1. Introduction

An increasing effort to commercialize research pervades science today, including university-based science. While some philosophers have started to identify some of the epistemological and ethical consequences of such a transition (Irzik Reference Irzik, Dorato and Suàrez2010; Biddle Reference Biddle, Carrier and Nordmann2011; Carrier Reference Carrier, Carrier and Nordmann2011; Fernández Pinto Reference Fernández Pinto2015; Jukola Reference Jukola2016), we still lack a proper understanding of the extent and depth of the effects of commercial interests for the production of scientific knowledge. Important contributions have been made regarding financial conflicts of interest generated by industrial ties to science (Elliott Reference Elliott2008; de Melo-Martín and Intemann Reference de Melo-Martín and Intemann2009; Resnik Reference Resnik2015). Similarly, case studies have clarified the ways in which industrial interests can corrupt the processes of scientific knowledge production and dissemination to the detriment of human and environmental health, as the cases of the tobacco industry (Proctor Reference Proctor2012), the petrochemical industry (Oreskes and Conway Reference Oreskes and Conway2010), and the pharmaceutical industry (Sismondo Reference Sismondo2009; Goldacre Reference Goldacre2012) illustrate. More recently, de Melo-Martín and Intemann (Reference de Melo-Martín and Intemann2018) have identified the commercialization of science as one of the main factors eroding the public trust of science, leading, for instance, to inappropriate dissent with negative consequences for policy making and society at large. Given de Melo-Martín and Intemann’s project of finding strategies to promote trust in science, identifying and understanding the mechanisms through which commercial interests erode such trust becomes especially relevant.

I examine the idea that commercialized science is a central factor in the erosion of trust in science. While concurring with the general claim that the current commercialization of research has contributed to increasing public mistrust in science, I argue that the relation between commercially driven research and trustworthiness is more complex. I claim that commercial interests have a particularly negative impact on the trustworthiness of science by leading the public to mistakenly trust or distrust scientific results. While this can be achieved in many ways, I identify two mechanisms through which this has recently occurred in clinical research. First, commercial interests can directly encourage the production of consensus or dissent that is favorable to the industry but not epistemically reliable. In order to show this, I make a distinction between the public perception of science as trustworthy and the actual trustworthiness of science. This distinction helps to clarify cases in which the public might perceive the science as trustworthy when the science is not actually so. A good example of this is the way the pharmaceutical industry positioned favorable results on antidepressant research in top medical journals and direct-to-consumer advertisement, while keeping unfavorable results in the dark, leading to a misperception that the results were favorable in the aggregate (Goldacre Reference Goldacre2012).

Second, commercial interests support a narrow understanding of trustworthiness that allows the industry to claim that their research is reliable when we have good reasons to believe it is not. In order to show this, I make another distinction between narrow and broad epistemic trustworthiness. Research might be epistemically trustworthy according to a narrow view but not from a broader perspective. For instance, many industry-driven clinical trials are considered methodologically as good as, or even better than, independently funded trials, although at the macrolevel one can detect an industry bias in the results of the former but no bias in the latter (Lexchin et al. Reference Lexchin, Bero, Djulbegovic and Clark2003). Accordingly, I claim that whether the science is considered epistemically trustworthy can be a matter of focus, narrow or broad, which in turn makes the conditions for epistemic trust more difficult to assess. I conclude by claiming that the relation between commercially driven research and trustworthiness needs a more fine-grained analysis.

2. The Relation between Science and Public Trust

Ideally, the relation between scientists and the public is one of trust. The general public depends on scientific knowledge to be appropriately produced, and scientists are trusted insofar as their efforts to produce knowledge are reliable. Society counts on the consistency of scientific practice, and scientists strive to warrant such consistency. According to this traditional view, as Scheman explains, “Science is a complex set of social practices governed by a complex set of norms that share little more than the aim precisely of ensuring that the results of scientific investigation will be trustworthy, meaning that those results will be credible not only to the scientists who came up with them but also to others who may not share their particular experiences, biases, perspectives, interests, and the like” (Reference Scheman2015, 212).Footnote ¹ To simplify, let us assume that science is a social practice ruled by a set of norms or standards that ensure the reliability or trustworthiness of the results. If this is so, the public would correctly trust scientific results when scientists follow the standardized processes of knowledge production and correctly mistrust the results when scientists fail to follow the process. Similarly, prima facie it would be a mistake to mistrust results that have followed the standard process, as well as to trust those that have failed to follow it.Footnote ²

