A comparison of statistical methods for meta‐analysis

Abstract

Meta-analysis may be used to estimate an overall effect across a number of similar studies. A number of statistical techniques are currently used to combine individual study results. The simplest of these is based on a fixed effects model, which assumes the true effect is the same for all studies. A random effects model, however, allows the true effect to vary across studies, with the mean true effect the parameter of interest. We consider three methods currently used for estimation within the framework of a random effects model, and illustrate them by applying each method to a collection of six studies on the effect of aspirin after myocardial infarction. These methods are compared using estimated coverage probabilities of confidence intervals for the overall effect. The techniques considered all generally have coverages below the nominal level, and in particular it is shown that the commonly used DerSimonian and Laird method does not adequately reflect the error associated with parameter estimation, especially when the number of studies is small.

Comorbidities, Exposure to Medications, and the Risk of Community-Acquired Clostridium difficile Infection: A Systematic Review and Meta-analysis

“…Moreover, the assumption of normally distributed random effects is not easily verified. 43 The use of a moment-based common variance 16 within this model is in the redistribution of the weights from larger to smaller studies. 18 The QE and IVhet models have both been created to do away with the problems that affect the RE model and both have coverage of the CI at or above the nominal level.…”

Cannabis use and symptom severity in individuals at ultra high risk for psychosis: a meta‐analysis

“…Standardised mean differences (SMD) were calculated to assess differences in overall positive and negative symptom severity between UHR cannabis users and non‐users using Hedges’ g. SMDs were also conducted on individual positive symptoms if reported among three or more samples. Random‐effects models were used throughout to account for heterogeneity between studies 26, 27. Heterogeneity across studies was quantified using the I 2 statistic 28.…”

Soy isoflavone intake inhibits bone resorption and stimulates bone formation in menopausal women: meta-analysis of randomized controlled trials

“…Thus, a statistical method of combining these diverse data is needed to evaluate the usefulness of isoflavone therapy. Meta-analysis combines or integrates the results of several studies to provide increased statistical power for the quantitative identification of trends (Brockwell and Gordon, 2001). In the RCTs, urinary deoxypyridinoline (Dpyr) was generally used as a bone resorption marker, and serum bone-specific alkaline phosphatase (BAP) was the most commonly used index of bone formation.…”