Knowledge Synthesis – Systematic Reviews

Systematic reviews are protocol-driven research projects that include:

• A thorough search for the best available research evidence
• Explicit inclusion and exclusion criteria
• Reproducible judgments about eligibility and quality
• Systematic data collection
• Quantitative and qualitative summaries that relate conclusions to the available data

If reviewers opt to produce a quantitative summary using statistical pooling then the systematic review is said to include a meta-analysis. Meta-analyses have the advantage of offering decision-makers with a quantitative summary estimate, an estimate of the associated precision, and an opportunity to evaluate how that estimate varies across predefined subgroups.

Systematic reviews yield strong inferences to the extent that they summarize the systematically-assembled totality of the available literature about a focused question with high quality primary data that yield consistent findings. Strong reviews have two additional advantages: they offer greater precision in the estimate of the measure of association than any of the primary studies included, and the results are applicable to a greater range of patients (at a minimum those included in the range of trials included in the review) than the results of each individual study.

Researchers at the KER Unit:

• Conduct systematic reviews and meta analysis, and develop clinical practice guidelines, on behalf of the American Association of Blood Banks
• Conduct systematic reviews and meta analysis, and develop clinical practice guidelines, on behalf of the Endocrine Society
• Conduct systematic reviews and meta analysis, and develop clinical practice guidelines, on behalf of the Vascular Society
• Conduct systematic reviews and meta analysis, and develop clinical practice guidelines, on behalf of the Sleep Society
• Bias in trials stopped early for benefit

One example of a knowledge synthesis project being conducted in the KER Unit currently:

The KER Unit is the coordinating center for the STOP IT 2 project. A collaboration of investigators from nine countries, the study seeks to determine the extent to which stopping trials early for benefit overestimates the intervention’s effect. The approach described has served this project well with over 2,500 trials evaluated in duplicate and more than 80 meta-analyses awaiting conduct. Soon, analysis of the data will indisputably influence the stopping early of clinical trials to come. We will be reporting on these efforts sometime in 2009.

STOP IT 2 Flow Chart