Abstract
Purpose:
To assess whether publication of Comparison of Age-related macular degeneration Treatment Trial (CATT) results and introduction of aflibercept to the marketplace affected intravitreal bevacizumab and ranibizumab utilization.
Design:
Retrospective analysis of treatment patterns.
Methods:
We calculated weekly bevacizumab and ranibizumab utilization during 3 timeframes: (1) before CATT publication, (2) between CATT publication (4/28/2011) and assignment of a unique aflibercept billing code (1/1/2013), and (3) afterwards for 164188 Medicare beneficiaries with neovascular macular degeneration receiving ≥1 anti-Vascular Endothelial Growth Factor injections from 1/1/2008 to 12/31/2014. We identified ophthalmologists who predominantly (≥80%) administered bevacizumab or ranibizumab, and evaluated changes in preferences over the 3 periods. We replicated analyses on 881381 commercially-insured beneficiaries.
Results:
Among 317 ophthalmologists administering predominantly ranibizumab to Medicare beneficiaries pre-CATT, 221 (69.7%) reduced ranibizumab utilization post-CATT, whereas 96 (30.3%) continued using ranibizumab ≥80% of the time. Findings were reversed among 1041 ophthalmologists who predominantly administered bevacizumab pre-CATT—777 (74.6%) continued bevacizu mab-predominant use while 264 (25.4%) reduced bevacizumab utilization post-CATT. Among the 145 ophthalmologists who predominantly administered ranibizumab before aflibercept’s availability, 77 (53.1%) reduced ranibizumab utilization and 68 (46.9%) continued using ranibizumab ≥80% of the time after aflibercept became available. Corresponding numbers among the 909 ophthalmologists who predominantly administered bevacizumab pre-aflibercept were 381 (41.9%) reducing and 528 (58.1%) continuing bevacizumab-predominant use. Similar results were observed for commercially-insured patients.
Conclusions:
Many ophthalmologists who favored ranibizumab switched to bevacizumab after CATT publication while most who favored bevacizumab prior to CATT publication continued favoring it afterwards. Aflibercept’s introduction had little impact on preferences for ranibizumab or bevacizumab.
INTRODUCTION
Anti-Vascular Endothelial Growth Factor (anti-VEGF) agents have dramatically transformed the landscape of retinal disease management over the past decade. They have undergone rapid and widespread adoption across health systems and insurance plans for the treatment of neovascular age-related macular degeneration (AMD), diabetic eye disease, retinal vein occlusion (RVO), and other off-label indications.1–3 Unlike prior treatments for these conditions, anti-VEGF agents have not only helped stabilize vision and prevent further vision loss, but many patients have experienced improvement in vision and health-related quality of life from the use of these agents. There are 3 commonly used anti-VEGF agents to treat retinal diseases—bevacizumab (Avastin; Genentech, South San Francisco, CA) and ranibizumab (Lucentis; Genentech, South San Francisco, CA) which became available in 2005–2006, and aflibercept (Eylea; Regeneron, Tarrytown, NY) which became available in 2011. While there are many similarities among these agents, there are several notable differences. Ranibizumab and aflibercept are both U.S. Food and Drug Administration (FDA) approved for intraocular injection, while bevacizumab does not have FDA approval to treat retinal diseases. The risk of endophthalmitis may be higher for bevacizumab since its use requires repackaging.4–9 Another key difference is cost. Ranibizumab and aflibercept cost approximately 40 times more than bevacizumab. This results in higher out-of-pocket costs to patients and higher societal costs for utilization of the more expensive 2 agents.2
From 2008 to 2010, researchers conducted the Comparison of Age-related macular degeneration Treatments Trial (CATT), a large multicenter randomized controlled trial designed to rigorously evaluate the safety and efficacy of bevacizumab and ranibizumab for neovascular AMD. This landmark study involved 1208 patients with neovascular AMD who were randomized to receive injections of ranibizumab or bevacizumab either monthly or pro re nata.10–11 The results of the trial demonstrated similar efficacy between the 2 agents and some suggestion of differences in the systemic safety profile of these agents, of unclear clinical significance. Despite the publication of the results of this trial, and several complementary trials performed in Europe which generated similar findings,12–16 there continues to be debate in the retina community regarding which agent to use for management of retinal diseases. While the CATT results have been widely cited in the literature, little is known the extent by which the publication of these results affected ophthalmologists’ actual utilization of these 2 agents.
Aflibercept received FDA approval and was introduced in November 2011. Though similar in cost to ranibizumab, one potential advantage of this agent over ranibizumab and bevacizumab is that it may remain in the eye longer, resulting in the possibility of less frequent dosing. On January 1, 2013, aflibercept received its own specific CPT code, facilitating its inclusion on many health plan drug formularies and making it easier to identify its administration in claims data.
While researchers have used claims data to assess trends in utilization of anti-VEGF agents1–3, 17–23 and identified differences in utilization of these agents based on sociodemographic factors and a patients’ location of residence,2,17,21–23 little is known about what factors contribute to ophthalmologists’ decisions to treat their patients with one versus another of these agents. The purpose of this study is to explore whether publication of the results of the CATT trial or the introduction of aflibercept to the marketplace may have affected ophthalmologists’ preferential utilization of bevacizumab and ranibizumab for patients with neovascular AMD.
METHODS
Data Sources
Data for these analyses came from two sources, a national Medicare claims database and the Clinformatics DataMart database (OptumInsight, Eden Prairie, MN), which captures patients in a large nationwide managed care network.
