To describe the usage patterns of pharmacological treatments for neovascular age-related macular degeneration (AMD) in Medicare fee-for-service beneficiaries.
Retrospective review of all Medicare fee-for-service Part B claims for neovascular AMD during 2008.
Medicare beneficiaries having undergone treatment were identified. The data collected for each visit for a given beneficiary included age, race, gender, Medicare region, state/zip code of residence, date of visit, whether or not the beneficiary had a treatment, the type and amount of drug, and dollars paid by Medicare. The main outcome measures were the number and rate of treatments, the types of drugs used for treatment, and the payments for these drugs.
Of the 222 886 unique beneficiaries, 146 276 (64.4%) received bevacizumab and 80 929 (35.6%) received ranibizumab. A total of 824 525 injections were performed with 480 025 injections of bevacizumab (58%) and 336 898 injections of ranibizumab (41%). National rates of injections per 100 000 fee-for-service Part B Medicare beneficiaries for bevacizumab and ranibizumab were 1506 and 1057, respectively. Total payments by Medicare were $20 290 952 for bevacizumab and $536 642 693 for ranibizumab. In 39 out of 50 states, the rate of injection was higher for bevacizumab compared with ranibizumab.
In 2008, bevacizumab was used at a higher rate than ranibizumab for the treatment of neovascular AMD. Even though bevacizumab accounted for 58% of all injections, Medicare paid $516 million more for ranibizumab than bevacizumab. These data suggest that despite its off-label designation, intravitreal bevacizumab is currently the standard-of-care treatment for neovascular AMD in the United States.
Age-related macular degeneration (AMD) is the leading cause of irreversible blindness among the elderly in the United States (US) and other industrialized countries. In the past, most of the severe vision loss in AMD has been associated with the neovascular form of the disease, but with the advent of anti-angiogenic therapies, in particular drugs that inhibit vascular endothelial growth factor (VEGF), the prognosis for these patients has significantly improved. Current pharmacologic treatments for neovascular AMD include verteporfin photodynamic therapy (Visudyne; QLT Pharmaceuticals, Vancouver, BC, Canada), intravitreal pegaptanib sodium (Macugen; Eyetech Pharmaceuticals, Palm Beach Gardens, Florida, USA), intravitreal bevacizumab (Avastin; Genentech/Roche, South San Francisco, California, USA), and intravitreal ranibizumab (Lucentis; Genentech/Roche).
In 2000, verteporfin photodynamic therapy was approved by the Food and Drug Administration (FDA) and covered by the Centers of Medicare and Medicaid Services (CMS) as a treatment for neovascular AMD. In late 2004, pegaptanib sodium was FDA approved as the first anti-VEGF drug for the treatment of neovascular AMD, and by early 2005, CMS was covering claims for pegaptanib. By mid-2005, intravitreal bevacizumab was introduced as an off-label anti-VEGF treatment for neovascular AMD, but CMS coverage lagged throughout the country as clinical evidence accumulated suggesting that bevacizumab was safe and effective. Ranibizumab was approved in mid-2006 and CMS coverage followed closely thereafter. Over the past 3 years, both bevacizumab and ranibizumab have emerged as the most common therapies for neovascular AMD as reported by the Patterns and Trends (PAT) surveys conducted by the American Society of Retinal Specialists (P. Rosenfeld; personal communication, February 2010). While the comparative efficacy of bevacizumab and ranibizumab continues to be debated and large comparative clinical trials are underway in the US and Europe, it is apparent that the off-label use of bevacizumab is driven by its low cost and molecular similarity to ranibizumab, while the use of ranibizumab is driven by extensive level 1 evidence documenting its efficacy and safety in well-controlled, prospective clinical trials. Both drugs have revolutionized the care of patients with neovascular AMD, and both drugs are locally covered by CMS for the treatment of neovascular AMD.
