Conflict of Interest Policies at Medical Schools and Teaching Hospitals: A Systematic Review of Cross-sectional Studies

Background

In the past decades, academics’ financial ties with commercial companies have been highly debated. Studies suggest that such relationships can introduce biases in research and education.¹ For example, both industry funding of medical research and financial relationships between academic researchers and commercial companies are associated with conclusions favourable to the companies’ products.^2-5 Moreover, industry funding can impose constraints on publication rights and access to the raw data which can potentially lead to selective reporting of favourable outcomes and withholding of unfavourable results.^6,7 Conflicts of interest (COI) can also influence medical education.¹

These concerns have led some medical schools to implement institutional policies to regulate interactions with commercial entities in order to protect the integrity of the research and educational process. Some studies conducted in North America have examined how medical schools vary in their implementation of COI policies.^8-11 More recently, analyses of COI policies have been conducted in some European countries.^12,13

COI policies likely have regional differences and have changed over time. However, to our knowledge no previous systematic reviews have comprehensively identified and analysed studies on COI policies of medical schools and teaching hospitals to protect the research and educational environment from commercial influences.

The objectives of this review are:

1. To describe the proportion of medical schools and teaching hospitals with COI policies for health research and education.

2. For the medical schools and teaching hospitals with COI policies:

   1. To describe the provisions included in the policies and to whom they apply.

   2. To describe the impact of COI policies on research outputs, educational quality or content.

Methods

Protocol, Search Strategy, and Study Selection

The protocol was published in PROSPERO (CRD42020197500). We searched the following databases (from inception to March 9, 2020): Ovid MEDLINE, Scopus, Ovid Embase, and the Cochrane Methodology Register. The search strategy was developed in consultation with an information specialist (Supplementary file 1). We also searched reference lists of included studies and Web of Science for publications that cited any of the included studies. For included studies available in unpublished form (eg, pre-prints), in October 2020 we checked whether the study had been published and updated the information we had previously extracted.

Eligibility Criteria

We included experimental and observational studies investigating COI policies of medical schools or teaching hospitals. For the purpose of this study, we considered COI policies any document or regulation on interactions with industry both at the individual level (eg, researchers, medical students) and institutional level. We focused on COI policies to protect research and education and did not restrict to policies related to the pharmaceutical industry. We included any study with one or more of the following aims:

• Estimating the prevalence of COI policies.

• Describing the content and scope of COI policies (eg, which activities should be disclosed, which activities are prohibited or constrained).

• Estimating the impact of COI policies on research output (eg, number of trials conducted after the implementation of the policy), educational quality or content. By impact we mean an objective measure, not a subjective one (eg, opinions of staff about the policy).

We excluded editorials, letters, commentaries, descriptive case studies, systematic and narrative reviews; studies on schools not in the medical sciences field (eg, engineering); studies that included different types of institutions, without separate data reported for medical schools or teaching hospitals. We excluded studies solely focusing on policies to protect clinical practice and studies that focused on one single element of a COI policy (eg, disclosure only, gifts only) rather than the breadth of policies in medical schools or teaching hospitals. We did not exclude studies based on language, publication date, or study setting.

Study Inclusion

One assessor (AF) screened the titles and abstracts of retrieved records. Two assessors (AF, KRH) independently assessed the full texts of the remaining records for inclusion. Discrepancies were resolved by discussion. If consensus could not be reached, a third assessor adjudicated (AL).

Data Collection

For each study we collected information on: general study information (author, publication year, funding source and authors’ financial COI), study design, setting, and population, study period, methods of data collection, and outcomes (any quantitative data was extracted on the above mentioned aims: prevalence of COI policies, content, scope and impact of the policies).

One assessor (AF) extracted information from records and a second assessor (KRH) checked the accuracy and completeness of extracted information. Discrepancies were resolved by discussion. When information reported in included records was unclear or if relevant data was missing, we contacted the study authors for clarifications or unpublished data. We contacted the authors of 13 studies: 11 replied and nine of these supplied the requested information.