Medicare Claims Data
Data on Medicare enrollees were obtained from a random 20% sample of Medicare beneficiaries with Parts A, B, and D coverage between January 1, 2008, and December 31, 2014. Medicare master beneficiary files were used to verify Medicare eligibility and obtain demographics and other summary patient characteristics. This database uses International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) billing codes to identify all ocular and non-ocular diagnoses and Current Procedural Terminology (CPT-4®) procedure codes and Healthcare Common Procedural Coding System (HCPCS) drug codes to identify all diagnostic and therapeutic procedures performed on each beneficiary during the time s/he was covered by the health plan. This database has been used in the past to study patients with ocular diseases.1–2
Clinformatics DataMart Database
The Clinformatics DataMart database contains records of all enrollees in a large U.S. managed-care network including persons with dual enrollment in Medicare Advantage plans or Medicaid managed care. The data available include beneficiaries receiving any form of eye care from January 1, 2005, through December 31, 2015. This subset of beneficiaries had ≥1 ICD-9-CM codes for any eye-related diagnosis (360–379.9) or CPT-4 code for any eye-related visit, diagnostic or therapeutic procedure (65091–68899 or 92002–92499), or any other claims submitted by an ophthalmologist or optometrist during the beneficiary’s time in the medical plan. Similar to the Medicare database, the Clinformatics DataMart contains information on medical claims for all ocular and non-ocular conditions and demographic information (age, sex, race/ethnicity) for each enrollee. The rationale for replicating our analyses on this managed-care sample was to determine how generalizable the results are to persons with a different health insurance from traditional fee-for-service Medicare.
The data from both sources were de-identified prior to our licensing it and the study was approved by the University of Michigan Institutional Review Board.
Sample Selection
We identified all enrollees who received ≥1 intravitreal anti-VEGF injection for neovascular AMD (ICD-9-CM code 362.52). In the Medicare dataset, patients enrolled in Medicare Advantage plans were excluded as we did not have all claims for these enrollees.
Classifying Anti-VEGFs
To identify anti-VEGF injections, we queried the databases for all records of intravitreal injections (CPT code 67028) coupled on the same date with CPT/HCPCS codes for one of the three anti-VEGF agents of interest. The majority of anti-VEGF agent injections had been assigned a unique HCPCS code for bevacizumab (J9035), ranibizumab (J2778), or aflibercept (J0178, Q2046, C9291). A smaller subset of injections (28% in Medicare and 24% in Clinformatics) were associated with a nonspecific drug CPT code (J3490, J3590, or C9399). We categorized these injections as bevacizumab, ranibizumab, or aflibercept based on their cost and date of administration, as others have done (see Supplemental Methods available online).1
Statistical Analysis
We performed statistical analyses using SAS software, version 9.4 (SAS Institute, Cary, NC). Enrollee characteristics were summarized using means and standard deviations (SDs) for continuous variables and frequencies and percentages for categorical variables. For all analyses, p<0.05 was considered statistically significant.
Changes in Anti-VEGF Utilization Before versus After Publication of CATT Results and Introduction of Aflibercept to the Marketplace
For each insurance plan, we calculated the total number of injections of bevacizumab, ranibizumab, and aflibercept, respectively, for neovascular AMD performed each week over the study period. Next, we divided the entire study period into 3 time periods. Period 1 corresponded to the timeframe before the CATT results were published (January 1, 2008 to April 28, 2011 for Medicare and January 1, 2005 to April 28, 2011 for the Clinformatics sample). April 28, 2011 was the date CATT results were first published online. Period 2 corresponded to the timeframe from publication of CATT results (April 29, 2011) until aflibercept received its own unique HCPCS billing code (January 1, 2013). Period 3 was the timeframe after aflibercept received its own code (January 2, 2013, until the last date we had access to data— December 31, 2014 for Medicare and December 31, 2015 for the Clinformatics dataset). Although aflibercept gained FDA approval on November 18, 2011, we wanted to provide time for health plans to approve reimbursement for this agent and clinicians to adopt it, if they chose to do so.
Linear regression analyses were performed with the weekly injections as the outcome and time (in weeks) as the predictor. There were separate regression models each for the Medicare and Clinformatics datasets and for bevacizumab and ranibizumab, leading to 4 models in all. Since we were expecting different injection patterns in the three different time periods, we allowed for discontinuities (in magnitude and direction) at the ends of Period 1 and Period 2. The resulting segmented linear regression model allowed significance tests to compare the slopes from Period 1 vs. 2 and Period 2 vs. 3 within each model to determine whether the changes evident in the time points could have resulted from chance variation alone. T-tests were used to compare slopes in different time periods for a given model.
Ophthalmologist Utilization of Each Type of Anti-VEGF Agent
Next, we assessed the respective influence of the publication of CATT findings and the introduction of aflibercept to the marketplace on relative use of bevacizumab and ranibizumab for neovascular AMD by individual ophthalmologists. Each ophthalmologist in the two datasets has a unique identifier, allowing us to track his/her use of specific agents and how it changed over time. We identified all ophthalmologists who performed 10 or more anti-VEGF injections for neovascular AMD in each of the three time periods of interest. For each of the time periods, we then determined whether each ophthalmologist’s utilization pattern was bevacizumab-predominant (defined as administering bevacizumab for ≥80% of the injections s/he performed during that period), ranibizumab-predominant (administering ranibizumab for ≥80% of injections s/he performed), or no strong predominance (using each agent < 80% of the time). To evaluate whether the publication of CATT trial results and the introduction of aflibercept affected ophthalmologists’ preferences for bevacizumab versus ranibizumab, we identified “stayers” as those ophthalmologists who did not change their proportional drug use relative to the 80% threshold between Period 1 and 2 or Period 2 and 3, respectively. Similarly, we identified “switchers” as ophthalmologists who switched from being lighter users of a given agent (< 80% of injections performed were with that specific agent) to predominant users of that agent (or vice versa) in consecutive time periods. Finally, we assessed ophthalmologists who predominantly administered bevacizumab or ranibizumab during all 3 time periods, and those who switched their preferred agent at any point (including whether the switch occurred from Period 1 to 2, Period 2 to 3, or both timeframes).
