What is Multiple-Treatments Meta-Analysis

mtmnetworkMultiple-Treatments Meta-analysis or MTM (also called ‘mixed-treatment meta-analysis’ and ‘Network meta-analysis’) reflects the network of comparisons that arises when collating studies involving different selections of competing treatments (Caldwell et al., 2005, Lu & Ades, 2004, Salanti et al., 2008a, Welton et al., 2008). Only recently, however, MTM approaches have become more widely demanded, with the increased complexity of analyses that underpin clinical guidelines and health technology appraisals. Several applications of the methodology have depicted the benefits of a joint analysis. These include improvement in precision for the estimated effect sizes and the ability to compare treatments that have not been directly compared in any trial. Despite their usefulness, MTM approaches are far from being an established practice in the medical literature. The majority of health care practitioners are sceptical towards this new technique. The ‘unclear’ assumptions underlying MTM, the uncertain role of bias and the lack of an interpretable and useful measure to summarize the results contribute to the scepticism.

Figure 1 shows a network from a recently published project comprising studies comparing different antidepressants (Cipriani et al. 2009). Εach link represents at least one study and the size of each node is proportional to the total sample size. An obvious difficulty with networks of studies is that simple description is difficult and flexible measures need to be used in order to describe the nature of such an entity. Moreover, the shape and the characteristics of the network are closely linked to the evaluation of underlying assumptions of MTM and unexplained asymmetry might be associated with publication bias.

The statistical methods involved in MTM go beyond the standard meta-analysis as they involve sophisticated modelling, including Bayesian approaches and specialised software, an outline of which can be found in (Salanti, Higgins, Ades, & Ioannidis, 2008a).  The basic idea is to build the network of trials (such as the one in Figure 1) in terms of relative effect sizes (e.g. odds-ratios or mean differences), select a number of key contrasts and express all other relationships as linear functions of those key contrasts.Then, assuming that different pieces of evidence fit together (assumption of consistency), direct and indirect information ‘flow’ within the network strengthening the precision in the estimates of treatment effects.

Concerns have been expressed about the validity of MTM methods as they rely on assumptions that are difficult to test. The consistency assumption might be violated when comparing two treatments through a third common comparator  and examples are outlined in recent research papers (Cooper et al., 2009, Lu & Ades, 2004, Salanti et al., 2009). There is some indication however that methods based on indirect evidence (such as MTM) can address biases that cannot be addressed in a single meta-analysis, such as sponsorship bias and I believe that MTM techniques may counterbalance bias under certain circumstances (Salanti et al., 2010c, Song et al., 2008)


Caldwell, D. M., Ades, A. E., & Higgins, J. P. (2005). Simultaneous comparison of multiple treatments: combining direct and indirect evidence. BMJ 331, 897-900

Cipriani, A., Furukawa, T. A., Salanti, G., Geddes, J. R., Higgins, J. P., Churchill, R., Watanabe, N., Nakagawa, A., Omori, I. M., McGuire, H., Tansella, M., & Barbui, C. (2009). Comparative efficacy and acceptability of 12 new-generation antidepressants: a multiple-treatments meta-analysis. Lancet 373, 746-758

Cooper, N. J., Sutton, A. J., Morris, D., Ades, A. E., & Welton, N. J. (2009). Addressing between-study heterogeneity and inconsistency in mixed treatment comparisons: Application to stroke prevention treatments in individuals with non-rheumatic atrial fibrillation. Statistics in Medicine 28, 1861-1881

Lu, G. & Ades, A. E. (2004). Combination of direct and indirect evidence in mixed treatment comparisons. Statistics in Medicine 23, 3105-3124

Salanti, G., Higgins, J. P., Ades, A. E., & Ioannidis, J. P. (2008a). Evaluation of networks of randomized trials. Statistical Methods in Medical Research 17, 279-301

Salanti, G., Marinho, V., & Higgins, J. P. (2009). A case study of multiple-treatments meta-analysis demonstrates that covariates should be considered. Journal of Clinical Epidemiology 62, 857-864

Salanti, G., Dias, S., Welton, N. J., Ades, A. E., Golfinopoulos, V., Kyrgiou, M., Mauri, D., & Ioannidis, J. P. (2010c). Evaluating novel agent effects in multiple-treatments meta-regression. Stat. Med. 29, 2369-2383

Song, F., Harvey, I., & Lilford, R. (2008). Adjusted indirect comparison may be less biased than direct comparison for evaluating new pharmaceutical interventions. Journal of Clinical Epidemiology 61, 455-463

Welton, N. J., Cooper, N. J., Ades, A. E., Lu, G., & Sutton, A. J. (2008). Mixed treatment comparison with multiple outcomes reported inconsistently across trials: Evaluation of antivirals for treatment of influenza A and B. Statistics in Medicine 29, 5620-5639

Additional information