Methodological Quality Assessment

We used an adapted version of the Checklist for Studies Reporting Prevalence Data developed by Joanna Briggs Institute to assess methodological quality.¹⁴ The original checklist assesses the quality of a study across nine items. We excluded items related to statistical issues and reporting quality and only included five items strictly related to methodological quality: sample frame, methods used to recruit participants, methods for identification of outcomes, measurement of outcomes and missing data. We changed the possible answers for each item from yes, no, unclear, or not applicable to high quality, low quality, unclear, or not applicable to reflect methodological quality (ie, ‘yes’ for appropriate sample frame was changed to ‘high quality’ for sample frame). We developed guidance for how to assess each item (Supplementary file 2) drawing inspiration from a previous systematic review where the guide for item assessment had been amended to reflect the focus on a policy issue rather than a clinical condition.¹⁵ Studies with high quality in all domains were assessed as overall high quality and other studies as overall low quality.

Analysis

We undertook an initial descriptive analysis of the included studies, presenting their characteristics, settings, and populations. We grouped studies based on which of our study aims they addressed. Any commonalities between studies was synthesised and discussed.

Prevalence of Policies

For studies estimating prevalence of COI policies, we calculated confidence intervals for individual studies using the Clopper-Pearson method.¹⁶ Since the studies on prevalence were very heterogeneous, for example in relation to study setting, methodology and study period, no summary estimate was calculated. However, we undertook meta-analysis (random effects meta-analysis using the DerSimonian-Laird estimate of single proportions with prevalence estimates that had been transformed using the Freeman-Tukey double arcsine transformation)^17,18 in order to assess heterogeneity using I² and explored reasons for heterogeneity in the following prespecified subgroup analyses: setting (North America versus other settings), timing (before versus after median publication year), and methodological quality (high versus low quality). Statistical analyses were conducted in MedCalc 19.6.

Content of Policies

For studies investigating content of COI policies, we grouped the studies that addressed this question in four different categories based on the assessment methods and the target institutions: (1) studies assessing the strength of the content of COI policies of medical schools using various assessment tools; (2) studies assessing the strength of the content of COI policies of teaching hospitals; (3) studies examining the content of COI policies without assessing their strength, and (4) studies examining the content of medical school and hospital-based Institutional Review Boards’ (IRBs’) COI policies. Moreover, for studies that used assessment tools (eg, checklists, scales) to analyse the strength of the content of COI policies (ie, category 1), we coded the items included in each tool and grouped them into different domains to synthesize their commonalities. In order to compare the results of these studies, for tools including a summary score (ie, scales), we also standardised the overall scores by dividing the mean score with the maximum score possible for a given tool and reported this as percentages. We ensured directionality of scales (ie, that high scores indicated high strength of policy). In our main analysis we included both institutions with a COI policy and institutions without a COI policy in the denominator for calculating mean standardised summary scores (ie, institutions without a COI policy will automatically receive a score of 0). In a sensitivity analysis, we restricted our analysis to institutions with available COI policies.

Association Between Policies and Research Output or Educational Quality or Content

We planned to calculate pooled estimates for the association between COI polices and research output, or educational quality or content, but no relevant studies were identified.

Results

Description of Included Studies

We identified 8657 records for screening and included 22 studies (from 24 records) (Figure 1). The most common reason for exclusion was type of publications (not primary research articles, n = 382), followed by aims of interest not investigated (n = 63). For two records, we could not retrieve full text versions despite contacting four medical libraries. None of the records had abstracts, none of the titles suggested that they were original research, and one was only one page long. We therefore judged these records as unlikely to report on primary research.^19,20 For one study a preprint version was initially included, but a journal publication was also included during the additional search run in October 2020.¹²

Close

Figure 1. PRISMA Flow Diagram of Included Studies. Abbreviations: COI, conflicts of interest; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Table 1 summarises the characteristics of the included studies. The 22 studies were published between 2000 and 2020 and all were cross-sectional. All the studies were conducted in high-income countries, primarily North America (n = 17). Several studies used data collected from multiple sources, such as policies retrieved from both institutional websites and through contact with University officials. Nine studies were surveys of institutional members and their response rates ranged from 62% to 100% (median 69%).

Methodological Quality of Included Studies

Figure 2 shows the quality assessment of the included studies. Fifteen of 22 studies were assessed as overall high quality (ie, high quality for all domains). The domain on methods for identification of outcomes was the most frequent domain judged as low quality (n = 4/22).