RESULTS
Study Sample
In Medicare, there were 164188 enrollees who received ≥ 1 anti-VEGF injections (2026794 total anti-VEGF injections) for neovascular AMD between January 1, 2008, and December 31, 2014, and in the Clinformatics sample, 89753 beneficiaries received ≥ 1 anti-VEGF injections (881381 total anti-VEGF injections) for neovascular AMD between January 1, 2005, and December 31, 2015. Medicare enrollees (mean (SD) age, 78.4 (8.1) years) in our study were older than those in the managed care network (mean (SD) age, 76.2 (10.4) years) and there were more females and fewer racial and ethnic minorities in Medicare (Table 1).
Table 1:
Patient Demographics of Beneficiaries who Received Anti-VEGF Injections for Neovascular Age-Related Macular Degeneration
| Characteristic | Medicare | Commercial Health Insurance |
|---|---|---|
| Age at enrollment (years), mean (SD) | 78.4 (8.1) | 76.2 (10.4) |
| Sex, n (%) | ||
| Male | 60296 (36.7) | 35003 (39.0) |
| Female | 103885 (63.3) | 54750 (61.0) |
| Race / ethnicity, n (%) | ||
| White | 154887 (94.6) | 69703 (85.3) |
| Black | 3007 (1.8) | 4432 (5.4) |
| Latino | 2059 (1.3) | 5696 (7.0) |
| Asian | 2018 (1.2) | 1904 (2.3) |
| Other | 1682 (1.0) | -- |
“Other” race category not assessed for those with c ommercial health insurance.
SD = standard deviation; VEGF = vascular endothelial growth factor
In both Medicare and the managed care plan, utilization of bevacizumab consistently exceeded that of ranibizumab or aflibercept throughout the entire study period—in total, as well as proportional drug utilization. We consistently observed an increase in utilization of each agent near the end of each calendar year, followed by a marked decrease in the last week of the year corresponding to the week between Christmas and New Years, and then a spike in the early weeks of January of the following year.
Impact of Publication of CATT Trial Results on Bevacizumab and Ranibizumab Utilization
Among Medicare enrollees, the mean weekly number of bevacizumab injections rose considerably from 1617 injections/week in early 2008 to 3338 injections/week just prior to CATT result publication (Figure 1). On average, the number of weekly bevacizumab injections was estimated to rise by 10.0 injections per week (95% CI +8.7 – +11.3 additional injections per week) during Period 1. By two months after the CATT results came out, the mean weekly number of bevacizumab injections had further risen, to 3594 injections/week. During Period 2 the number of weekly bevacizumab injections was estimated to decline by 3.0 injections per week (95% CI −6.6 – +0.7, p<0.0001 compared to the rate of change in Period 1).
Figure 1. Influence of CATT Trial Result Publication and Availability of Aflibercept on Utilization of Bevacizumab and Ranibizumab for Neovascular Age-Related Macular Degeneration Among Medicare Enrollees.
A) Trajectory slope in Period 1 = +10.0 injections/week, slope in Period 2 = −3.0 injections/week (p<0.0001 for difference relative to Period 1), slope in Period 3 = −1.8 injections/week (p = 0.63 for difference relative to Period 2);
B) Trajectory slope in Period 1 = + 7.3 injections/week, slope in Period 2 = −6.7 injections/week (p<0.0001 for difference relative to Period 1), slope in Period 3 = −2.2 injections/week (p = 0.005 for difference relative to Period 2).
The mean weekly number of ranibizumab injections also rose over Period 1, from 1199 injections/week in early 2008 to 2375 injections/week just prior to CATT result publication. The number of weekly ranibizumab injections was estimated to rise by 7.3 injections per week (95% CI +6.5 – +8.2) during Period 1 and decline by 6.7 injections per week (95% CI −9.1 – −4.2) during Period 2 (p-value <0.0001 compared to the rate of change in Period 1).
Supplemental Figure 1, available at AJO.com, shows the mean weekly injections of bevacizumab and ranibizumab during the timeframes before and after the CATT results were published for those in the managed care network. In this data set, the number of weekly bevacizumab injections was estimated to grow by 6.3 injections per week (95% CI +5.7 – +7.0) and the number of weekly ranibizumab injections was estimated to grow by 2.0 injections per week (95% CI +1.7 – +2.2) during Period 1. The number of weekly bevacizumab injections was estimated to grow by 4.0 injections per week (95% CI +2.2 – +5.7) and the number of weekly ranibizumab injections was estimated to grow by 0.2 injections per week (95% CI −0.3 – +0.8) during Period 2.
Impact of Aflibercept Introduction to the Marketplace on Bevacizumab and Ranibizumab Utilization
Among Medicare enrollees, the mean weekly number of bevacizumab injections remained relatively stable from the start of Period 2 (3495 injections/week) through late 2012, just before aflibercept received its own unique billing code (3552 injections/week), and continued to remain relatively stable after aflibercept made it to the marketplace (3341 injections/week in 2013 and 3208 injections/week in 2014) (Figure 1). The mean number of weekly bevacizumab injections was estimated to decrease by 3.0 injections per week (95% CI −6.6 – +0.7) during Period 2, and estimated to continue declining at a similar rate (1.8 injections per week, 95% CI −4.6 – +1.0) during Period 3 (p=0.63 for Period 2 vs. 3). For ranibizumab, the weekly number of injections was 2261/week at the start of Period 2. We observed a spike in weekly ranibizumab injections near the end of 2011 (as high as 2882/week), followed by a decline in its weekly use during 2012. As Figure 1 depicts, the weekly use of ranibizumab was slightly higher during 2013 and 2014 compared to 2012. However, the rate of change in weekly ranibizumab use also declined during Period 2 (estimated at 6.7 injections per week (95% CI −9.1 – −4.2) and continued declining to a lesser extent (2.2 injections per week, 95% CI −4.1 – −0.3) during Period 3 (p=0.005 for Period 2 vs. 3).
