There has been significant public debate about the susceptibility of research to biases of various kinds. The dialogue has extended to the peer-reviewed literature, scientific conferences, the mass media, government advisory bodies, and beyond. Whereas biases can come from myriad sources, the overwhelming focus of the discussion to date has been on industry-funded science. Given the critical role that industry has played and will continue to play in the research process, the International Life Sciences Institute (ILSI) North America Working Group on Guiding Principles has, in this article, proposed conflict-of-interest guidelines regarding industry funding to protect the integrity and credibility of the scientific record, particularly with respect to health, nutrition, and food-safety science. Eight principles are enumerated, which specify the ground rules for industry-sponsored research. This article, which issues a challenge to the broader scientific community to address all bias issues, is only a first step; the document is intended to be dynamic, prompting ongoing discussion and refinement. In the conduct of public/private research relationships, all relevant parties shall 1) conduct or sponsor research that is factual, transparent, and designed objectively, and, according to accepted principles of scientific inquiry, the research design will generate an appropriately phrased hypothesis and the research will answer the appropriate questions, rather than favor a particular outcome; 2) require control of both study design and research itself to remain with scientific investigators; 3) not offer or accept remuneration geared to the outcome of a research project; 4) ensure, before the commencement of studies, that there is a written agreement that the investigative team has the freedom and obligation to attempt to publish the findings within some specified time frame; 5) require, in publications and conference presentations, full signed disclosure of all financial interests; 6) not participate in undisclosed paid authorship arrangements in industry-sponsored publications or presentations; 7) guarantee accessibility to all data and control of statistical analysis by investigators and appropriate auditors/reviewers; 8) require that academic researchers, when they work in contract research organizations (CRO) or act as contract researchers, make clear statements of their affiliation; and require that such researchers publish only under the auspices of the CRO.
It has been said that “scientific ‘truth’ is the primary aim that all should pursue in the jungle of academic-industry interactions” (1). The point of scientific endeavor, in the first place, is and should be, the pursuit of truth—nothing more, nothing less—irrespective of financial or other interactions. It goes without saying that seekers of truth must not impose preconceptions on the method or result of their search: they must not have ulterior motives. Throughout modern history, scientists have been guided by rules that ensure the integrity of the pursuit of truth, rules that continue to evolve as the research and communication landscapes change. The purpose of this article is to articulate, in the sophisticated, industrialized, modern world in which we find ourselves, principles defining and protecting the integrity and maintaining the credibility of the scientific record, particularly that part of it devoted to health, nutrition, and food-safety science.
The agricultural, food, and nutrition sciences have come to be a crucial part of evolving health research, which, in turn, plays an ever-growing role in improving the human condition. Although regarded as important determinants of human health, agricultural practices, food processing and safety, and nutritional status do not receive the same attention and funding from the federal research agencies as biomedical research does. Federal funds allotted to agricultural, food, and nutrition research amount to ≈$1.8 billion annually (out of a total US Department of Agriculture research budget of $2.3 billion),
with most of this focusing on agricultural production; in contrast, $28.6 billion is appropriated to the National Institutes of Health (2). Industry-funded research projects, large and small, account for a large proportion of all food science and nutrition research (3–5), both for obvious and nonobvious reasons.† United States' law places the responsibility for product safety and for the truthfulness of label claims on the manufacturer. Clearly, it is in the food industry's interest to conduct the research necessary to meet the legal requirements as well as to improve food-product healthfulness, safety, accessibility, taste, cost, attractiveness, etc. Most of this research falls outside the mission of traditional federal funding agencies and would not be done without food industry support. Pursuant to an extensive web of laws and regulatory requirements concerning food and food ingredients that have evolved over the past century, industry scientists and academic researchers who work with industry strive to enhance food quality, studying everything from the safety of ingredients to the evidence in support of health claims that appear on food packaging.
The rationale for food industry funding of research may be less obvious in areas such as microbiology (6),‡ toxicology (7–9),§ nutrient bioavailability (10, 11), and fortification (12)—all of which lead to enhancement of human health and to research on animal breeding and agricultural efficiency, which helps to feed more people. Some such research will be conducted by industry, in-house, whereas other projects will be contracted out to academic institutions or government or contract research laboratories. Scientists, especially novice researchers, conducting investigations in any of these settings need principles on which to rely while conducting their research ethically and with integrity. Clearly, it is essential to preserve the integrity and credibility of food and nutrition science for the benefit of public health and understanding.
