To evaluate patterns of disease progression among individuals with age-related macular degeneration (AMD) and to compare costs over time.
Retrospective data analysis using 5% Medicare claims data from 1997 through 2009.
Beneficiaries were included if they had no diagnosis of AMD in 1997, were 65 years of age or older, had data through 2009, and had no major ophthalmic conditions. Two cohorts were identified: those who had dry AMD in 1998 (cases) and matched controls who never had AMD.
There were 52 607 beneficiaries who never had AMD and 1184 who were diagnosed with dry AMD in 1998. Among beneficiaries with dry AMD, the disease progressed in 20.4% to the wet form by 2009. From 1999 to 2009, average annual Medicare expenditures increased from $11 265 to $24 494 (cases whose disease did not progress) and from $11 712 to $34 308 (cases whose disease progressed). Among beneficiaries without AMD, expenditures also increased over time (from $4736 in 1999 to $17 473 in 2009), but consistently were lower than cases’ expenditures. Considering ophthalmic expenditures, the pattern was more pronounced: beneficiaries without AMD had annual expenditures less than $100, those with dry AMD had expenditures at least 3 times more, and wet AMD beneficiaries’ costs were at least 5-fold more than that of those with dry disease. A subgroup analysis of beneficiaries without hypertension revealed similar patterns, although expenditures were lower than in the general population.
AMD progression seems to be associated with increased annual Medicare expenditures. Findings suggest that halting or slowing disease progression using proven treatment such as Age-Related Eye Disease Study-endorsed vitamins or novel technologies could have a substantial positive impact by lowering public health expenditures.
Age-related macular degeneration (AMD) is an ophthalmic condition characterized by blurring, distortion, or loss of central vision. A recent report from the Age-Related Eye Disease Study estimated that approximately 8 million persons in the United States at least 55 years of age have monocular or binocular intermediate AMD or monocular advanced AMD. It is the leading cause of legal blindness in persons 65 years of age and older in the United States. In addition to increased age, other risk factors associated with AMD include smoking, cardiovascular disease, diet, light iris pigmentation, and possibly extensive ultraviolet light overexposure.
There are 2 forms of AMD: atrophic or dry (early or late) and exudative (wet). Dry AMD, the more common form of the disease, initially may be characterized by drusen and is generally slow to progress, although among individuals with AMD who are 43 to 86 years of age, the 15-year cumulative incidence of geographic atrophy (GA) is 14%. GA, an advanced form of dry AMD, is defined by the presence of a sharply demarcated area of apparent absence of the retinal pigment epithelium and photoreceptor cells, with visible choroidal vessels and no neovascular AMD. Although there is not a unique diagnosis code for GA, its clinical recognition suggests that dry AMD is not a monolithic entity. Wet AMD accounts for approximately 10% of cases; however, wet AMD accounts for the vast majority of cases of AMD-related severe visual loss. Wet AMD is characterized by choroidal neovascularization, the development of new blood vessels arising from the deeper layers of the eye. Before the appearance of either choroidal neovascularization or GA, patients typically exhibit clinical changes characteristic of at-risk AMD (ie, intermediate or large soft or confluent drusen with hyperpigmentation and no choroidal neovascularization or GA).
Both wet AMD and GA often initially develop in 1 eye, but AMD is ultimately a bilateral disease. Effective treatments for wet AMD include antibody fragments directed against vascular endothelial growth factor. Vascular endothelial growth factor inhibitors were approved by the Food and Drug Administration in 2006 for treatment of macular degeneration; monthly costs can exceed $2000. Progression of dry AMD may be slowed by nutritional supplementation. The costs associated with AMD and its treatment are substantial, and given the aging of the population, cases and costs are likely to increase. Thus, understanding the costs and cost drivers of the condition as it progresses could play an important role in public health decision making.
