Introduction

Trustworthiness is central to medical research, especially studies with human applications. Polices exist to ensure the integrity of research and merit of science. Among these polices, the disclosure of conflicts of interest (COI) is an important procedure. The Institution of Medicine defines COIs as “circumstances that create a risk that professional judgments or actions regarding a primary interest will be unduly influenced by a secondary interest” [1]. Potential, or actual, COIs mainly exist in research, editorial and peer-review processes, although the latter two receive less attention. COIs of authors, editors, referees, and other participants arise when their primary and secondary interests contradict each other. Authors’ declarations are key to establishing both their financial and nonfinancial COIs. Editors, referees and other participators should also be cautious about COIs that could bias their opinions. Comprehensive guidelines on COI disclosure have been published and promoted by the International Committee of Medical Journal Editors (ICMJE) and Committee on Publication Ethics (COPE) to enhance reproducibility and transparency and to help participants involved in medical publishing to distribute accurate, unbiased research.

There is anecdotal evidence indicating that authors who have relationships with industry tend to publish more positive results, which may favor the industry [2–4]. Negative events have highlighted the need to safeguard scientific integrity and public health. Stakeholders should make efforts to prevent bias caused by COIs and furthermore to discourage ill-considered relationships resulting in COIs [5].

Academic publishing is a cornerstone of the critical evaluation and dissemination of research findings. In efforts to preserve the transparency and integrity of medical research, the overarching role of medical journals cannot be ignored [6]. Efforts have been made to improve the awareness and engagement of authors, referees and publishing participants in COI disclosure and management. Top medical journals have adopted sophisticated COI policies in efforts to ensure the quality of published papers. From 1988 to 2008, the lack of funding disclosure in papers published in three high-impact medical journals: The Lancet, New England Journal of Medicine, and Journal of American Medical Association, decreased from 35% to 7%, and COI reporting increased from 2% to 84% (p < 0.0001) [7]. A study in 2014 showed that all but one of 117 core clinical journals indexed under Abridged Index Medicus had a COI policy [8].

High-impact medical journals have paid increasing attention to COI disclosure. Nevertheless, the awareness and enforcement of COI disclosure policy in Chinese medical journals remains unclear. This study is the first report of a comprehensive analysis of COI disclosure polices among Chinese medical journals, addressing characteristics and contents of the policies and details of financial and nonfinancial COI disclosure in research, editorial, and peer-review processes. We investigated the current status of COI policies applied by Chinese core journals in the field of medicine and health to provide an indication of how COI policies are read and implemented in China.

Methods

We conducted a cross-sectional study on journal instructions, journal websites and print issues of Chinese medical journals in the medical and health sector. The presence, completeness and related details of COI disclosure policies were examined.

Results

Characteristics and contents of conflicts of interest disclosure policies among journals

Of 248 eligible journals, 78 (31%) mentioned COI policies; 9 (4%) applied standardized disclosure forms; and 18 (7%) required disclosure statements in articles’ footnotes or at the end of the text. Of 496 articles examined, just 34 (7%) included a COI disclosure statement (Table 1). There was a statistically significant difference among journals with COI policies under categories of comprehensive medicine and health care (7 of 37 journals, 19%), preventive medicine and hygiene (7 of 27 journals, 26%), traditional Chinese medicine (1 of 19 journals, 5%), preclinical medicine (6 of 24 journals, 25%), comprehensive clinical medicine (52 of 116 journals, 45%), special medicine (4 of 10 journals, 40%), and pharmacy (1 of 15 journals, 7%) (p = 0.001, chi-square test). In multiple comparisons, a statistically significant difference was found in journals with COI policies between categories of traditional Chinese medicine and comprehensive clinical medicine (p = 0.001, Chi-square test; Bonferroni threshold: p < 0.002). There were more journals with COI policies in categories of comprehensive clinical medicine compared with comprehensive medicine (p = 0.005, Chi-square test) or pharmacy (p = 0.005, Chi-square test), but did not meet the prespecified Bonferroni threshold of p < 0.002. Whereas, the remaining 18 cases of pairwise comparisons between different disciplines of journals with COI policies showed no statistically significant difference. We noticed that 60 journals with COI disclosure policies had no disclosure statements in the published articles, while four journals with statements in print issues had no policies specified in the journal instructions or on their websites.