In recent years, a growing body of literature has evolved on the subject of conflicts of interest and their potential influence on the integrity of researchers and the scientific record. In these discussions, conflicts are typically treated as disqualifying factors in scientific papers and research; that is, scientists with conflicts of interest are viewed in the literature as being at least partially integrity-compromised, and, even with complete and open disclosure, are regarded, at least to some extent, as of suspect scientific credibility. It is hoped that this article will define and clarify the highly complex issues involved in questions of conflict and scientific bias, particularly with regard to the portion of research funding that originates from the food industry.
In the interest of beginning this crucial dialogue in a sharply defined and dispassionate manner, the focus of this article will be limited to only one very specific issue and its relation to bias: financial conflicts of interest, specifically funding-based conflicts. It must be pointed out that there is a potential for all funding, regardless of source (eg, public, private, government, or industry), to bias behavior, unconsciously or otherwise. The focus of the current article will be on the management of potential bias from industry funding of science. Our goal is to separate monetary considerations from the science—including research design, execution, reporting, publishing, and other factors.
From its beginning, the food industry has concerned itself with researching food products and ingredients from the perspective of safe and efficient delivery of food to a rapidly expanding population. Before World War II, the overwhelming bulk of food research was funded and carried out by food-industry scientists; there has been little public funding of food safety and nutrition research. It was the evolution of American society from the laissez faire environment that existed during the industrial revolution to the complex public/private sector mixed economy of the more recent past that transformed research funding and higher education in general.
Although the food industry first entered the era of managing financial conflicts in the late 18th century, with the development of proprietary technologies to enhance food preservation and safety, the post-World War II period saw an exponential increase in the administrative challenges of research funding. For example, the number of patents awarded to universities or academic researchers increased by a factor of 10 in the past 2 decades of the last century (13). Similarly, federal funding of research increased from $405 million to $1.7 billion in a single decade (1960–1970) after the launch of the space race between the United States and the Soviet Union (14).
In the decades after World War II, in addition to the significant increases in government funding of university research, the United States experienced, in general, rapid evolution of science and technology, transformation and consolidation of agricultural production, and the steady growth of industry, especially those companies involved in public health, eg, medical/pharmaceutical, chemical, and food industries. In late 1980, the US Congress passed the Bayh-Dole Act, with the specific intention of stimulating the transfer of technology from government-funded university research to the private sector (15). This legislation has not been without controversy—both over issues concerning the diversion of university faculty from basic research and conflict of interest concerns due to the resulting university-industry partnerships.
The research community and individuals involved in health communications and public policy advocacy became increasingly concerned about the possibility that exogenous interests might influence published results of scientific research (16, 17).ll By late 2000, this concern had become heightened around medical/pharmaceutical practice: a number of articles appeared in the major medical journals (18, 19) that explored the financial relations of the pharmaceutical industry and physicians and their possible effect on physicians' decisions about patient treatment, researchers' decisions concerning study design, companies' interference in publication, and public health policy in general. Medical and other scientific journals began establishing rules for disclosure of financial conflicts in an attempt to manage them.
In succeeding years, concern broadened to include other industries, more recently the food industry, with authorities questioning how financial conflicts might impinge on the outcomes of health, nutrition, and food-safety research. It was generally acknowledged that the issue was complex and not susceptible to narrow or inflexible remedies, but that has not deterred some groups from concluding that industry-funded science is inherently biased (20, 21). These groups demanded that all industry-funded research, whether conducted at contract research facilities or at universities, be denied consideration in the formulation of public policy and that scientists who have conducted industry-funded research be barred from serving on public policy advisory committees (22). It is this article's contention that such efforts are helpful neither to the public nor to the scientific community. Industry funding, although a major component of the scientific landscape, is only one piece of an extremely complex research environment. The twin issues of financial conflict and bias demand a more reasoned approach and skillful management.
DEFINING THE ISSUE
First, conflicts of interest are not inherently determinants of bias. Even a massive multiplicity of conflicts, in and of itself, carries with it no certainty of bias. Although many definitions exist for conflict of interest and bias, the simplest of definitions suffice.
Conflict of interest
“A conflict of interest is ‘a conflict between the private interests and the official responsibilities of a person in a position of trust.' A conflict of interest thus arises when a person has to play one set of interests against another” (23).
Per the online Oxford English Dictionary, bias is an “inclination or prejudice in favour of a particular person, thing, or viewpoint” (24). “A cognitive bias is something that our minds commonly do to distort our own view of reality” (25).
