American Journal of Law & Medicine

The money blind: how to stop industry bias in biomedical science, without violating the First Amendment.

The pharmaceutical and medical device industries use billions of dollars to support the biomedical science that physicians, regulators, and patients use to make healthcare decisions--the decisions that drive an increasingly large portion of the American economy. Compelling evidence suggests that this industry money buys favorable results, biasing the outcomes of scientific research. Current efforts to manage the problem, including disclosure mandates and peer reviews, are ineffective. A blinding mechanism, operating through an intermediary such as the National Institutes of Health, could instead be developed to allow industry support of science without allowing undue influence. If the editors of biomedical journals fail to mandate that industry funders utilize such a solution, the federal government has several regulatory levers available, including conditioning federal funding and direct regulation, both of which could be done without violating the First Amendment.

I. AN INTRODUCTION TO BIASED SCIENCE

Scholars estimate that between thirty and forty-five percent of the growth in medical spending is driven by the decisions of prescribers, patients, and payors to adopt new medical technologies produced by the drug and device industry. (1) This industry spends billions of dollars to create these innovative products, but also spends about as much to change the behavior of prescribers, consumers, and payors to ensure that they are purchased. (2) Against this onslaught, regulators restrict the industry's promotional efforts on behalf of these products, in order to protect patient welfare and to optimize the expenditure of public and private healthcare money, so that it is not wasted on products that are inefficient, ineffective, or even dangerous in a given application. These regulators of promotional efforts are, however, constrained by the First Amendment to the United States Constitution, which protects the industry's right to commercial speech. (3) Here, the battle lines are drawn. (4)

In some ways, however, the industry's efforts to influence biomedical science are more profound and more disconcerting than the industry's explicit promotional activities. When successful in its efforts to manipulate biomedical science, the industry transforms the very epistemological basis that scientists, regulators, juries, physicians, and patients rely upon to assess the safety and adequacy of industry products. Such influence literally changes what we think we know about these products. This section documents the problem of industry influence in biomedical science, and explains why status quo solutions are inadequate.

A. INDUSTRIAL SCIENCE AS COMMERCIAL SPEECH

Biomedical science is the boundary-setting precondition for industry promotional efforts. As a veteran of the industry writes, "in the pharmaceutical industry, there are two ways to market an approved drug for a new use: the 'indication' route--performing studies necessary for regulatory approval--or the 'publication' strategy, which stimulates off-label prescribing by using research 'to disseminate the information as widely as possible through the world's medical literature.'" (5) Both of these routes crucially turn on the industry's ability to procure scientific studies that purport to support their product. Thus, in a practical sense, biomedical science is the industry's first avenue of promotion. A longtime editor of the British Medical Journal provocatively titled his own article on the phenomenon, "Medical Journals Are an Extension of the Marketing Arm of Pharmaceutical Companies." (6) The industry apparently concurs in this assessment. In a strikingly candid document produced by Pfizer, the company asks, "What is the purpose of publications?" The answer: the "purpose of data is to support, directly or indirectly, the marketing of our product." Or in short: "Purpose of Publications: The Bottom Line." (7)

There is a growing recognition that the information presented in biomedical journal articles is distorted by these companies that fund the research and create financial relationships with the researchers as consultants or equity owners. (8) The former editor-in-chief of a major biomedical journal writes that "the public trust in research has been eroded and there is a perception that professional fidelity and honesty on the part of investigators and clinicians has deteriorated." (9) More bluntly, the editor of The Lancet states, "Journals have devolved into information laundering operations for the pharmaceutical industry." (10) The judicial system has begun to notice. Judge Jack Weinstein writes, "The pervasive commercial bias found in today's research laboratories means studies are often lacking in essential objectivity, with the potential for misinformation, skewed results, or cover-ups." (11)

Empirical evidence supports these conclusions. The drug and device industry is the single largest source of funding for biomedical research, both directly in its own research centers and indirectly through grants to academic investigators. (12) In the United States, for example, industry funds about seventy percent of the clinical trims of its drugs and devices. (13) And even when the industry is not funding the trial, its stockholders, consultants, officers, and directors are often conducting the study. (14) As the federal government cuts budgets, the industry's role as the primary benefactor of biomedical science is likely to grow. (15)

