Introduction

Generating scientific knowledge should be, in principle, a key consideration in the design of programmes to improve public health. Governments fund national agencies whose purpose is supporting science (e.g., the National Institutes of Health in the United States of America [USA], National Health and Medical Research Council (NHMRC) [Australia]), and researcher-initiated projects are routinely funded through such agencies. Research funding is also dedicated to addressing the priorities of funders with objectives typically relating to informing public policy or commercial imperatives [1]. Such strategic funding aims to address knowledge gaps important to funders thereby facilitating ‘research translation’ by ensuring relevance to end-users. However, these funding models have been shown to undermine the integrity of science by enabling funders to influence how research is done and reported [2–4].

As providers of publicly funded health and medical research, universities have a vital role in facilitating independent enquiry. The notion of academic freedom is that researchers bound by the scholarly conventions of peer review and ethical approval, are free to do research without interference or the threat of professional disadvantage [5]. Many see the preservation of such freedom as vital to safeguarding the reflection, critique, and innovation that academia can bring to society [3, 6]. However, academic integrity is increasingly undermined by the influence of vested interests on research [2, 7], and a reproducibility crisis [8], calling into question whether public research institutions actually serve the public interest [9]. That research funders who are also responsible for giving policy advice or implementing intervention programmes have a stake in study findings, puts pressure on the impartiality of the researchers who depend on the funding. This could include subtle pressure on researchers, unconsciously conveyed hopes for ‘positive’ findings, or total suppression or censorship of reports for political advantage [4]. Various mechanisms exist to regulate researcher behaviour, including codes of conduct and ethical review [10, 11]. In Australia, many government funding agreements require researchers to obtain funder approval to publish reports [12].

The suppression of public-good research by funders or other parties is neither well understood nor coherently regulated [13]. A 2006 survey of Australian public health researchers reported that 21% of participants (with a response rate of 46%) had experienced at least one incident in which a government funder suppressed their research in the preceding 5.5 years [14]. The most common forms of suppression reported were blocking, significantly delaying publication and requests to “sanitise” reports [14]. A survey of Australian ecologists and conservation scientists by Driscoll et al. [15] indicated that Government and industry respondents reported higher rates of suppression than university respondents (34%, 30% and 5% respectively) and this suppression was mainly in the form of internal communication and media. A 2015 Canadian survey of federal government scientists showed that within the last 5 years 24% of scientists had been asked to exclude certain findings from their reports, and 37% reported that they were prevented from responding to media enquiries within their area of expertise [16]. In the United Kingdom (UK) an enquiry into public-good research, commissioned by a science charity, presented nine case studies outlining the impact of significant delays in the publication of findings. In several cases delays appeared to be motivated by political considerations [4]. Knowing how often and in what circumstances the suppression of public health research occurs is important because of the potential impact of withholding, delaying, or misrepresenting findings. This is acutely apparent in the COVID-19 pandemic, where delays in releasing early research findings in China allowed significant outbreaks to occur in other countries [17–19].

The aims of this study were: (1) to ascertain the reported prevalence of efforts to suppress the findings of primary prevention trials that target nutrition, physical activity, sexual health, tobacco use, alcohol or substance use; and (2) to identify associations between trial characteristics and suppression events.

Methods

Results

From 42 eligible reviews we identified 208 trials and received survey responses from 104 (50%) of their corresponding authors. Papers were published from 2007–2016 and reported trials concerning physical activity and/or nutrition (55%), substance use and/or sexual health (47%). Two thirds were conducted in North America (34%) or Europe/United Kingdom (33%), with the balance in Oceania (17%) and other countries (16%). Examining democracy data, the majority of studies were from full democracy countries (61%). The majority of studies receive funding from Other Government agencies (39%).

S1 Table shows that the characteristics of trials whose authors did not complete the survey were not markedly different from those who did, in terms of the study design, full democracy, or publication date. However, the proportion conducted in North America was higher among non-respondents (53%) than respondents (34%). Too, non-responder reported more funding from an Other Government Agency (49% non-completers vs 39% completers).

Aim 1: Prevalence of suppression events

Eighteen percent (18/98, 6 unknown) of respondents reported at least one instance of suppression. Table 1 shows the number of respondents who reported each type of suppression event having occurred at least once, by funding source. Rates of suppression were highest for studies funded by Other Government agencies. The most commonly reported suppression event was that of the funder expressing reluctance for publication due to ‘unfavourable’ results: with six, two and one suppression events being reported from studies funded by other Government agencies, independent sources, and multiple funding sources, respectively. In comparison, researchers receiving industry or philanthropic funding did not report a single suppression event.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Researcher reports of funder efforts to suppress trial findings.

https://doi.org/10.1371/journal.pone.0255704.t001

Sensitivity analysis.

Under the extreme assumptions that none or all the non-respondents had experienced a suppression event, the prevalence estimates would be as low as 9% (18/208) or as high as 59% [(18+104)/208], respectively.