Or, more rigorously, bias is a deviation of either inferences or results from the truth, or any process leading to that kind of systematic deviation. This includes tendencies by which data are reviewed or analyzed or interpreted or published in a way that yields conclusions that deviate systematically from the truth (26, 27).¶
For example, for researchers, a conflict might describe a situation in which a funder has offered financial incentives for research and hopes for a particular research result; it might also describe a situation in which the researcher, for philosophical, religious, or professional reasons, wishes to achieve a certain result. Neither situation necessarily results in a biased result, which would depend on a measurable deviation of research results from “the truth,” although much of the literature regrettably confounds bias and conflict. For that matter, much of the literature confuses conflict with a particular kind of conflict—financial. Unfortunately, even if all conflicts were banished forever, there would still be myriad sources of bias.
For example, the following well-known forms of scientific and publication bias exist (28):# sample-selection bias, sample-size bias, data-collection bias, data-quality bias, statistical-analysis bias (29), confounding-variable bias, and publication bias (30). These are just a few of the more commonly encountered pitfalls leading to skewed research conclusions, but these scientific sources of bias may be easier to identify than other cognitive and emotional causes that have nothing to do with the formal research process. Consider the following possible sources of bias: one's previous body of work; one's desire for fame and respect among peers (or, alternatively, the desire to achieve iconoclastic stature); religious bias; ethical or values-based bias; philosophical bias; political bias; one's nationality or ethnicity; pressure to publish (31);** pressure to win prizes; fear of losing one's job or position; highly personal matters, such as one's physical or mental health issues or one's family's health; the pernicious effect of pack behavior or “group think” facilitated by social or professional networks, either in the physical world or in cyberspace (blogs, websites, chat rooms, list serves, and other communication tools of the Internet); financial or funding bias resulting from all kinds of financial incentives, including gratuities, bribes, grants, free trips, gifts, and cash prizes; and the desire to please one's source of funding, either unconsciously or deliberately.
The multiplicity and variety of sources of bias in research and in public health communications generally are extensive, complex, and yet of major importance to scientific research, the integrity of individual study, and the body of scientific literature as a whole. Strategies must be developed to address and manage all sources of bias, whether technical, statistical, cognitive, or emotional in origin. These are critically necessary, not just for the scientific community, but also for the well-being of the public. The interpretation of health research and the promotion of public policies resting on that research are far too important to be based on formulas that would address conflicts at the price of excluding the input of a large proportion of food-safety and nutrition scientists.
EXISTING CHECKS ON BIAS
As far as scientific research and communications are concerned, several checks exist to ensure adherence to good practice and to avoid biased conclusions. Of course, replication and coherence of scientific findings are the major mechanism by which bias in research is controlled. This section is intended to summarize postresearch control mechanisms. First and foremost is the system of scientific peer review that is built not only into publication in scientific journals, but also into the promotion and tenure decisions for individual faculty conducting research at colleges and universities. Governance and review processes of academe exercise oversight, particularly on industry-funded research projects. Charges of irregularities, errors, and outright scientific fraud are usually investigated by the academic institutions where the research is conducted. However, in one recent noteworthy case, a distinguished nutrition researcher resigned his university position 9 y after initial charges of fraud were filed in connection with his infant-formula study. In the university's subsequent report, the authors recommended that the government monitor scientific misconduct through a new national agency “charged with all aspects of science, irrespective of funding sources, public or industry [emphasis added]” (32).
Most importantly, peer pressure serves as a check on bias, ie, the peer pressure of meetings, conferences, e-mail listservs, and discussion boards run by scientific colleagues and, especially, the process of peer review, particularly relied on by the thousands of scientific journals around the world and other organizations (33).†† For more than a century, peer review has served to provide a rigorous framework by which research papers and articles can be evaluated before their general dissemination—although not foolproof, scientists regard the process as a reliable safeguard against errors, biases, and scientific misconduct. However, in recent months, a robust debate has been generated about peer review and whether it needs to be refined (34, 35). Donald Kennedy, the former Editor-in-Chief of Science for the Journal of the American Association for the Advancement of Science (AAAS), has offered an eloquent defense of the current peer review process as “a fair system of evaluating and publishing scientific work—one that offers high confidence in, though not an absolute guarantee of, the quality of the product” (36).
If all of these checks fail, a governmental oversight structure exists within the granting agencies. For example, the Office of Research Integrity in the Department of Health and Human Services sets policies for government research grants, establishes reporting standards, and investigates misconduct (37). National and local volunteer health organizations review health science as it unfolds. Finally, the following checks on bias exist: science writers and journalists, who attend scientific conferences, digest new studies, and communicate them to the public; science associations, such as the National Science Foundation and the National Academies of Science, which regularly review new research and publish articles that are, in turn, read and commented on by member scientists; Congressional hearings reveal and publicize the real or perceived biases arising from too-close relations between industry and academia; and, ultimately, public disgrace occurs when research is revealed as deeply flawed.
