Abstracts for reports of randomised trials of COVID-19 interventions had low quality and high spin

1. IntroductionIn December 2019, a novel RNA coronavirus, which was subsequently named severe acute respiratory syndrome coronavirus (SARS-CoV-2) caused an outbreak of pneumonia in Hubei province of China, and quickly transmitted to other countries, resulting in millions of infections and deaths globally in just a few months. Unfortunately, there were no efficient methods to kill this virus or to treat the coronavirus disease 2019 (COVID-19) it caused. To address this, a large number of randomized controlled trials (RCTs) emerged very quickly to test interventions to prevent or treat COVID-19, along with a vast number of other pieces of research. This rapid, unprecedented outpouring of research on a specific condition means that policy makers, patients and clinicians are under great pressure to identify the most useful and reliable information from amidst an overwhelming number of articles. They need relevant and valid material which can be accessed quickly and with minimal efforts [What clinical information do doctors need?.], and as a result, the abstracts of the full reports become key to supporting clinical decision making [Johnson HL Fontelo P Olsen CH Jones KD Gimbel RW. Family nurse practitioner student perception of journal abstract usefulness in clinical decision making: a randomized controlled trial.]. Clear reporting of a study's results in its abstract is likely to aid rational decision making, but selective outcome reporting and other biases in the abstracts may make readers disoriented. It will be also important for users of these RCTs to consider whether the studies were justified, given that this has been identified as a problem for RCTs in the past [Walters C Torgerson T Fladie I Clifton A Meyer C Vassar M. Are randomized controlled trials being conducted with the right justification?.].To help readers, the consolidated standards of reporting trials (CONSORT) statement, published in 2010, lists 16 essential items for inclusion in the abstracts for reports of RCTs [Moher D Hopewell S Schulz KF Montori V Gotzsche PC Devereaux PJ et al.CONSORT 2010 Explanation and Elaboration: Updated guidelines for reporting parallel group randomised trials.]. This provides the requirements for the content of a well reported abstract, but selectivity in how the authors present their results remains a problem, especially in studies without statistically significant outcomes where the authors might use spin to try to highlight results inappropriately.The concept of spin was originally used in fields such as culture, politic and commerce, and is defined as a form of propaganda that could affect public views on an organization or public figure [Fang X Hua F Riley P Chen F Zhang L Walsh T et al.Abstracts of published randomised controlled trials in Endodontics: Reporting quality and spin.]. In health care, it was defined by Boutron and colleagues in 2010 as “specific reporting strategies from whatever motive, highlighting the interventions are beneficial despite the statistically non-significant differences for the primary outcomes” [Boutron I Dutton S Ravaud P Altman DG. Reporting and interpretation of randomized controlled trials with statistically nonsignificant results for primary outcomes.]. Since then, it has been widely evaluated in several medical specialties [The SSSPIN study-spin in studies of spin: meta-research analysis.], including stomatology [Fang X Hua F Riley P Chen F Zhang L Walsh T et al.Abstracts of published randomised controlled trials in Endodontics: Reporting quality and spin., Wu X Yan Q Fang X Hua F Shi B Tu YK. Spin in the abstracts of randomized controlled trials in periodontology and oral implantology: A cross-sectional analysis.], otolaryngology [Cooper CM Gray HM Ross AE Hamilton TA Bea Downs J Wayant C et al.Evaluation of spin in the abstracts of otolaryngology randomized controlled trials.], surgery [Checketts JX Riddle J Zaaza Z Boose MA Whitener JH Vassar MB. An Evaluation of Spin in Lower Extremity Joint Trials., Arunachalam L Hunter IA Killeen S. Reporting of Randomized Controlled Trials With Statistically Nonsignificant Primary Outcomes Published in High-impact Surgical Journals.] and cardiology [Khan MS Lateef N Siddiqi TJ Rehman KA Alnaimat S Khan SU et al.Level and Prevalence of Spin in Published Cardiovascular Randomized Clinical Trial Reports With Statistically Nonsignificant Primary Outcomes: A Systematic Review.]