CMS databases have been useful in the past in assessing the incidence and prevalence of AMD in the US population, the association between AMD and other diseases, and the costs associated with the care of AMD patients. All these previous analyses using CMS data examined the standard 5% random sample of Medicare beneficiaries. The only previous analysis examining the cost of care associated with AMD patients was performed by Coleman and associates. In this study, Medicare costs over 5 years were analyzed from a 5% random sample of Medicare beneficiaries with both dry and neovascular AMD from 1995 to 1999. At the time of this previous report, the only treatment for neovascular AMD was laser photocoagulation. The authors determined that median eye-related Medicare costs were estimated at $1607 for neovascular AMD patients, $832 for dry AMD patients, and $658 for controls without the diagnosis of AMD. However, since 1999, 4 different drugs have been introduced for the treatment of neovascular AMD, resulting in escalating costs for Medicare.
With the availability of the complete claims file from the 2008 Medicare Part B fee-for-service database, it is now possible to obtain information on the use of particular treatments for particular diagnoses as long as these diagnoses and treatments are coded properly when a claim is filed. This report describes the Medicare payments associated with the use of different therapies for neovascular AMD throughout the country in 2008. To our knowledge, this is the first time that the entire 100% fee-for-service Part B claims file is being used for any eye-related analysis instead of the standard 5% analytic file.
The 100% fee-for-service Part B Medicare claims file for calendar year 2008, updated as of July 1, 2009, was used for this report. Since data from the entire fee-for-service Medicare population was used for this report and sampling was not performed, standard deviations and confidence intervals are not appropriate in this analysis.
This fee-for-service claims file includes beneficiaries who have their claims reported directly to CMS for 80% coverage of costs and does not include beneficiaries enrolled in Medicare managed health-care plans. Only those claims with an International Classification of Diseases, Clinical Modification, Version 9 (ICD-9-CM) line diagnosis of neovascular AMD (362.52) were used for all subsequent analyses. All persons having had a drug or infusion treatment for neovascular AMD were identified if the Healthcare Common Procedure Coding System ( HCPCS) code was “J3590” or “J9035” or “J3490” for bevacizumab; “J2778” for ranibizumab; “J3396” for verteporfin; or “J2503” for pegaptanib. Though “J3490” is the code for an unclassified biologic, we determined from the regional carrier medical directors who performed a review of this code for 2008 that almost 100% of coding with this HCPCS code was for a bevacizumab injection in patients with neovascular AMD. The data were analyzed as having been submitted by any eye care provider, regardless of subspecialty, when treating Medicare beneficiaries for the diagnosis of neovascular AMD. It was not possible to specifically associate claims data with retinal specialists since retinal specialists are not identified either in the claims data or in any known database that can be mapped to the claims data by provider number.
The data that were classified for each visit for a given beneficiary included age at first visit for neovascular AMD in calendar year 2008, race, gender, Medicare region, and state as well as zip code of residence. The 10 Medicare regions are Boston (ME, MA, NH, RI, VT), New York (NJ, NY, PR, VI), Philadelphia (DE, DC, MD, PA, VA, WV), Atlanta (AL, FL, GA, KY, MS, NC, SC, TN), Chicago (IL, IN, MI, MN, OH, WI), Dallas (AR, LA, NM, OK, TX), Kansas City (IA, KS, MO, NE), Denver (CO, MT, ND, SD, UT, WY), San Francisco (AZ, CA, HI, NV, AS, CNMI, GU), and Seattle (AK, ID, OR, WA).
Medicare codes that could change with each visit and were recorded for each visit included the date of visit, whether or not the beneficiary had an injection or infusion at that visit for neovascular AMD, the type of drug or infusion and relevant units, modifiers for the eye, dollars paid by Medicare for the specific HCPCS code, and visit number for that patient for neovascular AMD in calendar year 2008.
Data on the number of Medicare beneficiaries by state were obtained from CMS tables for 2008. These tables reported total number of fee-for-service Part A and Part B beneficiaries.