Close

Figure 2. Methodological Quality Assessment of Included Studies.

Prevalence of COI Policies

Twenty studies estimated the prevalence of COI policies with estimates ranging from 5 to 100% (median: 85%). We grouped the studies into categories according to their focus (Figure 3). Nine studies estimated prevalence of general COI policies in medical schools; prevalence 5% to 100% (median: 90%).^{8,9,12,13,21-25} Some of these studies only included policies of medical schools while two clearly stated that they also included COI policies of the parent University.^8,21 One French study reported that 53% (17/32) of teaching hospitals had a general COI policy.²⁶ Two studies estimated prevalence of COI policies for members of medical school-based IRBs in the United States; prevalence 74% and 81%.^27,28 Three studies estimated prevalence of COI policies for investigators from Universities or teaching hospitals in Canada and the United States; prevalence from 97% to 100%.^11,29,30 One study reported that 10% (10/99) of US medical schools had COI policies specifically targeting industry-student relationships.³¹

Close

Figure 3. Forest Plot of Prevalence of COI Policies. Abbreviation: COI, conflicts of Interest. *Data provided by the author. **Lieb, 2014 asked about the existence of a general COI policy and/or a policy on interactions between medical students and industry. I2: Heterogeneity described by calculating I2.

Finally, four studies were not included in Figure 3 because of their distinct focus: the first study reported that none of 16 Canadian Faculties of Medicine or their parent universities had policies regarding research funding or financial donations from tobacco companies³²; the second study focused on COI policies solely at the institutional level among 74 Canadian teaching hospitals, medical schools, and their parent Universities; prevalence of 55% (41/74).³³ The last two studies estimated prevalence of COI policies for US medical school and teaching hospital-based IRBs via a survey of their members; prevalence 46% (263/574) and 63% (311/493).^34,35 However, these estimates are not comparable to the other two studies on IRBs^27,28 reported in Figure 3 because prevalence was measured at member level and not at institutional level (ie, some members included in survey were from same institution).

Content and Strength of COI Policies

We grouped the 20 studies that assessed the provisions included in COI policies in four categories based on the assessment methods and the target institutions. Five studies across the four categories reported data on who the policies applied to and all studies found that policies often did not cover all relevant stakeholders (eg, only selected faculty) (Supplementary files 3 and 4). Nine studies assessed the presence of policy enforcement mechanisms such as oversight and sanctions for non compliance and found that they were not consistently mentioned in the included policies (Supplementary files 3 and 4).

Strength of the Content of COI Policies of Medical Schools

Nine studies analysed the strength of the content of COI policies of medical schools (in two cases also their parent universities) using assessment tools that looked at a range of policy domains.^{8-10,12,13,22-24,36} All assessment tools used a scale approach (ie, individual scores for each item transformed into an overall summary score). Table 2 presents the range of policy domains addressed by the various assessment tools. The number of items in each tool ranged from 7 to 14. We grouped the items into 14 common domains and one ‘other group.’ Seven of the nine studies used the American Medical Student Association (AMSA) scorecard or an adaptation of the scorecard as assessment tools.

The mean standardised summary score ranged from 2% to 73% (median: 30%). The European studies both had a summary score of 2% while in the North American studies the score ranged from 22% to 73%. The sensitivity analysis, restricting analysis to institutions with policies, increased the strength of policies from European institutions and thereby reduced the gap between European and US institutions, but still important differences were detected (Supplementary file 3, Table S1).

In Supplementary file 3 (Table S2) we report the number of policies that received maximum score in each domain for each study. There was variation across studies and domains. Most institutions ranked poorly in the studies conducted in Europe and Australia.^12,13,22 For example, in a study conducted among the 37 French Faculties of Medicine, only one was rated as having a model policy for two out of the 14 assessed policy domains (industry sponsorship of educational activities and industry funding of the medical school) and 36 Faculties did not have a model policy for any domains.¹³ In North America instead, more medical schools had model policies but still for several domains most institutions were not rated as having model policies.

