A Typology of Nonfinancial Conflict in Population Health Research

There is considerable evidence that opportunities for financial gain create conflicts of interest in research in general and health-related research in particular. Organizations that fund, administer, and publish research have done substantial work to minimize such conflicts. As a result, declarations of potential financial conflicts of interest have become a regular feature of investigators’ lives. But considerably less attention has been accorded to the role that nonfinancial conflicts can play in influencing scientific research, despite evidence that more than half of all conflicts of interest are nonfinancial conflicts¹ (although there is no agreed-on definition of nonfinancial conflicts).

Aiming to fill some of the gaps between the prevalence and recognition of nonfinancial conflict, I offer a typology of its manifestations. First, I define potential conflicts as situations in which there is a conflict between investigators’ scientific and nonscientific interests, jeopardizing their ability to carry out the best possible population health science. This definition is based on previous writing around understanding financial and nonfinancial conflicts of interest² but expands on this literature to recognize that such conflicts can arise from a broad range of factors. Second, I consider the best possible science as science that aims to advance our understanding free from unrecognized implicit or explicit bias that can influence the design, conduct, or interpretation of a study.

Third, I do not address the question of whether the conflict between advancing understanding and the multiple causes of systematic bias represents “conflict of interest,” leaving that to other authors.³ I also do not necessarily suggest that these “conflicts” are avoidable or that one should assume that the presence of such conflicts should lead us to discount the science.⁴ Fourth, I do not offer solutions; that would require a substantial discussion in the field about ways to grapple with these conflicts. My goal here is more modest. Insofar as we accept that conflicts of several kinds may pose problems, I offer a typology of nonfinancial conflict. Although I suspect that there may be other types of nonfinancial conflicts of concern to population health science, I articulate three types of conflict—career interest, network-based, and ideological conflicts—as a first-pass effort to inform an evolving conversation.

CAREER INTEREST CONFLICTS

Conflicts may arise when an investigator’s career interests are in conflict with the demands of the best possible science. Although the vast majority of researchers aspire to produce the best possible scholarship, they do so within a complex system of rewards and incentives. Researchers typically work for universities that decide whether to promote them according to mainly their success in publication and in attracting extramural funding. This funding comes from organizations—public, nonprofit, and commercial—that have priorities. Journals that publish research results have particular interests and often editorial preferences. Researchers seek rewards that are status enhancing (e.g., promotion) and also financial (e.g., raises). Anyone who works within the health research system learns that having good ideas may be necessary but is far from sufficient for publication and funding success.

These facts of life create potential conflict between the current dominant systems of rewarding research and the definition of best possible science offered here. The well-documented bias toward publication of positive research findings exemplifies this bias. Because career interest conflict is a rational response to the reward structure, it incentivizes scientific work that is heavily influenced by decisions made by a handful of dominant funders and journals.

NETWORK-BASED CONFLICTS

Conflict may occur when networks of investigators bias the funding and publication of studies of particular questions involving particular methods of investigation and interpretation at the expense of the best possible science. And networks can, not infrequently, become cliques. This form of conflict, although linked to and reinforced by career interest conflicts, stands alone. The disciplinary structure that is an important socializing force in academic life may contribute to network-based conflicts. Several branches of epidemiology—nutritional epidemiology, social epidemiology, and life-course epidemiology, for example—assume that particular aspects of life (the food we eat, our social experiences, or exposures throughout the life course) affect one’s health status. After all, there would be no need for social epidemiology if social factors were not causes of health. By contrast, we have no astronomical epidemiology, presumably because we do not think that the positions of the planets and stars are important drivers of population health.

Nutritional epidemiologists can advance their career through publishing articles that promote the interests of nutritional epidemiology. They survive and advance in their career by publishing work that reinforces what is normative within their discipline. This creates clear network-based potential conflict, with researchers having an incentive to design, conduct, and publish work that reinforces what is network normative and being less likely to publish work that does not do so. By way of illustration, researchers sitting on either side of the controversy around the role of salt intake in population health are 50% more likely to cite articles that agree with their view than articles that are in disagreement.⁵

IDEOLOGICAL CONFLICTS

Ideological conflicts arise when researchers’ convictions about political, economic, and even spiritual life influence their academic work. We all bring to our work biases and beliefs, and population health science inevitably addresses matters about which researchers have strongly held values and perspectives. Consider an article that showed that countries with democratic political systems are more likely to have positive health indicators than those without such systems.⁶ An astute commentator asked whether the British Medical Journal would have accepted the article “if health had been positively associated with political repression.”⁷ The answer, unfortunately, is probably not. Is it likely that the researchers thought to consider the question expecting that democracy harms health? And would the journal have given the article the same level of scrutiny if it had suggested this? Similar examples can be drawn from the literature on such contentious issues as the consequences of personally possessed firearms and the efficacy of prayer.

Ideological conflict may seem to pertain in particular to researchers who have been explicit about their biases in polemical articles. For example, researchers who have published opinion pieces advocating for or against the passage of laws restricting access to firearms are a potential target for claims that their empirical work is then conflicted. One might argue the opposite, however: that researchers who have not clearly articulated their biases face the most ideological conflict and could benefit from clearly articulating this conflict to help guide interpretations of their work.

IN SUMMARY

Nonfinancial conflicts pose particular challenges for population health science. Although financial sources—and their attendant potential conflicts—are easy to enumerate, it is much more difficult for investigators to articulate biases that are reinforced by their desire for professional advancement, membership in disciplinary networks, or ideology. Would articulating these biases improve science? Is it appropriate to even call these forces “conflicts,” or is some other label that acknowledges potential biases better in this context? These seem to me open questions. Identifying these challenges may be a first step, and I look forward to learning the views of colleagues who agree or disagree with this typology and, most important, to discussing how—if at all—we should tackle such issues in population health science.