Download:

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

Table 1. Characteristics of COI disclosure policies among journals [n (%)].

https://doi.org/10.1371/journal.pone.0219564.t001

Four journals (2%) mentioned the basis of their COI policies, and the bases were the guidelines of ICMJE or COPE; 62 (25%) and 55 (22%) journals noted financial and nonfinancial COIs specifically; 78 (31%), 7 (3%), and 12 (5%) journals highlighted COIs in research, editorial, and peer-review processes, respectively; no journal had any procedures to authenticate disclosed COIs; 2 (1%) journals mentioned how to deal with COI policy breaches; and 18 (7%) journals had policies to manage disclosed COIs. In several disciplines, most of the aforementioned contents of COI policies were not mentioned (Table 2). Of journals with IF ≥ 1.039, the presence of COI policies, financial COIs, nonfinancial COIs, COIs in research, editorial process, and peer-review process, did not differ from those with IF < 1.039, respectively (p > 0.05, Chi-square test or Fisher’s exact test) (S1 Table). It is worth noting that although the percentage of journals that required COI disclosure in research was low, 234 (94%) journals did require a disclosure related to research funding.

Download:

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

Table 2. Contents of COI policies among journals [n (%)].

https://doi.org/10.1371/journal.pone.0219564.t002

Specific items of COIs in research

All of the 78 journals (31%) that mentioned COI policies pointed out the authors’ COIs; 2 (1%) mentioned that it was necessary to disclose family members’ COIs; 6 (2%) pointed out institutional COIs; and no other COI subject was mentioned. Of 17 discipline categories of journal 15 did not mention family members’ COIs and 12 did not mention institutional COIs (Table 3).

Download:

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

Table 3. COI subjects involved in research [n (%)].

https://doi.org/10.1371/journal.pone.0219564.t003

Financial COIs mentioned relating to research covered grants (n = 22, 9%), personal fees (n = 13, 5%), indirect financial support (n = 3, 1%), stock shares (n = 10, 4%), employment (n = 14, 6%), and other types (n = 13, 5%). Other types included expert testimony, board membership, patent application/registration, and benefit-based relationships. Only seven journals (3%) under six discipline categories noted that the source of financial support should be disclosed, but none mentioned the amount (Table 4). Nonfinancial COIs relating to research covered intellectual COIs (n = 4, 2%), personal relationships (n = 7, 3%), academic competition (n = 5, 2%), belief (n = 1, 0%) and other types, i.e., politics (n = 2, 1%) (Table 5).

Download:

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

Table 4. Specific items of financial COIs in research [n (%)].

https://doi.org/10.1371/journal.pone.0219564.t004

Download:

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

Table 5. Specific items of nonfinancial COIs in research [n (%)].

https://doi.org/10.1371/journal.pone.0219564.t005

Discussion

Conscious or unconscious bias caused by COIs can compromise the public’s confidence in scientific research, and even cloud the truth, ultimately jeopardizing patient welfare [1]. Medical science has made dramatic contributions to human health and well-being, but the conscious or unconscious bias caused by COIs could unduly compromise further progress. A core goal of medical science has always been to guarantee public trust in medicine. Sophisticated policies established by ICMJE and COPE are aimed to ensure the proper declaration and interpretation of COIs. However, ensuring adequate reporting of all sources of COIs is becoming increasingly challenging as a result of the growing complexity of funding mechanisms and academic culture [9]. One way is to enhance participants’ awareness of COIs but also to not ignore medical journals’ critical role as guardians of propriety and transparency [10]. Leading journals are adopting and promoting COI disclosure policies as part of fulfilling this responsibility [11, 12].

An increasing number of domestic Chinese medical associations and journals’ affiliations and sponsors are recognizing the importance of publication ethics; adopting polices and guidelines from international organizations and aiming to promote policy enforcement in China. The published literature addressing this issue supports this drive. In this context, we present a cross-sectional study of Chinese medical journals, aiming to investigate the current status of COI disclosure policy enforcement, and reflect how COIs are currently disclosed and recognized.

