Background: Many systematic reviews appropriately restrict their inclusion to randomised trials. However, for systematic reviews that aim to synthesise the effects of organisational, policy change, or public health interventions or exposures, non-randomised studies may provide the only available evidence, or important additional evidence to that of randomised trials. The Interrupted Time Series (ITS) design is a type of non-randomised study where a series of measurements are collected at regular intervals before and after an interruption. The period before the interruption can be used to estimate the underlying time trend. This trend can then be projected into the post-interruption period to provide a counterfactual for what would have occurred in the absence of the interruption, allowing the calculation of different effect measures that quantify the immediate and long-term effects of the interruption. Meta-analysis can be used to combine effect estimates across ITS studies. However, the ITS design presents challenges for meta-analysts. To date there have been no reviews examining the approaches and methods used to meta-analyse effect estimates from ITS designs.
Objectives: In this review, we aim to: 1) investigate whether reviewers re-analyse primary ITS studies included in reviews, and if so, what re-analysis methods are used; 2) examine the meta-analysis method(s) used; 3) describe the effect measures reported and the completeness of their reporting; and 4) explore the tools and domains that are used to assess the risks of bias and/or methodological quality of the included ITS studies.
Methods: We searched eight electronic databases from a range of disciplines (e.g. public health, economics), between 2000 and 2019 to identify reviews that included a meta-analysis of at least two ITS studies. Study selection was undertaken independently by at least two authors of the review team. From eligible reviews, two authors will independently extract details at the review level: including discipline, and type of interruption; and at the meta-analytic level: type of outcome, effect measure(s), meta-analytic method(s), and any methods used to re-analyse the primary ITS studies. The characteristics of the reviews will be summarised with descriptive statistics.
Results: The search retrieved 4213 citations. After removing duplicates, we screened 2677 titles and abstracts published between 2000 and 2019. 2346 were excluded from the title and abstract screening. Full-text screening of 331 articles yielded approximately 60 reviews that include a meta-analysis of ITS studies. At the time of submission, the results are not available, but will be presented at the Colloquium.
Conclusions: Findings from this review will be used to inform future research examining how different meta-analysis methods for combining results from ITS studies perform, with a view to developing guidance for systematic review authors.
Patient or healthcare consumer involvement: No patients/consumers were involved in the design/reporting of this study.