3-D network plots - New Software Tool

Posted by do43 at May 24, 2017 10:25 AM |
3-D network plots – a new software tool to aid the exploration of covariate imbalances and effects in network meta-analysis

A novel web-based interactive 3-dimensional (3-D) tool has been developed by Alex Sutton in collaboration with the Decision Resources Group Abacus to facilitate the visualisation and exploration of covariate distributions and imbalances across evidence networks for network meta-analysis (NMA).

Network meta-analysis is becoming increasingly important for decision makers to assess the comparative efficacy and safety of interventions and is integral to health technology assessment (HTA). The exploration of covariate effects is important in NMA because the presence of unaccounted treatment covariate interactions can invalidate the assumptions that underlie NMA and bias results. Visually assimilating, exploring and interpreting the distribution of covariate values across trials in an NMA is challenging due to the complexities of representing the network structure alongside study-level covariate values.

In a recently published paper in the Journal of Clinical Epidemiology, DRG Abacus and Leicester University outline the rationale for and the features of the software tool and present a series of illustrative examples.

Key points from the research include:

  • The primary innovation that allows for extensions and improvements to evidence networks is the use of a 3-D graphical environment, incorporating the graphical representation of covariates on a third ‘z’ axis.
  • This work is believed to be the first application of a 3-D graphical environment to evidence networks in NMA.
  • The 3-D evidence network plot system will facilitate the exploration of covariate distributions and imbalances across evidence networks and will be of most value in the context of supporting NMA feasibility/validity assessments and aiding in the interpretation of NMA results by a wide audience.
  • The tool is freely available at http://3dnma.com

Share this page: