Sunday, June 2, 2013

Biased research, aggressive sales, harmful drugs

Approval from the U.S. Food and Drug Administration to market a new drug is a critical waypoint along the path to profits for pharmaceutical manufacturers. Unfortunately, recent case studies have illustrated that FDA approval does not necessarily provide assurances of effectiveness and safety. In last month's Georgetown University Health Policy seminar, we discussed two examples, anemia drugs and the diabetes drug rosiglitazone (Avandia), which were prominently featured in recent articles by Peter Whoriskey in The Washington Post.

Image courtesy of

The original impetus for the development of the anemia drugs Epogen, Procrit, and Aranesp, which mimic the  actions of the hormone erythropoietin, was to spare dialysis patients with severe anemia the inconvenience and risks associated with periodic blood transfusions. However, noted Whoriskey, pharmaceutical companies moved aggressively to market these drugs to a far larger patient population who were much less likely to benefit from them:

The trouble would arise as the drugmakers won FDA approval for vastly expanded uses, pushing it in larger doses, for milder anemia and for patients with a wider array of illnesses. Very quickly, the market included nearly all dialysis patients, not just the roughly 16 percent who required blood transfusions. The size of average doses would more than triple. And over the next five years, the FDA would approve it to treat anemia in patients with cancer and AIDS, as well as those getting hip and knee surgery.

Doctors were motivated to give more doses of these drugs due to generous financial incentives (estimated at between $100,000 and $300,000 annually for a typical oncologist) and the seductive thinking that if some drug was good, more was better. Even the publication of a 1998 study in the New England Journal of Medicine showing no survival advantage to boosting hematocrit levels to normal ranges in cardiac patients did little to discourage overprescribing. Not until 14 years later did an independent researcher obtain access to the complete study report from the FDA and conclude that the NEJM authors had used statistical slight-of-hand to obscure an increased risk of heart attacks and death in the normal-hematocrit group. In the meantime, lobbyists working for the drug manufacturers successfully blocked efforts by Medicare administrators to stop paying for the higher (harmful) doses.

Similarly, the evidence that rosiglitazone (Avandia) increased the risk of heart attacks was slow to come to light, due in part to the drugmaker's research emphasis on the surrogate outcome of glycemic control. Although a 2007 meta-analysis first sounded the alarm about rosiglitazone's cardiovascular risks, the manufacturer successfully stalled regulatory action in the U.S. for three more years, during which thousands of new patients were prescribed the drug.

Could the FDA and other U.S. government agencies do more to protect patients from the effects of biased research and aggressive sales tactics for newly marketed drugs? What concrete steps could health policymakers take to encourage research to identify unexpected harms earlier in the drug approval process?


The above post first appeared on The Health Policy Exchange.


  1. Addendum: Almost unbelievably, a New York Times article yesterday reports that the FDA is looking to reconsider the belated restrictions it placed on sales of Avandia a few years ago:

  2. I have to respectively disagree with you on this one. The implication of your post is that pharmaceutical companies bias research and inappropriately market products that do not help and often lead to harm. While the industry has had its share of bad examples in the past and must be monitored closely, you suggest this is an ongoing problem in need of stricter regulations.
    In the first example, Epogen and Procrit are safe and effective drugs when used in the appropriate patients. The problem was that they were over used in the wrong populations of patients. While the drug companies did promote the expanded use (as they have a fiduciary responsibility to do so), the main reason for the increased use according the Post article you site was “at the center of any explanation of the popularity of these drugs are the nation’s doctors, clinics and hospitals, and the choices they made for patients.” While you do cite this reason, your focus is on biased research and aggressive sales tactics, not greedy hematologists and nephrologists. If this drug were like other drugs (where doctors don’t receive additional compensation for prescriptions), use would have been far less.
    In your second example, your timing is pretty bad. Recent data shows that Avandia is likely to be safe after all. The FDA is meeting tomorrow to decide whether or not restrictions should be lifted.

    While on the one hand, I agree with you that it is incredibly important to have a strong regulatory process to ensure the medications that we prescribe to our patients are safe and effective, I am concerned when we slam the drug industry and imply that so many of the drugs we prescribe may not do any good and likely cause harm. The consequences of this kind of thinking are not insignificant. First, patients are now likely not to trust any medications we prescribe. I am sure you will agree that statins are the most well studied medication with the clearest risk benefit ratio for patients with cardiovascular risk. In addition, atorvastatin is now generic. Yet, so many patients who could benefit from this medication refuse to take this or any drug because of what they read on the internet and what they see on TV (especially laywer ads). Secondly, increasing regulations makes it harder for new treatments to come to market. Because of the Avandia scare (which has now been shown to be a hoax) all diabetes medications must go through an extra regulatory step to prove they don’t cause heart attacks. This has caused the delay of many good therapies.
    Our job as physicians is to do what’s best for our patients. We need to be cautious of the industry (pharmaceutical or otherwise) to ensure that they are providing safe and effective products for our patients, while at the same time being careful not to over state concerns which lead to fear and panic.

  3. Hi Dr. Mintz,

    Thanks for these thought-provoking comments. Like you, I'm a primary care physician and am also disgusted at the unscrupulous behavior of specialist physicians and hospitals that were overprescribing Epogen and Procrit to patients who were unlikely to benefit and often likely to be harmed. I certainly didn't intend to paint all drug companies with the same broad brush of corruption. Yes, the purpose of a business is to make money, and that applies to drug companies as well as medical practices. However, the "fiduciary responsibility" of drug companies doesn't exempt them from the obligation to practice ethical behavior. Slanting research findings to hide serious adverse effects and playing politics to sell more drugs regardless of the human consequences is unethical.

