|
|
@@ -1,3 +1,77 @@
|
|
|
-# umbrella-reviews-guide
|
|
|
+# Umbrella Reviews: Synthesising Systematic Reviews
|
|
|
|
|
|
-How to conduct umbrella reviews that synthesise systematic reviews, with source-linked extraction via https://yila.ai
|
|
|
+In many fields the number of systematic reviews has grown so fast that decision-makers now face a new problem. Instead of too few reviews, there are too many, often overlapping, sometimes contradictory and of variable quality. A clinician wanting to know which treatments work for a condition may find a dozen reviews, each covering a different subset of interventions, outcomes or populations. Umbrella reviews, also called overviews of reviews, address this by systematically identifying, appraising and synthesising existing systematic reviews. This guide explains when an umbrella review is appropriate, how to conduct one rigorously, the particular challenges of overlap and quality, and how a source-linked assistant such as [Yila](https://yila.ai) can help you extract and compare findings across many reviews.
|
|
|
+
|
|
|
+## What an umbrella review is
|
|
|
+
|
|
|
+An umbrella review is a systematic review whose unit of analysis is the systematic review rather than the primary study. It uses transparent methods to search for, select, appraise and synthesise reviews addressing a broad question, such as the effects of all interventions for a condition, or all health outcomes associated with an exposure.
|
|
|
+
|
|
|
+Umbrella reviews provide a high-level summary of the evidence and can highlight where reviews agree, where they conflict and where evidence is missing.
|
|
|
+
|
|
|
+## When an umbrella review is appropriate
|
|
|
+
|
|
|
+Umbrella reviews suit questions where several systematic reviews already exist and a broad summary would be useful. Typical aims include comparing the evidence for different interventions for the same condition, summarising evidence on the same intervention across different conditions or outcomes, exploring reasons for discrepancies between reviews, and providing a rapid overview of a broad field for guideline developers or policymakers.
|
|
|
+
|
|
|
+They are not appropriate when few reviews exist, when existing reviews are outdated or of poor quality, or when a precise answer to a narrow question is needed. In those cases a new systematic review of primary studies may be better.
|
|
|
+
|
|
|
+## Define the question and protocol
|
|
|
+
|
|
|
+As with any systematic review, start with a clear question and a protocol. Define the population, interventions or exposures, comparisons, outcomes and the types of reviews to include. Decide whether you will include only systematic reviews with meta-analysis, or also reviews without quantitative synthesis.
|
|
|
+
|
|
|
+Register or publish the protocol to increase transparency.
|
|
|
+
|
|
|
+## Search for reviews
|
|
|
+
|
|
|
+Search major bibliographic databases using filters designed to identify systematic reviews, as well as specialised databases of reviews. Check registries of review protocols to identify ongoing reviews. Search reference lists and consult experts.
|
|
|
+
|
|
|
+Document the search fully so it can be reproduced.
|
|
|
+
|
|
|
+## Select reviews
|
|
|
+
|
|
|
+Apply eligibility criteria consistently, with two reviewers screening independently. Consider how you will handle multiple reviews on the same question, updated versions of reviews and reviews that partially overlap with your scope.
|
|
|
+
|
|
|
+## Manage overlap
|
|
|
+
|
|
|
+Overlap is a defining challenge of umbrella reviews. Different reviews often include many of the same primary studies. If you simply count reviews, or combine their results, studies included in several reviews will be counted multiple times, exaggerating the apparent strength of evidence.
|
|
|
+
|
|
|
+Map the primary studies included in each review using a citation matrix. Measures such as the corrected covered area quantify the degree of overlap. Decide in advance how to handle overlapping reviews: for example, selecting the most recent, most comprehensive or highest-quality review for each question, or presenting all reviews while noting overlap.
|
|
|
+
|
|
|
+## Appraise review quality
|
|
|
+
|
|
|
+The quality of systematic reviews varies widely. Use an established tool designed for appraising systematic reviews to assess each included review. Key domains include whether the review had a protocol, whether the search was comprehensive, whether study selection and data extraction were duplicated, whether risk of bias in primary studies was assessed and whether meta-analysis methods were appropriate.
|
|
|
+
|
|
|
+Consider how quality will inform the synthesis, for example by giving more weight to high-quality reviews or excluding critically flawed ones.
|
|
|
+
|
|
|
+## Extract data from reviews
|
|
|
+
|
|
|
+Extract information about each review's question, included studies, methods and findings. Extract results for each outcome of interest, along with measures of heterogeneity and the review's own assessment of certainty of evidence where available.
|
|
|
+
|
|
|
+Reviews report findings in different ways, and key results can be spread across text, tables, forest plots and appendices. A source-linked assistant can help. Asking Yila what pooled effect a review reported for a particular outcome, how many studies contributed or how the authors rated certainty returns answers linked to the passages that support them, so the extractor can confirm values quickly and a second reviewer can check them against the same text.
|
|
|
+
|
|
|
+## Synthesise findings
|
|
|
+
|
|
|
+Umbrella reviews usually present findings in structured tables summarising, for each intervention and outcome, the reviews that addressed it, their results, their quality and the certainty of evidence. Narrative synthesis describes patterns, agreements and discrepancies.
|
|
|
+
|
|
|
+Some umbrella reviews reanalyse data from included reviews, for example recalculating effect sizes using consistent methods or grading the credibility of evidence using predefined criteria. Such reanalysis requires access to study-level data from the reviews.
|
|
|
+
|
|
|
+## Explore discrepancies
|
|
|
+
|
|
|
+When reviews reach different conclusions, investigate why. Differences may arise from different eligibility criteria, search dates, included studies, outcome definitions, analytic methods or quality. Understanding these differences is often one of the most valuable contributions of an umbrella review.
|
|
|
+
|
|
|
+## Assess certainty
|
|
|
+
|
|
|
+Certainty of evidence can be assessed using structured approaches adapted for umbrella reviews. Consider the certainty ratings reported by the original reviews, but check whether they were applied appropriately and whether newer evidence might change them.
|
|
|
+
|
|
|
+## Report clearly
|
|
|
+
|
|
|
+Follow reporting guidance for overviews of reviews. Report the search, selection, overlap assessment, quality appraisal, data extraction and synthesis methods. Present findings in clear tables and figures, and discuss limitations, including those inherited from the included reviews.
|
|
|
+
|
|
|
+## Build the right team
|
|
|
+
|
|
|
+Umbrella reviews demand a mix of skills. You need an information specialist to design searches for reviews, methodologists familiar with appraising systematic reviews and handling overlap, and content experts who understand the interventions and outcomes well enough to judge whether reviews are comparable. Because umbrella reviews often cover broad areas, the volume of included reviews can be large, so plan screening and extraction capacity carefully. Agree early on decision rules for overlap and quality, document them in the protocol and apply them consistently across the team.
|
|
|
+
|
|
|
+## Recognise limitations
|
|
|
+
|
|
|
+Umbrella reviews are limited by the reviews they include. If reviews are outdated, of poor quality or miss important studies, the umbrella review inherits those weaknesses. Umbrella reviews also depend on how reviews reported their findings, and important details may be missing.
|
|
|
+
|
|
|
+Umbrella reviews are a powerful tool for making sense of a crowded evidence base. With careful handling of overlap, rigorous quality appraisal, accurate extraction and thoughtful synthesis, they can give decision-makers a clear and trustworthy overview of what the evidence shows.
|
