Article type
Year
Abstract
Background: The Cochrane Effective Practice and Organisation of Care Group (EPOC) accepts inclusion of several study designs in addition to randomised trials, including quasi-randomised controlled trials; interrupted time series; and controlled before-after designs. But how often are these study designs included and is it worth the effort?
Objectives: To explore: 1) How many EPOC reviews include non-randomised studies; 2) How many and what type of non-randomised studies they find; 3) Whether there is a connection between the proportion of non-randomised trials identified and the review topic.
Methods: We examined the 65 EPOC reviews published before March 2010, and recorded the types of study designs considered. For reviews that included nonrandomised studies, we calculated the proportion of included studies that were nonrandomised. We categorised the reviews according to topic, using the categories adopted by the Health Systems Evidence Database, and explored whether there were differences across categories in the proportion of nonrandomised studies.
Results: We found that: 75% of all EPOC reviews include nonrandomised studies in their inclusion criteria; the proportion of non-randomised studies in each review varied widely (median: 33%, range: 0 to 100%). The proportion of non-randomised studies varied according to intervention type: Healthcare delivery interventions, median 18.5% (range 0 to 100%); implementation strategies, median 33% (range 0 to 100%); financial interventions, median 50% (range 33 to 100%); and governance interventions, median 100% (range 66 to 100%).
Conclusions: The majority of EPOC reviews do search for non-randomised studies. However, the degree to which they find such studies varies greatly, and may be influenced by intervention type. Given that non-randomised studies are often at higher risk of bias, and that including them entails extra effort, review authors should consider whether the benefits justify this decision. Research should explore whether it is more useful to include non-randomised studies in reviews of some intervention types than others.
Objectives: To explore: 1) How many EPOC reviews include non-randomised studies; 2) How many and what type of non-randomised studies they find; 3) Whether there is a connection between the proportion of non-randomised trials identified and the review topic.
Methods: We examined the 65 EPOC reviews published before March 2010, and recorded the types of study designs considered. For reviews that included nonrandomised studies, we calculated the proportion of included studies that were nonrandomised. We categorised the reviews according to topic, using the categories adopted by the Health Systems Evidence Database, and explored whether there were differences across categories in the proportion of nonrandomised studies.
Results: We found that: 75% of all EPOC reviews include nonrandomised studies in their inclusion criteria; the proportion of non-randomised studies in each review varied widely (median: 33%, range: 0 to 100%). The proportion of non-randomised studies varied according to intervention type: Healthcare delivery interventions, median 18.5% (range 0 to 100%); implementation strategies, median 33% (range 0 to 100%); financial interventions, median 50% (range 33 to 100%); and governance interventions, median 100% (range 66 to 100%).
Conclusions: The majority of EPOC reviews do search for non-randomised studies. However, the degree to which they find such studies varies greatly, and may be influenced by intervention type. Given that non-randomised studies are often at higher risk of bias, and that including them entails extra effort, review authors should consider whether the benefits justify this decision. Research should explore whether it is more useful to include non-randomised studies in reviews of some intervention types than others.