This, of course, is a huge simplification. The general public might have other reasons (political, ethical, cultural, etc.) to distrust scientific results, even if research standards have been properly followed. For example, traditionally the inclusion of women as research subjects was seen as methodologically inappropriate. One important rationale behind this methodological standard was the fact that women have more hormonal variability, due to menstrual cycles, pregnancy, and menopause, than men, which made confounders harder to control for in women and made them more expensive test subjects (Liu and DiPietro Mager Reference Liu and DiPietro Mager2016). Today we know that the exclusion of women from clinical trials led to an inappropriate understanding of pathologies in women, as well as inappropriate diagnosis and treatment. Thanks to the Women’s Health Movement, among others, who advocated for women’s equal rights to health care, the standard was challenged and transformed (Auerbach and Figert Reference Auerbach and Figert1995). Thus, in some cases, political or other social factors can very well lead to society’s distrust of scientific standards and promote change.

The public, in addition, is not homogenous. As Master and Resnik claim, “different publics may have different expectations of scientific research, and therefore have different levels of trust” (Reference Master and Resnik2013, 327). Many cases in the history of science illustrate that different social groups have had different relations with the scientific community, which have led to differing levels of trust in science. A clear example of this is the relation between clinical researchers and the African American community in the United States. After a history of patient exploitation and abuse—remember, for instance, the infamous Tuskegee syphilis study (Thomas and Quinn Reference Thomas and Quinn1991)—African Americans are less likely to trust clinical researchers and doctors than other members of the US population (Dancy et al. Reference Dancy2004; Branson, Davis, and Butler Reference Branson, Davis and Butler2007). Yet, because of a history of strong colonialism and deference to authority, Latin American patients frequently hold strong relations of trust with their doctors and are thus more likely to participate as subjects in clinical trials (Fernández Pinto Reference Fernández Pinto2019). Recent empirical research also shows that political affiliation can be related to the level of trust in science and that such relations can vary over time. Political conservatives’ trust in science, in particular, has been in decline since the 1970s, while political liberals’ overall trust in science has been more or less unchanged during the same period (Gauchat Reference Gauchat2012).

Having acknowledged that the relation between science and the public is much more complex than the ideal view allows, let us assume, for the sake of argument, that we can speak of the “public trust in science” in terms of the trust that society places in science as a social institution, particularly with respect to research results that are relevant for social matters (Resnik Reference Resnik2011). Without undermining its importance, the relation of particular social groups to science and the level of trust of such relations will not be a matter of concern here.

De Melo-Martín and Intemann (Reference de Melo-Martín and Intemann2018) identify commercial interests as a fundamental contributor to the erosion of trust in science. They detect at least two mechanisms through which commercial interests can affect the trustworthiness of science: first, they put into question the common good as a primary goal of research, and second, they create conflicts of interest that provide reason to question the integrity of research. Notice, however, that while both of these mechanisms can in fact affect the trustworthiness of science, they do not necessarily do so.Footnote ³ As Carrier (Reference Carrier, Carrier and Nordmann2011) has argued, commercial interests might be well aligned with the social and epistemic goals of research, properly fomenting the common good. It is only when commercial interests clash or are misaligned with the social or epistemic goals of science that a problem arises, making research unreliable (Fernández Pinto Reference Fernández Pinto2018). In a similar vein, financial conflicts of interest might lead to research results that tend to favor commercial sponsors, but this does not necessarily compromise the quality of the research, or at least it is not obvious in which way they do so, as will become clear in what follows. Hence, more needs to be said about how exactly commercial interests might contribute to the erosion of trust in science.

