# "Network meta-analysis combining survival and count outcome data: A simple frequentist approach"
# by Hisashi Noma and Kazushi Maruo
# https://doi.org/10.1101/2025.01.23.25321051

#  R example code for implementing the network meta-analysis

# The "survNMA" package
# CRAN webpage: https://doi.org/10.32614/CRAN.package.survNMA

###

# Install the R package file from CRAN (only at the first time):

install.packages("survNMA")

###

library("survNMA")				# load the "survNMA" package
library("netmeta")				# load the "netmeta" package

##

# (1) Preparing a survival outcome data
# In multi-arm trials, hazard ratio (HR) estimates and their standard error estimates are not reported for several contasts of comparative groups.
# Using the approximation method of Woods (2010), they can be calculated, especially using "calcse" function.

-0.154 - 0.055							# log HR estimate can be calculated simply as the difference of log HRs.
calcse(0.096,0.092,1521,1534,1524)		# the SE estimate can be calculated using "calcse" function.


# (2) Calculating HR statistics from dichotomized outcome data

data(woods2)
woods2		# dichotomized dataset

woods3 <- pairwiseHR(treat, studlab=study, event=d, n, data=woods2)


# (3) Combining the HR statistics from survival and dichotomized outcome data

data(woods1)
woods1

woods4 <- combine2(woods1,woods3)

woods4		# the combined dataset (this dataset object can be directly used in the analysis of "netmeta" package


# (4) Network meta-analysis using "netmeta" package

net1 <- netmeta(TE=TE, seTE=seTE, treat1=treat1, treat2=treat2, studlab=studlab, data=woods4, common=FALSE, ref="Placebo", sm="HR")
net1		# network meta-analysis; synthesis analysis

netpairwise(net1)		# pairwise meta-analysis for all possible treatment pairs

ran1 <- rankogram(net1,nsim = 5000)
netrank(ran1)			# calculating SUCRA

netgraph(net1, seq = "optimal", plastic = FALSE, number.of.studies = TRUE)			# network plot

forest(net1)		# forest plot

netleague(net1, seq = netrank(net1), ci = TRUE)			# league table

netsplit(net1)		# sidesplitting (analysis of inconsistency)