The industry's expenditure on this publication strategy seems to be a worthwhile investment. In a landmark review of the literature, an Institute of Medicine report concluded that: "Several systematic reviews and other studies provide substantial evidence that clinical trims with industry ties are more likely to have results that favor industry." (16) Indeed, one meta-study showed that industry-funded research is eight times less likely to reach unfavorable conclusions compared to independent studies. (17) Industry-sponsored studies can be biased in favor of the product being studied due to choice of design and methodologies, selective analysis and interpretation of data, and conclusory statements in the resulting journal abstracts and articles that might not be supported by the data. Indeed, there are no less than eighteen such opportunities for motivated investigators to consciously or subconsciously bias their research discussed below. (18)

Admittedly, there may be benign explanations for some portion of the apparent biases. Perhaps industry-funded studies are more likely to reach favorable results simply because the industry is more conservative in deciding which studies it funds, compared to government and foundation funders who have the luxury of pursuing more conjectural hypotheses. (19) The industry's strategy could be conceived of as an efficient use of research dollars, or it may degrade into a company's willful ignorance of unfavorable results, as they "decline[] to fund clinically important studies at least partly because the results might reduce sales of the drug." (20) The benign explanations can go only so far, because the evidence also shows that industry-funded studies are also biased towards being rather weak methodologically. A recent review of the methodological quality of 886 published studies in one field of medicine (orthopedics) found that "the level of evidence of industry-funded studies was lower than that for studies funded by governments, foundations, or universities." (21)

The problem is also one of trust. Even if all this industry money did not in reality create pernicious biases in science, it has clearly undermined the perceived legitimacy of this important institution. (22) The flood of industry funding creates an appearance of impropriety, one that is leading towards a "systematic distrust and devaluation of expertise" in this context. (23) A former editor-in-chief of the New England Journal of Medicine (NEJM) has lamented, "Physicians can no longer rely on the medical literature for valid and reliable information." (24) If physicians cannot rely on the medical literature, what are they doing instead? The very profession of medicine is at stake.

Together then, we have evidence showing that industry funds a huge portion of biomedical science, that industry studies tend to be favorable to industry (a seeming bias in their conclusions), but that the studies are relatively weak methodologically (a seeming bias in their evidentiary strength). As a result, some physicians may careen towards complete skepticism of industry science, but the remainder who must proceed to practice in this flood of industry science will rely upon it. Such physicians will be swayed to use drugs or medical devices in contexts where they might not be effective, where they might present unnecessary risks to patients, or where they simply are not economical compared to treatment alternatives. Thus, as the industry succeeds in warping biomedical science to represent industry interests rather than physiological reality, it degrades the practice of medicine, harms patient welfare, and raids the treasuries of state and national governments.

B. THE FAILURE OF STATUS QUO SOLUTIONS

What can be done to reduce the potential corrupting influence of industry money in biomedical research? Current regulatory mechanisms include litigation, peer review, and mandatory disclosure. This section explains why they fail to solve the problem.

1. Litigation

To date, except for ex post reactions in the most egregious cases, the law has been largely silent with respect to the problem of biased science. (25) In theory, if the industry's manipulation of science rose to the level of outright fraud, a plaintiff could recover under state tort laws, the federal qui tam act, and federal racketeering statutes--assuming that the plaintiff had standing and that he or she could prove causation of a specific injury. Except in the most egregious cases, these barriers are nearly insurmountable. (26)

When drug- and device-makers promote their products beyond the uses approved by the Food and Drug Administration (FDA), such violations can lead to Department of Justice enforcement actions and seemingly-large settlements. (27) Such lawsuits do not, however, reach the fundamental problem of biased science, but instead focus on the downstream problems that arise when companies go too far in promoting their products.

There are also more creative theories available for litigators. In the recent case of Merck v. Reynolds, the U.S. Supreme Court heard a case involving a securities fraud class action against the manufacturer of rofecoxib (Vioxx), an anti-inflammatory drug approved to treat arthritis pain. (28) The plaintiffs alleged that the company had made various misrepresentations about the drug in order to inflate its stock price, including a March 2000 study supported by the company and published in the NEJM. The data showed a four-fold increased risk of adverse cardiovascular events with Vioxx over naproxen, but the industry-affiliated authors put the finding on its head. The authors wrote that the adverse event rate "was significantly lower in the naproxen group than in the rofecoxib group (0.1 to 0.4)," (29) a statement that implies that the difference was due to a cardioprotective effect of naproxen rather than a toxic effect of Vioxx. In addition, the investigators had predetermined an endpoint for the study, and it was later revealed that additional post-endpoint cardiovascular adverse events occurred in Vioxx-treated patients in that trial, but these results were not included in the published article. (30) After the publication of this and other similarly-biased trials, annual sales of Vioxx reached billions of dollars, but the drug was later removed from the market due to concerns over its cardiovascular safety. The plaintiff stockholders argued in the securities fraud case that the biased biomedical journal articles misrepresented the safety of the product, and thus the financial security of the company. After a win for the plaintiffs in the Supreme Court on a preliminary procedural issue, this case remains pending. (31)