Strength of the Content of COI Policies of Teaching Hospitals

One study analysed COI policies of 32 teaching hospitals in France²⁶ using an assessment tool adapted from AMSA and other previous studies.^8,13 Fifteen hospitals had no COI policies available on their websites and did not reply to the authors’ request for information. For 17 hospitals, policies were present for some items only (Supplementary file 4, Table S1). The mean standardised summary score was 6%; when considering only the 17 hospitals with available policies in sensitivity analysis, the score was 11%.

Content of COI Policies With No Assessment of Their Strength

Eight studies examined the content of COI policies without assessing their strength. Key findings of these studies are reported in Supplementary file 4, Table S2.

In three studies the assessment of whether the policies covered specific domains was done by the study authors based on copies of COI policies.^11,29,33 Two of them assessed whether the policies covered specific domains using checklists designed by the authors; the checklists contained 16 and 61 items.^11,33 For example, Lexchin assessed the content of policies that dealt with COI of investigators in Canada. All included teaching hospitals, Faculties of Medicine and their universities had policies, but they varied widely with regard to the domains addressed. Important areas were not addressed; for example, 61% of the institutions had no policies on investigators’ publication rights.¹¹

In four studies the content of the policies was based on survey responses by Deans of medical schools.^10,25,28,30 One investigated COI policies at German medical schools, while the other three were conducted in the United States. For example, Chimonas asked Deans whether their schools had policies on 11 key items. Most of the institutions (from 55% to 70%) had policies concerning consulting, honoraria, gifts, meals, vendor access, and continuing medical education. Other policy areas were less frequently addressed; for example, only 23% of the institutions had a policy on ghostwriting.

Finally, one study used both methods, namely survey of University officials and authors’ assessment of policies identified via institutional websites.²¹ Some discrepancies were found between the two assessment methods. For example, in the survey, most institutions (90%) reported having a policy for consulting, continuing medical education and industry-funded speaking bureaus. Instead, according to the policies identified on institutional websites, these domains were only covered by 75%, 65%, and 75% of the institutions, respectively.

Content of COI Policies of IRB

Two studies analysed the content of IRB COI policies of medical schools and hospitals.^27,35 Detailed results are reported in Supplementary file 4, Table S3. The studies found that there is still a lack of clear guidance on how to disclose and manage COI. For example, Wolf and Zandecki reported that while 75% of the policies prohibited members from discussing or voting on a research protocol in which they had a COI, only 14% prohibited them from being reviewers of the protocol.²⁷

Impact of COI Policies

None of the included studies assessed the impact of COI policies on research outputs or educational quality or content.

Exploration of Heterogeneity

We found large heterogeneity of prevalence estimates with I² of 98% (Figure 3). We explored reasons for heterogeneity in relation to setting, timing, and methodological quality for prevalence of general COI polices for medical schools. We did not explore heterogeneity for other types of COI policies due to the limited number of included studies for those groups. We found no indication that timing or methodological quality could explain differences in prevalence estimates (Supplementary file 5, Figures S2 and S3). However, when we stratified our analysis according to study setting, heterogeneity decreased from I² of 98% in overall analysis to I² of 69% for studies conducted in North America and 95% for studies conducted outside North America. When we limited the latter group to the three studies conducted in Europe, by excluding the study by Mason conducted in Australia, I² decreased to 0%. Similarly, COI policies from North American institutions scored better on strength of content than policies from European institutions. As Table 2 shows, the European studies both had a summary score of 2% while in the North American studies the score ranged from 22% to 73%.

Discussion

Key Findings

Of the 22 included studies, 20 reported prevalence of COI policies (from 5% to 100%). Twenty studies assessed the content of COI policies and found that they varied widely and still had substantial shortcomings. The majority of North American institutions had COI polices in contrast to European institutions where the majority did not have COI policies. Similarly, for institutions with policies, North American policies scored better on strength of content than European institutions. None of the included studies assessed the impact of COI policies on research outputs or educational quality or content and none of the included studies investigated institutional policies in Africa, Asia or South America.

