There’s Still Too Much Uncertainty About Conflict of Interest Bias

Image credit

Cartoon by Hilda Bastian, CC BY-NC-ND 4.0

A paper published this month got me frustrated, yet again, about this situation: There really is still too much uncertainty about conflict of interest bias in clinical research.

The paper is by Philip Kranz and colleagues from the German health technology agency, IQWiG. It might not be instantly obvious why this had me looking for research on COI, but I’ll explain.

The authors use a drug for a rare heart condition to demonstrate some flow-on effects of drugs having extremely high prices. The drug is for a rare heart condition. It’s called tafamidis, and it costs over a quarter of a million dollars a year in the US. They write that the manufacturers have not been providing the drug for clinical trials of potential competitor drugs. As a result, new drugs for the condition are either being tested against placebo in countries where tafamidis is not available or reimbursed—or in countries where healthcare systems reimburse it, then the system is effectively bearing huge commercial costs for the development of these new drugs. Ethics require that people in comparison groups in randomized trials get the best available treatment.

Now, the IQWiG authors point out, agencies like theirs can’t know which of these drugs is better for patients, because there simply isn’t adequate head-to-head data. The cost of running useful, ethical trials has become astronomical. For example, in Germany, the authors estimate buying tafamidis for a medium-sized trial would add €40 million to the study’s costs each year. Thus, they write, “high drug prices incentivize trial designs that prioritize cost containment over clinical relevance. Beyond unequal treatment access, high drug prices affect ethical and meaningful research by undermining clinical trial design.”

Trial sponsors, Kranz and co argue, should ensure that participants in control groups have access to the currently best available treatment. And secondly, the price of patented high-price drugs should be capped for research purposes.

This is where my thoughts turned to the COI question. Some people aim to discredit trials if manufacturers have provided access to medications. Up to the last time I looked for evidence, though, there wasn’t a strong research basis to support the claim that this biases research.

A review published in 2022 by Scott Graham and colleagues examined this, among other issues. They wanted to see if there was evidence to support policy decisions about the amount or type of industry sponsorship that constitutes a risky conflict. Both those questions are important because people set thresholds of minor or major amounts of sponsorship that could bias researchers. And the type of interest is thorny too—does it mean just complete sponsorship of research? An author that is (or was) an employee of a manufacturer with an interest? An author that has accepted speaking or consultancy fees from a manufacturer?

Graham and co found no studies that adequately assess the impact of different magnitudes of sponsorship, and very little on the type. There was only one study assessing provision of medication, and it didn’t show an effect—not that a single small study could settle the question anyway. The Cochrane review that is usually cited as evidence of industry sponsorship bias was last updated in 2017, and it has no precise definition of sponsorship, or granular assessment of amount or type. I hunted a bit, and as near as I can see, the 2022 Graham review is still the best evidence we have.

Why does this frustrate me so? I think the arguments for not using public research funding to pay for drugs in trials are stronger than the arguments against, and I don’t consider provision of medications alone compromises publicly funded trials. Reading the new Franz paper—good grief!—strengthened my belief. But I’d really like to know whether or not it biases research results. And the same goes for the other unanswered questions about types of potential conflicts.

Without a strong evidence base we’re left to muddle through, without much objective guidance on fairness. In 2024, Scott Graham and others published another paper, discussing the ways that the lack of research has blocked progress on addressing COI in research, especially in comparison to addressing COI in guideline development and medical education. They argue, in part, that the individualized focus of research on COI in research is too limited an approach, given that “the structural conditions of financial influence persist regardless of any individual’s COIs and regardless of any individual’s COIs’ biasing effects.” They point to the advantages of considering other frameworks, such as the work being done on commercial determinants of health.

I agree we need a lot more than data with an individualist focus, but I also hope we’re not stuck in the “no evidence” zone on those questions permanently. The status quo leaves the concept of COI infinitely malleable for those who seek to discredit research results they don’t agree with.

All my Absolutely Maybe posts on conflicts of interest here.

You can keep up with my work at my newsletter, Living With Evidence. And I’m active on Mastodon: @hildabast@mastodon.online and less so on BlueSky (hildabast.bsky.social).

~~~~

Disclosures: I sit on two committees advising on conflicts of interest (Australia’s National Health and Medical Research Council Dietary Guidance Governance Committee and Gender Guidelines Governance Committee). From 2004 to 2011 I worked at IQWiG in Germany, and was a colleague and friend of two of the authors of the IQWiG paper I discuss in this post.

The cartoon is my own (CC BY-NC-ND license). (More cartoons at Statistically Funny.)