Writing in the New England Journal of Medicine (Identifying and Eliminating the Roadblocks to Comparative-Effectiveness Research) three authors share their experience in running a head-to-head trial of Avastin (bevacizumab) versus Lucentis (ranibizumab) for wet age-related macular degeneration (AMD). They describe the barriers they faced and suggest that the barriers will need to be removed for comparative effectiveness research –as envisioned under ARRA– to succeed. They make good points and may well be correct in their policy recommendations.
However the case of Avastin and Lucentis is unusual. The products are made by the same manufacturer and are essentially identical. Avastin and Lucentis are marketed separately by Genentech mainly to allow the company to capture a return on investment from its R&D. The issue is that a regular dose of Avastin (e.g., for lung cancer) can be divided up into many doses for the eye. Since the products are sold by volume it turns out that Avastin is cheap when used for wet AMD, even though it’s pricey when used for cancer. As I’ve suggested previously, Genentech should be able to charge Lucentis prices for Avastin when it’s used in the eye. So there are quite a lot of people –starting with the manufacturer itself– who didn’t really want this study to go forward. That’s less likely to be the case with other studies.
With those caveats, here are the issues that were encountered:
- Initially CMS did not want to pay for routine Lucentis use by the study population. This policy was changed in 2007 to provide coverage of drugs under investigation if they were normally covered outside the trial. So this should no longer be an issue.
- Some patients were responsible for co-pays, and the differentials could be large because of the difference in the prices for the drugs. That could discourage participation or bias the results. NIH was able to make an exception to its policy and cover the difference. Ordinarily the differences won’t be so stark, but this could come up again.
- The most significant issue from my standpoint is how to mask the drugs so study participants don’t know which drug they’re getting. That’s actually harder to accomplish than it may sound when a trial population is mixed in with a clinical population. For example, how do you bill for a drug that is not identified? And how do you prevent an Explanation of Benefits from being printed that contains the name of the drug? It sounds as though a couple of attempts have been made to address this issue but that they have come to naught.
The authors rightly point out that studies in non-Medicare populations will be even more complex, because of all the private insurers involved. I agree this is an important area to address.
There are plenty of political barriers to the conduct and use of comparative effectiveness research. I’m glad to see people thinking about the practical barriers, even if I don’t agree that they are completely generalizable.