pipette
ENEnglish

Visualizing Joint Ranking Distributions in Network Meta-Analysis with Hammock Plots

Carmen Alicia Liang Zhen, Audrey B\'eliveau, Matthias Schonlau

Preprint

En palabras de los autores

Treatment ranking provides a summary of the relative effectiveness of competing treatments and is a key output of network meta-analysis (NMA). Existing visualizations primarily summarize marginal ranking distributions or present individual treatment hierarchies with limited insight into uncertainty in the overall hierarchy. We introduce hammock plots as a visualization framework for the joint posterior distribution of treatment rankings. Although motivated by treatment ranking in NMA, the proposed framework is broadly applicable to any setting in which uncertainty in posterior rankings is of interest. Specifically, we develop separate hammock plots to display relationships between (a) adjacent treatments or (b) adjacent ranks. Using highlighting, we show how to assess posterior support for selected treatment hierarchies, to evaluate hierarchies identified by ranking metrics against the posterior distribution, and to examine uncertainty within selected regions of the treatment hierarchy. We apply the proposed visualizations to a thrombolytic treatment network, which revealed patterns in the posterior ranking distribution and the location of uncertainty in treatment hierarchies that were difficult to identify using existing ranking graphics. We also provide a web application to generate all of the proposed hammock plot variations for any data set.

Resultado principalEl resumen no menciona limitaciones.

Apareció: miércoles, 21 de octubre. arXiv. Preprint, todavía sin revisión por pares.