Why Lisa Rosenbaum Gets Conflict of Interest Policies Wrong

May 28th, 2015

May 28, 2015

By Vinay Prasad, MD MPH


Before I take on arguments offered by Dr. Lisa Rosenbaum in her three-part New England Journal of Medicine series regarding conflicts of interest (COI), let me begin with some facts we all agree on. First, the biomedical industry has developed some great drugs and devices over the last 30 years. Consider drugs like traztuzumab or ace-inhibitors, and devices like implantable cardiac defibrillators or coronary stents (for acute myocardial infarction—not stable angina, another debate). The industry has also developed and promoted some OK (or less than OK) drugs and devices. Many cancer drugs that offer very marginal improvements in survival fit this bill, as do many devices, such as the aortic balloon pump.[1] Dr. Rosenbaum also concedes that “some interactions [between industry and physicians] do threaten professional judgment and, ultimately, public health…. [For instance] withholding data, falsely advertising, and securing physicians’ loyalty with Hawaiian vacations are egregious and should be prohibited.” We can all agree with both of these things: Industry has created some good products, and the industry has also engaged in practices that threaten professional judgment.

Similarly, we can all agree that the purpose of medicine, particularly academic medicine, is to promote human health—maximize human health in so far as possible. The literal meaning of the words ‘conflict of interest’ means that some relationships, goals, biases, and sentiments that academic physicians have occasionally conflict with the goal of maximizing human health. For instance, the primary purpose of the biomedical industry is to maximize profits, and a secondary goal is to improve human health. Nearly all the well-publicized examples of industry malfeasance have to do with cases where the industry is forced to choose between pursuing those two goals, and pursuing profits wins [2]. Those decisions help make the pharmaceutical industry often among the most profitable sectors of the economy [3].

Of course, we all agree that there are interests besides financial ones. The person who develops a cancer drug in his/her lab may really want to see it work; so much so that they may unconsciously advocate for a skewed trial design. So I agree with Dr. Rosenbaum on this point, “scientists who develop novel treatments should not be the ones testing them in humans.” But simply because physicians can have other competing interests doesn’t mean we should ignore financial conflicts. Rather, it means we should try to tackle all conflicts, including financial conflicts.

Finally, the simple fact is that, as of 2015, our conflict of interest policies remain weak. In nearly all cases, simply disclosing a financial conflict is considered sufficient. In other cases, disclosure is not even required. Many academics consult for many drug companies, and collect tens of thousands of dollars (or more). Many academics have large research funds that they can control, built from industry dollars, and some of those funds support travel, staff and other services that make life more pleasant for the academic (often very pleasant). Dr. Rosenbaum criticizes the critics who object to recent cholesterol guidelines, written by a panel where seven out of 15 panelists had financial conflicts of interest. But the fact remains that these panelists were included, and that tells us how weak the rules currently are. Whether right or wrong, we must acknowledge that COI policies are currently lax. With this background, let’s turn to Dr. Rosenbaum’s series.

Turning to the specifics—As Larry Husten has noted, “Responding to Rosenbaum isn’t easy, since the articles, though beautifully written, do not present a coherent and cogent position.[4]”  As such, here is my best stab at what her thesis is: Restrictions on financial conflicts between academic medicine physicians and the biomedical industry have not been shown to improve patient outcomes and may stifle innovation. As she puts it, “the key question: are any of these interactions, or efforts to curtail them, beneficial or harmful to patients?” And, “I think the desire for retribution against “bad pharma” informs our management of industry interactions in a way that obscures the possibility that we are obstructing medical advances.”

One thing here is true—no, we don’t have prospective randomized trials showing that conflict of interest policies improve patient outcomes. But, such a stance is absurd: as Husten notes, “By that logic, we should repeal all laws prohibiting bribery of public officials until we can empirically demonstrate the disastrous effects.” At the same time, we have abundant data showing that the influence of the industry leads to more favorable conclusions for trials and cost-effective analyses [5], which serve as the bedrock for clinical decisions. Moreover, industry sponsored studies ask certain questions (and ignore others) that appear to prioritize market share over making the best medical decisions. [6]  Time and again, research has shown that industry-influenced research presents the most favorable impression of their products, which may or may not reflect the true risk-benefit ratio. It is not surprising this is the case, as it appears to fit with the fact that industry prioritizes profit over health.

Rosenbaum’s idea that even the weak COI rules we have now has hindered novel therapies is incredibly speculative, and lacks any empirical evidence. Consider another equally speculative claim, and imagine if instead this formed the basis of a series of NEJM articles: Had Congress created a federal agency to develop drugs to compete against for-profit pharmaceutical companies, who knows how many more transformative drugs we would have had, how cheap they would be, and how many more lives we would have saved!

This statement is also speculative, but motivated by a different ideology. The fact that the NEJM articles speculate one way, but not the other tells us little more than the attitudes and ideologies of the writers and editors. Without further empirical data, neither sentence belongs in a major medical journal.


The Watchman Device

Moreover, Dr. Rosenbaum’s call for high evidence showing COI policies improve patient outcomes runs contrary to the standard of evidence she demands from actual medical devices. In one example of too-stringent COI policing, she writes:

“At issue is the Watchman, an atrial appendage closure device made by Boston Scientific, a company for which the cardiologist had previously consulted on an unrelated product. In keeping with the data, the cardiologist deemed the device less effective than warfarin in reducing the risk of stroke in patients with atrial fibrillation. Yet he voted in favor of the Watchman, believing it should be available for patients whose bleeding risk precludes warfarin use. Despite this appropriate clinical justification, however, the article’s conclusion insinuates that his true motive was financial: “Following the advisers’ vote, Boston Scientific told analysts it expected the Watchman to win FDA approval in the first half of 2015 and eventually reach $500 million in yearly sales.”