    As for Avandia, I strongly disagree with your assertion that the findings of harm have "been shown to be a hoax"; the convening of an FDA meeting to reexamine the evidence hardly proves that the substantial body of literature showing an association with increased heart attacks and deaths is suddenly null and void. Even if a drug has harmful side effects, it shouldn't necessarily be pulled from the market - as long as the benefits outweigh the harms for a group of patients. In my mind, Avandia doesn't pass this test, as it's never been shown to improve patient-oriented outcomes (CVD, death) and there are safer alternatives available. Finally, speaking on behalf of my patients with diabetes, I think it's great that the FDA is now requiring manufacturers of new diabetes medications to demonstrate that they don't cause heart attacks BEFORE these drugs go on the market, given the alternative.

  4. Thanks for responding to my response.
    We are probably on the same page with Epogen and Procrit. Their manufacturers' practices are certainly partly to blame, I am not about to defend them.
    My problem is that when you put this example with Avandia and conclude with "Could the FDA and other U.S. government agencies do more....," you imply that this is a pervasive problem that must be dealt with. As someone who lives around the Beltway and knows about the mad dash for toilet paper at the first mention of a snow flake, I am sure you will agree that over-dramatizing a problem can lead to negative consequences.
    We are likely to disagree about Avandia. There is no "substantial body of literature showing an association with increased heart attacks and deaths" with Avandia. There is essentially one meta-analysis done by a self-promoting cardiologist that purposely created attention by going around the FDA and straight to the NEJM. Since I know you are a fan of evidence based medicine, I am sure you will agree that a large, randomized trial specifically designed to look at safety trumps a meta-analysis any day. With this particular meta-analysis: 1) someone else did the same analysis with different techniques and found different results and 2) the event rate was IDENTICAL in the Avandia and non-Avandia groups, but some how the relative risk showed a 42% increase in heart attacks? AHRQ's review, ACP's review, and ACC's review of the data all saw no increase rates of heart attacks with the exception of the single meta-analysis (and maybe one or two of many claims analysis that showed variable results).
    Regardless of the outcome of the advisory committee, based on what has been presented today, my conclusion is that the FDA got it right on Avandia. When GSK presented the FDA with the meta-analysis of their own data that suggested a risk of myocardial ischemia, the FDA decided to wait until other randomized, large prospective trials reported out, include RECORD which was designed to specifically look at CV safety. Turns out they were correct, since all other data seems to negate the meta-analysis. This suggests that the policies we have in place are not all that terrible.
    Finally, in regards to CV outcomes, no diabetes drug has ever been shown to improve CV outcomes, despite the fact that diabetes is so closely associated with CV risk. There are likely multiple reasons for this including the fact that patients must be on statins and it may take up to a decade to show any real difference. Nonetheless, most clinicians assume that it is wise to treat diabetes to prevent both micro and macrovascular complications. Given that only diabetes drugs have to prove CV safety (no other drugs are required to do this) and we now know that Avandia doesn't cause heart attacks, requiring new diabetes medications to prove CV safety is unnecessary and only leads to delays in newer medications that could help patients.

  5. In general, I agree that a large, well-designed and high-quality RCT trumps a meta-analysis of smaller studies not designed to answer the question being asked (in this case, does Avandia increase risk of heart attacks / mortality)? However, I do not believe that the RECORD trial was well-designed or high-quality, despite the findings of the independent re-analysis commissioned by GSK and presented to the FDA. Based on the result of today's vote, it appears that the FDA will lift existing restrictions on Avandia, but I will not prescribe it and will discourage other clinicians from doing so.

  6. I guess we will just have to agree to disagree on this one. I believe RECORD was actually a well designed study. The Duke independent analysis shows that the study was well done, so the only issue here is that it was not blinded. While the rationale given at the FDA panel was that rosiglitazone was not indicated for use with insulin, I contend that even that doesn't matter. Because glycemic control is associated with CV outcomes (even if treatment hasn't been proven to change that outcome), in order to show a difference in safety, you would want glycemic control to be similar in both groups. Thus, if you blinded patients not controlled on SU or metformin to placebo or Avandia, those on placebo would initially have poorer glycemic control, giving an unfair safety advantage to Avandia. Designing the study as open label allows for randomization to a world with Avandia and a world without Avandia. While RECORD showed no difference in CV safety, it also showed that in a world without Avandia, there was more insulin use and more hypoglycemic events.
    Finally, blinding patients in an efficacy study makes sense. If you know you are taking a medicine, due to the placebo effect, you are likely to get better just by thinking the medication works (even if it does not). However, in a safety study, this should theoretically work in reverse. In other words, if you know you are taking a medicine, due to the nocebo effect, you are likely to get a side effect just by thinking the medication might be dangerous. In other words, there was clearly bias given the open label nature of RECORD, but that bias was against Avandia not in favor of it. Yet despite this unfair disadvantage, Avandia still showed no increase risk of heart attacks or CV death.
    Nissen's meta-analysis simply doesn't hold up. No one has ever been able to explain to me how the actual event rate is the same in both groups but the relative risk is 43%. Others have analyzed the same data set with different techniques and found no difference. Essentially, it is my belief that in the Avandia case, politics trumped science-something I know you are very familiar with.