When I tell people about my research group’s systematic review of systematic reviews, also known as an ‘overview’, my colleagues often reply “why not do a ‘regular’ review‘?”
In 1979 Archie Cochrane gives us the answer when he scolded the medical community for not having a system in place to produce a ‘critical summary, by speciality or subspeciality, adapted periodically, of all relevant randomised controlled trials’.1 At that time, the rate of trial publication was increasing nearly exponentially. The same pattern has emerged for published systematic reviews2 and it is a problem. For researchers trying to understand the state of the literature in an area, the volume of published reviews can be difficult to manage. For clinicians and policy-makers trying to use reviews to inform their decisions, it is impossible.
Overviews can be used to manage the multitude of reviews. Broadly speaking, they serve the same purpose as ‘traditional’ reviews: to map the evidence base and to compare, contrast, and synthesise findings from multiple studies. The only difference is that the included articles are systematic reviews instead of primary studies. However, overviews can also be used to answer novel questions that can’t feasibly be answered using traditional reviews.
The Cochrane Collaboration has identified five types of overviews, classified based on their objective.3 One type in particular holds huge potential to support the success of health systems: outcomes-based overviews.
An outcomes-based overview compares the effectiveness of different interventions to achieve a single outcome, and is well-suited to answer questions of relevance to health systems. For example, what are the most effective, evidence-based interventions to reduce unscheduled hospital admissions? Decision-makers want to know the answer to this broad and ambitious question. It usually wouldn’t be feasible to conduct a traditional review as the number of randomized trials may run into the thousands. Furthermore, if we were to conduct a review of RCTs , meta-analysing ‘like with like’ would result in hundreds of sub-groups and ultimately mirror the results of the individual systematic reviews anyway.
By comparing and contrasting existing reviews we can utilize the meta-analysed results of thousands of trials to prioritize interventions that may be targets for implementation in-practice.
“Our overview4”, I tell my colleagues, “has utilised thousands of trials via hundreds of systematic reviews to find evidence-based approaches to reduce hospital admissions. In the end, we will have a list of interventions ranked by absolute reduction in admissions and the quality of evidence that researchers, clinicians, and decision-makers can use to guide pilot projects and resource allocation.”
“So what’s at the top of the list?” they ask.
“It’ll have to wait until Evidence Live where I am presenting the results.”
Niklas Bobrovitz is a PhD candidate in the Nuffield Department of Primary Health Care Sciences at Oxford University, a Clarendon Scholar, a member of the Centre for Evidence-Based Medicine, and a member of the Evidence Live 2016 steering committee which this year includes a theme on translating Evidence into Better-Quality Health Services.
You can follow him on twitter at @nikbobrovitz
Competing interests: I have read and understood BMJ policy on competing interests. I have no other competing interests to declare.
Disclaimer: The views expressed are those of the author and not necessarily of any of the institutions or organisations mentioned in the article.
Cite as Bobrovtiz N (2016): Are Outcomes-Based Overviews the Best Kept Secret in Health Research? CEBMJ /are-outcomes-based-overviews-the-best-kept-secret-in-health-research/ DOI: 10.13140/RG.2.1.3027.4966