Supplemental Figure 1, available at AJO.com shows how for enrollees in the managed care network, the rate of change in bevacizumab utilization continued to grow (estimated to increase by 0.9 injections per week (95% CI +0.2 – +1.6)) after aflibercept became available, while comparable ranibizumab utilization declined slightly (by 0.6 injections per week, 95% CI −0.8 – −0.3) during this timeframe.
Ophthalmologists’ Preferences for Using Bevacizumab versus Ranibizumab for Neovascular Age-Related Macular Degeneration
There were 2226 eligible ophthalmologists who cared for Medicare enrollees and 1474 eligible ophthalmologists who cared for patients in the managed care network. Among the ophthalmologists caring for Medicare beneficiaries, nearly one quarter (21.2%) used bevacizumab ≥80% of the time during all 3 time periods. Only 2% of all eligible ophthalmologists treating Medicare enrollees administered ranibizumab ≥80% of the time during all 3 of the time periods. Many of the ophthalmologists did not administer either agent ≥80% of the time during any of the 3 time periods, and the proportion of ophthalmologists who used either agent ≥80% of the time successively declined (61% in Period 1, 47% in Period 2, and 33% in Period 3). Very few ophthalmologists switched from using one agent ≥80% of the time to using another agent ≥80% of the time. Similar administrating patterns were observed for the ophthalmologists caring for the managed care enrollees (Table 2).
Table 2:
Ophthalmologists’ Preferences for Bevacizumab or Ranibizumab for Neovascular Age-Related Macular Degeneration Before and After CATT Trial Results and Aflibercept Availability
| Bevacizumab Treatment Patterns | Ophthalmologists treating Medicare patients N (%) |
Ophthalmologists treating Commercially-insured patients N (%) |
Ranibizumab Treatment Patterns | Ophthalmologists treating Medicare patients N (%) |
Ophthalmologists treating Commercially-insured patients N (%) |
|---|---|---|---|---|---|
|
Period 1 / Period 2 / Period 3 |
Period 1 / Period 2 / Period 3 |
||||
| ≥80% / ≥80% / ≥80% | 471 (21.2) | 305 (20.7) | ≥80% / ≥80% / ≥80% | 46 (2.1) | 43 (2.9) |
| ≥80% / ≥80% / <80% | 306 (13.8) | 280 (19.0) | ≥80% / ≥80% / <80% | 50 (2.3) | 43 (2.9) |
| ≥80% / <80% / ≥80% | 14 (0.6) | 17 (1.2) | ≥80% / <80% / ≥80% | 6 (0.3) | 9 (0.6) |
| <80% / ≥80% / ≥80% | 57 (2.6) | 53 (3.6) | <80% / ≥80% / ≥80% | 22 (1.0) | 17 (1.2) |
| ≥80% / <80% / <80% | 250 (11.2) | 169 (11.5) | ≥80% / <80% / <80% | 215 (9.7) | 113 (7.7) |
| <80% / ≥80% / <80% | 75 (3.4) | 79 (5.4) | <80% / ≥80% / <80% | 27 (1.2) | 19 (1.3) |
| <80% / <80% / ≥80% | 28 (1.3) | 22 (1.5) | <80% / <80% / ≥80% | 20 (0.9) | 17 (1.2) |
| <80% / <80% / <80% | 1025 (46.1) | 549 (37.3) | <80% / <80% / <80% | 1840 (82.7) | 1213 (82.3) |
| CATT Result Publication | CATT Result Publication | ||||
| Pre/Post | Pre / Post | ||||
| ≥80% / ≥80% | 777 (34.9) | 585 (39.7) | ≥80% / ≥80% | 96 (4.3) | 86 (5.8) |
| ≥80% / <80% | 264 (11.9) | 186 (12.6) | ≥80% / <80% | 221 (9.9) | 122 (8.3) |
| <80% / ≥80% | 132 (5.9) | 132 (9.0) | <80% / ≥80% | 49 (2.2) | 36 (2.4) |
| <80% / <80% | 1053 (47.3) | 571 (38.7) | <80% / <80% | 1860 (83.6) | 1230 (83.4) |
| Bevacizumab Stayer | 1830 (82.2) | 1156 (78.4) | Ranibizumab Stayer | 1956 (87.9) | 1316 (89.2) |
| Bevacizumab Switcher | 396 (17.8) | 318 (21.6) | Ranibizumab Switcher | 270 (12.1) | 158 (10.7) |
| Aflibercept-Specific CPT Code Release | Aflibercept-Specific CPT Code Release | ||||
| Pre / Post | Pre / Post | ||||
| ≥80% / ≥80% | 528 (23.7) | 358 (24.3) | ≥80% / ≥80% | 68 (3.1) | 60 (4.1) |
| ≥80% / <80% | 381 (17.1) | 359 (24.4) | ≥80% / <80% | 77 (3.5) | 62 (4.2) |
| <80% / ≥80% | 42 (1.9) | 39 (2.6) | <80% / ≥80% | 26 (1.2) | 26 (1.8) |
| <80% / <80% | 1275 (57.3) | 718 (48.7) | <80% / <80% | 2055 (92.3) | 1326 (90.0) |
| Bevacizumab Stayer | 1803 (81.0) | 1076 (73.0) | Ranibizumab Stayer | 2123 (95.4) | 1386 (94.1) |
| Bevacizumab Switcher | 423 (19.0) | 398 (27.0) | Ranibizumab Switcher | 103 (4.7) | 88 (6.0) |
Eligible ophthalmologists were required to have performed at least 10 anti-vascular endothelial growth factor injections for neovascular age-related macular degeneration in all 3 time periods for that insurance plan. Stayer = Ophthalmologists who continued to a) administer bevacizumab (or ranibizumab) ≥ 80% of the injections both before and after CATT results were published or aflibercept became available; b) administer bevacizumab (or ranibizumab) < 80% of the injections before and after CATT results were published or aflibercept became available. Switcher = Ophthalmologists who administered bevacizumab (or ranibizumab) ≥ 80% of the time and then switched to administering that agent a lesser proportion of the time or vice versa comparing their utilization before and after CATT results were published or aflibercept became available. Period 1 = pre-CATT (before 4/27/2011); Period 2 = post-CATT (4/27/2011 – 1/1/13 when the aflibercept-specific billing code was introduced); Period 3 = post-aflibercept (after 1/1/2013). CATT = Comparison of Age-related macular degeneration Treatments Trial
Before Versus After Publication of CATT Trial Results