Data were analyzed from the 1997 through 2009 Medicare Limited Data Set (LDS) files. The LDS represents a random 5% sample of all Medicare enrollees and is representative of United States citizens age 65 years of age and older. The random sample used for this claims data set is selected based on the same algorithm each year, meaning that the same patients are included in the LDS data each year (unless they die); therefore, longitudinal treatment patterns can be evaluated. The LDS data consist of 7 claims components: inpatient; outpatient; durable medical equipment; hospice; home health agency; skilled nursing facility (nursing home); and physician/supplier (Part B) claims. The data do not contain personally identifiable information, and institutional review board is not required. Medicare data are useful for evaluating clinical events (eg, diagnoses, rates of surgeries) and expenditures (both billed charges and reimbursed amounts) for conditions common in older patients. The sample size, adjudicated nature of the claims, and longitudinal follow-up are important advantages to using Medicare data; these data arguably are the most complete source of objective health information on older populations in the United States. Institutional review board approval is not necessary for this type of analysis. The data were analyzed and reported in compliance with a data use agreement developed by the Centers for Medicare and Medicaid Services.
Beneficiaries were eligible for this study if they were covered and enrolled through 2009, were receiving benefits based on their age (ie, not because of disability or end-stage renal disease), had no diagnoses for AMD (International Classification of Diseases—Clinical Modification [ICD-9-CM] codes 362.51 [dry AMD], 362.52 [wet AMD], and 362.57 [drusen]) in 1997, had no more than 1 claim associated with a diagnosis code for diabetic retinopathy (ICD-9-CM codes 362.01 through 362.06), or glaucoma (ICD-9-CM code 366) during the study period and had no evidence of cataract surgery (Current Procedural Terminology codes 66982 through 66984).
The remaining beneficiaries were stratified into 2 cohorts. The first cohort included beneficiaries who were not diagnosed with drusen or AMD during the study period (potential controls), and the second cohort included those diagnosed with drusen or dry AMD in 1998 (cases). Cases could have progressed to wet AMD in 1999 or later. Requiring a year of so-called clean claims, that is, without AMD, was an effort to identify incident cases. Among the possible controls, an age-matching approach was used to compensate for the higher age among the case group. For each AMD case, up to 50 patients of the same age were selected randomly from the group of potential controls to be his or her comparison. For certain older ages, there were not enough AMD-free patients of that age to provide a 50-to-1 sample. In those cases, all available AMD-free patients of that age were selected. Simple random sampling without replacement was used so that the same AMD-free patient would not appear more than once as matching comparisons for an AMD patient.
Demographic and clinical characteristics at baseline were described. The presence of chronic nonophthalmic disease was defined as the presence of 1 or more health care encounters with selected diagnosis codes in 1997. Progression was identified by the presence of the ICD-9-CM code for wet AMD.
Total expenditures and ophthalmic expenditures (defined as those with ICD-9-CM diagnosis codes 360 through 379) were assessed. For these analyses, 3 groups were compared: (1) the no AMD control group, (2) the subset of cases who did not progress to wet AMD, and (3) the subset of cases that progressed to wet AMD. The second and third groups were defined based on their diagnosis in a given year, and the sample size decreased and increased over time, respectively. For example, in 1998, there were 1184 cases, but starting in 1999, when 31 patients converted to wet AMD, the dry AMD group decreased to 1153 and the wet AMD group included 31 patients. The number of beneficiaries in the dry AMD only group decreased over time (to 943), whereas the number in the wet group increased (to 241). Accordingly, expenditures could not be calculated for the wet group for 1998 because by definition no one progressed to wet AMD until 1999.
For these analyses, costs were inflated to 2011 United States dollars using the medical care index published by the Bureau of Labor Statistics. All data analysis was performed using SAS software version 9.2 (SAS Institute, Cary, North Carolina, USA).
