This Drug Didn’t Work for Me. May I Have My $88,000 Back, Please?
If you bought a new toaster, carried it home, plugged it in, and it didn’t work, what would you do? Suppose you purchased a pineapple at the grocery store, sliced it open, and found it was inedible. In both cases you’d seek the same remedy: you’d return the item and ask for your money back. With reputable merchants, the chances are extremely high that you would receive both an apology and your refund. If the merchant wouldn’t return your money, you could likely get the charges reversed anyway if you paid by credit card.
Now let’s look at one more variation on this theme. Imagine that you bought (or had someone buy for you) a product that cost a whopping $88,000. You used it according to the manufacturer’s directions, but not only did this product fail to work, it significantly raised your blood pressure or caused a hemorrhage. You’d want to get your money back for sure. But in this final scenario, you are a breast cancer patient, and the thing you (or more likely your insurance company, or Medicare) paid all that money for is a cancer drug called bevacizumab (Avastin). Genentech, the medicine’s manufacturer, follows a general convention in the industry and offers no refunds if this drug doesn’t work. My question is: why not? Wouldn’t it make sense for manufacturers to refund the cost of uber-expensive drugs that provide no demonstrable benefit to a majority of patients?
The use of Avastin in breast cancer patients has been mired in controversy. The FDA created an “accelerated approval” program in 1992 to speed the arrival of critical drugs onto the market for those with life-threatening diseases, like AIDS or cancer. Avastin, Genentech’s best selling drug, won FDA approval in 2004 for treating metastatic colon cancer and in 2006 for non-small cell lung. It received conditional approval in 2008 for use in breast cancer patients, even though it failed to extend their survival in a clinical trial. The preliminary approval was given with the concomitant requirement that there be a later review to see if the drug conferred either a survival benefit or improved quality of life for patients. The subsequent analysis showed that in this particular disease indication bevacizumab provided neither benefit, and its approval for use in breast cancer patients was rescinded in December 2010. Genentech appealed the decision, but a just completed review by an FDA advisory panel led to the same conclusion. The panel voted unanimously that breast cancer patients should no longer use the drug.
This decision does not prevent doctors from still prescribing the drug “off label” for their breast cancer patients. However, it reduces the likelihood that your insurance company will pay for the drug to near zero, meaning that you will have to pay for it out of pocket. This medicine is not cheap, with doses for the typical breast cancer patient costing $88,000 a year. The practical result is that only the wealthiest patients will be able to afford this (and other) expensive medication for “off-label” treatments.
A recent Wall Street Journal editorial opined that it was bad medicine for the FDA to rescind the drug’s approval for breast cancer patients. This was based on the fact that while bevacizumab demonstrated no overall statistical survival benefits to breast cancer patients, anecdotal reports suggest that a small (but unknown) percentage of patients responded favorably to the drug. I can readily sympathize with patients who may feel that a potential lifeline is being taken away from them. However, the advocacy group Breast Cancer Action agreed with the FDA decision to withdraw the treatment’s approval for this specific disease indication. As their Executive Director, Karuna Jaggar, succinctly stated, “…anecdotes are not science….”. Christine Brunswick of the National Breast Cancer Coalition echoed these comments “The FDA’s decision on Avastin must be based on scientific evidence from well-done trials and cannot be based on any one individual story, no matter how compelling.” The FDA itself was unable to find any evidence for the existence of a population of “super-responders” to the drug.
The problem with the WSJ argument, of course, is that it could just as easily be applied to nearly all drugs. It’s possible (maybe even likely) that you could find a tiny fraction of patients tucked away in the tail end of the bell-shaped curve that might somehow benefit from any rejected medicine if you looked hard enough. However, the vast majority may not be helped, and many might even be harmed by side effects. The risk to benefit ratio for any medication is evaluated based upon the seriousness of the disease being treated. Society is willing to tolerate more serious side effects in a cancer medication than a cold remedy, but these still need to be outweighed by the benefits. And since most patients can’t afford to pay for these medicines out of pocket, then insurance companies must bear the cost. This is ultimately passed along to consumers in the form of higher premiums. Insurance companies are not in the habit of paying for drugs that are ineffective, have serious side effects, or both, which is exactly what the FDA concluded in their review of this treatment for breast cancer. There is a longstanding historical rationale for having the FDA serve as a societal shield that protects consumers from dangerous and ineffective drugs.
I’ll suggest a potential solution for those rare patients who might be helped by the drug, but whose insurance companies won’t pay for it. Why not set up a system where patients that may benefit from the drug get to take it on a trial basis? Their insurance plan will pay for its use in this indication under strict guidelines. If the patient shows no clear benefit from the drug in a specified period of time, then the drug’s seller will reimburse the insurance company (or the patient, if they paid for it out of pocket) for the expense.
