Baseline synthesis models — baseline

Runs a baseline synthesis model, placing a random effect on the study-specific intercepts to obtain a pooled estimate of the absolute outcomes on the network reference treatment (or the baseline treatment specified by baseline_trt). This can be combined with a regression model to obtain an adjusted estimate. Estimates of both the overall mean and standard deviation (baseline_mean and baseline_sd) as well as the predictive distribution (baseline_new) are returned.

Usage

baseline_synthesis(
  network,
  baseline_trt = NULL,
  prior_intercept = .default(normal(scale = 100)),
  prior_intercept_sd = .default(half_normal(scale = 5)),
  ...
)

Arguments

network: A multinma network object.
baseline_trt: Baseline treatment on which to place the random effect for baseline synthesis. If not specified, the network reference treatment will be used.
prior_intercept: Prior for the mean of the random effects distribution on the study-specific intercepts.
prior_intercept_sd: Prior for standard deviation of the random effects distribution on the study-specific intercepts.
...: Additional arguments passed directly to nma().

Value

A stan_baseline object, inheriting from stan_nma.

Details

The baseline synthesis model places a random effect on the study-specific intercept parameters of the NMA or ML-NMR model fitted by nma(): $$\mu_j \sim \mathrm{N}(m, \tau_\mu)$$ The prior distribution on the mean $m$ is specified by the prior_intercept argument, and the prior for $\tau_\mu$ by prior_intercept_sd.