. Considering that such a large number of COVID-19 articles are being published with shorter periods for peer review and editorial oversight, this raises the possibility that more spin is making its way into the literature and that readers are being misled. Therefore, we have undertaken what we believe to be the first study of spin in reports of RCTs of interventions for COVID-19. This study investigates (1) the reporting quality of the abstracts for published RCTs of interventions for COVID-19 and their use of spin strategies, and the extent and level of spin in abstracts with statistically non-significant primary outcomes, and (2) potential predictors for reporting quality and the severity of spin.4. DiscussionTo the best of our knowledge, this is the first study to systematically assess the reporting quality and presence of spin in the abstracts of reports of RCTs testing interventions for COVID-19. The CONSORT statement for abstracts provides authors with guidance on the necessary details and clarity required for good reporting. It is intended to improve the reporting quality of abstracts, and recently, has been used as a tool to evaluate the reporting quality of abstracts [Song SY Kim B Kim I Kim S Kwon M Han C et al.Assessing reporting quality of randomized controlled trial abstracts in psychiatry: Adherence to CONSORT for abstracts: A systematic review., Speich B Mc Cord KA Agarwal A Gloy V Gryaznov D Moffa G et al.Reporting Quality of Journal Abstracts for Surgical Randomized Controlled Trials Before and After the Implementation of the CONSORT Extension for Abstracts., Narayan VM Cone EB Smith D Scales CD Dahm P Improved Reporting of Randomized Controlled Trials in the Urologic Literature., Ghimire S Kyung E Lee H Kim E Oncology trial abstracts showed suboptimal improvement in reporting: a comparative before-and-after evaluation using CONSORT for Abstract guidelines., Chhapola V Tiwari S Brar R Kanwal SK. Reporting quality of trial abstracts-improved yet suboptimal: A systematic review and meta-analysis.]. Our results show that the overall reporting quality of the abstracts of these COVID-19 RCTs is far from satisfactory, with adherence of reports to the CONSORT items ranging from 2.5% to 97.5%. Notably, the CONSORT items most inadequately reported were those related to trial design, randomization, blinding, and trial status. Other studies in the medical literature have also shown a high prevalence of non-adherence to the CONSORT guidelines in RCT abstracts [Reporting quality for abstracts of randomized controlled trials in cancer nursing research., Kumar S Mohammad H Vora H Kar K. Reporting Quality of Randomized Controlled Trials of Periodontal Diseases in Journal Abstracts-A Cross-sectional Survey and Bibliometric Analysis., Khan MS Shaikh A Ochani RK Akhtar T Fatima K Khan SU et al.Assessing the Quality of Abstracts in Randomized Controlled Trials Published in High Impact Cardiovascular Journals., Mozetic V Leonel L Leite Pacheco R de Oliveira Cruz Latorraca C Guimaraes T Logullo P et al.Reporting quality and adherence of randomized controlled trials about statins and/or fibrates for diabetic retinopathy to the CONSORT checklist., Gallo L Wakeham S Dunn E Avram R Thoma A Voineskos S The Reporting Quality of Randomized Controlled Trial Abstracts in Plastic Surgery., Reporting quality of randomized controlled trial abstracts in the seven highest-ranking anesthesiology journals.]. For example, Mozetic et al found that the most underreported items were related to the methods items of CONSORT, such as trial design, allocation concealment, implementation of randomization sequence, and blinding [Mozetic V Leonel L Leite Pacheco R de Oliveira Cruz Latorraca C Guimaraes T Logullo P et al.Reporting quality and adherence of randomized controlled trials about statins and/or fibrates for diabetic retinopathy to the CONSORT checklist.]. Gallo et al also found this, with limited adherence to the CONSORT for abstracts checklists among RCT abstracts published in the top plastic surgery journals [Gallo L Wakeham S Dunn E Avram R Thoma A Voineskos S The Reporting Quality of Randomized Controlled Trial Abstracts in Plastic Surgery.]. They concluded that the most poorly reported items were trial registration (4%), method of randomization (2.4%), and source of trial funding (0%). Furthermore, Janackovic and Puljak evaluated 622 RCT abstracts in anesthesiology and observed the lowest consistency in trial design (18%), recruitment status (9%), number of participants analyzed (8%), randomization (3%) and funding (0.2%) [Reporting quality of randomized controlled trial abstracts in the seven highest-ranking anesthesiology journals.]. Even in the top five highest-impact general medical journals, there was lack of adherence to the CONSORT statement for abstracts [Hays M Andrews M Wilson R Callender D O'Malley PG Douglas K. Reporting quality of randomised controlled trial abstracts among high-impact general medical journals: a review and analysis.]. Overall, RCT abstracts in medical journals are poorly reported, providing readers with insufficient information, which means that readers might not get useful information in a short time and might even be misled by the incomplete results. One possible explanation for this failure to follow the guidelines is that some researchers are not familiar with the CONSORT statement, and that they may repeat the structure and content they have seen in other abstracts when they draft their report. In addition, the editorial office has certain responsibility, because abstracts are often structured in accordance with the author guidelines of academic journals. To remedy this, we recommend that authors, peer reviewers and editors make better use of reporting guidelines such as CONSORT and its extensions when preparing, appraising and editing research articles.