There were 1041 ophthalmologists caring for Medicare enrollees who administered bevacizumab ≥80% of the time during Period 1. Among these 1041 ophthalmologists, 777 (74.6%) continued administering bevacizumab ≥80% of the time during Period 2 while 264 ophthalmologists (25.4%) used bevacizumab <80% of the time during Period 2. By contrast, there were 317 ophthalmologists caring for Medicare enrollees who administered ranibizumab ≥80% of the time during Period 1. Among these 317 ophthalmologists, only 96 (30.3%) continued administering ranibizumab ≥80% of the time during Period 2, while 221 ophthalmologists (69.7%) used ranibizumab <80% of the time during Period 2. Among the 771 ophthalmologists caring for patients in the managed care network who predominantly administered bevacizumab during Period 1, 585 (75.9%) continued to use bevacizumab predominantly during Period 2 while 186 (24.1%) used it <80% of the time during Period 2. Among the 208 ophthalmologists caring for patients in the managed care network who predominantly administered ranibizumab during Period 1, 86 (41.3%) continued to use ranibizumab predominantly during Period 2 while 122 (58.7%) used ranibizumab <80% of the time during Period 2.
Of the 2226 ophthalmologists caring for Medicare enrollees, 1830 (82.2%) maintained the same preferences for using bevacizumab when comparing Periods 1 and 2. In other words, as “bevacizumab stayers,” they either admini stered bevacizumab ≥80% of the time during both timeframes or administered bevacizumab <80% of the time during both time periods. There were 396 ophthalmologists (17.8%), who were “bevacizumab switchers,” meaning they administered bevacizumab ≥80% of the time during Period 1 and then administered bevacizumab a lesser proportion of the time afterwards or, alternatively, administered bevacizumab <80% of the time during Period 1 and ≥80% of the time in Period 2. Less than 1% of the ophthalmologists who switched away from being bevacizumab-predominant began using ranibizumab ≥80% of the time in Period 2.
The corresponding numbers of “ranibizumab stayers” and “ranibizumab switchers” were 1956 (87.9%) and 270 (12.1%), respectively. Most switchers (81.9%) switched away from predominantly using ranibizumab during Period 2. Of these, 3.2% switched to using bevacizumab ≥80% of the time. Very similar patterns of “stayers” and “switchers” for each medication were observed among the ophthalmologists caring for persons in the managed care network (Table 2), and 6.6% of the managed care ophthalmologists who switched away from using ranibizumab ≥80% of the time began using bevacizumab ≥80% of the time during Period 2.
Before Versus After Aflibercept Came to the Market
There were 909 ophthalmologists caring for Medicare enrollees with neovascular AMD who administered bevacizumab ≥80% of the time during Period 2. Among these 909 ophthalmologists, 528 (58.1%) continued administering bevacizumab ≥80% of the time during Period 3 while 381 ophthalmologists (41.9%) used bevacizumab <80% of the time during Period 3. There were 145 ophthalmologists caring for Medicare enrollees who administered ranibizumab ≥80% of the time during Period 2. Among these 145 ophthalmologists, 68 (46.9%) continued administering ranibizumab ≥80% of the time during Period 3 while 77 ophthalmologists (53.1%) used ranibizumab <80% of the time during Period 3. The corresponding numbers and proportions of ophthalmologists caring for managed care network enrollees who maintained their preference for using bevacizumab and ranibizumab from Period 2 to Period 3 can be found in Table 2.
Of the 2226 ophthalmologists caring for Medicare enrollees with neovascular AMD, 1803 (81.0%) maintained the same preferences for using bevacizumab when comparing utilization during Periods 2 and 3. In other words, they were “bevacizumab stayers” from Period 2 to 3. There were 423 ophthalmologists (19.0%) who were “bevacizumab switchers” from Period 2 to 3. Less than 1% of those who switched away from using bevacizumab ≥80% of the time began using ranibizumab or aflibercept ≥80% of the time during Period 3. A large majority of ophthalmologists (95.4%) were “ranibizumab stayers” (mostly preferring <80% utilization during both timeframes), while only 103 ophthalmologists (4.7%) were “ranibizumab switchers” from Period 2 to Period 3. Only 5.2% of those who switched away from predominantly using ranibizumab began using bevacizumab (2.6%) or aflibercept (2.6%) ≥80% of the time. Among ophthalmologists caring for the managed care network enrollees, there were very similar proportions of “stayers” and “switchers” fo r bevacizumab and ranibizumab. (Table 2)
DISCUSSION
Building on findings from previous studies looking at trends in anti-VEGF utilization for retinal diseases and differences in drug adoption between health plans,1,3 the focus of this study was to explore how new knowledge generated from randomized clinical trials and how a new drug entering the marketplace each may have affected patients’ receipt of bevacizumab and ranibizumab and ophthalmologists’ preferences for administering bevacizumab and ranibizumab for neovascular AMD. Using a large nationally-representative sample of more than 200,000 Medicare enrollees who received anti-VEGF injections for neovascular AMD, and replicating our analyses in a managed care patient population, we found that the publication of CATT results was associated with a decline in patients’ receipt of ranibizumab as well as ophthalmologists’ preferences for administering this agent. By comparison, bevacizumab utilization remained relatively stable after the CATT results came out and ophthalmologists’ preferences for administering it continued to remain quite high. We also learned that utilization of both bevacizumab and ranibizumab slowed following the introduction of aflibercept to the marketplace and ophthalmologists’ preferences for continuing to utilize each drug for neovascular AMD relative to one another remained relatively stable.
