Display options
Cite
Kabali C, Ghazipura M. Transportability in Network Meta-analysis. Epidemiology. 2016;27(4):556-61doi: 10.1097/EDE.0000000000000475.
Kabali, C., & Ghazipura, M. (2016). Transportability in Network Meta-analysis. Epidemiology (Cambridge, Mass.), 27(4), 556-61. https://doi.org/10.1097/EDE.0000000000000475
Kabali, Conrad, and Ghazipura, Marya. "Transportability in Network Meta-analysis." Epidemiology (Cambridge, Mass.) vol. 27,4 (2016): 556-61. doi: https://doi.org/10.1097/EDE.0000000000000475
Kabali C, Ghazipura M. Transportability in Network Meta-analysis. Epidemiology. 2016 Jul;27(4):556-61. doi: 10.1097/EDE.0000000000000475. PMID: 26963292.
Copy Download .nbib Format: NLM
Share it on

Epidemiology. 2016 Jul;27(4):556-61. doi: 10.1097/EDE.0000000000000475.

Transportability in Network Meta-analysis.

Epidemiology (Cambridge, Mass.)

Conrad Kabali, Marya Ghazipura

Affiliations

  1. From the aEvidence Development and Standards Branch, Health Quality Ontario, Ontario, Canada; bDivision of Epidemiology, Dalla Lana School of Public Health, University of Toronto, Ontario, Canada; and cDepartment of Environmental Health Sciences, New York University Langone Medical Center, New York, NY.

PMID: 26963292 DOI: 10.1097/EDE.0000000000000475

Abstract

Network meta-analysis is an extension of the conventional pair wise meta-analysis to include treatments that have not been compared head to head. It has in recent years caught the interest of clinical investigators in comparative effectiveness research. While allowing a simultaneous comparison of a large number of treatment effects, an inclusion of indirect effects (i.e., estimating effects using treatments that have not been randomized head to head) may introduce bias. This bias occurs from not accounting for covariates differences in the analysis, in a way that allows transfer of causal information across trials. Although this problem might not be entirely new to network meta-analysis researchers, it has not been given a formal treatment. Occasionally it is tackled by fitting a meta-regression model to account for imbalance of covariates. However, this approach may still produce biased estimates if covariates responsible for disparity across studies are post-treatment variables. To address the problem, we use the graphical method known as transportability to demonstrate whether and how indirect treatment effects can validly be estimated in network meta-analysis. See Video Abstract at http://links.lww.com/EDE/B37.

MeSH terms

Publication Types