Let me start by noticing that, according to the ideal relation between science and the public clarified earlier in this section, the public would be correct in mistrusting scientific research in cases in which scientists have failed to follow the standardized processes because of commercial interests. In fact, I would argue that public mistrust in science in such cases is a good symptom, insofar as it reflects that the general public is attuned to the signals of what counts as appropriate scientific practice.Footnote ⁴ The public is well justified in mistrusting research in such cases because the science is in fact untrustworthy. If this is the case, then the more complex, and perhaps the more worrisome, cases appear when commercial interests lead the general public to incorrectly trust or mistrust science. As de Melo-Martín and Intemann (Reference de Melo-Martín and Intemann2018) acknowledge, people can be confident about scientific results that are not reliable, or they can be skeptical about scientific claims that are in fact trustworthy. Let us analyze such cases and some of their operating mechanisms in the following sections.

3. Public Perception versus Actual Trustworthiness

Commercial interests might affect the reliability of results in different ways. In order to counteract this negative impact, a more nuanced understanding of the mechanisms operating in commercially driven science is urgently needed. Research funded by the private industry may very well lead to public mistrust insofar as it has socially questionable goals and perhaps encourages scientific misconduct, as de Melo-Martín and Intemann (Reference de Melo-Martín and Intemann2018) have emphasized. However, I would claim that the most efficient mechanisms through which commercial interests contribute to the erosion of trust in science are much less transparent.

A first way in which commercial interests reduce the general trustworthiness of research is by directly encouraging the production of consensus or dissent that is favorable to the industry, which in turn leads the public to mistakenly trust or distrust scientific results. Private industry has successfully manufactured doubt in the general public, using scientific research as a key tactic, in crucial policy controversies involving tobacco smoking, climate change, a number of pharmaceuticals, the financial crisis, the ozone hole, asbestos, among many others (Michaels Reference Michaels2008; Oreskes and Conway Reference Oreskes and Conway2010; Proctor Reference Proctor2012; Mirowski Reference Mirowski2013). As I have suggested elsewhere (Fernández Pinto Reference Fernández Pinto2017), the industry has successfully misguided the public in these cases precisely by reshaping the epistemic and social aims of scientific practice, while pretending not to do so. This last part is crucial: it is because commercially driven science is able to claim that it follows the traditional standards of research that the industry is successful in convincing the public of incorrectly trusting or distrusting certain results. And this happens both in cases of normatively inappropriate dissent (NID), as de Melo-Martín and Intemann (Reference de Melo-Martín and Intemann2018) call it, such as the public disagreement regarding anthropogenic climate change, as well as in cases of what we might call normatively inappropriate consensus (NIC), that is, public agreement with scientific results based on merely apparent or ill-grounded scientific consensus. Let me illustrate the latter with an example.

A good example of NIC is the case of research on antidepressants. In 2008, Erick Turner and his colleagues conducted a study examining selective publication of clinical trials on 12 antidepressants. They obtained reviews from the FDA on studies involving 12,564 patients and then compared these reviews with matching publications through a literature search. The results were surprising. According to the FDA reviews, among 74 studies reported, 38 had positive outcomes (51%), while 36 (49%) had negative outcomes, and 24 (12%) were considered questionable. The published literature revealed a quite different scenario: of the 74 studies reported to the FDA only 51 were published, 48 of which revealed a positive outcome (94%). Among these 48 studies, 37 were also considered by the FDA to have a positive outcome, but 11 conflicted with the FDA’s assessment: they were not considered to have a positive outcome by the FDA, but they were published in medical journals as conveying a positive outcome. Contrasting the published literature against the studies reported by the FDA resulted in a 32% overestimation of effect size in the published literature (Turner et al. Reference Turner, Matthews, Linardatos, Tell and Rosenthal2008, 255–56). Therefore, the aggregate of published results, to which prescribing doctors and patients had access, supported the idea that antidepressants were a rather effective treatment for depression, but the same conclusion would not have been reached if unpublished trial results had been considered.