It should be noted that poor reporting quality of abstracts cannot be misinterpreted as poor study design. Limited length of the abstract means that it is often not possible to show all details of the research. Therefore, consistent with our results, items on RCT methods tend to have lower reporting rates in abstracts while, at least in journal papers with high impact, methodological details are well formulated in the full text.

The concept of spin, applying to clinical research, means selectively reporting significant findings while neglecting non-statistically significant results. The role of spin is to make imperfect research results more meaningful, and thus stand out from similar studies. Up to now, there is no completely objective evaluation method for spin. In addition to the spin strategies mentioned by Boutron [Boutron I Dutton S Ravaud P Altman DG. Reporting and interpretation of randomized controlled trials with statistically nonsignificant results for primary outcomes.], more unrecognized strategies exist. We assessed the most common and widely used eleven spin strategies, and nearly 60% of the RCT abstracts with statistically non-significant primary outcomes in our study contained spin. High prevalence of spin has also been reported by other studies, ranging widely from 17% to 86% [Fang X Hua F Riley P Chen F Zhang L Walsh T et al.Abstracts of published randomised controlled trials in Endodontics: Reporting quality and spin., Wu X Yan Q Fang X Hua F Shi B Tu YK. Spin in the abstracts of randomized controlled trials in periodontology and oral implantology: A cross-sectional analysis., Cooper CM Gray HM Ross AE Hamilton TA Bea Downs J Wayant C et al.Evaluation of spin in the abstracts of otolaryngology randomized controlled trials., Checketts JX Riddle J Zaaza Z Boose MA Whitener JH Vassar MB. An Evaluation of Spin in Lower Extremity Joint Trials., Arunachalam L Hunter IA Killeen S. Reporting of Randomized Controlled Trials With Statistically Nonsignificant Primary Outcomes Published in High-impact Surgical Journals., Khan MS Lateef N Siddiqi TJ Rehman KA Alnaimat S Khan SU et al.Level and Prevalence of Spin in Published Cardiovascular Randomized Clinical Trial Reports With Statistically Nonsignificant Primary Outcomes: A Systematic Review., Lockyer S Hodgson R Dumville JC Cullum N. "Spin" in wound care research: the reporting and interpretation of randomized controlled trials with statistically non-significant primary outcome results or unspecified primary outcomes., Jellison S Roberts W Bowers A Combs T Beaman J Wayant C et al.Evaluation of spin in abstracts of papers in psychiatry and psychology journals., Latronico N Metelli M Turin M Piva S Rasulo FA Minelli C. Quality of reporting of randomized controlled trials published in Intensive Care Medicine from 2001 to 2010., It's All How You "Spin" It: Interpretive Bias in Research Findings in the Obstetrics and Gynecology Literature.]. We speculate that high spin prevalence in abstracts of COVID-19 RCTs is due to the specific background at that time, that is, on the one hand, lower standards in medical journals and rapid peer review might lead to lax assessment of manuscripts, while on the other hand, high mortality and morbidity of COVID-19 needed prompt evidence on therapies, resulting in the emergence of large numbers of RCTs of interventions which were subsequently found to be ineffective. Consequently, the existence of spin in these early reports seems not surprising.Among various forms of spin strategy, the most common one in the results section of the abstract was focusing on statistically significant secondary outcomes to claim benefit, while no consideration of the statistically non-significant primary outcomes was the most common strategy found in the conclusion sections. Similar findings were reported by Jellison et al [Jellison S Roberts W Bowers A Combs T Beaman J Wayant C et al.Evaluation of spin in abstracts of papers in psychiatry and psychology journals.]. In another study, however, focusing on statistically significant within-group analysis was the most common spin strategy used in the result section of abstracts, and claiming equivalence or non-inferiority of results with non-significance was the most common in the conclusion section [Fang X Hua F Riley P Chen F Zhang L Walsh T et al.Abstracts of published randomised controlled trials in Endodontics: Reporting quality and spin.]. Moreover, Turrentine analyzed 83 scientific publications with spin in abstracts and concluded that the more common types of spin strategies in general obstetrics and gynecology were: emphasizing statistically significant secondary results (40%), interpreting non-statistically significant primary outcomes as equivalent or similar effectiveness (37%) and claiming beneficial effects of treatment despite the statistical non-significance (15%) [It's All How You "Spin" It: Interpretive Bias in Research Findings in the Obstetrics and Gynecology Literature.]. Lockyer et al indicated that there was potential for spin in wound care trials emphasizing study results of significance rather than the importance of outcomes [Lockyer S Hodgson R Dumville JC Cullum N. "Spin" in wound care research: the reporting and interpretation of randomized controlled trials with statistically non-significant primary outcome results or unspecified primary outcomes.]. This is concerning because clinicians are prone to misunderstand the outcomes of a trial when spin is present and make inappropriate clinical decisions [Boutron I Dutton S Ravaud P Altman DG. Reporting and interpretation of randomized controlled trials with statistically nonsignificant results for primary outcomes.]. This makes it especially important to find ways to identify and mitigate spin [Checketts JX Riddle J Zaaza Z Boose MA Whitener JH Vassar MB. An Evaluation of Spin in Lower Extremity Joint Trials.]. Readers should keep the concept of spin in mind when reading abstracts and be aware of the diversity and heterogeneity of spin strategies, and researchers should properly report their results and conclusions within the limited word count for the abstract, instead of giving space to only those results that they wish to highlight to show the importance of their research. Journal reviewers and editors also need to be rigorous in their assessment of manuscripts. Our analyses of the predictors potentially associated with reporting quality show that a larger word count was significantly associated with better quality reporting in RCT abstracts. For this reason, allowing authors more words in their abstracts might be a simple method to improve the reporting quality of abstracts. In addition to larger word counts, reporting of trial registration and funding were also positively correlated with high-quality reporting [Shaqman M Al-Abedalla K Wagner J Swede H Gunsolley JC Ioannidou E. Reporting quality and spin in abstracts of randomized clinical trials of periodontal therapy and cardiovascular disease outcomes.]. In the multivariate analysis of relevant factors to spin severity, we found that research from non-Asian regions might be relevant to fewer spin strategies. Both Cooper and Reynolds-Vaughn demonstrated that a majority of abstracts with spin were funded by industry [Cooper CM Gray HM Ross AE Hamilton TA Bea Downs J Wayant C et al.Evaluation of spin in the abstracts of otolaryngology randomized controlled trials., Reynolds-Vaughn V Riddle J Brown J Schiesel M Wayant C Vassar M. Evaluation of Spin in the Abstracts of Emergency Medicine Randomized Controlled Trials.], while Jellison et al found no relationship between industry funding and spin in abstracts [Jellison S Roberts W Bowers A Combs T Beaman J Wayant C et al.Evaluation of spin in abstracts of papers in psychiatry and psychology journals.], and use of statistician and article section were further confirmed to be unrelated to the presence of spin in another study [Arunachalam L Hunter IA Killeen S. Reporting of Randomized Controlled Trials With Statistically Nonsignificant Primary Outcomes Published in High-impact Surgical Journals.]. Furthermore, numbers of research centers were reported associated with presence of spin by Wu et al [Wu X Yan Q Fang X Hua F Shi B Tu YK. Spin in the abstracts of randomized controlled trials in periodontology and oral implantology: A cross-sectional analysis.], and Checketts et al found that word count limit promoted the prevalence of spin [Checketts JX Riddle J Zaaza Z Boose MA Whitener JH Vassar MB. An Evaluation of Spin in Lower Extremity Joint Trials.]. Whatever, these findings raise concerns about the reporting specification of abstracts, and it is of vital importance for researchers to objectively and accurately report their findings.

