The kind of thing that makes industry ‘science’ look bad.

In a post last week, I mentioned a set of standards put forward by Carol Henry (a consultant and former vice president for industry performance programs at the American Chemistry Council), who says they would improve the credibility of industry-funded research.
But why does industry-funded research have a credibility problem in the first place? Aren’t industry scientists (or academic scientists whose research is supported by money from industry) first and foremost scientists, committed to the project of building accurate and reliable knowledge about the world? As scientists, aren’t they just as hard-headed and devoted to objectivity — indeed, to truth — as the rest of their professional community?
I have no doubt that many industry (and industry-funded) scientists do take good knowledge-building as their most important job. And this means that some of those who depart from this commitment are making things harder for those scientists whose loyalties to their industry benefactors do not extend to misrepresenting the truth. Plus, of course, they may be misleading policy makers and the public by passing off as reliable scientific knowledge something that is not.
In the article “Tobacco Industry Influence on Science and Scientists in Germany,” [1] Thilo Grüning, Anna B. Gilmore, and Martin McKee draw on internal tobacco industry documents (released in 1998 as part of the settlement of litigation by the state of Minnesota against tobacco companies) to identify the strategies tobacco companies used to influence scientists and to distort science.

The tobacco industry strategy, it seems, was to fund the research of scientists who were not themselves employed by the tobacco companies but worked instead, say, as scientists conducting research at medical universities. The hope was that this research would counter the existing scientific findings that linked smoking to various diseases. Moreover, if these contrary findings were produced by independent scientists, rather than by scientists employed by tobacco companies, they would be more likely to be taken seriously as credible scientific findings.
Why wouldn’t findings generated by tobacco company employees be as credible? Because the companies have a clear interest in generating information that would help (rather than hurt) the image of the product that they sell. But the documents studied in the Grüning article indicate that this interest played a clear role in the recruitment and management of the independent scientists whose work was funded — to the extent that their “independence” was less than surface appearances would suggest.

Philip Morris documents outline the rationale: that research “should be ‘at arm’s length,’ in order to protect the Industry or individual Companies from litigation.” Reasons for sponsoring such research included “1. To secure scientists who could act as potential experts for Industry, 2. To secure goodwill support on critical issues, 3. To push scientific extremists into isolation, 4. To have work published which is suited to reestablish a balanced view in the scientific community, i.e., defuse critical issues.” Philip Morris also emphasized that “the professional quality of the research scientists,” as well as “the capabilities of the research institution” in which the commissioned work was to be carried out, were important for the “legal effectiveness,” “credibility,” and “bargaining value with authorities” of the industry’s sponsored research.

This was not a plan to recruit unbiased scientists to conduct research and report the results, whatever they might be, all the while hoping that these results would vindicate cigarettes. Rather, this was an effort to buy “scientific credibility” — or at least enough of it to sway policy makers and the public.
It is fascinating how important the appearance of independence was, given that the companies putting up the money for the research needed to be sure that they would be able to exercise tight control over the science produced. The Grüning article includes a detailed section on the various players and an org chart on how they are connected to each other. One of the central entities that keeps popping up is the Research Council on Smoking and Health (Forschungsrat Rauchen und Gesundheit, or “the Forschungsrat” for short), established by the VdC (a German tobacco trade organization) in 1975. Composed of leading German scientists (many of them department heads in “a relevant medical specialty”), the Forschungsrat apparently struck some tobacco company heads as too independent. The president of the VdC sought to reassure them:

As regards the many proposals you made in order to influence the projects in their details and to exclude scientists who may have been troublesome in the past,. . . We fear. . . that the concept of our Smoking and Health policy would be spoilt if we tried to limit the independence of the Forschungsrat too severely from the start. . . What we can do at this stage is to take safeguards. . . against uncontrolled publication of results that could give rise to misinterpretation. . . The presence of Herr Schlenker, former chairman of our Association [VdC], and Dr Schenzer, manager of our Association until recently, in the Forschungsrat will ensure effective industry representation in this body. . . Moreover, our Association has engaged an eminent young scientist [Franz Adlkofer] whose function it will be to supervise the execution of the research programme in close contact with the scientists and coordinate the work done by the Forschungsrat.

That supervision and coordination was key. Not only did the VdC decide (based on the Forschungsrat’s recommendations) which projects would be funded, but it also exerted significant control over which results from the funded projects would be published. And, Grüning et al. point out,

A number of additional steps were taken to ensure industry control of this “independent” research council. Only the chairman could make public statements on its behalf; members required his permission to do so. Rather than advertising for research proposals, selected scientists were to be informed about funding opportunities. Furthermore, the bylaws guaranteed the attendance of 2 VdC representatives as guests and of Adlkofer, director of the VdC’s Scientific Department, as an observer. The documents suggest that Adlkofer’s role extended beyond that of an observer. Attending every meeting, he wrote the minutes and was referred to — and referred to himself — as the Forschungsrat’s scientific secretary.

It’s worth noting that the Forschungsrat’s scientific secretary, Franz Adlkofer, had a long relationship with the VdC and that his own laboratory was directly funded by the VdC. That might be seen to undermine the independence of the Forschungsrat from the tobacco industry.
The heart of the Grüning et al. article is the analysis of the particular strategies the VdC used to control the “independent” research they were funding:

Suppression. Just as the VdC closed the German Industry Research Institute when its head published results unfavorable to the industry, so it aimed to suppress the dissemination of unfavorable results. Documents report how data on the co-carcinogenicity of nicotine was to be kept confidential. They also reveal that the VdC would “hide” some of its tumor studies and that Adlkofer, examining the effect of “sidestream smoke” (i.e., passive smoking) on animals, “guaranteed the results of the study will not be published [emphasis in original].” Conversely, the industry encouraged the publication of favorable findings.