Turner et al. (Reference Turner, Matthews, Linardatos, Tell and Rosenthal2008) acknowledge that the reasons for not publishing research results can vary and that their study does not reveal the motivations of the actors involved. However, they also claim that “since the protocols were written according to international guidelines for efficacy studies and were carried out by companies with ample financial and human resources, to be fair to the people who put themselves at risk to participate, a cogent public reason should be given for failure to publish” (259), and they conclude the paper warning about the social consequences of such concealment of information: “By altering the apparent risk-benefit ratio of drugs, selective publication can lead doctors to make inappropriate prescribing decisions that may not be in the best interest of their patients and, thus, the public health” (259).

In a similar vein, published research of successful treatment with antidepressants reports only a clinically meaningful difference between drug and placebo treatments for patients with severe forms of depression. As a meta-analysis on antidepressant benefits concluded, “the new-generation antidepressants do not produce clinically significant improvements in depression in patients who initially have moderate or even very severe depression, but show significant effects only in the most severely depressed patients” (Kirsch et al. Reference Kirsch2008, e45). However, this information has not been properly communicated to doctors and patients in drug ads, which do not make clear that antidepressants have only been proved to work on the most severe cases of depression (Fournier et al. Reference Fournier2010, 52).

As has been shown, selective publication has led in the case of antidepressant research to inappropriate consensus, prompting doctors, patients, and the public at large to believe that the evidence supported the efficacy of antidepressants for the treatment of depression and for all stages of depression, when in fact the evidence pointed to opposite conclusions (Ioannidis Reference Ioannidis2008; Every-Palmer and Howick Reference Every-Palmer and Howick2014).

So even if we do not know the actual motivations behind selective publication, we do know that withholding results alters the apparent risk-benefit ratio of drugs, which in turn leads to doctors and patients believing inaccurate information about prescriptions. And we also know that believing this inaccurate information can benefit pharmaceutical companies, at least in the short run, insofar as they increase profits on drugs that are not as effective or as safe as the published literature claims. In this sense, commercial interests reduce the general trustworthiness of research by misleadingly encouraging public trust or distrust.

4. Narrow and Broad Views on Standards for Trustworthiness

A second way in which commercial interests reduce the general trustworthiness of research is by endorsing a narrow understanding of the standards for the trustworthiness of scientific research. To make the point, let me introduce the distinction between a narrow and broad view on the methodological standards that warrant the epistemic trustworthiness of research. As previously said, scientific research is considered trustworthy insofar as scientists follow the standardized processes of knowledge production. However, this can be understood in at least two ways. Narrowly, it can be understood as complying with a check list of formalized or standardized procedures. Broadly, it can be understood as a self-regulatory process through which scientists are constantly weighing their microdecisions during the research process against the broad epistemic and social goals of research as understood by their scientific community, as well as against standards for scientific integrity.

Commercially driven research has every incentive to narrowly comply with scientific standards, so that scientific results appear trustworthy. However, it is possible for research to narrowly comply with scientific standards and still fail to be trustworthy from a broader perspective. Let me illustrate this case going back to one of the main examples used by de Melo-Martín and Intemann (Reference de Melo-Martín and Intemann2018): the consequences of financial conflicts of interests in clinical research.

As the latest meta-analyses have systematically showed, industry-funded studies are significantly more likely to obtain results favoring sponsors than is independently funded research (Bekelman, Li, and Gross Reference Bekelman, Li and Gross2003; Lexchin et al. Reference Lexchin, Bero, Djulbegovic and Clark2003; Sismondo Reference Sismondo2008; Lundh et al. Reference Lundh, Lexchin, Mintzes, Schroll and Bero2017). Surprisingly, the same meta-analyses have also shown that industry-sponsored studies have a lower risk of bias, for example, from blinding (Lundh et al. Reference Lundh, Lexchin, Mintzes, Schroll and Bero2017, 16), and that their methodological quality is at least as good as, sometime even better than, the quality of independent studies (Lexchin et al. Reference Lexchin, Bero, Djulbegovic and Clark2003, 1170).