Demographic and Clinical Characteristics
There were a total of 3 787 786 unique beneficiaries in the database in 1997. After all exclusions were applied (eg, presence in the data set through 2009, age older than 65 years, lack of particular ophthalmic diagnoses), there were 146 887 eligible beneficiaries. Of these, 1184 were diagnosed with dry AMD in 1998 but did not progress to wet AMD until 1999 or later. For the remaining 145 703, the age-matching approach describe above was applied to determine which would be designated as controls. After matching to cases at a 50-to-1 ratio, 52 607 of the available 145 703 beneficiaries remained in the no AMD control group.
In all groups, most beneficiaries were female. Mean age among cases was 74.9 ± 6.0 years; mean age in the no AMD control group was 73.8 ± 5.3 years. In general, beneficiaries in the no AMD control group were healthier than their peers with dry AMD in terms of diagnoses of hypertension, diabetes, and hypercholesterolemia. Table 1 presents detailed demographic and clinical characteristics.
|Sex (% female)||61.5||72.8|
|65 to 69||24.4||21.7|
|70 to 74||31.0||27.5|
|75 to 79||29.8||27.3|
|80 to 84||12.1||17.1|
The Figure shows cumulative patterns of progression of cases. For example, in 1999, 2.6% (31 of 1184) of the cases had progressed to wet AMD. By the end of the observation period, 20.4% (241 of 1184) had progressed to wet AMD.
As shown in Table 2 , controls with no AMD throughout the study period had lower average annual Medicare expenditures, ranging from $4769 in 1998 to $17 473 in 2009. As expected, beneficiaries with AMD had higher average annual expenditures. For beneficiaries with dry AMD, expenditures ranged from $10 555 in 1998 to $24 494 in 2009, whereas for those with wet AMD, they ranged from $11 713 in 1999 to $34 308 in 2009.
|Total||None||$4769||$4736||$4956||$5867||$7152||$7954||$9255||$9758||$10 502||$12 181||$14 467||$17 473|
|Dry||$10 555||$11 265||$10 574||$13 051||$15 538||$18 036||$18 787||$20 172||$19 358||$22 348||$21 555||$24 494|
|Wet||NA b||$11 713||$14 644||$16 165||$21 895||$22 948||$22 032||$24 275||$25 412||$26 908||$29 437||$34 308|
A similar pattern was evident for ophthalmic expenditures. The highest average annual expenditures were found again for beneficiaries with wet AMD, approaching $3000 per patient in some years; the ophthalmic expenditures for those with dry AMD were less than $500 with 1 exception. The average annual ophthalmic expenditures for controls consistently were less than $100.
Given the differences in rates of hypertension across cohorts, we undertook 1 additional analysis. We redefined the population and required that they did not have a diagnosis of essential hypertension (ICD-9-CM code 401.0) during the baseline observation year (ie, 1997), although we did not require that they be free of the diagnosis for the entire observation period. The 2 cohorts then included 628 with dry AMD, 128 of whom progressed to wet AMD, and 30 189 age-matched controls without AMD. The population remained mostly female, but less so than in the overall population (59.5% in the control group and 68.6% in the case group) and was almost the same age as the primary study population, including the beneficiaries with hypertension. The rate of progression was identical: 20.4% of the dry AMD patients progressed to wet AMD by 2009. Beneficiaries without hypertension tended to have lower average annual expenditures, also. Table 3 shows annual overall and ophthalmic expenditures for both cohorts. The patterns of expenditures, with those with no AMD having the lowest costs, and those with dry AMD and wet AMD having increasingly greater costs, respectively, also was evident in terms of ophthalmic expenditures. However, because the number of beneficiaries progressing to wet AMD was small, there may be some year-to-year variation because of small cell sizes.
|Dry||$5793||$5848||$4766||$6581||$7590||$8851||$7825||$9035||$8501||$10 002||$8790||$10 108|
|Wet||NA b||$6163||$12 416||$13 035||$17 624||$24 418||$17 961||$17 311||$28 007||$26 081||$30 152||$30 794|