new patients and at 2.4% for established patients.
Changes in procedure or service volume did not always move in a uniform fashion. For example, ECCE/IOL increased in 1992, followed by a decrease in 1993, suggesting that cases that normally would have been operated on in 1993 were moved up to 1992—perhaps in an attempt to maintain practice income in the face of a decrease in the Medicare fee schedule for cataract surgery.(Unadjusted average allowed charges for the surgical procedure dropped from$1274 in 1991 to $1074 in 1992 and $1061 in 1993.) Reimbursement levels may also have had an influence on the coding of examination and evaluation visits. For example, the decrease in ophthalmologic examinations (Current Procedural Terminology codes 92012 and 92014) between 1991 and 1992 was nearly offset by a corresponding increase in office evaluations(Current Procedural Terminology codes 99211-99215). The subsequent leveling off of office evaluations beginning in 1993 was interrupted in 1997 and again in 1998 by increases corresponding roughly to the decreases in ophthalmic examinations during these 2 years. The back-and-forth switching between ophthalmic examination and more general office evaluation and management codes was consistent with changes in charges allowed for such services. For example, comprehensive office evaluations and management for new patients had an unadjusted average allowed charge of $80.55 in 1992, up from $60.05 in 1991, whereas the average allowed charge for a comprehensive ophthalmic examination increased less, from $50.23 to $56.81.
Table 6 shows the distribution of charges for procedures and evaluation and management services in each of 3 representative years; itemization was limited to procedures and services that exceeded 0.5% of total allowed charges in any 1 year. The ECCE/IOL category accounted for a major portion of all charges for eye care but experienced a relative decrease during the 8-year period. Ophthalmic examinations and evaluation and management services for new and established patients were also of major significance, increasing from 16.5% of allowed charges in 1991 to 25% in 1998. This increase relative to other procedures or services was largely the consequence of changes in the Medicare fee schedule: allowed charges for evaluation and management services increased during this 8-year period while those for surgical procedures decreased. The dramatic drop in charges for laser trabeculoplasty was the result of a reduction in the charge allowed for the procedure, from an unadjusted average of $787 in 1991 to $325 in 1998, coupled with a more than 40% decrease in procedure frequency during this period(Table 5). The reduction in the use of laser trabeculoplasty was also, no doubt, motivated by the introduction of new, more effective medications to reduce ocular pressure. Anterior segment photography also experienced a substantial reduction in the charge allowed for the procedure and in its utilization: allowed charges decreased from an average of $100 in 1991 to $68 in 1998, and procedure frequency decreased 70%.
Table 7 shows that more than 18% of eye care visits in each year were to providers other than ophthalmologists and optometrists, including 2% or more to ASCs. ("Cataract surgeon," which was established as a separate provider type by HCFA from October 1991 to May 1992, is included with the data for ophthalmologists.) Changes in the distribution of eye care visits by provider type were amplified in the distribution of allowed charges. Allowed charges for optometrists and ASCs doubled, reflected as an 11–percentage point decrease in ophthalmologist charges. If one were to ignore the services of ASC facilities in calculating the distribution of charges across provider types, ophthalmologist involvement would appear more favorable—a decrease from 86.1% in 1991 to 80.3% in 1998. The increasing popularity of ASCs for cataract surgery coupled with no significant decrease in the charges allowed by such facilities resulted in ASCs receiving an increasing percentage of eye care charges. The relative increase in charges by optometrists reflects the predominance of evaluation and management services within the profession—with increases in the fee schedule for such services during the 1991 to 1998 period—and increased involvement with ophthalmologists in the comanagement of cataract cases. Ophthalmologists were adversely affected by decreases in the fee schedule for ECCE/IOL and other ophthalmic procedures during the 8-year period.
Active specialties within the "other provider" type were internal medicine(cardiology, neurology, dermatology, and allergy), anesthesiology and certified nurse anesthetists, and general or family practice, along with multispecialty clinics and group practices. In 1998, these other providers accounted for 4.2%, 5.7%, 2.4%, and 1.4% of eye care visits, respectively. (To the extent that ophthalmologists participated in multispecialty group practices or billed under a group practice identifier, they, too, were represented by the other provider category.) Pathology and clinical laboratories, radiology, and surgery specialties (otolaryngology and plastic and reconstructive surgery) accounted for 1.0%, 0.5%, and 0.6% of eye care visits, respectively.
The use of eye care services is increasing within the FFS Medicare population, from 41.4% of beneficiaries in 1991 to 48.1% in 1998. Average eye care charges per beneficiary decreased by 25% during this period, in contrast to overall Part B charges, which decreased by 10%. Accordingly, eye care represented 12.5% of Part B charges in 1991 but only 10.4% by 1998. Much of the decrease in the cost of eye care was associated with a decrease in cataract-related charges, despite cataract cases increasing from 23.4% of the beneficiary population to 27.3%. Visits for retinal diseases, such as macular degeneration and diabetic retinopathy, increased rapidly: 11.4% of the beneficiary population had such visits in 1998, up from 7.8% in 1991; associated charges increased from 11% of all eye care charges to 15%. Glaucoma-related visits also increased substantially: 9.5% of beneficiaries had such visits in 1998 compared with 6.8% in 1991, while charges remained nearly constant at almost 10% of eye care charges in each year. Less than 4% of the beneficiary population was affected by any of the other disease categories.