We found a huge variability in prevalence of COI policies across countries. It is possible that some institutions that lack formal policies, still address COI on a case by case basis, and conversely the presence of a formal policy does not necessarily mean that its provisions are appropriately implemented. Enforcement mechanisms, including active oversight to ensure compliance and sanctions in case of violations are vital to ensure that policies reach their intended objectives. It is also worth mentioning that in some countries there could be laws that regulate or prohibit certain interactions with industry and these were not captured by the analyses of institutional policies. For example, in France there is a “Sunshine Policy” that requires disclosure of payments to health professionals in a publicly available database.³⁷ Similarly, the analysis of institutional policies does not comprehensively capture the environment in which faculty and students operate and the difference between North American and European institutions may therefore not necessarily imply that COI have greater influence in European institutions. However, the variability we detected means that academic researchers and students are likely facing very different institutional environments when dealing with industry. Interactions with industry pose important ethical challenges and the presence of explicit institutional policies could provide a roadmap for academic researchers and students and may help to eliminate potential grey areas.

Our findings that policies in US tended to be stricter compared to Europe could in part be explained by the AMSA scorecard. In 2007 AMSA started publishing an annual scorecard ranking US medical schools on their COI policies and the considerable media attention generated by the scorecard influenced the development of policies in several US Schools.³⁶ The AMSA scorecard has also had an international impact prompting similar analyses in other countries.^{8,12,13,22,23} Similarly to the AMSA experience, the publication of the French scorecard in 2017 prompted important changes in the country with the publication of a Code of Ethics by the National Conference of Deans of Medicine and Odontology Schools³⁸ that was subsequently adopted by several medical schools across the country.³⁹

Strengths and Limitations

To our knowledge, this is the first systematic review, to examine the prevalence, content, and impact of COI policies of medical schools and teaching hospitals. We registered our protocol before conducting the review and undertook a comprehensive search strategy. However, our review also has limitations. First, most of the studies were conducted in North America and this limits the generalisability of our findings. Second, the included studies varied in many aspects (eg, types of institutions, study period, assessment tools used) making comparison across studies challenging. While our analyses to some degree indicate that prevalence estimates are similar within geographical regions, our results are limited by few studies conducted outside North America. Third, for the North American studies, there could have been overlap between samples; however, since in many cases individual institutions and scores could not be identified, it was not possible to investigate whether differences in estimates could be due to different assessments by study authors. Fourth, the majority of studies were assessed as being of high quality, but there are no available methodological quality assessment tools specifically tailored for these types of studies and this may have affected our ability to measure the quality of included studies. Fifth, it is important to acknowledge that there are many other studies that focus on IRBs and their practices in the medical literature,^40,41 but we only identified three that specifically dealt with their COI policies at medical schools and teaching hospitals. Sixth, while our search strategy was comprehensive we excluded 17 studies not reporting data separately for medical schools or teaching hospitals. It is possible that the study authors might have collected relevant data, but we did not contact them for additional data as this would likely require extensive recoding and analyses which was beyond the scope of our study.

Implications for Research

None of the included studies assessed the impact of COI policies on research outputs or educational quality or content. The introduction of institutional COI policies on pharmaceutical promotion has been associated with reduced prescribing of newly approved psychotropic drugs⁴² and of heavily marketed and brand reformulated antidepressants.⁴³ Similarly, future studies could investigate the impact of COI policies on institutional research and educational activities.

The majority of included studies were from the United States, five from Canada, four from Europe, and one from Australia. This exposes an important knowledge gap as little research has been conducted outside of North America and no data is available from middle- or low-income settings.

Finally, despite the great impact that the AMSA scorecard had on this field of research, it is worth mentioning also some of its limitations, which could have implications for future research. The AMSA scorecard went through a rigorous process of revisions between 2012 and 2014 thanks to the work of two committees.³⁶ However, with regard to its content, the focus of the tool is mainly on interactions with industry at the individual level without considering COI at the institutional level²³ and apart from ghostwriting the tool does not address COI and research, for example industry sponsors’ influence on study design, analysis and reporting.⁴⁴ Moreover, the adaptation of the AMSA tool to the different local contexts – although in some cases necessary – introduced variations in the domains addressed or in their definition which made it difficult to compare the results of these studies in our review. In light of these methodological considerations, there could be a need for a new tool to assess institutional COI policies. The tool could be developed starting from the current AMSA scorecard using methodological approaches ensuring high validity and reliability.⁴⁵

Conclusion

Of 22 cross-sectional studies, 20 showed that prevalence of COI policies at medical schools and teaching hospitals varied greatly in high-income countries. We found no studies estimating the prevalence of COI policies in low or middle-income countries. The content of COI policies varied widely and while most European institutions scored poorly, US medical schools have taken actions in adopting and strengthening COI policies, although many still have room for improvement. None of the included studies assessed the impact of COI policies on research outputs or educational quality or content. Interactions with industry pose important ethical challenges and the presence of explicit institutional policies could provide a roadmap for researchers and students.

