Basic & Clinical Pharmacology & Toxicology Policy for Experimental and Clinical studies

Abstract

Inconsistencies in the reproducibility of research findings are of pertinent concern throughout the scientific community. Initially addressed in pre-clinical biomedicine studies 1, 2, the discussion has expanded to fields outside the natural sciences, questioning the reliability of data and subsequent published findings, proposing a ‘Crisis in Confidence’ or ‘Reproducibility Crisis’ within several (if not all) research areas 2-6. The disclosure of an apparent and low degree of reproducibility has fuelled investigations of causal factors, ranging from behavioural preferences in decision-making/’human nature’ and conviction bias to experimental constituents 5, 7, 8. In its wake is the ongoing discussion on how to move forward and restore confidence in the scientific literature as a whole. The increased awareness has identified several aspects in data collection, analysis and subsequent interpretation to be carefully considered. Goodman et al. 9 suggest three target descriptors of research reproducibility: methods reproducibility, results reproducibility and inferential reproducibility. These refer to the ability to reach similar results by applying equal methodology (e.g. in data analysis), the ability to reproduce findings when conducting a novel study in similar experimental conditions (data replicability) and, lastly, to reach concurrent conclusions when reproducing data either by repeating the data analysis or the study 9. This proposed subdivision in terminology illustrates the complexity and multifaceted origin of random or systematical error/bias easily incorporated in experimental design, analytical methods and decision-making. Counteracting the virtues of randomization, blinding and control, factors such as HARKing (Hypothesizing After the Results are Known) 5, 10, P-hacking (choosing statistical methods to reach specific levels of significance) 9, 11 and publication bias 11 all contribute to misinterpretation and overestimation of findings. Although likely unintentional, the combined impact of these shortcomings undermines the validity and credibility of the scientific literature. A range of preventive measures and considerations has been proposed to improve reproducibility. In this, research transparency has been highlighted as instrumental 4, 8. Enhancing transparency enables independent replication of published findings, and for findings to be reviewed in greater detail, including comparisons of methodology, statistical applications and experimental protocols, raising the credibility and reducing the risk of unnecessary multiplications of performed studies. In this aspect, scientific journals can contribute by defining criteria to improve reproducibility as part of their pre-submission guidelines, in this way recognizing publication standards as an important tool in expanding documentation and encouraging authors to report and share their research in greater detail. Several journals and research institutions have recently taken steps towards such measures, but though concerns are shared, standardized guidelines have not yet been implemented on a global scale 8. Basic & Clinical Pharmacology & Toxicology (BCPT) receives an increasing number of manuscript submissions each year. Through the efforts of reviewers assisted by the editorial team, the submitted manuscripts are scrutinized as part of the selection process prior to a putative acceptance for publication. Despite the high quality of many of the submitted manuscripts, the editorial team is frequently faced with rejecting submissions due to insufficient disclosure of experimental details and/or failure to comply with generally accepted standards of experimental design and data analysis. The editorial team has therefore decided to revise the BCPT guidelines for authors to ensure that research transparency and standards are uniformly communicated for both experimental/pre-clinical (Table 1) and clinical studies (Table 2). Extensive efforts have been made to implement an increased focus on the validity of pre-clinical and clinical studies and to comply with ethical principles, such as the 3R principles for animals applied in research 12, 13. This has resulted in several initiatives outlining good research practice, for example ARRIVE guidelines 14, CIOMS-ICLAS principles (https://www.icsu.org/members/iclas-international-council-for-laboratory-animal-science visited April 2018) and the experimental design assistant (EDA) 15. The recommendations propose several measures to improve reliability during all phases of experimentation starting in the early planning phase. This includes generating a hypothesis prior to designing the experiment, determining statistical power and group sizes based on pre-experimental considerations and diminishing sources of bias (e.g. blinding and randomization) to reduce the risk of flawed statistical analysis and the overestimation of a hypothesized effect. Specifically for studies including experimental animals, considerations of animal welfare, such as husbandry, pain management and humane end-points, are crucial in preserving animal health, hereby complying with welfare legislations and ensuring reproducibility of in vivo studies. The guidelines for experimental/pre-clinical studies are shown in Table 1. A plethora of guidelines exists for clinical studies, that is studies with human participants. The EQUATOR network (Enhancing the QUAlity and Transparency Of health Research) is an international initiative that seeks to promote accurate reporting of health science globally. BCPT welcomes this initiative. The network has gathered knowledge and links to guidelines for main study types such as randomized trials (CONSORT), observational studies (STROBE), systematic reviews (PRISMA), study protocols (SPIRIT) and clinical practice guidelines (AGREE). The EQUATOR website (available at http://www.equator-network.org/– accessed April 2018) currently links close to 400 reporting guidelines (Table 2). The General Data Protection Regulation (GDPR) implemented on 25 May 2018 covers human individuals within the European Union (EU) and the European Economic Area (EEA) (https://ec.europa.eu/commission/priorities/justice-and-fundamental-rights/data-protection/2018-reform-eu-data-protection-rules_en, accessed 8 June 2018). The reform offers stronger rules and gives people more control over their personal data. However, it does not go beyond the principles of The Declaration of Helsinki, which generally aims at promoting and ensuring research subjects’ health, dignity, integrity, right to self-determination, privacy and confidentiality of personal information. It falls within the responsibility of any scientific journal to endorse that the published studies are reliable, for example that experimental groups include sufficient numbers to achieve an acceptable power, hereby allowing for confidence in the reported findings. The revised guidelines aim to aid authors in providing such sufficient information. Reflecting on the debate on transparency in research and means to regain confidence in published results, it is the hope of the BCPT editorial team that the issued guidelines may also be considered helpful in the early phases of an experimental design, providing input as to how reproducibility may be strengthened.