To the extent that science is supposed to be directed at generating accurate answers to questions — as opposed to what you wish the answers to those questions were — such suppression of findings is problematic.

Dilution. The selective funding of research and the recruitment of scientists who had doubts about the adverse health effects of smoking, or whose previous work had found no links, led to the funding of research projects designed to find no association between smoking and disease. This probably caused dilution of genuine studies, introducing severe bias into the evidence base, especially when meta-analyses were later undertaken.

The scientists funded with VdC were not chosen strictly on the basis of their scientific skills, nor were they even a representative sample of the scientific community in Germany. Rather, they were a sample skewed in favor of those who had already expressed skepticism that smoking was detrimental to health. Given their starting assumptions, it would not be surprising if their research tended to show that smoking was not detrimental to health.
In the ongoing scientific discourse, the hope is that scientists can move past their own individual biases by engaging with other scientists (and their results) with different biases. Looking at all the findings together and trying to be aware of (and critical of) your own biases, you’re supposed to be able to move closer to the truth. But if one set of biases is overrepresented in the pool of scientific knowledge — not because it conforms better to the physical and biological effects but because the scientists who have these biases are more generously funded — how much harder does it become to correct for these biases?

Distraction. The industry selected and supported a large number of research projects that aimed to distract attention from smoking by investigating other potential causes of smoking-related diseases — so-called “confounder studies.”

On its face, funding research into other causes of cancers (for example) doesn’t seem like a bad thing. The information generated by such studies could point to environmental factors besides cigarettes that are bad for us. But none of this would render cigarettes harmless.
While I was reading this article, I was getting a “where have I seen this name before?” vibe. It turns out that Franz Adlkofer is an author on those two questionable papers on EMFs and DNA damage. If anyone has access to these papers and can figure out the funding sources for the research behind them, I’d be interested to know whether VERUM (the current incarnation of the Forschungsrat) is on that list.

Concealment. It seems that in order to increase the credibility and impact of the studies presented, whenever possible, favorable scientific results were presented and published by a “third party”–a scientist whose connection to the industry could be hidden, with the industry’s involvement often actively concealed.

Of course, if people know about industry involvement, the results don’t come across as independent. On the other hand, if industry is so heavily involved as to have articles written by industry scientists published under the names of other scientists, then the research in question isn’t independent.

Manipulation. Some articles and presentations were vetted by the industry before publication or presentation. A Philip Morris document reports, “The VdC has influenced Dr Schmähl and his group to speak out against a poor publication which is hurting the industry. . . . The VdC is also influencing publications which will be presented at the Fourth World Health Conference that deals with the cost to the economy due to smoking.”

I’m guessing here that the “poor publications” in question here were judged lacking not on the basis of the science behind them so much as on the basis of reporting results that made cigarettes look like health hazards.
These five strategies — suppression, dilution, distraction, concealment, and manipulation — make short work of Carol Henry’s standards for credible industry-funded research. In fact, it is precisely this kind of faux-independence and the distortion of science that can result that make Henry’s standards look like a necessary safeguard.
It may be tempting here to blame the scientists who accepted the tobacco research funding, casting them as willing accomplices to VdC’s public relations campaign. I think this would be a mistake. As Grüning et al. note,

[I]t is apparent that some scientists (especially those whose links to the industry were in the 1970s or perhaps even the 1980s) did not realize the implications of accepting this funding, or that they would be working within a system so tightly controlled by an industry that was assuring them–often falsely–that they would have full independence. This is illustrated by the letter of a scientist who did not receive a reply from the industry to his research proposal. Years later, he wrote to Adlkofer, “Retrospectively, I am even grateful to you, that you never came back to me. . . since through this I did not load my conscience. . . with the burden of accepting research funding that I would regret today [translated from the German].”

Scientists want to answer scientific questions. They want to conduct good research and to contribute to a solid body of knowledge. However, they operate in a world where it costs money to do research, and where that money sometimes comes with strings.
Captains of industry are not, by and large, scientists. They are interested in putting out a product, finding a market for that product, and making the shareholders happy. Science may strike them as one tool they can use to help them achieve these goals, but they are interested in science as a tool rather than an end in itself:

Finding the scientific truth was not the aim of the tobacco industry. Instead, it sought to manipulate and distort the evidence. The documents suggest it achieved this through the selective recruitment and funding of scientists and projects likely to produce favorable results, the suppression of unfavorable findings, the promotion of favorable findings, and the promulgation of alternative explanations for diseases associated with tobacco use. Importantly, major and often complex efforts were made to hide industry links at each stage of the process–from recruitment to publication. However, when RJR directly approached researchers, it attempted to reassure them of their “complete ‘freedom’ regarding the results and their publication,” something that other documents suggest was far from likely.
The evidence presented in this essay suggests that the industry introduced serious bias into published research that probably influenced scientific consensus and public opinion in Germany. This is likely to have increased the social acceptability of smoking, influenced the policy context, and undermined efforts to control tobacco use, just as the industry desired.

Science, used as a tool by industry to achieve other ends, sometimes results in a distortion of the science. In this case, that distortion of science ended up being a better tool to achieve tobacco industry ends than did science played by the rules that the scientists (including those funded by industry) recognized.
Avoiding this kind of distortion — establishing safeguards against it — is crucial if scientists working on industry’s dime want their work to be taken seriously by other scientists. Having utter transparency about funding sources and about the extent of industry influence would be a good first step.
_______
[1] Thilo Grüning, Anna B. Gilmore, and Martin McKee, “Tobacco Industry Influence on Science and Scientists in Germany,” American Journal of Public Health (January 2006) Vol 96, No. 1, 20-32.
DOI: 10.2105/AJPH.2004.061507