jarbes (version 1.7.2)

gesmix: Generalized Evidence Synthesis with Indirect Penalization of Observational Studies

Description

This function performers a Bayesian meta-analysis to jointly combine different types of studies. The random-effects follows a finite mixture of normals.

Usage

gesmix(data, mean.mu = 0, sd.mu = 10^6, scale.sigma = 0.5,
  K.lower = -10, K.upper = 0, nr.chains = 2, nr.iterations = 10000,
  nr.adapt = 1000, nr.burnin = 1000, nr.thin = 1, be.quiet = FALSE,
  r2jags = TRUE)

Arguments

data

A data frame with at least three variables with the following names: 1) TE = treatment effect, 2) seTE = the standard error of the treatment effect 3) design = the study type. The other columns of the data frame may correspond to the study name and other covariates that characterize the study and could influence internal validity bias (e.g. evaluation of risk of bias).

mean.mu

Prior mean of the overall mean parameter mu, default value is 0.

sd.mu

Prior standard deviation of mu, the default value is 10^-6.

scale.sigma

Prior scale parameter for the standard deviation between studies, the default value is 0.5.

K.lower

Lower bound of the bias parameter K, the default value is -10.

K.upper

Upper bound of the bias parameter K, the default value is 0.

nr.chains

Number of chains for the MCMC computations, default 2.

nr.iterations

Number of iterations after adapting the MCMC, default is 10000. Some models may need more iterations.

nr.adapt

Number of iterations in the adaptation process, defualt is 1000. Some models may need more iterations during adptation.

nr.burnin

Number of iteration discared for burnin period, default is 1000. Some models may need a longer burnin period.

nr.thin

Thinning rate, it must be a positive integer, the default value 1.

be.quiet

Do not print warning message if the model does not adapt. The default value is FALSE. If you are not sure about the adaptation period choose be.quiet=TRUE.

r2jags

Which interface is used to link R to JAGS (rjags and R2jags), default value is R2Jags=TRUE.

Value

This function returns an object of the class "gesmix". This object contains the MCMC output of each parameter and hyper-parameter in the model and the data frame used for fitting the model.

Details

The results of the object of the class gesmix can be extracted with R2jags or with rjags. In addition a summary, a print and a plot functions are implemented for this type of object.

References

Verde, P. E. (2017) Two Examples of Bayesian Evidence Synthesis with the Hierarchical Meta-Regression Approach. Chap.9, pag 189-206. Bayesian Inference, ed. Tejedor, Javier Prieto. InTech.

Verde, P.E. and Curcio, D. (2019) Hierarchical Meta-Regression Modelling: The Case of The Pneumococcal Polysaccharide Vaccine. Technical Report.

Examples

# NOT RUN {
 library(jarbes)
 
out = gesmix(data = ppvipd)
attach.jags(out, overwrite = TRUE)

par(mfrow = c(1,2))
hist(lambda[,1],  breaks = 100, probability = TRUE,
    ylim = c(0, 3),
    xlim = c(-2, 0.85), main = "",
    xlab = "Treatment Effect: log(Odds Ratio)")

lines(density(lambda[,2]), lwd = 3, col = "red")
lines(density(lambda[,1]), lwd = 3, col = "blue")
abline(v = 0, lty = 2)
legend(-2, 3, legend = c("RCTs' Component",
                        "Obsevationals' Component"),
      col = c("blue", "red"),
      lty = rep(1, 2),
      lwd = rep(2, 2))
hist(c(lambda[,1],lambda[,2]+0.337), 
    breaks = 150,
    probability = TRUE,
    ylim = c(0, 3.5),
    xlim = c(-2.1, 0.2), main = "",
    xlab = "Treatment Effect: log(Odds Ratio)")
lines(density(c(lambda[,1],lambda[,2]+0.337)), col = "blue", lwd =3)
abline(v = quantile(c(lambda[,1],lambda[,2]+0.337), 
                   probs = c(0.025,0.5,0.975)), col ="blue",
      lty = 2)
m.pool <- log(0.43)
sd.pool <- (log(0.54) - log(0.34))/(2*1.96)
curve(dnorm(x, m = m.pool, s = sd.pool), 
     from = -3, to = 0.5, add = TRUE, col = "black", lwd =3)
legend(-2.1, 3, 
      legend = c("Bias-adjusted",
                 "Biased result"),
      col = c("blue", "black"),
      lty = rep(1, 2),
      lwd = rep(2, 2))
abline(v = 0, lty = 2)
arrows(-1.5, 1.1, -1, 1.55, lwd =2, length=0.1, angle=20)
text(-1.8, 1, "Ignoring study types")
par(mfrow = c(1,1))
# }