Surgery highly effective for seizure cessation in drug-resistant epilepsy

Abstract

A systematic review is a methodologically rigorous overview of the medical literature designed to answer a specific clinical question. MA is the statistical application to a systematic review's individual studies' combined results. There are 2 basic approaches to combining individualstudy results: combining individual patient data and combining individual study-summary data using weighted averaging. The latter is notably more common, as individual patient results across multiple studies is difficult to obtain. What emerges from the MA results is essentially a single, larger "study" with more statistical power and a more precise effect estimate. Perhaps the most important underlying problem in MA is assuming that the individual studies' methodologies are sufficiently identical. In other words, are the individual studies' results measuring the identical or almost-identical effect? To answer this, statistical methods provide an estimate of heterogeneity (the less heterogeneity, the more reliable the final MA summary result). One such method is the I test, discussed in an earlier EBM Pearl. 1 Among the various controversies surrounding MA is the realization that an MA result and a result from an equivalently-large (study patients), well-designed single study-on the exact same clinical questionmay not be numerically the same and have differing clinical implications. The controversy lies in which approach-an MA or a large study-is a better estimate of the "true" result. In any case, MA, a methodologically-rigorous, statistically well-powered, enhanced-precision numeric-effect summary of the entire literature on a single clinical question, affords clinicians a tool to inform clinical decision making, in the hope that patients are better served thereby.