Such an approach, though admittedly complicated, has several benefits. Patients will gain “off label” access to a drug that their doctors feel may benefit them, and have the expense covered by their insurance if it works. Insurance companies only pay for the drug if it works effectively; otherwise the manufacturer reimburses them. The drug company’s potential pool of patients will be expanded with a proviso that their sales are directly tied to the ability of the medicine to benefit this population. This reimbursement approach links company profits from “off-label” sales to how frequently the drug actually works in this population, which would help promote more effective medicines.
This idea is not as radical as it sounds. A very similar approach is already being employed in England. Britain’s drug oversight group, the National Institute for Health and Clinical Excellence (NICE), originally rejected covering the multiple myeloma drug bortezomib (Velcade) for patients in the U.K. Andrew Dillion, the head of NICE, put forth a “money back” proposal to the drug’s seller “If the drug’s manufacturer accepts the proposals . . . it will mean that when the drug works well the NHS pays but when it doesn’t the manufacturer should bear the cost. All patients suitable for treatment will get the chance to see if the drug works well for them.” Janssen-Cilag, the drug’s manufacturer, accepted the proposal under which NICE would now recommend the drug (which costs $49,450 per year). Patients that show a minimal response, or none at all, will have their treatment stopped and the drugs costs will be refunded. Why not bring such a system here to the U.S?
This isn’t to say that all drugs should be sold on a refundable basis if the patient derives no benefit. While a merchant can readily establish that a toaster is defective or a pineapple is bad, proving that an expensive drug failed to work would likely require costly and possibly invasive tests, or at least a fair amount of paperwork. For relatively cheap medicines, these expenses would likely cost more than the drug itself. Drug manufacturers would also want to protect themselves from scammers who benefited from the drug, but seek refunds anyway to get it for free. Therefore, the burden would be on the patient (or more likely, his or her physician) to provide assurances that the drug did not work. This policy need not be mandated, either, but a money-back guarantee would certainly help convince an insurance company to pay for a particular treatment.
The primary difficulty here, of course, is coming up with an agreed-upon definition of whether or not the patient has “responded well” to the drug. Getting a clear, unambiguous agreement that would be acceptable to patients, the drug seller, and the insurance company will certainly be a challenge. While this is not a trivial issue, it should be a solvable problem. It should certainly be easier than the challenge of coming up with the drug in the first place. And suppose the initial diagnosis was wrong? Drug companies wouldn’t want to eat the expense of refunding medicines that didn’t work because the patient was misdiagnosed. However, this expense would be wrapped into the overall cost of doing business, lumped in with unproductive research programs and failed clinical trials.
Would linking drug performance to cost reimbursement result in an increase in the price of our drugs? One could argue that as the percentage of patients paying for a drug decreases, then the cost of the drug would necessarily increase. This would be balanced, to some degree, by an increase in the number of patients (financed by their insurance companies) willing to try the drug knowing that the expense would be reimbursed if it didn’t work. In countries where there are strict cost controls on the health care system, expensive drugs are effectively rationed and must demonstrate great benefit in order to be sold there.
Another question would be whether to apply this reimbursement policy to drugs sold for approved indications as well as “off label” uses. For example, many expensive drugs (e.g. TNF inhibitors for rheumatoid arthritis patients) are quite effective in a majority of patients, but provide no benefit to a smaller percentage. Venture capitalist Steve Burrill was recently cited as stating that 90 percent of drugs work in just 30 to 50 percent of patients. Why not refund the cost of their drugs to this latter group as well?
I have no hard answers here, but believe the subject is worth examining in greater detail as more and more high priced medicines come to market. Drug makers are currently focused on creating medicines that are more likely to be effective against smaller, well defined subsets of patients (e.g. those identified by using biomarkers) with particular diseases. Smaller patient populations will translate into higher priced drugs, but the model is viable. Some of Genzyme’s medicines, for example, can cost $200,000 dollars a year and up. Furthermore, it is increasingly likely that treating some diseases, such as cancer, will be done by prescribing a combination of drugs that hone in on specific targets, raising the potential cost of treatment to what many fear will be prohibitive rates.
Many drug are given away to doctors as free samples, so the idea of giving out medications on a trial basis is well established in clinical practice. And virtually all drug makers are happy to obtain the extra revenues earned when their drugs are prescribed for either legitimate (e.g. testing an anti-cancer medicine against a rare tumor) or illegitimate (e.g. use of epo by Tour de France cyclists) “off label” uses. We like to think that in a fair society, payment for a job is directly linked to performance. Perhaps the time has come for drug payments to be linked in the same manner.