There are some limitations to our study. Although we included all abstracts for RCTs of interventions for COVID-19 that we identified in PubMed up to the end of October 2020, the number of included RCTs is still relatively small, which means that our estimates might change if the study was expanded to include more abstracts. Second, although we evaluated spin strategies with a predesigned 11-item form used in other studies, some other potential spin strategies might have been omitted, leading to an underestimate of the presence of spin. Third, only RCTs testing interventions for COVID-19 were included when evaluating the reporting quality and spin of abstracts, that is to say, our analyses was focused on abstract sections, and our conclusions of poor reporting quality and high spin should not be extended to the full text. Fourth, evaluations of reporting quality and spin are subjective, and although we adopted an approach of double, independent and calibrated assessment to control the magnitude of subjectivity, if others repeated our assessments they might obtain different results. Despite these limitations, our study provides important new insights for the reporting quality of RCT abstracts which may have implications to research reporting more generally, as well as its specific relevance to the reporting of RCTs of interventions for COVID-19.

Creadit author statement

Dongguang Wanga: Data extraction, statistical analysis and interpretation; Drafting the manuscript.

Lingmin Chen: Data extraction, statistical analysis and interpretation; Drafting the manuscript.

Lian Wang: Data extraction, statistical analysis and interpretation.

Fang Hua: Data extraction, statistical analysis and interpretation; Revising the manuscript.

Juan Li: Data extraction, statistical analysis and interpretation.

Yuxi Li: Data extraction, statistical analysis and interpretation.

Yonggang Zhang: Concept and design; Data extraction, statistical analysis and interpretation; Drafting the manuscript.

Hong Fan: Concept and design; Supervision.

Weimin Li: Concept and design; Supervision.

Mike Clarke: Data extraction, statistical analysis and interpretation; Revising the manuscript.

留言 (0)

沒有登入
gif