abstract
- Systematic error or bias in a review is hard to avoid when the review is executed without a clearly defined research question and methodology. The methods of systematic review and meta-analysis are designed to address these and other sources of bias with the goal of producing the most valid and precise estimate of effect or risk possible. The guiding principle is the a priori establishment of a protocol which details the research question, study inclusion criteria, and methods for quantitatively combining data to provide an overall estimate of effect or risk. Randomized clinical trials generally provide the least biased estimates of treatment effect and are therefore more suitable for meta-analysis. The strengths and weaknesses of this approach are discussed in detail. Finally, a description of the goals and structure of the Cochrane Collaboration is included.