Projects per year
Abstract
Background: Systematic reviews and metaanalyses of binary outcomes are widespread in all areas of application. The odds ratio, in particular, is by far the most popular effect measure. However, the standard metaanalysis of odds ratios using a randomeffects model has a number of potential problems. An attractive alternative approach for the metaanalysis of binary outcomes uses a class of generalized linear mixed models (GLMMs). GLMMs are believed to overcome the problems of the standard randomeffects model because they use a correct binomialnormal likelihood. However, this belief is based on theoretical considerations, and no sufficient simulations have assessed the performance of GLMMs in metaanalysis. This gap may be due to the computational complexity of these models and the resulting considerable time requirements.
Methods: The present study is the first to provide extensive simulations on the performance of four GLMM methods (models with fixed and random study effects and two conditional methods) for metaanalysis of odds ratios in comparison to the standard random effects model.
Results: In our simulations, the hypergeometricnormal model provided less biased estimation of the heterogeneity variance than the standard randomeffects metaanalysis using the restricted maximum likelihood (REML) estimation when the data were sparse, but the REML method performed similarly for the point estimation of the odds ratio, and better for the interval estimation.
Conclusions: It is difficult to recommend the use of GLMMs in the practice of metaanalysis. The problem of finding uniformly good methods of the metaanalysis for binary outcomes is still open.
Methods: The present study is the first to provide extensive simulations on the performance of four GLMM methods (models with fixed and random study effects and two conditional methods) for metaanalysis of odds ratios in comparison to the standard random effects model.
Results: In our simulations, the hypergeometricnormal model provided less biased estimation of the heterogeneity variance than the standard randomeffects metaanalysis using the restricted maximum likelihood (REML) estimation when the data were sparse, but the REML method performed similarly for the point estimation of the odds ratio, and better for the interval estimation.
Conclusions: It is difficult to recommend the use of GLMMs in the practice of metaanalysis. The problem of finding uniformly good methods of the metaanalysis for binary outcomes is still open.
Original language  English 

Article number  70 
Journal  BMC Medical Research Methodology 
Volume  18 
DOIs  
Publication status  Published  4 Jul 2018 
Keywords
 Generalized linear mixedeffects models
 Random effects
 Hypergeometricnormal likelihood
 Transformation bias
 Metaanalysis
Profiles

Elena Kulinskaya
 School of Computing Sciences  Professor in Statistics (AVIVA)
 Business and Local Government Data Research Centre  Member
 Norwich Epidemiology Centre  Member
 Data Science and Statistics  Member
Person: Research Group Member, Research Centre Member, Academic, Teaching & Research
Projects
 1 Finished

Smart Data Analytics for Business and Local Government
Hancock, R., Sena, V., Coakley, J., Cornford, J., De La Iglesia, B., Fasli, M., Fearne, A., Forder, J., Harwood, A., Hviid, M., Jones, A., Kulinskaya, E., Laurie, H., Lovett, A., Schofield, G., Appleton, K., Morciano, M. & Sunnenberg, G.
Economic and Social Research Council
31/01/14 → 31/10/20
Project: Research