Meta-analysis in evidence

EDITORIAL

Meta-analysis in evidence

Alvaro Nagib Atallah

MD, PhD, MCE Chaiman, Department of Internal Medicine, Escola Paulista de Medicina, Editor, São Paulo Medical Journal

Recently the New England Journal Medicine an original article that may mislead some readers¹. The study compared 12 randomized trials to 19 meta-analyses performed previously and state that the meta-analysis would lead to the adoption of ineffective treatment or rejection of useful treatment in most of the treatments studied. The analysis was based only on statistically significant effects, ignoring the relevance and meaning of the confidence intervals.

It is important to emphasize that a meta-analysis is a mathematical syntheses of the results of two or more studies that have addressed the same hypothesis, and a systematic review is an overview using explicit and reproducible methods, and may be followed by a meta-analysis or not.

The most important objective of a systematic review or a well-executed meta-analysis is to obtain data, based on the best evidence existing (published or not), with which we can attempt to check what we know with some level of certainty about a subject (for example, a treatment); what we really do not know, due to the absence of good evidence; and what we are uncertain about². Therefore, this refereed study¹ may have significant bias because, usually, large trials are developed after a meta-analysis that has left some level of uncertainty which is clinically important.

It is possible to observe this kind of disagreement between systematic review based on small trials and the necessary subsequent large trials without any surprise^3,4,5. Cappelleri et al⁶ in an article published in the JAMA, evaluated the results of large clinical trials vs the pooled results of smaller trials. They found, by random effects calculations, agreement between large and smaller trials in 90% of the meta-analysis selected by the sample size approach and in 82% of the meta-analysis selected by the statistical power approach.

Publications of the Cochrane Collaboration or of the Antithrombotic Trialists Collaboration, or others could help provide better understanding of systematic review for the reader. We would like to refer those interested to a recent issue of the British Medical Journal⁷, where Naylor, in the Editorial, starts with the subtitle "Meta-analyses is an important contribution to research and practice but it is not a panacea" and concludes: "In summary, meta-analysis has made and continues to make major contributions to medical research, clinical decision-making and standards of research reportage"... In the same issue there are more interesting information on systematic review interpretation and evidence based medicine⁸.

• 1. Lelerier J., Gregorie G., Benhaddad A., Lapierre J., Derderiana F., Discrepancies between meta-analyses and subsequent large randomized controlled trials. New England Journal of Medicine1997; 337:536-42.
• 2. Atallah A.N.. Systematic Review of Medical Literature and Meta-analyses. São Paulo Medical Journal 1996; 114(5):1285-876.
• 3. Atallah A.N.. ECPPA Randomized trial of low dose aspirin for the prevention of maternal and fetal complications in high risk pregnant women. British Journal of Obstetrics and Gynaecology. 1996; 103:39-47.
• 4. Atallah A.N.. Meta-analyses Calcium supplementation reduce blood pressure and pre-eclampsia during pregnancy. Evidence Based Medicine. 1996;1(6):173.
• 5. Levine RI; Hauth JC; Curet LB; Sibai, BM; Catalano PM; Morris CD; DerSimonian R; Esterlitz JR; Raymond EG; Bild DE; Clemens JD; Cutler JÁ. Trial of calcium to prevent preeclampsia. New England Journal of Medicine. 1997; 337(2):69-76.
• 6. Cappelleri JC; Ioannidis JP; Schmid CH; de Ferranti SD; Aubert M; Chalmers TC; Lau J. Large trials vs meta-analysis of smaller trials: how do their results compare? JAMA 1996 Oct 23; 276(16):1332-38.
• 7. Naylor D. Meta-analyses and the meta-epidemiology of clinical research. British Medical Journal. 1997;315(7109):617-619.
• 8. Greenhalgh T. Papers that summarize other paper. British Medical Journal. 1997; 315(7109):672-675.

Publication Dates

• Publication in this collection
  05 Nov 2008
• Date of issue
  Aug 1997