In any case, given the increasingly broad and complex nature of scientific research and communications, additional recommendations are appropriate for managing the extremely complex issues of financial conflicts and potential bias.
PROPOSED GUIDELINES ON INDUSTRY FUNDING OF RESEARCH
Although funding, whether through the private or public sector, does not automatically introduce bias into scientific research, it is nonetheless prudent to address both the possibility of bias and the perception of it through explicit guidelines. On the basis of work commissioned by the ILSI North America Working Group on Guiding Principles, a series of proposals was developed to manage potential biases resulting from conflicts of interest between research investigators and companies wishing to fund their work.
It is our view that disclosure is an essential, but no longer a sufficient, measure to safeguard research from undue influence exerted by funding organizations. Managing conflicts, case by case, is the requisite step, ie, procedures need to be established, such as the following guidelines, to ensure research integrity. This should apply across the array of mechanisms through which research is funded currently: in intramural industry and government laboratories; in sponsored grants and contracts; and in cooperative agreements, Cooperative Research and Development Agreements (CRADAs), and “platforms” funded jointly by governments and industry, as is the case in the European Union and Australia. Whereas there may be a multitude of mechanisms by which research is funded, designed, conducted, and communicated, these guidelines should be adhered to by all parties, in all respects, in the spirit of openness and honesty that are the aim of this article (see the footnote to guideline 2 below).
It is also our view that industry participation in the effort to disclose and manage financial conflicts of interest is crucial. Future university-level science students will find their way into either private-sector research occupations or public-sector careers. All need a set of principles to guide their interaction with funding organizations, whether public or private, just as those organizations need principles to guide them in their interactions with academic scientists. Consequently, we propose the following guidelines to serve as a checklist to achieving unbiased research results from industry-funded activities—just as they might be useful guidance in public- or foundation-funded projects (38).‡‡
In the conduct of public/private research relations, all relevant parties shall:
1) Conduct or sponsor research that is factual, transparent, and designed objectively, and, according to accepted principles of scientific inquiry, the research design will generate an appropriately phrased hypothesis and the research will answer the appropriate questions, rather than favor a particular outcome;
2) Require control of both study design and research itself to remain with scientific investigators;§§
3) Not offer or accept remuneration geared to the outcome of a research project;
4) Ensure, before the commencement of studies, that there is a written agreement that the investigative team has the freedom and obligation to attempt to publish the findings within some specified time frame;llll
5) Require, in publications and conference presentations, full signed disclosure of all financial interests;
6) Not participate in undisclosed paid authorship arrangements in industry-sponsored publications or presentations;
7) Guarantee accessibility to all data and control of statistical analysis by investigators and appropriate auditors/reviewers;¶¶
8) Require that academic researchers, when they work in contract research organizations (CRO) or act as contract researchers, make clear statements of their affiliation; and require that such researchers publish only under the auspices of the CRO.
IMPORT AND IMPLICATIONS OF THE GUIDELINES
Obviously, guidelines are just … guidelines. They are not law, but if the research community embraces them, or even embraces their spirit, we believe there will be a profoundly beneficial effect on the quality and integrity of research that will encourage responsible oversight and stewardship of scientific research by all funding organizations. Following the guidelines will undoubtedly lead to closer and more open communication between funding bodies and researchers, resulting in a new spirit of collaboration. Still, it must be stressed that each organization wishing to adopt these guidelines needs to develop its own quality-control mechanism to ensure good compliance.
A strong peer-review system coupled with open declarations of research sponsorship in all scientific communications is a mandatory prerequisite for these guidelines to be effective. The second prerequisite is that university and industry policies be promulgated to address the issues raised in these guidelines regarding control of the design and conduct of the research and its publication. It is the responsibility of both the funding entity and the researchers being funded to adhere to the guidelines; existing oversight structures are also encouraged to endorse and adhere to them. Furthermore, it should be understood that failure to embrace the guidelines will raise serious questions about any research project so conducted.
It has been suggested that, in the past, industry-funded research may have had a bias toward results favored by the food industry (21, 43). The authors of one publicized study (4) who reached this conclusion proposed several explanations: 1) food industry companies may wish to demonstrate the superiority of their products to those of their competitors, 2) investigators are influenced by their funding when formulating their research design and/or hypotheses, 3) industry sponsors of research may suppress unfavorable results, 4) authors of scientific reviews may deliberately bias their searches and interpretations to the benefit of their industry funders, and 5) scientific reviews may disproportionately represent studies “arising from industry-supported scientific symposia.” Such criticism overlooks the fact that most university research is basic in nature and that companies frequently enter into research agreements with university faculty at a point at which preliminary experiments (whether conducted in the faculty member's laboratory or in the company's laboratory) have established the proof of concept and, therefore, the likelihood that the research will have positive results is enhanced.