Dr. Rosenbaum states that approving the Watchman would be “appropriate” for patients who cannot take warfarin, yet there are no randomized trials showing benefit of the device in this setting. This is a different question entirely, one that is wholly unproven, and one for which the device ultimately was not approved. In fact, we may legitimately ask to see a trial of the Watchman versus investigator choice (typically aspirin) among patients who cannot take coumadin—something the panelist was not asking for. In short, the panelist thought the device should be available for an indication for which it had no data, rather than an indication for which it had unfavorable data. Such a position is absurd, but Dr. Rosenbaum does not question it.



Dr. Rosenbaum begins her discussion criticizing the critics of recent statins guidelines for including panelists with financial conflicts with statins makers. Rosenbaum writes, “One could argue that people also deserve to know that statins are, in many cases, the best drugs we have to prevent cardiovascular disease and that the committee had spent 5 years reviewing the evidence to identify the patients who would benefit most.”

But Dr. Rosenbaum does a disservice to the debate about statins because she does not make an attempt to fairly articulate the concerns of critics. First, everyone loves statins for patients post-MI, so everyone can be persuaded by data when it is strong. But at the same time, recent guidelines represent an unprecedented promotion of statins for healthy people, and may lead up to a billion people world wide to take statins [7].

Whether or not a healthy person should take a preventive medicine is justified only by improvement in survival or quality of life. For statins, the best data is unclear whether it improves survival[8]. It’s also up for debate whether a decrease in some events (MI), but possible increases in other side effects constitutes an improvement in quality of life. Hence, the ongoing statin controversy. Dr. Rosenbaum erects a straw man in claiming that critics’ sole objection to the guidelines is based on the rampant COI. In fact, the reason many wonder about whether COI played a role is because the data are not clear. This is a common theme of COI: The worry is that they affect medicine not in the black and white cases, but the gray. And in modern medicine, there is a lot of gray.

Finally, it must be worth noting that the largest set of data to adjudicate this question is not being shared by the Cholesterol Treatment Trialists, in part due to contractual agreements they have with industry. Rosenbaum herself called concealing data “egregious” – what makes not sharing the statin data acceptable?


Calcium channel blockers (CCBs)

As others have noted, Dr. Rosenbaum misunderstands the history of calcium channel blockers. They were widely promoted and marketed as superior to other anti-hypertensives, but ultimately proved to be no such thing. CCBs were not better than diuretics in the ALLHAT trial, and we really should have stuck with diuretics until CCBs proved superiority in trials and not just in expert opinion. Moreover, those who were most supportive of CCBs had financial conflicts. An interesting modern study would be to examine patterns of recommendation and prescription for nebivolol (Bystolic). Her choice of CCBs as an example was telling: Even when one selects examples to make the cases against COI policies, one finds COI biasing medical decisions.


COI system

I think Dr. Rosenbaum needs to consider one aspect of how we deal with COI more. She writes, “I think oversimplified conflict narratives pose a similar threat to medicine, allowing true experts to be replaced — on advisory panels, as authors of reviews and commentaries, in other capacities of authority — by people whose key asset is being conflict-free.”

This is an important question: Who should be on guidelines and advisory panels [9]?  I think surely not people with financial conflicts of interest. I don’t even think we will miss the so-called “content experts.” Likely non-conflicted people with training in research methodology should be the primary deliberators, with experts serving as consultants [10]. Others have considered this at length.  In the era of clinical trials, the expert is the person who best understands trial design and bias, and not the one with the best knowledge of the purported pathophysiology of disease.



These were just a few focused criticisms of the series, which overall does not present a convincing narrative. I don’t have energy to comment on the many pop-psychology references, or her pervasive caustic language (“pharmascold”). I find the article challenging to rebut because it doesn’t take clear positions, and generally laments as draconian our current COI policies, which honestly are not restrictive at all. If anything, our present challenge is that we are drowning amid conflicts of interest. For some fields and topics, it is nearly impossible to get the opinion of someone not consulting for industry.  For some fields, the only trials we have are bad trials, with bad choices of controls and endpoints—and their interpretation is led by people with financial conflicts.  Such a system does not appear to seek to maximize the health of our patients.

The first step towards managing financial conflict of interest will be greater transparency and reporting, which is coming. Transparency will allow researchers to show just how much COI affects medical care. Over time, I hope to see a medical system try to improve the incentive structure such that the goal of maximizing human health remains paramount. Although other biases exist, financial conflicts are a special problem, and warrant immediate and swift action. The path forward will require impartial and scientific design of clinical trials, and careful weighing of evidence. Doing so, will require stronger policies to mitigate conflicts of interest.

More than anything, I am personally disappointed that a journal that I love to read, so much so that I visit the website at 5:01 PM EST every Wednesday, has lowered their standards and published this long series of ultimately speculative articles. My reaction to the series can be summed up simply:  The articles are on the wrong side of history.


Dr. Prasad is a physician who lives in Washington, DC


[2] https://www.youtube.com/watch?v=YQZ2UeOTO3I  & Avorn, Jerry. Powerful Medicines: The Benefits, Risks, and Costs of Prescription Drugs & Angell, Marcia. The truth The Truth About the Drug Companies: How They Deceive Us and What to Do About It