Skip to content

[R-meta] Help with estimate adjustment

3 messages · Massimo Baudo, Guido Schwarzer

Massimo Baudo
# 
Hi,

I am writing for help.
In a meta-analysis using metamean in meta package I was trying to adjust with a covariate the obtained estimate to assess the impact.
This is the code I use:

MIR <- metamean (n = N..of.pts, 
              mean = MAP..mean..mmHg, 
              sd = MAP..DS..mmHg, 
              median = MAP..median..mmHg,
              q1 = MAP.Q1.mmHg,
              q3 = MAP.Q3.mmHg,
              #min = min,
              #max = max,
              studlab=author.year,
              sm = "MLN",
              method.tau = "DL",
              data=data); MIR

How can I perform adjustment (if even possible)?

Thank you in advance for your help.

Best regards,

Massimo
Guido Schwarzer
# 
Massimo,

You could either conduct a subgroup analysis (if your covariate has few different values) or a meta-regression. Here is a fictitious example.

library(meta)

dat <- data.frame(n = rep(100, 6), mean = 1:6, sd = rep(1, 6), age = runif(6, 40, 60))
dat$age.c <- dat$age - mean(dat$age)
dat$agecat <- factor(dat$age > 50, levels = c(FALSE, TRUE), labels = c("<= 50 years", "> 50 years"))

m <- metamean(n, mean, sd, data = dat)
m

# Subgroup analysis
update(m, subgroup = agecat)

# Subgroup analysis assuming a common between-study variance in subgroups
update(m, subgroup = agecat, tau.common = TRUE)
metareg(m, ~ agecat) # same result

# Meta-regression using continuous covariate
metareg(m, ~ age.c)


Best,
Guido
Massimo Baudo
# 
Thank you Guido for your precious and valuable answer.

Best,
Massimo