All the above data were used to obtain the appropriate numerators and denominators for calculations presented in this report, most of which involved the fee-for-service Part B Medicare population. The data were treated with appropriate integrity, security, and confidentiality, as detailed in the Data Use Agreement required by CMS.
SAS 9.1 (SAS Institute Inc, Cary, North Carolina, USA), Microsoft Office Excel 2007, and Microsoft MapPoint 2010 (Microsoft, Redmond, Washington, USA) were used to calculate the results in this report.
A total of 222 886 unique Medicare beneficiaries were identified who had the diagnosis of neovascular AMD and who received 1 or more injections or infusions for this diagnosis among 31 879 444 Part B Medicare beneficiaries in 2008 ( Figure 1 ). Of these beneficiaries, 9041 had injections in both right and left eyes, though not necessarily at the same time during 2008. Nine hundred ninety-seven unique beneficiaries had bevacizumab injected in one eye and ranibizumab injected in the other eye at some point during the year for neovascular AMD, and 3686 had ranibizumab and bevacizumab injected in the same eye. These statistics regarding injections in both eyes are lower than expected since each beneficiary was exposed in time an average of 6 months of the calendar year 2008.
The mean age of the beneficiaries identified with neovascular AMD was 81.4 years, 64.2% of whom were female. Individuals with the race classification of white comprised over 96% of these beneficiaries diagnosed with neovascular AMD who received at least 1 treatment for that diagnosis that involved an injection or infusion in 2008 ( Table 1 ).
|Mean age ± SD||81.40 ± 7.30||—|
Of the 222 886 unique Medicare beneficiaries with the diagnosis of neovascular AMD receiving treatment, there were a total of 824 525 injections and 17 257 verteporfin infusions performed. For beneficiaries receiving at least 1 treatment for neovascular AMD, there were 480 025 total injections of bevacizumab in 146 276 different beneficiaries, 336 898 total injections of ranibizumab in 80 929 different beneficiaries, 17 257 infusions of verteporfin in 13 812 different beneficiaries, and 7602 injections of pegaptanib in 2521 different beneficiaries ( Figure 2 ). Among the individuals who received injections, 146 276 (64.4%) received bevacizumab and 80 929 (35.6%) received ranibizumab. This total number of unique beneficiaries receiving a treatment is greater than the number of 222 886 unique beneficiaries receiving any treatment because 14 652 beneficiaries received more than 1 type of injection or infusion for neovascular AMD during the year.
National rates of injections/infusions for neovascular AMD per 100 000 fee-for-service Part B Medicare beneficiaries were 1506 for bevacizumab, 1057 for ranibizumab, 54 for verteporfin, and 24 for pegaptanib. These findings also indicate that the average number of injections of bevacizumab given to a Medicare beneficiary treated with this drug during calendar year 2008 was 3.3 while the comparable number for ranibizumab was 4.2. Since our data are calendar year specific, these persons, on average, were exposed to treatment for less than 12 months since treatment could be initiated anytime during the calendar year. Therefore, the actual average number of injections per year is lower than would be expected for each beneficiary with the diagnosis of neovascular AMD for an entire year. Yet, the two drugs can still be compared using this metric given there was no difference in the availability of either drug during 2008.
Total expenditures for all neovascular AMD treatments during 2008 are shown in Figure 3 . The difference in expenditures between ranibizumab and bevacizumab is attributable to the difference in cost of ranibizumab per injection (average paid by Medicare is $1593 per injection) as compared with the cost of bevacizumab per injection (average paid by Medicare is $42 per injection), and the number of injections given. The expenditures for verteporfin and pegaptanib were much lower by comparison because of the infrequent use of these therapies for the treatment of neovascular AMD according to the CMS data. The 100% Part B fee-for-service database contained 100% of all the intravitreal injections performed in 2009. However, this database did not include all the names of the drugs used in hospital and other outpatient settings. While 98% of all drug names were identified, approximately 2% of drugs used for intravitreal injections in the US that were purchased by outpatient facilities other than private practice offices (eg, hospital pharmacies) could not be identified.