Most eye care consisted of evaluation and management services and ophthalmic examinations, representing nearly 60% of all procedure or service code usage in each year. Procedure-based eye care was dominated by ophthalmic biometry, lens surgery anesthesia, ECCE/IOL, and laser capsulotomy for cataract; visual field studies for glaucoma; fundus photography and fluorescein angiography for retinal diseases; and extended ophthalmology. Ophthalmologists were associated with a decreasing percentage of Part B eye care charges during the 8-year period, while charges for visits to optometrists and ASC facilities increased. The relative decrease in charges by ophthalmologists was attributable to a general decline in the Medicare fee schedule for cataract and other surgical procedures.
The methodology used here has much in common with that of an earlier study of 1991 Medicare data.4 The way in which beneficiaries were identified and the way eye care was defined differed, however. Specifically, the previous study did not exclude HMO enrollees or railroad retirees from the beneficiary population, and the 65-year age criterion was based on beneficiary age at the beginning of the calendar year, rather than at the end. Of greater significance was the earlier use of a generally more inclusive list of diagnoses in identifying eye care claims. This had the effect of overstating visits to providers other than ophthalmologists and optometrists in the earlier study (for example, for visits associated with eyelid dermatitis and tumor diagnoses).
In general, annual case incidence as reported here is an underrepresentation of the actual prevalence of ocular disease in the community. Case incidence will correspond with prevalence, as typically obtained through examination of population-based samples, when the disease or condition of interest is in a symptomatic stage and treatment was sought during the year, or when the eye disease is asymptomatic but detected during a routine or other eye examination. A beneficiary with eye disease but without an eye care visit, whether symptomatic or not, will not be reflected in case incidence tabulations. Diseases or conditions with visits in previous years will also go unreported, unless there was also a visit during the current year. Minority populations and the poor elderly may be differentially affected by underreporting associated with reduced access to care. Case incidence may be particularly useful in providing a perspective on prevalence for chronic diseases or conditions where annual or more frequent visits to eye care providers are common, such as for glaucoma. Similarly, case incidence data should have merit in appraising the prevalence of acute conditions for which treatment is commonly sought, for example, conjunctivitis, corneal infections, and retinal detachments.
It should be recognized that use of a procedure code (Table 5) is not necessarily an accurate reflection of the frequency with which the procedure is actually performed. This applies, in particular, to surgical procedures performed in ASC facilities, such as trabeculectomy, laser capsulotomy, and ECCE/IOL. (Approximately 25% of the code usage for these 3 procedures in 1998 was for ASC facility services, in addition to an identically coded professional service claim for the actual surgical procedure.) Code usage also overstates procedure frequency when a second physician provider uses the same code for support services, such as for providing assistance at surgery,4 or when the code is used for the comanagement of patients. For example, in 1998, 2.6% of the laser capsulotomy coding and 7.8% of that for ECCE/IOL was by optometrists, reflecting participation in patient management but not the actual surgical procedure. Code usage can also understate procedure frequency, eg, when during a single visit the procedure is performed on both eyes. (A modifier field in the claim is used to activate the increased allowed charge associated with bilaterality.)
Because the objective of this study was to present descriptive analyses and trends of eye care utilization and cost on an annual basis, along with the incidence of ocular diagnoses, cross-sectional data were not adjusted for year-to-year changes in age, sex, race, or geographic composition. Any in-depth investigation of factors underlying the observed trends would require these demographic adjustments, as well as others dealing with reimbursement policies and financial incentives, provider service settings,7 and workforce supply.15 Indeed, considering the increasing popularity of HMO enrollment among Medicare beneficiaries, it is likely that some of the observed increase in eye care was because of selective enrollment of healthier beneficiaries in Medicare HMOs and possible disenrollment by sicker patients.16 Such enrollment biases were shown to have affected cataract extraction rates.7
If FFS Medicare beneficiaries were representative of the US population, approximately 16 million of 35 million Americans 65 years of age or older(34 991 753 in the 2000 census) are receiving eye care on an annual basis, at a cost of more than $6 billion for physician and ambulatory surgery services, excluding correction of refractive error. The majority of this eye care is cataract related, accounting for 60% of eye care costs. This current projected volume of eye care represents a substantial increase over the estimated 12.5 million persons who received such services in 1991 (of a population of 31 779 000). Considering that the unadjusted cost for these services in 1991 was on the order of $5.5 billion, the increase during the past decade was at less than inflationary levels. Eye care utilization is increasing, but without any corresponding increase in cost.
Submitted for publication September 5, 2001; final revision received January 23, 2002; accepted January 31, 2002. Ms Urato was supported by contract 290-95-2002 from the Agency for Healthcare Research and Quality, Rockville, Md.
We thank Rita S. Hiller, MS, and Robert D. Sperduto, MD, of the National Eye Institute, Bethesda, Md, and Nancy T. McCall, ScD, and Joyce H. Huber, PhD, of Health Economics Research Inc, Waltham, Mass, for their assistance and comments on data preparation and analysis.
Corresponding author and reprints: Leon B. Ellwein, PhD, National Eye Institute, 31 Center Dr, Bethesda, MD 20892-2510 (e-mail: email@example.com).