TimelyScholarFree Tools › Network Meta-Analysis Helper
Prepare a network meta-analysis: draw the network of treatment comparisons, check that it is connected, count the studies behind each direct comparison and get data and code for R netmeta.
FreeNo account neededYour data stays in your browser
| Treatments | 4A, B, C, D |
|---|---|
| Studies | 242 multi-arm |
| Direct comparisons | 6A vs C (15); A vs D (2); C vs D (4); B vs C (2); B vs D (2); A vs B (3) |
| Connected? | Yes: one connected networkAll treatments can be compared |
library(netmeta)
dat <- read.csv("network-data.csv")
# arm-level data to one row per pairwise comparison
p1 <- pairwise(treat = treatment, event = events, n = n, studlab = study, data = dat, sm = "OR")
nm <- netmeta(p1)
summary(nm)
netgraph(nm)
forest(nm)Edge width and labels show the number of studies making each direct comparison; node size reflects the number of participants. Before running a network meta-analysis, check transitivity (similar populations and effect modifiers across comparisons) and, after fitting, inconsistency between direct and indirect evidence.
One row per arm: study, treatment, events and number of participants.
Look for disconnected treatments and comparisons supported by few studies.
Download the diagram and copy the netmeta code.
Network meta-analysis compares several treatments at once, combining direct evidence (trials comparing two treatments head to head) with indirect evidence through common comparators.
It relies on transitivity: the studies making different comparisons should be similar in the factors that modify treatment effects. After fitting, check for inconsistency between direct and indirect evidence.
Treatments in separate sub-networks cannot be compared. Analyse the sub-networks separately or look for studies that link them.