11.Meta analysis04.Randomized trials - sporedata/researchdesigneR GitHub Wiki
- At least two, previously published, interventional clinical trials with adequate quality and focusing on the same intervention (including times when interventions were measured), sample characteristics (inclusion/exclusion criteria), and outcome measures.
The goal of meta-analyses for clinical trials is to augment systematic reviews as a way to quantitatively bring together the results of multiple previous clinical trials, both evaluating their heterogeneity as well as defining a single measure of effect when that measure might be both clinically and statistically plausible.
The number needed to treat (NNT) is commonly used as a mechanism to communicate with clinicians and patients, but it has a number of issues such as very large confidence intervals (often unreported), the fact that it is a group measure used for an individual - see Problems with NNT.
- CRAN taskview for meta-analyses
- Number needed to treat
- statcheck: Extraction of information from plots and tables
- Can serve as prior information for the simulation of Bayesian adaptive trials or as input for the sample size calculation for frequentist trials.
- Are often one of the main pillars behind clinical practice guidelines.
-
Books
-
Articles
- Random-effects meta-analysis: summarizing evidence with caveats [4].
[1] Moher D, Liberati A, Tetzlaff J, Altman DG, Prisma Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS med. 2009 Jul 21;6(7):e1000097.
[2] Harrer, M., Cuijpers, P., Furukawa, T.A, & Ebert, D. D. (2019). Doing Meta-Analysis in R: A Hands-on Guide.
[3] Chen DG, Peace KE. Applied meta-analysis with R. CRC Press; 2013 May 3.
[4] Serghiou S, Goodman SN. Random-effects meta-analysis: summarizing evidence with caveats. Jama. 2019 Jan 22;321(3):301-2.