Acknowledgements

We thank Lasse Østengaard (University of Southern Denmark) for helping us to develop the search strategy.

Ethical issues

Not applicable.

Competing interests

Authors declare that they have no competing interests.

Authors’ contributions

All authors contributed to study idea and design. AF conducted the literature search and screened abstracts. AF and KRH screened full texts, acquired the data, and assessed the methodological quality of included studies. AF and AL analysed the data. AF wrote the first draft of the manuscript. All authors edited drafts of this article and approved the final version.

Funding

No external funding. All authors were financed through institutional salaries.

Supplementary files

References

1. Lo B, Field MJ. Conflict of Interest in Medical Research, Education, and Practice. Washington, DC: National Academies Press; 2009.
2. Ahn R, Woodbridge A, Abraham A. Financial ties of principal investigators and randomized controlled trial outcomes: cross-sectional study. BMJ 2017; 356:i6770. doi: 10.1136/bmj.i6770 [Crossref] [ Google Scholar]
3. Lundh A, Lexchin J, Mintzes B, Schroll JB, Bero L. Industry sponsorship and research outcome. Cochrane Database Syst Rev 2017; 2:MR000033. doi: 10.1002/14651858.MR000033.pub3 [Crossref] [ Google Scholar]
4. Hansen C, Lundh A, Rasmussen K, Hróbjartsson A. Financial conflicts of interest in systematic reviews: associations with results, conclusions, and methodological quality. Cochrane Database Syst Rev 2019; 8(8):MR000047. doi: 10.1002/14651858.MR000047.pub2 [Crossref] [ Google Scholar]
5. Nejstgaard CH, Bero L, Hróbjartsson A. Association between conflicts of interest and favourable recommendations in clinical guidelines, advisory committee reports, opinion pieces, and narrative reviews: systematic review. BMJ 2020; 371:m4234. doi: 10.1136/bmj.m4234 [Crossref] [ Google Scholar]
6. Gøtzsche PC, Hróbjartsson A, Johansen HK, Haahr MT, Altman DG, Chan AW. Constraints on publication rights in industry-initiated clinical trials. JAMA 2006; 295(14):1645-1646. doi: 10.1001/jama.295.14.1645 [Crossref] [ Google Scholar]
7. Lundh A, Krogsbøll LT, Gøtzsche PC. Access to data in industry-sponsored trials. Lancet 2011; 378(9808):1995-1996. doi: 10.1016/s0140-6736(11)61871-0 [Crossref] [ Google Scholar]
8. Shnier A, Lexchin J, Mintzes B, Jutel A, Holloway K. Too few, too weak: conflict of interest policies at Canadian medical schools. PLoS One 2013; 8(7):e68633. doi: 10.1371/journal.pone.0068633 [Crossref] [ Google Scholar]
9. Chimonas S, Evarts SD, Littlehale SK, Rothman DJ. Managing conflicts of interest in clinical care: the “race to the middle” at US medical schools. Acad Med 2013; 88(10):1464-1470. doi: 10.1097/ACM.0b013e3182a2e204 [Crossref] [ Google Scholar]
10. Chimonas S, Patterson L, Raveis VH, Rothman DJ. Managing conflicts of interest in clinical care: a national survey of policies at US medical schools. Acad Med 2011; 86(3):293-299. doi: 10.1097/ACM.0b013e3182087156 [Crossref] [ Google Scholar]
11. Lexchin J, Sekeres M, Gold J. National evaluation of policies on individual financial conflicts of interest in Canadian academic health science centers. J Gen Intern Med 2008; 23(11):1896-1903. doi: 10.1007/s11606-008-0752-4 [Crossref] [ Google Scholar]
12. Grabitz P, Friedmann Z, Gepp S. Quantity and quality of conflict of interest policies at German medical schools: a cross-sectional study and survey. BMJ Open 2020; 10(9):e039782. doi: 10.1136/bmjopen-2020-039782 [Crossref] [ Google Scholar]
