Article type
Year
Abstract
Background: Overviews of reviews are 'systematic reviews of systematic reviews' that examine all possible interventions for a certain condition. To date, no overviews published in The Cochrane Library or by the Child Health Field have included non-Cochrane reviews. Our most recent overview on eczema prevention is the first overview to include both Cochrane and non-Cochrane reviews.
Objectives: To discuss the methodological challenges we faced, how we overcame these challenges, and what we learned from the process of including non-Cochrane reviews in overviews. Challenges: 1) Two of our three potentially included non-Cochrane reviews did not contain data that enabled calculation of risk ratios and were therefore excluded; 2) many non-Cochrane reviews analyzed data from Cochrane reviews we had already included in our overview; and 3) we debated using scores from the AMSTAR tool (Assessment of Multiple Systematic Reviews) as an inclusion criterion. Our main challenge, however, centred on the fact that Cochrane and non-Cochrane reviews often examined the same interventions and contained many of the same trials. Deciding which reviews to include in our overview involved a trade-off between inclusivity and ensuring the project did not become too unwieldy or large for the end user. Solution: We used a consensus-based process to concurrently develop two sets of inclusion criteria: one assessed the eligibility of each review individually, and the other ensured that only one eligible review per topic area was included in our overview (Table1).
Conclusions: Six Cochrane reviews and one non-Cochrane review were included in our overview. Data extraction proceeded largely without difficulty, mostly due to our development of logical a priori inclusion criteria which other groups undertaking overviews might find helpful. Including non-Cochrane reviews allowed us to consider a broader range of interventions, although this was limited by our inability to extract usable data from two potential non-Cochrane reviews.
Objectives: To discuss the methodological challenges we faced, how we overcame these challenges, and what we learned from the process of including non-Cochrane reviews in overviews. Challenges: 1) Two of our three potentially included non-Cochrane reviews did not contain data that enabled calculation of risk ratios and were therefore excluded; 2) many non-Cochrane reviews analyzed data from Cochrane reviews we had already included in our overview; and 3) we debated using scores from the AMSTAR tool (Assessment of Multiple Systematic Reviews) as an inclusion criterion. Our main challenge, however, centred on the fact that Cochrane and non-Cochrane reviews often examined the same interventions and contained many of the same trials. Deciding which reviews to include in our overview involved a trade-off between inclusivity and ensuring the project did not become too unwieldy or large for the end user. Solution: We used a consensus-based process to concurrently develop two sets of inclusion criteria: one assessed the eligibility of each review individually, and the other ensured that only one eligible review per topic area was included in our overview (Table1).
Conclusions: Six Cochrane reviews and one non-Cochrane review were included in our overview. Data extraction proceeded largely without difficulty, mostly due to our development of logical a priori inclusion criteria which other groups undertaking overviews might find helpful. Including non-Cochrane reviews allowed us to consider a broader range of interventions, although this was limited by our inability to extract usable data from two potential non-Cochrane reviews.
Images
