Graphical tools for network meta-analysis in STATA

Abstract

Network meta-analysis synthesizes direct and indirect evidence in a network of trials that compare multiple interventions and has the potential to rank the competing treatments according to the studied outcome. Despite its usefulness network meta-analysis is often criticized for its complexity and for being accessible only to researchers with strong statistical and computational skills. The evaluation of the underlying model assumptions, the statistical technicalities and presentation of the results in a concise and understandable way are all challenging aspects in the network meta-analysis methodology. In this paper we aim to make the methodology accessible to non-statisticians by presenting and explaining a series of graphical tools via worked examples. To this end, we provide a set of STATA routines that can be easily employed to present the evidence base, evaluate the assumptions, fit the network meta-analysis model and interpret its results.

Figure 1. Network plot of the acute mania network (efficacy outcome).

Nodes are weighted according to the number of studies including the respective interventions. Edges are weighted according to the mean control group risk for comparisons between placebo and active treatment. Edges connecting two active treatments have been given minimal weight.

Figure 2. Plot of the acute mania network (efficacy outcome) using coloured edges according to adequacy of allocation concealment estimated as the level of bias in the majority of the trials and weighted according to the number of studies in each comparison.

Figure 3. Contribution plot for the coronary artery disease network.

The size of each square is proportional to the weight attached to each direct summary effect (horizontal axis) for the estimation of each network summary effects (vertical axis). The numbers re-express the weights as percentages. (MT =  medical therapy, PTCA =  percutaneous transluminal balloon coronary angioplasty, BMS =  bare-metal stents, DES =  drug-eluting stents).

Figure 4. Inconsistency plot for the acute mania network (for the efficacy outcome) assuming loop-specific heterogeneity estimates using the method of moments estimator.

(PLA =  placebo, ARI =  aripiprazole, ASE =  asenapine, CARB =  carbamazpine, DIV =  divalproex, HAL =  haloperidol, LAM =  lamotrigine, LITH =  lithium, OLA =  olanzapine, QUE =  quetipaine, RIS =  risperidone, TOP =  topiramate, ZIP =  ziprasidone, PAL =  paliperidone).

Figure 5. Comparison-adjusted funnel plot for the rheumatoid arthritis network.

The red line represents the null hypothesis that the study-specific effect sizes do not differ from the respective comparison-specific pooled effect estimates. The green line is the regression line. Different colours correspond to different comparisons.

Figure 6. Predictive interval plot for the rheumatoid arthritis network on a logarithmic scale.

The black solid lines represent the confidence intervals for summary odds ratios for each comparison and the red dashed lines the respective predictive intervals. The blue line is the line of no effect (odds ratio equal to 1).

Figure 7. Plots of the surface under the cumulative ranking curves for all treatments in the rheumatoid arthritis network.

Black solid lines correspond to the unadjusted model and red dashed lines to the adjusted for small-study effects model.

Figure 8. Ranking plots for the rheumatoid arthritis network.

Treatments have been ranked (a) according to the surface under the cumulative ranking curves (SUCRA) and (b) according to the unique dimension estimated from multidimensional scaling (MDS) approach. Red points correspond to treatments ranked in different order by the two approaches. (PLA =  placebo, ABA =  abatacept, ADA =  adalimumab, ANA =  anakinra, ETA =  etanercept, INF =  infliximab, RIT =  rituximab).

Figure 9. Clustered ranking plot of the acute mania network based on cluster analysis of SUCRA values for two different outcomes: efficacy and acceptability.

Each colour represents a group of treatments that belong to the same cluster. Treatments lying in the upper right corner are more effective and acceptable than the other treatments.