Aim 2: Association between trial characteristics and suppression events

Table 2 summarises associations between trial characteristics and suppression events. Researchers receiving Other Government Grants, or who conducted studies in Europe, had higher odds of reporting a suppression event compared to those reporting on studies with independent dedicated research funding or conducting studies in North America, respectively. Researchers who had conducted sexual health/substance use trials more commonly reported a suppression event than those who had conducted nutrition/physical activity trials. Whether the publication came from a democratic country or not did not seem to change the odds of reporting suppression. As the age of the publication increased, so too did the odds of reporting suppression. The confidence intervals for the odd ratios of all the comparisons were wide and included 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Associations between trial characteristics and suppression events.

https://doi.org/10.1371/journal.pone.0255704.t002

Discussion

In a sample of authors of prevention intervention trial reports published over a decade, of whom 50% responded, we found that one in five reported at least one suppression event. A simple sensitivity analysis suggests that rates of suppression could be as low as 9%, and as high as 59%, depending on the proportion of non-respondents who were subjected to suppression events. Our overall estimate of 18% is similar to estimates from previous studies in Australia (21% to 34%) [15, 32] and Canada (24%) [16]. Notably, we asked specifically about what occurred in relation to a single trial, while other studies evaluated suppression events across many projects over the course of five [16] or 5.5 years [14]. As such, rates of suppression may be much higher.

The sampling frame provides for greater international representation and broader coverage of health research than previous studies [14, 32]. However, relying on published studies to direct us to authors means that we would not have identified authors of studies whose publication was supressed entirely. It is also likely that some authors would not disclose suppression events, even in the confidential context of a research study, fearing repercussions from the funder, or being negatively evaluated by the researchers. The implication of this being that the actual rate of suppression is much higher. Finally, the small numbers in our study constrain the precision of our estimates of prevalence and association. Accordingly, we suggest that 18% is an under-estimate of the true prevalence of studies subject to some form of suppression by funders.

It is hard to determine why older publications or those published in different geographical regions were suggestive of having to have greater odds of suppression (though the confidence intervals for all the comparisons were wide and included 1). A possible explanation may be that older studies which have existed longer may have had more ‘chance’ to experience suppression on their findings. That studies published outside North America were found to have greater odds of suppression is hard to elucidate. It may be that instances of suppression in North America are only reported by individual researchers when it is ‘more severe’ than those reported in other countries, that they were more afraid to report suppression, or that there are tighter regulations on what types of suppression can be enacted on grant holders. More research would be needed to explain the differences noted.

Our results, along with those of previous investigations, suggest that government funders interfere with public-good research. In addition to curtailing independent scientific enquiry, such practices deny the public access to the findings of research paid for through taxation, which in some cases, could have informed policy decisions. On this point, in his ‘Missing Evidence’ report, former High Court Judge Sir Stephen Sedley observed that the UK Parliament made a critical decision concerning the merits of minimum unit pricing of alcohol without the benefit of key findings whose publication had been deliberately delayed by the Department of Health [4]. In addition to the loss borne by taxpayers due to ill-informed policy, is the damage done to democracy when such perversions come to light.

This research has its limitations. For example, we did not attempt to ‘deep dive’ on the different types of suppression. We found that reluctance to publish research findings due to unfavourable results was the most reported type of suppression experiences. However, we did not seek to determine what reluctance meant in this context–whether it meant funders prevented publication or just tried to influence it. Too we did not give participants the opportunity to describe other forms of suppression they may have felt. Further research should investigate this in order to determine all type of suppression experienced in order to develop the most practical ways to overcome them.

Attention is urgently needed to protect the integrity of public health research from the influence of vested interests, whether private or official in origin. Preventive actions are required of all actors involved in the generation of research findings:

1. Government agencies must ensure appropriate terms in funding agreements formed with research providers that protect academic freedom for e.g. the removal of clauses which require the approval of results prior to publication [32];
2. Research institutions must not accept funding on terms that permit funders to interfere in public-good research;
3. Government agencies should establish a registry of studies funded by government agencies including the terms of the funding to encourage openness;
4. Research Ethics Committees must consider the source and terms of research funding to determine if there are any ethical implications of the funding source;
5. Scientific journals must require authors to declare the terms of research funding, and potential conflicts of interest;
6. Researchers must be held to Code of Conduct provisions concerning acceptable terms of funding;
7. Audits of contracting and research practices of tertiary academic institutions should be undertaken by an independent body with appropriate powers (e.g. an independent government department); and
8. Universities should consider establishing a mechanism for reporting instances research suppression, and the management of funders/individuals/etc. who are known to attempt to suppress research findings.

Suppression still exists among public-good researchers, with this study suggesting rates of suppression for a single trial may be as high as one in five. Prevention is key and these suggestions, similar to those previously described [32], need to be adopted in order to thwart the occurrence of suppression of public-good health research. As it is unlikely that instances of suppression will ever truly be stopped, further research needs to be conducted to determine ways of handling suppression when they do happen at the researcher level (e.g. what can a researcher do if someone tries to delay their publication), as well as reporting procedures in place if instances of suppression do occur.

Supporting information