13. Scheffer P, Guy-Coichard C, Outh-Gauer D. Conflict of interest policies at French medical schools: starting from the bottom. PLoS One 2017; 12(1):e0168258. doi: 10.1371/journal.pone.0168258 [Crossref] [ Google Scholar]
14. Critical Appraisal Checklist for Studies Reporting Prevalence Data. The Joanna Briggs Institute website. https://joannabriggs.org/sites/default/files/2019-05/JBI_Critical_Appraisal-Checklist_for_Prevalence_Studies2017_0.pdf. Published 2017.
15. Fabbri A, Parker L, Colombo C. Industry funding of patient and health consumer organisations: systematic review with meta-analysis. BMJ 2020; 368:l6925. doi: 10.1136/bmj.l6925 [Crossref] [ Google Scholar]
16. Newcombe RG. Two-sided confidence intervals for the single proportion: comparison of seven methods. Stat Med 1998; 17(8):857-872. doi: 10.1002/(sici)1097-0258(19980430)17:8<857::aid-sim777>3.0.co;2-e [Crossref] [ Google Scholar]
17. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986; 7(3):177-188. doi: 10.1016/0197-2456(86)90046-2 [Crossref] [ Google Scholar]
18. Freeman MF, Tukey JW. Transformations related to the angular and the square root. Ann Math Statist 1950; 21(4):607-11. doi: 10.1214/aoms/1177729756 [Crossref] [ Google Scholar]
19. Seeking disclosure. GMHC treatment issues: the Gay Men’s Health Crisis newsletter of experimental AIDS therapies 2001;15(11-12):16.
20. Steiner DJ. Pharmaceuticals and medical devices: business practices. Issue Brief Health Policy Track Serv. 2012:1-35.
21. Klein T, Graves J. The scope of inclusion of academic conflict of interest policies. J Acad Ethics 2018; 16(2):103-116. doi: 10.1007/s10805-017-9298-6 [Crossref] [ Google Scholar]
22. Mason PR, Tattersall MHN. Conflicts of interest: a review of institutional policy in Australian medical schools. Med J Aust 2011; 194(3):121-125. doi: 10.5694/j.1326-5377.2011.tb04193.x [Crossref] [ Google Scholar]
23. Mathieu G, Smith E, Potvin MJ, Williams-Jones B. Conflict of Interest Policies at Canadian Universities and Medical Schools: Some Lessons from the AMSA PharmFree Scorecard. BioéthiqueOnline; 2012.
24. Yeh JS, Austad KE, Franklin JM. Association of medical students’ reports of interactions with the pharmaceutical and medical device industries and medical school policies and characteristics: a cross-sectional study. PLoS Med 2014; 11(10):e1001743. doi: 10.1371/journal.pmed.1001743 [Crossref] [ Google Scholar]
25. Lieb K, Koch C. Conflicts of interest in medical school: missing policies and high need for student information at most German universities. GMS Z Med Ausbild 2014; 31(1):Doc10. doi: 10.3205/zma000902 [Crossref] [ Google Scholar]
26. Guy-Coichard C, Perraud G, Chailleu A, Gaillac V, Scheffer P, Mintzes B. Inadequate conflict of interest policies at most French teaching hospitals: a survey and website analysis. PLoS One 2019; 14(11):e0224193. doi: 10.1371/journal.pone.0224193 [Crossref] [ Google Scholar]
27. Wolf LE, Zandecki J. Conflicts of interest in research: how IRBs address their own conflicts. IRB 2007; 29(1):6-12. [ Google Scholar]
28. Ehringhaus SH, Weissman JS, Sears JL, Goold SD, Feibelmann S, Campbell EG. Responses of medical schools to institutional conflicts of interest. JAMA 2008; 299(6):665-671. doi: 10.1001/jama.299.6.665 [Crossref] [ Google Scholar]
29. Lo B, Wolf LE, Berkeley A. Conflict-of-interest policies for investigators in clinical trials. N Engl J Med 2000; 343(22):1616-1620. doi: 10.1056/nejm200011303432206 [Crossref] [ Google Scholar]