The utilization of ranibizumab and bevacizumab was evaluated based upon the 10 different Medicare regions throughout the US. Each region includes several states, and is identified by a major city in the region. There were only two regions, identified by the cities of Boston and Kansas City, where the rate of injection for ranibizumab exceeded that for bevacizumab per 100 000 fee-for-service Part B Medicare beneficiaries. The remainder of the regions injected bevacizumab as their preferred treatment. Because of the low rate of utilization, both verteporfin and pegaptanib therapies were excluded from the plot. Specific data for regional differences are shown in Table 2 . The highest regional rates of bevacizumab injection for neovascular AMD among Medicare fee-for-service Part B beneficiaries were in the Seattle and Denver regions, at 2062 and 1996 injections per 100 000 beneficiaries, respectively. The highest regional rates of ranibizumab injection were in the Kansas and New York regions, at 1442 and 1372 injections per 100 000 beneficiaries, respectively.
|Total||Per 100 000||Total||Per 100 000||Total||Per 100 000||Total||Per 100 000|
|USA||480 025||1506||336 898||1057||17 257||54||7602||24|
|01-Boston||20 540||1173||23 749||1356||442||25||164||9|
|02-New York||42 527||1394||41 850||1372||1565||51||200||7|
|03-Philadelphia||49 194||1547||41 212||1296||1659||52||776||24|
|04-Atlanta||102 023||1421||84 444||1176||2652||37||1402||20|
|05-Chicago||94 911||1654||53 845||938||3815||66||1349||24|
|06-Dallas||55 909||1501||26 431||710||2115||57||547||15|
|07-Kansas||22 065||1278||24 892||1442||1726||100||2242||130|
|09-San Francisco||50 711||1468||25 129||728||1655||48||692||20|
For the entire US, the rate for any injection/infusion was 2641 per 100 000 fee-for-service Part B Medicare beneficiaries for neovascular AMD. However, notable variations in treatment patterns were observed when the CMS data were reported by state ( Table 3 ; Figure 4 ). Connecticut and Florida had the highest rates of any type of injection/infusion per 100 000 fee-for-service Part B Medicare beneficiaries, at 4138 and 4010 respectively. The comparable lowest rates were in Vermont and DC, at 763 and 1127 respectively per 100 000 fee-for-service Part B Medicare beneficiaries. Among states, the highest rates of bevacizumab injection per 100 000 fee-for-service Part B Medicare beneficiaries were 2632 in Minnesota and 2622 in Oregon, while the lowest rates were in Vermont and Hawaii, at 349 and 671 respectively. For ranibizumab in the US, the highest rates of injection per 100 000 fee-for-service Part B Medicare beneficiaries were 2689 in Connecticut and 2281 in Iowa, compared with the lowest rates in Alaska and DC of 56 and 252 respectively.
|Total||Per 100 000||Total||Per 100 000||Total||Per 100 000||Total||Per 100 000|
|USA||480 025||1506||336 898||1057||17 257||54||7602||24|
|CA||36 205||1372||20 915||792||1368||52||375||14|
|FL||37 803||1735||47 706||2190||1282||59||562||26|
|IL||24 994||1685||16 592||1118||632||43||128||9|
|MI||14 738||1279||12 268||1065||852||74||272||24|
|NJ||15 602||1494||21 097||2020||626||60||95||9|
|NY||26 660||1430||20 685||1109||936||50||105||6|
|NC||16 949||1520||11 200||1005||309||28||369||33|
|OH||19 619||1553||11 353||899||766||61||332||26|
|PA||18 511||1487||18 713||1504||990||80||268||22|
|VA||13 123||1514||11 044||1274||426||49||205||24|