Industry funding of research has always been a rather fraught topic. In the immediate aftermath of what we might reasonably call the “Coca-Cola calorie debacle,” or perhaps “GEBNgate,” it is all the more so. But in this domain, as in so many, the rush to summary judgment (in either direction) obliterates nuance fundamental to real understanding, and obscures a middle path conducive to important and even crucial advances.
Consider, for instance, that were we to take admonishments against industry funding to their logical extreme, 1 thing they would yield for certain is a world with no pharmacies. Virtually everything in the modern pharmacopoeia is a product of industry-funded research.
For some few of you, devoted, perhaps to natural remedies, and thus far spared by happenstance or youth the need for modern medical intercession, a world with no pharmacies may sound good. But most of us, even those of us very devoted to lifestyle as medicine, and very disinclined to take actual pharmaceuticals, have at times unavoidably been their beneficiary. I am, and I have, and thus have cause to be grateful that somebody funded the studies of the drugs I happened to need. So, too, for surgery, the tools of which also advance preferentially through the advent of industry funding.
So here, as in virtually all such scenarios where a complex topic warrants a good scrubbing, we have baby and bathwater in the same tub, and a perilous tendency to overlook the distinction.
Are there important liabilities of industry funding, of pharmaceuticals and anything else? Of course. Interested parties infuse a research agenda with a predictable bias. That bias in turn can corrupt methods from the outset, sometimes rather subtly, and foster the propagation of systematically misleading results.
That is not the only risk. When research costs are covered by private rather than public dollars, the funder might choose to suppress unfavorable findings, or to release findings selectively, or slowly. We might thus wind up with the truth, and maybe even nothing but the truth, but something so far removed from the whole truth as to constitute a falsehood. The famous case of Vioxx is a precautionary tale occupying just such terrain.
But on the other hand, were there no industry funding of pharmacotherapeutic advance, the noteworthy declines in cancer mortality would almost certainly not have been seen. The variety of antibiotics that, thus far at least, defend us admirably against all but the most exceptionally resistant microbes, would not be at our disposal.
Of comparable relevance, funding by the most ostensibly unimpeachable sources is no assurance against just such liabilities. The premier source of biomedical research funding in the U.S., the National Institutes of Health, is by no means devoid of its own native biases, and even potential for conflicts of interest.
Just like any other funder, the NIH favors positive results. The occasional study may be aimed at debunking, but most studies are infused by the hope of positive outcomes. This means that the researchers funded by the NIH are biased as well, hoping to show positive results. There are rather salient reasons why.
For 1 thing, all funders, ultimately, have funders of their own. In the case of NIH, it is the U.S. Congress, processing the money we, the people, provide in our tax dollars. NIH must remake the case for its slice of the pie as the federal budget is renegotiated every year, and breakthrough results certainly help. The expenditure by NIH last year on lackluster results does not help this cause.
Another problem is that our statistical conventions make positive results generally more reliably positive (albeit, far from perfectly so) than negative results are reliably negative. I will spare you the analytical weeds here, and keep it at a very high level: one of the most common explanations for a negative result at odds with hope and expectation is lack of statistical power to find the effect being sought. This, in turn, results from too small a sample.
So, you think, just use a larger sample! Good idea, but human clinical trial costs are related directly to sample size. I have a precautionary tale to tell.
My lab received funding from the NIH some years ago to run the first ever, placebo-controlled, randomized trial of an intravenous nutrient infusion in widespread, but unproven, use for fibromyalgia (among other things). The funding was for a preliminary, rather small, pilot study, with the understanding that the conduct of the trial, and outcome, would determine next steps.
Our treatment group did improve significantly. In fact, the effect size we saw was greater than reported for FDA-approved drugs for the same condition. However, our intravenous placebo produced a very robust placebo effect, so our control group improved substantially as well, albeit less. The result was that the between-group differences favored active treatment, but were not statistically significant. These results were negative, but not reliably so.
This seemed to be a perfect argument for a larger, definitive study. The pilot had proven our ability to apply appropriate methods, and run the study in this challenging population, but left efficacy in doubt. But we could not secure funds for a definitive trial. Reviewers at the NIH simply referred to the “negative” pilot results, and said: no thanks. We struggled to get our results published as well, despite the robustness of the study methods, because journals were no more favorably disposed to such murky outcomes than were funders. We prevailed, but only in a far from top-tier journal.
Such is the peril of pilot studies and negative results: One may not be able to say whether there is a true lack of effect, or merely lack of evidence of effect. In the case of intravenous nutrients for fibromyalgia, the result was a very unfortunate dead-end, leaving the matter unresolved to this day.