30. Weinfurt KP, Hall MA, Hardy NC, Friedman JY, Schulman KA, Sugarman J. Oversight of financial conflicts of interest in commercially sponsored research in academic and nonacademic settings. J Gen Intern Med 2010; 25(5):460-464. doi: 10.1007/s11606-010-1264-6 [Crossref] [ Google Scholar]
31. Sierles FS, Brodkey AC, Cleary LM. Medical students’ exposure to and attitudes about drug company interactions: a national survey. JAMA 2005; 294(9):1034-1042. doi: 10.1001/jama.294.9.1034 [Crossref] [ Google Scholar]
32. Kaufman PE, Cohen JE, Ashley MJ. Tobacco industry links to faculties of medicine in Canada. Can J Public Health 2004; 95(3):205-208. doi: 10.1007/bf03403650 [Crossref] [ Google Scholar]
33. Rochon PA, Sekeres M, Lexchin J. Institutional financial conflicts of interest policies at Canadian academic health science centres: a national survey. Open Med 2010; 4(3):e134-138. [ Google Scholar]
34. Campbell EG, Vogeli C, Rao SR, Abraham M, Pierson R, Applebaum S. Industry relationships among academic institutional review board members: changes from 2005 through 2014. JAMA Intern Med 2015; 175(9):1500-1506. doi: 10.1001/jamainternmed.2015.3167 [Crossref] [ Google Scholar]
35. Campbell EG, Weissman JS, Vogeli C. Financial relationships between institutional review board members and industry. N Engl J Med 2006; 355(22):2321-2329. doi: 10.1056/NEJMsa061457 [Crossref] [ Google Scholar]
36. Carlat DJ, Fagrelius T, Ramachandran R, Ross JS, Bergh S. The updated AMSA scorecard of conflict-of-interest policies: a survey of US medical schools. BMC Med Educ 2016; 16(1):202. doi: 10.1186/s12909-016-0725-y [Crossref] [ Google Scholar]
37. Base transparence santé. https://www.transparence.sante.gouv.fr/flow/main;jsessionid=26597601D207A4E8DA579E8E8D9525DB?execution=e1s1.
38. Ethical and Deontological Charter of the Deans’ Conferences of Medicine and Odontology Schools. Formindep website. Available from: https://formindep.fr/wp-content/uploads/2017/11/dean-charter-english.pdf. Published 2017.
39. Classement des facultés françaises en matière d’indépendance. https://facs2018.formindep.fr/intro.html. Published 2018.
40. Wright MS, Robertson CT. Heterogeneity in IRB policies with regard to disclosures about payment for participation in recruitment materials. J Law Med Ethics 2014; 42(3):375-382. doi: 10.1111/jlme.12153 [Crossref] [ Google Scholar]
41. Resnik DB, Babson G, Dinse GE. Minor changes to previously approved research: a study of IRB policies. IRB 2012; 34(4):9-14. [ Google Scholar]
42. King M, Essick C, Bearman P, Ross JS. Medical school gift restriction policies and physician prescribing of newly marketed psychotropic medications: difference-in-differences analysis. BMJ 2013; 346:f264. doi: 10.1136/bmj.f264 [Crossref] [ Google Scholar]
43. Epstein AJ, Busch SH, Busch AB, Asch DA, Barry CL. Does exposure to conflict of interest policies in psychiatry residency affect antidepressant prescribing?. Med Care 2013; 51(2):199-203. doi: 10.1097/MLR.0b013e318277eb19 [Crossref] [ Google Scholar]
44. Rasmussen K, Bero L, Redberg R, Gøtzsche PC, Lundh A. Collaboration between academics and industry in clinical trials: cross-sectional study of publications and survey of lead academic authors. BMJ 2018; 363:k3654. doi: 10.1136/bmj.k3654 [Crossref] [ Google Scholar]
45. Whiting P, Wolff R, Mallett S, Simera I, Savović J. A proposed framework for developing quality assessment tools. Syst Rev 2017; 6(1):204. doi: 10.1186/s13643-017-0604-6 [Crossref] [ Google Scholar]