In general, industry-sponsored research complies with methodological standards that can be statistically assessed because tools have been created to account for risk of bias. The Cochrane Collaboration, for example, has developed a tool for assessing risk of bias in randomized trials (Higgins et al. Reference Higgins2011). Methodological aspects whose risk of bias can be assessed, and are contemplated in the Cochrane’s tool, include random sequence generation, allocation concealment, blinding of participants and personnel, and selective outcome reporting (6). Tools have also been developed to assess the risk of publication bias, for example, through funnel plots, but not without limitations (Lau et al. Reference Lau2006). Thus, researchers have some tools available to account for risk of bias with respect to certain methodological decisions made during the research process. Research that passes a risk of bias evaluation can be regarded as trustworthy, at least under the narrow view.

As the results of the previously mentioned meta-analyses suggest, financial conflicts of interest do not lead to overt cases of scientific fraud, insofar as the methodological quality of industry-funded studies is good, or even excellent, according to the available tools to evaluate risk of bias. Nevertheless, meta-analyses also show that research funded by the industry is significantly more likely to obtain results favorable to the sponsors. Why we see this tendency in the results and how is it introduced is still a puzzle.

Notice, however, that there are many other decisions that scientists make during the research process whose compliance has not been formalized through a risk of bias evaluation tool. Hence, in order to understand the mechanisms through which industrial sponsorship affects results, we have to look beyond the standardized methodological procedures that have been formalized. Meta-analyses suggest that plausible mechanisms behind the biased outcomes include inferior comparators, overinterpretation of results, inappropriate trial durations, poor surrogate endpoints, and publication bias, among others (Bekelman et al. Reference Bekelman, Li and Gross2003; Lexchin et al. Reference Lexchin, Bero, Djulbegovic and Clark2003; Lundh et al. Reference Lundh, Lexchin, Mintzes, Schroll and Bero2017).

The possibility emerges for commercial interests to bias the results of clinical research, as meta-analyses confirm, while following standard procedures, narrowly conceived. The results are not epistemically trustworthy, however, if one holds a broader understanding of this concept. For results to be trustworthy according to this broader view, scientists need to pay attention not only to formalized methodological standards but also to standards guiding microdecisions during the research process regarding, for example, choice of comparators, reporting of results, trial durations, surrogate endpoints, and decisions to publish, standards that are accepted as appropriate, even if implicitly, by the scientific community.

As the case of financial conflicts of interest in clinical trials illustrates, to evaluate the epistemic trustworthiness of research, that is, to determine whether research results are reliable, can be a matter of focus, narrow or broad. Research that might be considered trustworthy under a narrow conception might be considered untrustworthy under a broader view. The problem is that processes to evaluate scientific research, such as peer review, rely on a narrow conception of trustworthiness. Hence, if commercial interests encourage scientists to make microdecisions that bias results, but that are not detected by formalized methods for quality checks, as the meta-analyses mentioned here seem to suggest, then commercial interests are effectively reducing the general trustworthiness of research by endorsing a narrow understanding of the standards for the trustworthiness of scientific research and then claiming that commercially driven research is methodologically sound and, thus, trustworthy.

5. Conclusion

Commercial interests can have a negative impact on the trustworthiness of science. However, the mechanism through which commercial interests have such an impact need to be better understood. To begin with, when commercial interests explicitly affect research in ways that render results unreliable, for example, explicit cases of scientific fraud or misconduct, then policy makers and the general public are well justified to distrust such results. In fact, public distrust is an appropriate reaction toward unreliable results. How to prevent such cases will certainly contribute to strengthening the public trust in science, but it will not target the more complex mechanisms through which commercial interests affect the reliability of scientific results. As I have argued in this article, commercial interests have a negative impact on the trustworthiness of science by leading the public to mistakenly trust or distrust scientific results. Among the ways this can be achieved, I have identified two mechanism: first, by directly encouraging the construction of consensus or dissent that is favorable to industrial interests but not epistemically reliable and, second, by subscribing to a narrow understanding of trustworthiness that allows the industry to claim that their research is reliable when we have good reasons to believe it is not. Further research of these strategies is still needed if we want to understand to what extent commercially driven research is hurting the relation between science and the public and contributing to the erosion of trust in science.