Exploring trial publication and research waste in COVID-19 randomised trials of hydroxychloroquine, corticosteroids, and vitamin D: a meta-epidemiological cohort study

Session Type
Fincham L1, Hohlfeld A2, Clarke M3, Kredo T4, McCaul M1
1Division of Epidemiology and Biostatistics, Department of Global Health, Faculty of Medicine and Health Sciences, Stellenbosch University, South Africa
2Cochrane South Africa, South African Medical Research Council, South Africa, South Africa
3Northern Ireland Methodology Hub, Centre for Public Health, Queen’s University Belfast, Northern Ireland, Northern Ireland
4Health Systems Research Unit, South African Medical Research Council, South Africa, South Africa

Background: The global research response to the COVID-19 pandemic was impressive, but also led to an infodemic and considerable research waste. Registered, but unpublished trials add to these problems.
Objectives: We aimed to determine the publication status of registered randomised trials of common COVID-19 treatments, to describe the characteristics of these trials and to examine the association between these characteristics, publication and research waste.
Methods: This meta-epidemiological cohort study used a sample of randomised trials that tested corticosteroids, hydroxychloroquine or vitamin D as treatments for COVID-19, registered between 1 November 2019 and 31 December 2021 and available via the World Health Organisation ICTRP portal. We searched for each trial’s published results up to 20 October 2022. We analysed the trials’ characteristics with descriptive statistics and performed univariate logistic regression to examine the association between these characteristics and publication status. We followed this with multiple logistic regression using significant characteristics to assess the association between trial characteristics and publication status.
Results: We identified 357 eligible trials on ICTRP. Of these, 107 (30%) had published or made their results available publicly by 20 October 2022. We could not find publications or other publicly available results for 250 (70%). Multiple logistic regression analysis showed that a larger target sample size was a significant positive predictor of publication, with target sample sizes above 300, almost tripling the odds of publication (aOR: 2.75, 95% CI: 1.35 to 5.62).
Conclusions: Less than -one-third of this sample of registered trials made their results public. Our findings identified that many trialists had not updated their trial’s status, results or both on the trial registry. Failure to share trial results publicly is a disservice to patients, clinicians and policymakers and adds to research waste.
Patient, public and/or healthcare consumer involvement: no direct involvement in the conduct of our research but, given the significant interest in these medicines, their impact on patient’s lives (including the potential for harm) and their costs, we would welcome public partnership in drawing attention to our findings.