Notwithstanding the obvious observation that scientific reviews conducted by nonindustry-supported authors are also subject to many potential biases, the 8 principles articulated in this article address all of these possible sources of skewed research. Indeed, if these principles are vigorously adopted as the guidelines they are intended to be, there would be virtually no reason to quarrel with a research conclusion except to dispute the science itself.
In fact, the 8 principles articulated herein are intended to provide a clear statement of responsibility on all sides—those that are funding activities as well as those being funded—when academic institutions or academicians are recipients of industry funding for research, publication, or presentation. The principles are intended to offer guidance for the food industry and academic researchers who work with industry, when industry-funded research projects are involved. They may be thought of as a checklist to help ensure insulation of any research project from the provision of the resources enabling the project.
Finally, the guidelines are offered as only a first step in creating a firewall against bias in research: this article is intended to be a dynamic document, prompting ongoing discussion and refinement of the guidelines it presents.
A CHALLENGE TO THE BROADER COMMUNITY
The objectives outlined above may be worthy, though not easy to achieve. However, these principles can also serve as an invitation to the broader scientific, science communications, and public policy communities to embrace similar pledges to immunize their work against the myriad potential sources of bias—nonfinancial as well as financial conflicts. The present article was necessarily confined to one relatively small aspect of an extremely complex issue. However, future discussions could be much wider ranging and much more comprehensive if they embrace all sources of bias and expand the focus from the very narrow issue of potential bias due to financial conflicts of interest.
Consider the extensive list of biases touched on at the end of the section on definitions above: how constructive might it be for the broader scientific, communications, and public policy communities to adopt guidelines to ensure that their work is free from bias? For example, such guidelines might include pledges of transparency (eg, voluntary disclosure of all previous research, published articles, and policy positions that might influence present research, published articles, and policy positions), disclosure of sources of funding (both of the project at hand and overall funding), and disclosure of other potential biases (eg, philosophical, religious, ethical, or political orientation; intention to publish or otherwise garner public or political authority or power through publicity; and previously announced public positions that might be relevant to the work at hand).
Other researchers or groups that are not supported by the food industry (eg, nongovernmental organizations, foundations, and advocacy and consumer groups) might include in their public communications appropriate promises that their work, to the extent possible, is open and objective (not skewed to a particular conclusion or philosophical view) and is controlled by the researcher or cited authority (rather than by a hidden funder or interested party). The checklist provided in the section above on the guidelines' import and implications might prove helpful in designing similar guidelines for other groups.
It is important to state explicitly what this paper has excluded from consideration. Notwithstanding that all scientific research, whether funded by industry or not, should be subject to the same ethical rules, discussion of all of the following potential institutional sources of bias that can affect the integrity of the published scientific record was specifically excluded from this article: foundation-funded research, government-funded research, and work by academicians on advisory panels to industry, grant panels, government advisory panels, nongovernmental organization panels, and voluntarism on behalf of professional societies.
This is a short list of organizational work and funding situations that routinely pose profound challenges to the independence and integrity of scientific research—the list could certainly be lengthened. All of these potential sources of bias are outside and beyond the scope of this article, but it is suggested that future articles explore the ramifications of inappropriate influence of such organizational bias on research or public policy. It is strongly urged that future investigations into this area be sufficiently broad to include the many nonscientific and other institutions that routinely play a communications role in science-based public policy.
We could lament that this entire effort to manage conflicts of interest and to banish bias in science, is, alas, insufficient. It would be easy to complain that the financial and other pressures on research are too great to channel them neatly. Furthermore, some will argue that a mere set of guidelines cannot immunize science from error, misinterpretation, or deliberate miscalculation. We deliberately left aside, for the time, the matter of enforcement mechanisms for these or any guidelines, believing instead that achieving a consensus on best practices in managing conflicts must certainly come before establishing sanctions for failing to adhere to best practices. As professional scientific societies, industry groups, and other organizations that engage regularly with researchers adopt a common set of rules by which to manage these difficult issues, enforcement of guidelines will automatically become increasingly less problematic.
In the end, management of conflicts of interest, and, for that matter, management of scientific biases altogether, is a matter of consensus building, not enforcement. Should we indulge in more of the self-recriminations that have gone on for far too long or should we construct a workable start to a solution? The choice is obvious: it is time to act. The interpretation of health research and the promotion of public policies resting on that research are far too important for us not to address and manage the myriad potential biases that can intrude. Let this effort be a start.