We found 248 eligible medical journals in which 78 (31%) mentioned COI policies. On the other hand, journals cited by Abridged Index Medicus presented sound COI policies [8]. Furthermore, a study published in 2011 showed that of 52 oncology journals listed in the Thomson Institute for Scientific Information, 51 (98%) identified COI disclosures, and 31 (61%) included disclosure statements [12]. In our investigation, only one out of nine oncology journals had COI disclosure statements. While a cross-sectional study carried out in 2012 showed that 9 (90%) Indian and 25 (92.5%) British dental journals required COI declaration [13]. Only 2 out of 5 stomatology journals in our study managed to do so. Compared with these English-language medical journals, some even being reported from more than a decade ago, Chinese medical journals left much to be desired in adherence to guidelines of COI disclosure policies, despite four journals claiming that their COI policies were based on ICMJE or COPE guidelines. Nevertheless, clinical journals that were more likely to be linked with industry generally paid more attention to COI policies (45%).

It is worth noting that 60 journals mentioned COI policies but had no statement in print, and articles with COI statements were published in four journals that did not mention COI policies. This phenomenon was also observed in another study [14]. On the other hand, in two journals, only one of two articles that were examined included a COI declaration statement. This might suggest that the policies were implemented arbitrarily or that authors might be reluctant to disclose COIs publicly. In this circumstance, medical journals should not only specify COI policies but also define them. Authors should be provided with clear guidance on understanding and managing COIs to enhance their engagement with COI disclosure policies.

A sound COI disclosure policy should specify details such as subjects, financial and nonfinancial COIs, and their types. We found that 78 journals (31%) stated that authors should disclose COIs; 2 (1%) and 6 (2%) journals mentioned family members and institutional COIs, respectively. Sixty-two journals (25%) noted financial COIs, and 22 (9%) specified at least one form of financial COI type. Only seven journals (3%) noted that the source of financial support should be disclosed, but no journals mentioned the amount. Fifty-five journals (22%) noted nonfinancial COIs, and 11 (4%) specified nonfinancial COI types. Compared with individual COIs, institutional COIs were less frequently required to be reported. Fortunately, of the journals that mentioned financial COIs, most also pointed out nonfinancial COIs. Comparing periodicals under other indices, in 2014, all 117 core clinical journals indexed under Abridged Index Medicus required the disclosure of financial COIs regarding authors, but only a minority required the disclosure of family members’ (35%) or authors' institution’s (29%) COIs; 57% required nonfinancial COI types, of which two (3%) referred to intellectual COIs [8]. Of 399 biomedical journals of the JCR in 2011, 358 (89.7%) required financial COI disclosure and 280 (70.2%) required nonfinancial COI disclosures [14]. This study also highlighted that clinical journals paid more attention to COI policies than those in fundamental disciplines [14], and this result was consistent with our findings, although it was not statistically significant. Investigation across various fields could provide more insight into the COI disclosure issue [15] and promote journals’ responsibilities to encourage COI declaration policies across different academic cultures.

We also found there was no association between IF and the presence COI policies. This might be due to small differences in journals’ IF, and it might indicate there was a general lack of policies. Nevertheless, it is worth noting that although the number of journals that required COI disclosure in our study was low, 234 journals (94%) did require a disclosure relating to research funding representing a major step forward in the adoption of COI policies.

While journals adopt policies to enhance authors’ understanding of COI declaration, they themselves carry considerable responsibility in implementing policies. They are both supervisors and participators; but only seven journals (3%) noted COIs related to the editorial process. In 2017, a study showed that 34% of 72 health policy and services journals described COIs in the editorial process [16]. In a cross-sectional study of 399 biomedical journals of the JCR in 2011, 155 (38.8%) required editors’ COI disclosures [14]. These studies suggest that there is a generic lack of accessible editorial COI policy among medical journals. COIs involve not only individual but also intuitional behavior. However, institutional COIs have generally received less attention than individual COIs [11]. We found that no journals stated that there should be disclosure of institutional COIs.