Prior to the CATT trial, there was no strong level 1 evidence evaluating the safety or efficacy of ranibizumab versus bevacizumab for retinal disease. Factors that may have influenced ophthalmologists’ decisions to use one agent versus the other included the results of earlier studies, costs of purchasing and storing these agents, reimbursement for administration of one agent versus the other, financial incentives for making bulk purchases or purchasing these medications using credit cards, the medicolegal environment where the ophthalmologist was practicing (given that only ranibizumab was FDA approved), input from pharmaceutical representatives, the location where the ophthalmologist did his or her fellowship training, preferences of the attendings who trained him or her, preferences of other ophthalmologists in the same practice or geographic community, insurance tiering and other policies, and/or whether the ophthalmologist’s patients could afford the co-pays for the more expensive agent.1–3,17–23
Publication of CATT results offered strong evidence of similar efficacy and relatively similar safety profiles between bevacizumab and ranibizumab. From our analysis of over 2000 ophthalmologists who frequently performed anti-VEGF injections on Medicare enrollees with neovascular AMD both before and after the publication of CATT results, we identified three key trends in anti-VEGF utilization patterns that may be attributable to publication of the CATT results. First, we learned that the majority of ophthalmologists did not predominantly administer either anti-VEGF agent ≥80% of the time either before or after the CATT results came out. Second, we found that the majority of the ophthalmologists were “stayers,” meaning that their preference for administering bevacizumab or ranibizumab did not substantially change after the publication of the CATT results. Third, in the subset of ophthalmologists who predominantly administered 1 agent or the other before publication of the CATT results, we found that more than two-thirds of ophthalmologists who predominantly used ranibizumab prior to publication of the CATT results did not continue to do so afterwards, while three-quarters of ophthalmologists who predominantly administered bevacizumab for neovascular AMD before CATT results were published still predominantly used bevacizumab afterwards.
Not surprisingly, for many ophthalmologists who had no strong preference regarding choice of agent before CATT results were published, the trial results appear to have had little or no impact on their treatment patterns. These findings are consistent with previous studies which have suggested that the effect of randomized controlled trials on altering medical practice has been suboptimal and often slow (e.g., underutilization of beta-blockers after myocardial infarction).24–30 However, among ophthalmologists who had a strong pre-existing preference for one anti-VEGF agent over the other, we found that the vast majority (75%) of ophthalmologists who favored bevacizumab for neovascular AMD before the CATT results came out continued to favor it afterwards, whereas many ophthalmologists who favored use of ranibizumab prior to the publication of CATT results may have been swayed away from ranibizumab-predominant use based on the results of the trial.
A variety of factors may have led to the shift away from ranibizumab utilization after CATT results were published. Some ophthalmologists may have predominantly used ranibizumab during Period 1 because it was FDA approved or because of medicolegal concerns with administering bevacizumab. A subset of these ophthalmologists may have felt that the results of this landmark trial were sufficient to support use of a non-FDA approved product and/or offered adequate evidence of standard of care to protect them from potential lawsuits for using bevacizumab off label. Other ophthalmologists who preferred bevacizumab before CATT results were published may have had their preferences for this agent during Period 1 supported by the trial findings of similar efficacy of the 2 agents. Similarly, CATT findings of no statistically-significant difference in endophthalmitis rates between bevacizumab and ranibizumab may have reinforced these ophthalmologists’ preexisting preferences for the lower-cost bevacizumab and potentially convinced some ophthalmologists who previously favored ranibizumab to switch to preferring bevacizumab. It is also possible that other factors, such as corporate marketing or other anti-VEGF manuscripts that were published around the same time at CATT may have also influenced ophthalmologists’ utilization patterns. Other considerations, such as cost and financial incentives were unlikely to contribute to the differences we observed from Period 1 to Period 2, given that these factors remained relatively constant during both periods.
In contrast to publication of CATT results, marketplace entry of aflibercept appeared to be associated with a shift away from preferences for bevacizumab and ranibizumab for neovascular AMD. Approximately 40% of the ophthalmologists who predominantly administered bevacizumab and 53% of the (few) ophthalmologists who predominantly administered ranibizumab for neovascular AMD before aflibercept came out were no longer predominantly using these drugs afterwards. Part of this change may have represented ophthalmologists switching some of their patients to aflibercept in the hope of potential greater effectiveness, especially in nonresponders to other agents. However, since all of the considerations distinguishing bevacizumab from ranibizumab continued to exist after aflibercept came out, some ophthalmologists who previously favored ranibizumab (due to its FDA approval, compounding concerns with bevacizumab, or financial considerations) might have recognized similarities in aflibercept and been persuaded to switch given the added benefit of potentially less-frequent aflibercept dosing. Other ophthalmologists who previously favored bevacizumab may have also been swayed by the potential for fewer injections, but at a slightly lower rate, since presumably these ophthalmologists recognized bevacizumab’s substantially lower cost and may have been less concerned regarding lack of FDA approval of bevacizumab, its need for compounding, or financial incentives.
