Do conflicts of interest negate scientific consensus on GMO safety?

By Mark Lynas

November 3, 2016

In denying the existence of a scientific consensus about the safety of crop genetic engineering, anti-GMO activists often reject the weight of published scientific literature because, they claim, a large number of studies are compromised by industry conflicts of interest.

This narrative is a familiar part of the GMO debate. Activist groups such as US Right to Know have recently fired volleys of Freedom of Information requests at public sector biotechnology academics aiming to find evidence of industry collaboration. Such collusion, USRTK suggests, would invalidate their scientific research findings.

But how compromised by conflicts of interest is the overall scientific literature? Apparently not very, according to a fascinating new quantified analysis published in the journal Nature Biotechnology.

Miguel Sanchez

Chilean researcher Miguel Sanchez examined 698 peer-reviewed papers on the health or environmental safety of GMOs published between 1993 and 2014, and manually categorized funding sources disclosed by the authors.

Conflict of Interest (COI) was purposefully defined very broadly in the study. As Sanchez reports:

“I decided to analyze our set of papers for financial COIs those that arise when research is fully or partially funded by a party with a stake in the development of GM crops; and also for professional COIs those that arise when at least one author is affiliated with a company developing GM crops, even if the research is supported through public funding. To gain insight about COIs, individual reports were manually checked, making it possible to assess information regarding authors’ affiliation and funding sources.”

Even with this broad definition of COI, 58.3% of published papers “have no financial or professional COIs, as the authors were not affiliated with companies that develop GM crops and also declared that the funding sources did not come from those companies,” Sanchez reports. On the other hand, 25.8% did have COIs according to his definition, while 15.9% of papers although authored by scientists not affiliated with companies developing GM crops did not provide funding information.

Sanchez who discloses, appropriately enough, that he “is employed by ChileBio, which is funded by CropLife International and companies that develop GM crops” concludes that his analysis gives further support to the existence of a scientific consensus on GM crop safety.

“Overall, the analysis of all 698 reports collected here makes it clear that GM crops have been extensively evaluated for potential risks and that genetic modification technologies based on recombinant DNA do not carry a greater risk than other types of genetic modification. Claims either that there is not sufficient peer-reviewed literature evaluating GM food/feed safety issues or that COIs prevail in the published literature are not supported by this analysis.”

Although the purpose of Sanchez’s analysis was not to quantify the proportion of negative/positive conclusions in the literature on GMO health and environmental safety, he does include the following rather intriguing line: “It is worth noting that fewer than 5% of all reports published reported negative outcomes.”

In other words, the 698 papers he examined had an implicit 95% support for the existence of a scientific consensus on GMO safety. This is very close to the famous 2013 Cook et al. study on climate change, which found a 97% consensus in the literature on the existence of human-caused climate change.

It is also of particular interest to us at the Cornell Alliance for Science as we are currently carrying out an analysis of the literature to quantify the extent of any scientific consensus on GMO health issues. This review is being conducted with the assistance of John Cook and is employing broadly the same methodology. (Call out: We are taking a crowd-source approach to rating the papers we’re evaluating. If you have a couple of hours to help out, please sign up here.)

There are also some caveats to add to Miguel Sanchez’s useful study. The first that occurs to me is that he does not evaluate whether COIs go the other way; for example, if any researchers have funding from companies that market or promote organic food, or have some other financially-motivated reason to negatively present GMOs.

A recent controversial example might be Charles Benbrook, whose (now terminated) employment at Washington State University where he published several papers critical of GMOs and herbicides was largely funded by the organic industry. (According to the Supplementary Info table, Benbrook’s papers did not come up in the analysis.)

Another important thing to note is that the existence of a COI does not per se invalidate the research finding, however important it is that full transparency and disclosure standards are maintained. This is particularly the case because the burden of proof for product safety naturally falls on the developers of GM seeds, rather than being carried by the state. So industry-sponsored studies, which are typically then independently evaluated by regulators in different countries, are a crucial part of the process.

It should also be noted that many public sector academics are specifically required to work with industry, and doing so does not negate their work or mean that their conclusions should be rejected outright. As Kent Bradford at University of California, Davis (one of the targets of USRTK’s Freedom of Information emails attacks) wrote recently:

“As you know by working with us, our opinions on policy issues are based on facts and science, not on funding sources. We are not a sock-puppet or shill for anyone. We strongly reject USRTK s fundamental premise that public-private collaborations are conspiracies and that public researchers modify or bias data or opinions to suit the objectives of their funders.”

Bradford continued:

“Consider that concept for just a moment: Is it really believable that groups sponsoring research at the top agricultural university in the world expect to receive false or misleading information? Would you want to base the future success of your company or organization on false or biased research, and pay for the privilege of getting it? This is inherently absurd. These broad FOIA requests are an attack on science and on the trust that the public properly has in its academic institutions.”

This is the most important caveat of all, and I recommend bearing it in mind while evaluating Sanchez’s work. While it is a useful thing to know that close to 60% of the published literature has no conflicts of interest, it does not mean that the 25% with declared COIs should necessarily be rejected. Publication in a bona fide peer-reviewed journal (as opposed to pay-for-play) remains the gold standard across science, and we should not forget that.