There is yet another problem with federal funding. While individual grants receive excruciating review, there is concern that the overall allocation of biomedical research dollars may be misdirected. This concern has been voiced by colleagues in Preventive Medicine, and echoed by none other than a Nobel Prize-winning, former NIH director.
In addition, politics may introduce bias directly. Colleagues directly affected tell me that during the Bush years (the second Bush), NIH grants addressing harm reduction strategies, such as needle exchange, were summarily rejected on ideological grounds, despite their clear, epidemiologic promise.
There is one more handicap to NIH, or other federal agency funding: these sources are none too enthusiastic about any proprietary advance, and tend not to fund studies of same. But what if there is legitimate hope that a proprietary, branded app, technology, device, program, algorithm, or even food has specific, favorable health effects? The only likely source of proof one way or the other is funding from the intellectual property owner, because no one else is apt to pay.
Having received funding from the NIH, the CDC, private foundations, and industry sources over the years, I can attest to personal bias in every instance, i.e., with every study, my team and I were “hoping” for a particular outcome. Why on earth would we even bother to do a study (they are rather tedious to design, fund, and run) if we didn't care about the outcome? The answer is, we wouldn't; no one would. As soon as one does care about the outcome, bias is inescapable.
But that is biased preference. Biased methods, in contrast, are avoidable, or nearly so. Methods, then, and safeguards against bias in the study protocol, are the more robust measure of a study's merits than the funding source, or the native preferences of funder or investigators.
We actually have a very good template in place for the research equivalent of having our cake and eating it too, i.e., allowing for industry funding and protecting against systematic distortion. That template is the very matter with which we began: funding of drug studies. The for-profit pharmaceutical companies pay these (often colossal) bills, but the FDA standardizes the criteria. A variation on that same theme could work for all industry funded research.
My initial, fairly obvious standards for industry-funded research include:
1) the customary defenses against bias and confounding, whenever practicable: randomization; double blinding; concomitant control
2) contractual protection of the right to publish, independent of study outcomes
3) complete research team autonomy with regard to study management
4) complete research team autonomy with regard to data analysis, interpretation, and manuscript preparation
A final consideration might be disclosure at the time of publication regarding satisfaction of these criteria. Funding sources are already disclosed routinely. We might add, without much fuss: “This study does/does not satisfy Standard Criteria for Research at Low Risk for Undue Influence by the Funding Source.” The papers that did should be more trusted, and indeed published preferentially, as compared to those that did not. The same check list could be applied to all entries at clinicaltrials.gov.
To the best of my knowledge, every research contract at my lab over the past 20 years has met the criteria above. As a result, we certainly have published industry-funded studies with negative results the funder did not love.
Absent industry funding, research in pharmacotherapeutic advances would be set back decades. I think few propounding the perils of industry funding, of which there clearly are, pause to consider what its abolition would mean to the offerings of any modern pharmacy; they would be all but annihilated. One might even argue that the problem is less industry funding, and more the lack of a level playing field. The pockets of some funders, namely those who deal routinely in patents, are much deeper than those of others, and this in turn produces a distinct lack of parity in the capacity to generate evidence. When evidence-based medicine becomes revenue-based medicine, we all have a problem.
That problem will not be solved by summary judgments, dismissive of such subtleties. We have in our culture, arguably, killed expertise. We do the same to dialogue at our peril.
Use of the word “conflict“ has become an almost reflexive disparagement of industry funded research; a meme, of sorts, invoked presumably to highlight one's own virtue. Those addressing the matter thoughtfully, however, have begun to note that such funding may portend confluence rather than conflict, and the distinction should be made.
That distinction can be all but self-evident. Coca-Cola, clearly, should not be subsidizing our understanding of “energy balance.” Trade organizations representing nuts, or produce, or oils funding legitimate examinations of specific health effects of their products, or wearable health tech companies funding studies of their proprietary offerings, are another matter. But for the confluence of their interest and funding, research in such areas may simply lag behind.
Propriety can lead to credible funders, and clear criteria for research findings we can trust. Sanctimony leads reliably only toward the domain of history's greatest blunders.
It is also, predictably, parent to hypocrisy when the thing you assert with self-righteous zeal is wrong, becomes the very thing you need. I have yet to meet anyone with acute ureteral colic, or pancreatitis, holding forth in the Emergency Department about who funded the trials of the drugs they need. Their message tends to be much more primal: Help me!
We generally can do so, and industry funded research is much of the reason. In a world with no pharmacies, we would look on at such misery rather helplessly. We should take steps to foreclose on true conflicts in research. But we should allow for confluence, too, and keep the pharmacies open.
This post by John H. Schumann, MD, FACP, originally appeared at GlassHospital. Dr. Schumann is a general internist. His blog, GlassHospital, seeks to bring transparency to medical practice and to improve the patient experience.