Thus, the pharmaceutical and medical device industries may face potential liability in the extreme cases that rise to outright fraud, but only on the rare occasion that it can be detected by plaintiffs, proven in ex post litigation, and where causation of a specific and tangible harm can be demonstrated. The concern here, on the other hand, is about a ubiquitous industry influence on biomedical science, a biasing pressure that is strong enough to change prescribing and consumer behavior, but not so blatant as to be prosecuted as outright fraud. That sort of manipulation appears to enjoy something near legal impunity.

2. Peer Review

The biomedical journals utilize a peer review process to police the methodological rigor of biomedical journal articles, and thus may be a bulwark against industry efforts to manipulate biomedical science. Biomedical journal editors have two primary tools in their arsenal: peer review and disclosure. (32) Peer review is in one sense an extreme form of non-governmental regulation, not unlike censorship, saying to rejected authors, "you cannot say that here." (33) Thus, peer review is in theory a strong bulwark against the industry's ability to manipulate science. (34)

Still, peer review focuses merely on the methods and data reported in the text of journal article drafts, and only indirectly addresses industry influence in a long chain of decisions that produce those drafts. (35) And, unlike a grant funding agency or an Institutional Review Board, the peer reviewers must assess the methodological rigor of the submitted articles after the studies have been designed and the research has been completed. Peer reviewers' choices are limited to accepting an article (with or without textual revisions) or rejecting an article (which effectively demotes it to another journal, which will then have the opportunity to publish the groundbreaking results, notwithstanding the limitations).

Thus, peer reviewers' decisions are necessarily pragmatic and comparative--weighing the clinical significance of the findings against the study's apparent methodological rigor, and asking whether there is likely to be a more significant or more rigorous article in the queue to fill the journal's pages instead. To the extent that industry-influenced studies dominate the medical literature, they define the range of alternative articles and thus set their own benchmarks for methodological rigor.

Some have argued that peer review is "slow, expensive[,] ... something of a lottery, prone to bias[,] ... easily abused," and hopeless at spotting errors and fraud. (36) Without wading into the details of that debate, for our purposes the results speak for themselves; the foregoing evidence showing that industry-funded studies tend to be biased and methodologically weak are based on publications in peer-reviewed journals. (37) Thus, while one could speculate about how much worse the situation would be without peer review, it remains clear that peer review is not a complete solution.

3. Mandatory Disclosure

Another potential remedy is for biomedical journal editors to require authors to disclose industry funding and investigators' related financial interests. In theory, readers of biomedical journal article abstracts (i.e., physicians, payors, and regulators) would use disclosures of the authors' relationships with industry to calibrate their reliance on the abstracts they read. (38)

It is worthwhile to understand how this reliance-calibration mechanism is supposed to work in practice. Suppose that a physician is deciding whether to prescribe a certain drug for a given disease that is not listed on the label. (39) Since the FDA has not determined whether the drug is in fact safe and effective for the off-label indication, the physician must make her own epistemic assessment. (40) The physician knows that the chance that any random chemical would be useful for alleviating a given disease is quite low, and the physician has no particular physiological theory that would predict that this drug would be effective in treating the given disease. Nonetheless, the physician has heard anecdotes that patients with the disease have improved after receiving the drug. So all this information could form the basis for a Bayesian prior of non-effectiveness. She might conclude that there is a 0.1 probability of the drug being effective for the given disease. (41)

Now, if the physician were then presented with a scientific research study funded by the National Institutes of Health showing that the drug is effective for the given disease, she might then update her prior belief and now conclude that the probability of efficacy is greater than 0.5. Perhaps she would then prescribe the drug. (42) On the other hand, suppose that a disclosure mandate instead revealed that the second study was funded by the company that makes the drug. …

Log in to your account to read this article – and millions more.