Performance of several types of beta-binomial models in comparison to standard approaches for meta-analyses with very few studies.
Beta-binomial model
Few studies
Generalised linear mixed models
Meta-analyses
Simulation study
Journal
BMC medical research methodology
ISSN: 1471-2288
Titre abrégé: BMC Med Res Methodol
Pays: England
ID NLM: 100968545
Informations de publication
Date de publication:
13 12 2022
13 12 2022
Historique:
received:
31
05
2022
accepted:
29
10
2022
entrez:
13
12
2022
pubmed:
14
12
2022
medline:
16
12
2022
Statut:
epublish
Résumé
Meta-analyses are used to summarise the results of several studies on a specific research question. Standard methods for meta-analyses, namely inverse variance random effects models, have unfavourable properties if only very few (2 - 4) studies are available. Therefore, alternative meta-analytic methods are needed. In the case of binary data, the "common-rho" beta-binomial model has shown good results in situations with sparse data or few studies. The major concern of this model is that it ignores the fact that each treatment arm is paired with a respective control arm from the same study. Thus, the randomisation to a study arm of a specific study is disrespected, which may lead to compromised estimates of the treatment effect. Therefore, we extended this model to a version that respects randomisation. The aim of this simulation study was to compare the "common-rho" beta-binomial model and several other beta-binomial models with standard meta-analyses models, including generalised linear mixed models and several inverse variance random effects models. We conducted a simulation study comparing beta-binomial models and various standard meta-analysis methods. The design of the simulation aimed to consider meta-analytic situations occurring in practice. No method performed well in scenarios with only 2 studies in the random effects scenario. In this situation, a fixed effect model or a qualitative summary of the study results may be preferable. In scenarios with 3 or 4 studies, most methods satisfied the nominal coverage probability. The "common-rho" beta-binomial model showed the highest power under the alternative hypothesis. The beta-binomial model respecting randomisation did not improve performance. The "common-rho" beta-binomial appears to be a good option for meta-analyses of very few studies. As residual concerns about the consequences of disrespecting randomisation may still exist, we recommend a sensitivity analysis with a standard meta-analysis method that respects randomisation.
Sections du résumé
BACKGROUND
Meta-analyses are used to summarise the results of several studies on a specific research question. Standard methods for meta-analyses, namely inverse variance random effects models, have unfavourable properties if only very few (2 - 4) studies are available. Therefore, alternative meta-analytic methods are needed. In the case of binary data, the "common-rho" beta-binomial model has shown good results in situations with sparse data or few studies. The major concern of this model is that it ignores the fact that each treatment arm is paired with a respective control arm from the same study. Thus, the randomisation to a study arm of a specific study is disrespected, which may lead to compromised estimates of the treatment effect. Therefore, we extended this model to a version that respects randomisation. The aim of this simulation study was to compare the "common-rho" beta-binomial model and several other beta-binomial models with standard meta-analyses models, including generalised linear mixed models and several inverse variance random effects models.
METHODS
We conducted a simulation study comparing beta-binomial models and various standard meta-analysis methods. The design of the simulation aimed to consider meta-analytic situations occurring in practice.
RESULTS
No method performed well in scenarios with only 2 studies in the random effects scenario. In this situation, a fixed effect model or a qualitative summary of the study results may be preferable. In scenarios with 3 or 4 studies, most methods satisfied the nominal coverage probability. The "common-rho" beta-binomial model showed the highest power under the alternative hypothesis. The beta-binomial model respecting randomisation did not improve performance.
CONCLUSION
The "common-rho" beta-binomial appears to be a good option for meta-analyses of very few studies. As residual concerns about the consequences of disrespecting randomisation may still exist, we recommend a sensitivity analysis with a standard meta-analysis method that respects randomisation.
Identifiants
pubmed: 36514000
doi: 10.1186/s12874-022-01779-3
pii: 10.1186/s12874-022-01779-3
pmc: PMC9745934
doi:
Types de publication
Meta-Analysis
Journal Article
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
319Informations de copyright
© 2022. The Author(s).
Références
Stat Med. 2016 Jul 10;35(15):2503-15
pubmed: 26842654
Contemp Clin Trials. 2007 Feb;28(2):105-14
pubmed: 16807131
Res Synth Methods. 2018 Sep;9(3):366-381
pubmed: 29573180
J R Stat Soc Ser A Stat Soc. 2009 Jan;172(1):137-159
pubmed: 19381330
Stat Med. 2019 Nov 30;38(27):5197-5213
pubmed: 31583750
Stat Med. 2018 Mar 30;37(7):1059-1085
pubmed: 29315733
Res Synth Methods. 2018 Sep;9(3):382-392
pubmed: 29504289
Biometrics. 2012 Dec;68(4):1269-77
pubmed: 22845838
J Natl Cancer Inst. 1959 Apr;22(4):719-48
pubmed: 13655060
Control Clin Trials. 1986 Sep;7(3):177-88
pubmed: 3802833
BMC Med Res Methodol. 2020 Jun 11;20(1):152
pubmed: 32539721
Stat Med. 2017 Nov 10;36(25):3923-3934
pubmed: 28748567
Res Synth Methods. 2019 Mar;10(1):23-43
pubmed: 30129707
Int J Epidemiol. 2012 Jun;41(3):818-27
pubmed: 22461129
Stat Methods Med Res. 2016 Dec;25(6):2858-2877
pubmed: 24823642
Biom J. 2010 Feb;52(1):85-94
pubmed: 20140900
Prog Cardiovasc Dis. 1985 Mar-Apr;27(5):335-71
pubmed: 2858114
Stat Med. 2002 Nov 15;21(21):3153-9
pubmed: 12375296
BMC Med Res Methodol. 2002;2:3
pubmed: 11860606
Biom J. 2016 Nov;58(6):1428-1444
pubmed: 27546483
Cochrane Database Syst Rev. 2019 Oct 3;10:ED000142
pubmed: 31643080
Curr Control Trials Cardiovasc Med. 2002 Mar 19;3(1):6
pubmed: 11991807
Stat Med. 2014 Dec 10;33(28):4861-74
pubmed: 25244540
Stat Med. 2015 Mar 30;34(7):1097-116
pubmed: 25446971
BMC Med Res Methodol. 2014 Feb 18;14:25
pubmed: 24548571
Res Synth Methods. 2020 Jan;11(1):74-90
pubmed: 31348846
Res Synth Methods. 2019 Dec;10(4):515-527
pubmed: 31111673
Stat Med. 2001 Dec 30;20(24):3875-89
pubmed: 11782040
Biometrics. 2019 Jun;75(2):485-493
pubmed: 30430540
Stat Med. 2003 Sep 15;22(17):2693-710
pubmed: 12939780
Res Synth Methods. 2019 Sep;10(3):398-419
pubmed: 30854785
J Res Natl Bur Stand (1977). 1982 Sep-Oct;87(5):377-385
pubmed: 34566088
Res Synth Methods. 2019 Mar;10(1):83-98
pubmed: 30067315
Stat Med. 2019 May 20;38(11):2074-2102
pubmed: 30652356
Biom J. 2020 Nov;62(7):1597-1630
pubmed: 32510177
Stat Med. 2007 Jan 15;26(1):53-77
pubmed: 16596572