We observed similar utilization patterns among ophthalmologists treating neovascular AMD patients with commercial health insurance as among those treating Medicare patients. However, there was more switching of utilization behavior among ophthalmologists caring for patients with commercial health insurance, especially away from bevacizumab and ranibizumab after aflibercept became available and (to a lesser extent) toward predominantly utilizing bevacizumab after publication of the CATT results. Since many eye care providers treat both Medicare and commercially-insured patients, we hypothesize that apparent differences in treatment patterns between Medicare and commercial health insurance may be driven more by insurance plan incentives than by beliefs regarding drug efficacy, safety, or FDA approval, which should remain the same for a given provider regardless of the type of insurance his or her patients possess.
Our analysis is limited in that we were unable to assess differences in clinical outcomes such as visual acuity, ocular coherence tomography (OCT) findings, incidence of blindness, or patient quality of life between patients receiving bevacizumab versus ranibizumab and how these factors affected preferential use of a given agent. Because direct identifiers of the names or practices of the ophthalmologists were masked to us, we could not perform comparisons to assess whether specific ophthalmologists managed patients with Medicare differently than those with commercial health insurance. When interpreting our study findings, it is important to keep in mind that there are clearly many factors contributing to the overall utilization of anti-VEGF agents and changes over time. Factors such as the number of enrollees entering or leaving insurance plans in a given year, the number of ophthalmologists performing anti-VEGF injections, and the number of previous injections a given patient received, may all affect the overall number of injections performed for neovascular AMD. The intent of our analyses was to estimate the respective impacts of CATT results and the introduction of aflibercept on differential use of ranibizumab and bevacizumab for neovascular AMD, however, it is possible that some of these other factors also affected ranibizumab and bevacizumab use for neovascular AMD over the study period. Our findings are also inherently limited by reliance on administrative claims-based data, however, since physician reimbursement for these injections is tied to proper coding and documentation of care provided, missing or inaccurate data is likely minimal in this analysis. Finally, while our analyses help substantiate associations between changes in relative utilization of specific anti-VEGF agents with the timing of CATT Trial result publication and introduction of aflibercept to the marketplace, it is important to note that in an analysis such as this one, it is impossible to prove causality.
Despite these limitations, our study helps the ophthalmologic community and health policy-makers appreciate the influence that the publication of results of major ophthalmologic clinical trials and the introduction of new ophthalmic interventions to the marketplace can have on patients’ receipt of existing therapies as well as ophthalmologists’ preferences for selecting one intervention over another. Our research highlights the importance of the National Eye Institute and other funding agencies to support and expand funding for large-scale clinical trials such as CATT. By identifying what factors appear to influence decision-making, health-policymakers can come up with strategies to more effectively promote high quality, cost-effective care.
Supplementary Material
A) Trajectory slope in period 1 = +6.3 injections/week, slope in period 2 = +4.0 injections/week (p = 0.01 for difference relative to period 1), slope in period 3 = +0.9 injections/week (p = 0.002 for difference relative to period 2);
B) Trajectory slope in period 1 = +2.0 injections/week, slope in period 2 = +0.2 injections/week (p < 0.0001 for difference relative to period 1), slope in period 3 = −0.6 injections/week (p<0.01 for difference relative to period 2).
ACKNOWLEDGEMENTS
(a) Funding/Support: Funding to support this research comes from NEI R01 EY026641 (JDS), NIA R03 AG056453 (SP), and departmental support from Research to Prevent Blindness (SP and JDS). The funding organizations had no role in the design or conduct of this research.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Presented, in part, at the Association for Research in Vision and Ophthalmology meeting, May 10, 2017, Baltimore, MD.
Supplemental Material available at AJO.com
DISCLOSURES
(b) Financial Disclosures: 1) SP: consultant with equity interest in Verana Health, San Francisco, California, USA; consultant, Acumen, LLC, Burlingame, California, USA; 2) NT: none; 3) STA: none; 4) JG: none; 5) JMR: none; 6) VSD: none; 7) JDS: none
REFERENCES
- 1.Pershing S, Pal Chee C, Asch SM, et al. Treating age-related macular degeneration: comparing the use of two drugs among Medicare and Veterans Affairs populations. Health Aff (Millwood) 2015;34:229–238. [DOI] [PubMed] [Google Scholar]
- 2.Stein JD, Hanrahan BW, Comer GM, Sloan FA. Diffusion of technologies for the care of older adults with exudative age-related macular degeneration. Am J Ophthalmol 2013;155(4):688–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Parikh R, Ross JS, Sangaralingham LR. Trends of anti-Vascular Endothelial Growth Factor use in ophthalmology among privately insured and Medicare Advantage patients. Ophthalmology 2017. March;124(3):352–358. [DOI] [PubMed] [Google Scholar]
- 4.Corbett-Corren J FDA Warns on Eye Infections With Repacked Avastin. Wall Street Journal 2011. August 31; 2011. [Google Scholar]
- 5.Pollack A Avastin Injections Are Reported to Cause Blindness New York Times; 2011. August 31. [Google Scholar]
- 6.Pollack A Five More Reports of Avastin Injections Causing Blindness New York Times; 2011. September 2. [Google Scholar]
