Article
Connecting the disconnected: New statistical methodology or new clinical research?
Search Medline for
Authors
Published: | August 29, 2017 |
---|
Outline
Text
Network meta-analysis, like standard pairwise meta-analysis, is classically based on relative effects, i.e., contrasts between treatments that have been directly compared in a trial. This approach, called contrast-based or treatment effect model, automatically preserves randomisation. Recently, some authors proposed arm-based approaches (also called treatment response models) to network meta-analysis and also models combining both approaches (baseline response plus treatment effect models). This provoked a controversial discussion [1], [2], [3].
Disconnected networks may arise in a variety of situations, e.g., when there is no accepted standard of care, when there has been a major paradigm shift in treatment, or for particular outcomes, where the network is connected for other outcomes. Based on the approach by Hong 2015 [2], Goring 2016 [4] presented an approach for analysing disconnected networks using an arm-based approach, arguing that the data in a disconnected network do not provide enough information for modelling the relative effects for the disconnected treatments.
In my talk I want to critically discuss the arguments given by the proponents of connecting disconnected networks. Particularly, a so-called “major paradigm shift in treatment” is no reason for abandoning randomized trials. As an essential presumption in clinical research, such a shift should be justified by clinical trials comparing the new treatment to the best standard of treatment before. The right thing to do would be to set up a trial that closes the gap. This can be done based on information from network meta-analysis [5], [6], [7]. On the other hand, if such a trial seems unethical because the new treatment is thought better anyway, then there is no reason to conduct a network meta-analysis including the “old” treatments, unless they can serve as bridge comparators just in order to obtain a connected network.In conclusion, disconnected networks should be analysed as separate networks. Rather than being fixed by intricate methodology, gaps identified in a network should motivate new primary research.
Der Vortrag gehört zum Workshop "Methods for Generalized Evidence Synthesis".
Organisatoren: R. Bender, K.H. Herrmann, K. Jensen, D. Hauschke, F. Leverkus & T. Friede
Die Autoren geben an, dass kein Interessenkonflikt besteht.
Die Autoren geben an, dass kein Ethikvotum erforderlich ist.
References
- 1.
- Hawkins N, Scott DA, Woods B."Arm-based" parameterization for network meta-analysis. Res Synth Methods. 2016 Sep;7(3):306-13. DOI: 10.1002/jrsm.1187
- 2.
- Hong H, Chu H, Zhang J, Carlin BP. A Bayesian missing data framework for generalized multiple outcome mixed treatment comparisons. Res Synth Methods. 2016 Mar;7(1):6-22. DOI: 10.1002/jrsm.1153
- 3.
- Dias S, Ades AE. Absolute or relative effects? Arm-based synthesis of trial data. Res Synth Methods. 2016 Mar;7(1):23-8. DOI: 10.1002/jrsm.1184
- 4.
- Goring SM, Gustafson P, Liu Y, Saab S, Cline SK, Platt RW. Disconnected by design: analytic approach in treatment networks having no common comparator. Res Synth Methods. 2016 Dec;7(4):420-432. DOI: 10.1002/jrsm.1204
- 5.
- Nikolakopoulou A, Mavridis D, Egger M, Salanti G. Continuously updated network meta-analysis and statistical monitoring for timely decision-making. Stat Methods Med Res. 2016 Sep 1. pii: 0962280216659896.
- 6.
- Nikolakopoulou A, Mavridis D, Salanti G. Planning future studies based on the precision of network meta-analysis results. Stat Med. 2016. 35(7):978-1000. DOI: 10.1002/sim.6608
- 7.
- Nikolakopoulou A, Mavridis D, Salanti G. Using conditional power of network meta-analysis (NMA) to inform the design of future clinical trials. Biom J. 2014 Nov;56(6):973-90. DOI: 10.1002/bimj.201300216