Methods
Search dates
All searches took place between 2020-10-21 and 2020-10-21.
Results
Trial screening
As of 30 June 2020, the ICTRP COVID-19 database contained 3749 registered trials. Following all automated and manual exclusions, our final analysis dataset included 285 completed clinical trials. Details of all exclusions is available in Figure X. Brief descriptive details of the 285 trials are included in Table X.
Per protocol analysis
Using our per-protocol criteria, we located 47 full results published before 16 August 2020 for 41of the285(14%) trials registered as completed by 30 June 2020. The breakdown of results by dissemination route is detailed in Figure X [The venn diagram]. Figure Y details the reporting of results by registry.
Kaplan meier
Data for KM plots is available at kaplan-meier-time-to-pub.csv.
Figure Z shows time-to-publication using Kaplan-Meier (K-M) curves. Per protocol, we use the latest trial completion date reported in a publication when available (n = 4 of 41 trials, 9.8%), and otherwise the completion date reported in the registry, preferring the primary completion date to the study completion date, when available. For trials with multiple publications, we select the earliest publication date for the specified dissemination route. Trials with a publication date prior to the available completion date were considered reported at time 0 (across all results, n = 8 of 41 trials, 19.5%; n = 4 journal articles, n = 4 preprints). We show time from completion to: a) first publication across any dissemination route, b) journal publication, and c) preprint publication. As no method of reporting crossed 50%, we could not generate a median time to publication. As only two trials reported results to a registry, we did generate a K-M plot for this route. Summary results for these two trials were published on ChiCTR 3 and 20 days following completion. For the 4 trials that had both a preprint and a journal publication, the median days from first preprint publication to journal publication was 24.5 (14.2, 36.2). For the 21 trials that had a preprint only, preprints had been published a median of 90 days (76, 136) by completion of follow-up.
Regarding time-to-preprint, trials with only a preprint published the preprint with a median of 32 days (10, 53), whereas trials with both a preprint and a journal article published the preprint with a median of 31 days (19.5, 31).
Regarding time-to-journal-article, trials with only a journal article published the article with a median of 45.5 days (0.8, 83.5), whereas trials with both a preprint and a journal article published the article with a median of 46 days (24.2, 67.2).
Sensitivity Analyses
Whereas our per-protocol analysis relies on the trial completion date on the registry on 30 June 2020, we considered trial results reporting using the latest available completion date for the 285 trials included in the main analysis. The 30 June 2020 registry completion date was superseded by a later completion date reported in a publication (regardless of whether the publication was included in our analysis) for n = 8 trials and a later completion date found on the registry during manual searches for n = 40 trials. Considering these updated completion dates, 252 trials were completed by 30 June per their latest available completion date, of which 41 trials (16%) had full results available by 16 August 2020.
Restricting our sample from the 285 trials included in the main analysis to only those trials that reached full completion by 30 June 2020, showed 38 of 212 (18%) trials reported.
The inclusion of interim results adds an additional 1 result to our per-protocol analysis – a reporting rate of 15%.
Extending our reporting window beyond 6 weeks to 3 months (i.e., by 1 October 2020, prior to the start of our manual searches) adds an additional -59 results for a reporting rate of 54/285 (19%).
Data for 3-month follow-up KM plots is available at kaplan-meier-time-to-pub.csv.
Results excluded from per protocol analysis
Through our automated and manual publication search strategies, we found a total of 123 publications of complete results as either registry results, preprints, or journal articles. 47 results were associated with trials included in our per protocol analysis. Of the 76 excluded results, 5 could not be associated with a trial registration number, 8 were associated with a trial with no registered completion date, 23 were associated with a trial with a completion date following the 30 June 2020 cutoff, 1 has no publication date, 38 have a publication date following the 15 August 2020 cutoff, and 1 indicated that the trial was retrospective and was therefore post hoc excluded from our per protocol population.
Completion-date excluded trials
Data on the registrations for the 397 trials excluded from our cutoff 1 search based on completion date is available at cd-excluded-registrations.csv.
For cross-registered trials, our automated strategies relied on registry data from the preferred registry only. Based on the completion date provided on the preferred registry, we excluded 1724 trials. Of these, 105 trials had a total of 123 secondary cross-registrations. We post hoc retrieved completion dates from these registries, as they stood on 30 June 2020. X trials had discrepant completion dates, including X trials which would have been included in our search based on a secondary registration.
Figures and Tables
Table: Trials summary
Figure: Results by registry
Figure: Publication types (bar plot)
Figure: Publication types (upset plot)
Figure: Publication types (venn plot, ggVennDiagram)
Data for publication type sets for venn diagrams is available at pub-type-sets.csv.
Figure: Publication types (venn plot, eurlerr)
