Graphical evaluation of evidence structure within a component network meta-analysis

Component network meta-analysis (CNMA) compares treatments comprising multiple components and estimates the effects of individual components. For network meta-analysis, a standard network plot with nodes for treatments and edges for direct comparisons between treatments is drawn to visualize the evidence structure and the connections between treatments. However, the standard network plot does not effectively illustrate the connections between components for a CNMA. For example, the comparison between linear combinations of components within a trial is not shown directly in a standard network plot, and whether all components are identifiable cannot be deduced directly from the plot. Therefore, we need a new approach to visualizing the evidence structure of a CNMA. In this article, we proposed a new graph, a modified signal-flow graph representing a system of equations, to evaluate the evidence structure for CNMA. In our new graph, each node represents a component, and arrows are used to show linear relationships between components. We used two examples to demonstrate how to draw and interpret the graph and how to use it to identify components that require more evidence.