Peer-review is a key procedure in scholarly publishing practice; letting referees play a critical role in COI disclosure. Here, 12 journals (5%) mentioned COIs in the peer-review process, and 8 (3%) required referees to disclose COIs. Another survey published in 2015 reported that among 121 Japanese Association of Medical Sciences (JAMS) journals, 60 (49.6%) required editorial committee members to disclose COIs [17]. Encouraging medical journals to adopt comprehensive COI disclosure policies pertaining to the peer-review process, providing guidance on COIs for referees, adopting transparent peer-review, and building reliable referee pools based on disclosed COIs and other critical information might be effective in addressing these issues [18, 19].

Although disclosure has become one of the central approaches to managing COIs, critics contend that it does not serve as a reliable indicator of research credibility, and adopting sophisticated mechanisms to verify the accuracy and completeness of disclosed COIs and to address the breach of COI policies is key [20]. Here, no journal mentioned procedures to authenticate disclosed COIs, and only two journals (1%) mentioned how to deal with any breach of COI disclosure policy. Data were also less than satisfactory among journals in other indexes. As reported in 2017, 72 health policy and services journals had no policy checking the accuracy of disclosed COIs, and only 21% explicitly stated that inaccurate disclosures of COI might result in manuscript rejection [16]. A survey conducted in 2014 reported that, among 121 JAMS journals, only 24 (19.8%) verified the accuracy of declared COIs, and only 33 (27.3%) had clearly stated consequences for violators [17].

We found that 18 journals (7%) had policies to manage disclosed COIs. The proportion was low but still higher than that of the 72 health policy and services journals listed by Web of Science, among which no policy had criteria for COI management [16]. It is noteworthy that of the 18 journals in our study, 11 noted that all authors needed to be able to declare that no COI existed otherwise papers with COIs might be rejected. Some COIs might be unavoidable and it might be too conservative to require a zero tolerance of COIs, such as any link with industry, which may contain risks that could compromise the research results. However, successful collaboration facilitates achievement in medical research and we need to find an appropriate way to manage such collaboration, rather than simply denying it. The severity of COIs depends on the likelihood of undue influence and the seriousness of the potential harm. On the journals’ side, comprehensive criteria for assessing the severity of COIs are in demand. Maharaj [21] proposed a scale to quantify financial COIs on 11 levels. This represents a valuable contribution to the more objective review of research results. COI management should consider the value and scope of secondary interests and the seriousness of potential impacts and make all sources of information available to the public.

Our analysis has certain limitations. We enrolled Chinese core journals, but most Chinese journals were not listed by the index used. Nevertheless, our analysis indicates the main trends in academic publishing in this area. Second, our analysis is based on publicly available information. It could be incomplete as unpublished policies may exist for the processing of manuscript submissions and transcript revisions. Moreover, our study is cross-sectional, looking at a single moment in an ever-changing field but it still represents a contribution to the understanding of current application of COI policies in Chinese medical journals. Despite these flaws, our investigation is an in-depth evaluation of 17 fields of medicine and health, representing its generalizability. One of the aims of this study was to motivate journals to adopt comprehensive policies. Thereby, medical periodicals could become a force in promoting a culture in which COIs are taken seriously by stakeholders. Ultimately, COIs that could compromise primary interests should be discouraged.

Scholarly publishing has the strength of influencing science. To safeguard the integrity of research, journals should recognize the importance of, and explicitly state, comprehensive COI disclosure policies in research, editorial and peer-review processes to promote transparency in rapidly evolving medical areas. Approximately one-third of Chinese medical journals were found to have COI policies, however, the extent to which journals adopted accessible and comprehensive COI policies was insufficient. There was a generic lack of standardized disclosure forms and of the management of COIs in most journals, and the subjects and details of the COIs involved in editorial and peer-review processes received less attention than those in research. Fortunately, in contrast to financial COIs, nonfinancial ones were not neglected in the disclosure policies. There is much to be done, and there may be a long way to go, but medical journals, as one of guardians of medical research quality, should adhere to accepted COI disclosure guidelines, close gaps and loopholes in their COI policies, and make the best decisions for medical publishing practice.

Supporting information