- 7.Frost BA, Kainer MA. Safe preparation and administration of intravitreal bevacizumab injections. The N Engl J Med 2011;365:2238. [DOI] [PubMed] [Google Scholar]
- 8.Goldberg RA, Flynn HW, Isom RF, Miller D, Gonzalez S. An outbreak of Streptococcus endophthalmitis after intravitreal injection of bevacizumab. Am J Ophthalmol 2012;53(2):204–208.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Edison LS, Dishman HO, Tobin-D’Angelo M, et al. Endophthalmitis Outbreak Associated with Repackaged Bevacizumab. Emerging Infectious Diseases 2015;21(1):171–173. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.The CATT Research Group* Ranibizumab and Bevacizumab for Neovascular Age-Related Macular Degeneration. N Engl J Med 2011; 364:1897–1908. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Comparison of Age-related Macular Degeneration Treatments Trials (CATT) Research Group., Martin DF, Maguire MG, Fine SL, et al. Ranibizumab and bevacizumab for treatment of neovascular age-related macular degeneration: two-year results. Ophthalmology 2012;119(7):1388–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.IVAN Study Investigators, Chakravarthy U, Harding SP, Rogers CA, et al. Ranibizumab versus bevacizumab to treat neovascular age-related macular degeneration: one-year findings from the IVAN randomized trial. Ophthalmology 2012. July;119(7):1399–411. [DOI] [PubMed] [Google Scholar]
- 13.Chakravarthy U, Harding SP, Rogers CA, et al. ; IVAN study investigators. Alternative treatments to inhibit VEGF in age-related choroidal neovascularisation: 2-year findings of the IVAN randomised controlled trial. Lancet 2013. October 12;382(9900):1258–67. [DOI] [PubMed] [Google Scholar]
- 14.Berg K, Pedersen TR, Sandvik L, Bragadóttir R. Comparison of ranibizumab and bevacizumab for neovascular age-related macular degeneration according to LUCAS treat-and-extend protocol. Ophthalmology 2015. January;122(1):146–52. [DOI] [PubMed] [Google Scholar]
- 15.Krebs I, Schmetterer L, Boltz A, et al. A randomised double-masked trial comparing the visual outcome after treatment with ranibizumab or bevacizumab in patients with neovascular age-related macular degeneration. Br J Ophthalmol 2013;97:266–271. [DOI] [PubMed] [Google Scholar]
- 16.Schauwvlieghe AM, Dijkman G, Hooymans JM, et al. Comparing the Effectiveness of Bevacizumab to Ranibizumab in Patients with Exudative Age-Related Macular Degeneration. The BRAMD Study. PLoS One 2016. May 20;11(5):e0153052. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Gower EW, Stein JD, Shekhawat NS, Mikkilineni S, Blachley TS, Pajewski NM. Geographic and demographic variation in use of ranibizumab versus bevacizumab for neovascular age-related macular degeneration in the United States. Am J Ophthalmol 2017. December;184:157–166. [DOI] [PubMed] [Google Scholar]
- 18.McLaughlin MD, Hwang JC. Trends in Vitreoretinal Procedures for Medicare Beneficiaries, 2000 to 2014. Ophthalmology 2017. May;124(5):667–673. [DOI] [PubMed] [Google Scholar]
- 19.Hutton D, Newman-Casey PA, Tavag M, Zacks D, Stein JD. Switching to less expensive blindness drug could save Medicare Part B $18 billion over a ten-year period. Health Aff (Millwood) 2014;33(6):931–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lad EM, Hammill BG, Qualls LG, et al. Anti-VEGF treatment patterns for neovascular age-related macular degeneration among Medicare beneficiaries. Am J Ophthalmol 2014;158:537–43. [DOI] [PubMed] [Google Scholar]
- 21.Curtis LH, Hammill BG, Qualls LG, et al. Treatment patterns for neovascular age-related macular degeneration: analysis of 284 380 Medicare beneficiaries. Am J Ophthalmol 2012;153:1116–24. [DOI] [PubMed] [Google Scholar]
- 22.Brechner RJ, Rosenfeld PJ, Babish JD, Caplan S. Pharmacotherapy for neovascular age-related macular degeneration: an analysis of the 100% 2008 Medicare fee-for-service Part B claims file. Am J Ophthalmol 2011;151:887–95. [DOI] [PubMed] [Google Scholar]
- 23.Erie JC, Barkmeier AJ, Hodge DO, Mahr MA. High variation of intravitreal injection rates and Medicare anti-Vascular Endothelial Growth Factor payments per injection in the United States. Ophthalmology;123:1257–62. [DOI] [PubMed] [Google Scholar]
- 24.Goldberger JJ, Bonow RO, Cuffe M, et al. , for the PACE-MI Investigators. Beta-blocker Use Following Myocardial Infarction: Low Prevalence of Evidence-based Dosing. Am Heart J 2010. September; 160(3): 435–442.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Lenfant C Clinical Research to Clinical Practice —Lost in Translation? N Engl J Med 2003; 349:868–874. [DOI] [PubMed] [Google Scholar]
- 26.Boissel JP. Impact of randomized clinical trials on medical practices. Control Clin Trials 1989. December;10(4 Suppl):120S–134S. [DOI] [PubMed] [Google Scholar]
- 27.The Impact of Randomized Clinical Trials on Health Policy and Medical Practice: Background Paper (Washington, D. C: U.S. Congress, Office of Technology Assessment, OTABP-H-22, August 1983). [Google Scholar]
- 28.Lubloy A Factors affecting the uptake of new medicines: a systematic literature review. BMC Health Serv Res 2014. October 20;14:469. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Davis DA, Thomson MA, Oxman AD, et al. Evidence for the Effectiveness of CME: A Review of 50 Randomized Controlled Trials. JAMA 1992;268(9):1111–1117. [PubMed] [Google Scholar]
- 30.Davis D, O’Brien MA, Freemantle N, et al. Impact of formal continuing medical education: do conferences, workshops, rounds, and other traditional continuing education activities change physician behavior or health care outcomes? JAMA 1999. September 1;282(9):867–74. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
A) Trajectory slope in period 1 = +6.3 injections/week, slope in period 2 = +4.0 injections/week (p = 0.01 for difference relative to period 1), slope in period 3 = +0.9 injections/week (p = 0.002 for difference relative to period 2);
B) Trajectory slope in period 1 = +2.0 injections/week, slope in period 2 = +0.2 injections/week (p < 0.0001 for difference relative to period 1), slope in period 3 = −0.6 injections/week (p<0.01 for difference relative to period 2).