MBNMAdose-package

MBNMAdose for dose-response Model-Based Network Meta-Analysis

MBNMAdose provides a collection of useful commands that allow users to run dose-response Model-Based Network Meta-Analyses (MBNMA). package

Introduction

MBNMAdose allows meta-analysis of studies that compare multiple doses of different agents in a way that can account for the dose-response relationship.

Whilst making use of all the available evidence in a statistically robust and biologically plausible framework, this also can help connect networks at the agent level that may otherwise be disconnected at the dose/treatment level, and help improve precision of estimates \insertCite pedder2021MBNMAdose. The modelling framework is based on synthesising relative effects which avoids the necessity to adjust for baseline predictors, thereby making fewer assumptions than in typical Model-Based Meta-Analysis.

By modelling the dose-response, MBNMA avoids heterogeneity and inconsistency that can arise from "lumping" different doses together (a technique sometimes done in Network Meta-Analysis). All models and analyses are implemented in a Bayesian framework, following an extension of the standard NMA methodology presented by \insertCite lu2004;textualMBNMAdose and are run in \insertCite jags;textualMBNMAdose. For full details of dose-response MBNMA methodology see \insertCite mawdsley2016;textualMBNMAdose. Within this package we refer to a treatment as a specific dose or a specific agent .

Workflow

Functions within MBNMAdose follow a clear pattern of use:

1. Load your data into the correct format using mbnma.network()
2. Analyse your data using mbnma.run() with a wide range of dose-response functions
3. Examine model results using forest plots and treatment rankings
4. Check model fit and test for consistency using functions like mbnma.nodesplit()
5. Use your model to predict responses using predict()

At each of these stages there are a number of informative plots that can be generated to help understand the data and to make decisions regarding model fitting.

Examples

# Generate an "mbnma.network" object that stores data in the correct format
network <- mbnma.network(triptans)

# Generate a network plot at the dose/treatment level
plot(network, level="treatment")

# Generate a network plot at the agent level
plot(network, level="agent", remove.loops=TRUE)

# Perform "split" NMA to examine dose-response relationship
nma <- nma.run(network)
plot(nma)

# Analyse data using mbnma.run() with an Emax dose-response function
# and common treatment effects
result <- mbnma.run(network, fun=demax(),
  method="common")

# Generate forest plots for model results
plot(result)

# Rank results and plot rankograms
ranks <- rank(result)
plot(ranks, params="emax")

# Predict responses
pred <- predict(result, E0=0.2)

# Plot predicted response with "split" NMA results displayed
plot(pred, overlay.split=TRUE)

References

\insertAllCited

See Also

Useful links:

• https://hugaped.github.io/MBNMAdose/

Author(s)

Maintainer : Hugo Pedder hugopedder@gmail.com (ORCID)

Other contributors:

• Adil Karim [contributor]