Article type
Abstract
Background: Interrupted time series (ITS) studies are used to evaluate the effects of population-level interventions. ITS studies with statistically significant results may be published earlier than those with statistically non-significant results (time lag bias, a type of publication bias), which may influence conclusions drawn from systematic reviews of ITS studies. The extent of publication bias in ITS studies is unknown.
Objectives:
1.) To investigate the time from protocol submission of ITS studies to the publication of their results, and
2.) To examine whether the time to publication of ITS studies is influenced by their results.
Methods: We searched for published protocols of ITS studies and reports of their results in PubMed, MEDLINE, and Embase, up to December 31, 2022, using a search filter for ITS studies. We calculated the time to publication for each report of results, defined as the time from the protocol submission date (for ITS with data collected before protocol submission) or the end date of data collection (for ITS with data collected after protocol submission) to the publication date of the result report.
Results: We found 138 protocols of ITS studies, published between 2009 and 2022. Of these, 68 (49%, 68 of 138) were registered, either in clinical trial registries (46%, 64 of 138), post-authorization study registers (1%, 1 of 138), or as registered reports (2%, 2 of 138). We excluded 15 protocols for which data collection is ongoing and 42 protocols with insufficient information to determine the study’s progress. Of the remaining 81 protocols, we found 49 reports of results for 31 protocols (38%, 31 of 81). The median time to publication was 1,313 days (interquartile range 784–1,917) (Figure). We will present at the Summit findings from a Cox proportional hazards regression analysis examining whether the time to publication is influenced by the statistical significance of study results.
Conclusion: These findings will inform systematic reviewers about the potential extent of publication bias among ITS studies.
Relevance and Importance to Patients: Findings of systematic reviews that include ITS studies often underpin policy decisions that impact large populations. Publication bias compromises the validity of review findings; therefore, understanding the extent of publication bias is important.
Objectives:
1.) To investigate the time from protocol submission of ITS studies to the publication of their results, and
2.) To examine whether the time to publication of ITS studies is influenced by their results.
Methods: We searched for published protocols of ITS studies and reports of their results in PubMed, MEDLINE, and Embase, up to December 31, 2022, using a search filter for ITS studies. We calculated the time to publication for each report of results, defined as the time from the protocol submission date (for ITS with data collected before protocol submission) or the end date of data collection (for ITS with data collected after protocol submission) to the publication date of the result report.
Results: We found 138 protocols of ITS studies, published between 2009 and 2022. Of these, 68 (49%, 68 of 138) were registered, either in clinical trial registries (46%, 64 of 138), post-authorization study registers (1%, 1 of 138), or as registered reports (2%, 2 of 138). We excluded 15 protocols for which data collection is ongoing and 42 protocols with insufficient information to determine the study’s progress. Of the remaining 81 protocols, we found 49 reports of results for 31 protocols (38%, 31 of 81). The median time to publication was 1,313 days (interquartile range 784–1,917) (Figure). We will present at the Summit findings from a Cox proportional hazards regression analysis examining whether the time to publication is influenced by the statistical significance of study results.
Conclusion: These findings will inform systematic reviewers about the potential extent of publication bias among ITS studies.
Relevance and Importance to Patients: Findings of systematic reviews that include ITS studies often underpin policy decisions that impact large populations. Publication bias compromises the validity of review findings; therefore, understanding the extent of publication bias is important.