Research integrity in reporting health research : perceptions and experiences around plagiarism, conflict of interest and authorship criteria in low- and middle-income countries

[摘要] ENGLISH SUMMARY : Background: There is little research on research integrity in low- and middle-income countries (LMICs). This thesis investigates perceived and actual research reporting practices in relation to authorship, plagiarism, redundant publication and conflicts of interest amongst LMIC health researchers.Methods: To take stock of existing research, we summarised prevalence and causes of research misconduct amongst health researchers from LMICs in a systematic review. We then explored perceptions and awareness of poor practices through an online survey of Cochrane authors based in LMICs, using hypothetical scenarios to elicit responses. We gained more insight through follow-up, in-depth interviews with willing survey respondents. Next, we described actual practices in African biomedical journals indexed on the Africa Journals Online database. We measured plagiarism by screening articles with text-matching software, and analysed those with an overall similarity index (OSI) >10% with a pre-specified plagiarism framework. We analysed journal policies and author guidelines and assessed adherence to these in a random selection of articles. Lastly, we piloted a workshop to introduce good reporting practices in two academic institutions in Malawi and Nigeria. We used a variety of teaching approaches to facilitate learning, and based discussions on scenarios.Results: Existing studies from LMICs reported high prevalence of research misconduct. However, studies had limitations related to selection of participants and outcome measurements. One hundred and ninety-nine (34%) Cochrane authors from LMICs responded to the online survey. Of those, 77% reported that guest authorship occurred at their institution, 60% reported text-recycling, 43% reported plagiarism of ideas and 40% indicated that their colleagues had not declared conflicts of interest in the past. Four themes emerged from the qualitative data: 1) authorship rules are simple in theory, but not consistently applied; 2) academic status and power underpin behaviours; 3) institutions and culture fuel bad practices; and 4) researchers are uncertain about what conflicts of interest means, and how this may influence research. We screened 495 published articles from 100 journals for plagiarism. Of the 358 articles with an OSI >10%, we found plagiarism in 73% (95%CI 67 to 78), comprising one to two copied sentences in 26% (95%CI 22 to 31), three to six copied sentences in 25% (95%CI 20 to 29), and at least four linked or more than six copied sentences in 22% (95%CI 18 to 28). Journal policies and author guidelines were lacking, especially amongst non-commercial journals. Existing guidelines were poorly implemented. Workshop participants acknowledged the importance of research integrity and engaged in discussions and activities.Conclusions: Researchers across LMICs report that poor research reporting practices are common. They are mostly concerned about widespread guest authorship. Actual rates of plagiarism in African biomedical articles are very high. Conflicts of interest are poorly understood and not declared. The desire for academic status, institutional systems linked to promotions and organisational culture fuel bad practices. Efforts to promote research integrity should be multi-faceted and targeted at various stakeholders, including institutions and journals. Future research should identify effective interventions to promote research integrity in LMICs. Further testing of our plagiarism framework is needed.

[发布机构] Stellenbosch University