Requiring that industry funding of studies published in Epidemiology be declared could be justified on a duty ethics “right-to-know” basis or on results-based “do most good for most people” ethics. I agree with the duty ethics argument but want to discuss here the results that might flow from several types of declaration related to bias and conflicting interests.
I think that authors submitting papers should be asked to declare:
- all sources of funding,
- if the contract with the funder allowed suppression or forced alteration of the report (this acknowledgment should lead to a rejection of the work out of hand),
- how many previous or current contracts the author has had with this funder,
- if there were some sort of balanced advisory/steering committee, including “true believers” and “doubting Thomases” to oversee the study design, execution, and report,
- on the basis of the published literature, what the authors previous and posterior degree of certainty was that the alternative hypothesis was true (eg, virtually certain to be false, prone to disbelieve, close to the dividing line, prone to believe, etc), and
- that they will complete a standardized values questionnaire that allows readers to judge the authors' relative fears of false positives and negatives.
This last piece of information would not influence acceptance or rejection of the article but would be useful in subsequent meta-analyses to assess the role of world view as opposed to financial interests.
In my experience, a systematic bias in designing, analyzing, and reporting evidence rarely results from a conscious effort to present a conclusion that the author knows to be a lie. Rather, authors bring values systems that lead them to a particular balance of aversions to false-positive and false-negative results, and they may have reasons to strongly believe or disbelieve the truth of the alternative hypothesis. These factors have been shown to influence the interpretation of data1,2 and may lead authors to practices that seem deceptive to those who hold other views.
Powerful funders tend to recognize scientists who share their values and assessments, and funders can create ecological niches in which these scientists can flourish. The funders act in a systematic manner to see that “their” scientists are represented on government panels and other venues in which the funding and interpretation of research are influenced. Thus, funding can be a marker of inherent biases even if funding has not caused those biases by any crass conflict of interest.
All of us are influenced silently by interests and worldviews as we choose what to study and choose how we design, conduct, and analyze our studies and “spin” the results that we see. Wise scientists acknowledge this human proclivity and take a variety of measures to protect themselves and the scientific community from it. Inviting scientists with different proclivities to critique each stage of the research process is one good way to do this.
The existence of worldview, interests, or a particular previous degree of certainty about the hypothesis being studied represent risk factors for bias with a less-than-perfect sensitivity and specificity. This situation means that the predictive value positive is insufficient to definitely believe or disbelieve the results of any particular study. Judging by some recent studies of industry funding and bias,3 the added probability of supporting the funder's interests is something of the order of 20% points, although this difference statistically distinguishes the body of industry-funded studies from the body of studies with other kinds of funding. So that other than to satisfy a “right-to-know,” why should we ask that the aforementioned information be declared? I believe that such information would be useful as a variable on which to stratify in a meta-analysis and might lead to more self-knowledge.
Policy makers could look to see whether certain associations in well-conducted studies are found regardless of funding or worldview or only in nonindustry-funded studies or only in studies conducted by those with certain worldviews. Depending on the results, they may have to make a policy decision about which body of evidence to believe. One could see whether the source of funding predicted certain design flaws and “spinning” practices. Other benefits might be better industry contracts and procedures and more self-aware researchers.
There is a danger that conscientiously conducted studies declared to be funded by vested interests might be falsely denigrated and ignored. As more meta-analyses are performed and the positive predictive value of bias among industry-funded studies is better understood, the credence afforded such studies will become more rational.
ABOUT THE AUTHOR
RAYMOND RICHARD NEUTRA obtained his medical degree from McGill University in 1965 and his doctorate in public health from Harvard University in 1974. He taught epidemiology at Harvard and UCLA before joining the California Department of Health Services in 1980, where he is currently Chief, Division of Environmental and Occupational Disease Control. He has managed to steer clear of financial conflicts of interest and lawyers, weighs the penalties of false negatives and positives in each new situation, and is disinclined to demonize those who disagree with him. This comment does not necessarily reflect the position of his employer.
1. Slovic P, Finucane M, Peters E, et al. Risk as analysis and risk as feelings: some thoughts about affect, reason, risk and rationality. Risk Analysis
2. Weed DL. Underdetermination and incommensurability in contemporary epidemiology. Kennedy Inst Ethics J
3. Bekelman JE, Li Y, Gross CP. Scope and impact of financial conflict of interest in biomedical research: a systematic review. JAMA