2
$\begingroup$

After using the bivariate model I got a pooled sensitivity and false positive rate. How do I calculate the likelihood ratios, specificity etc with confidence intervals from here on? Can it be done using MADA on R or do I need to use something else?

$\endgroup$

1 Answer 1

1
$\begingroup$

You can indeed use the mada R package, by means of the madauni command for univariate meta-analysis and the SummaryPts command for bivariate meta-analysis (the latter being recommended for instance by Zwinderman and Bossuyt). A suitable example follows:

install.packages("mada")
library("mada")
data(AuditC)

# univariate meta-analysis
negLR.DSL <- madauni(AuditC, type = "negLR", method = "DSL")
summary(negLR.DSL)
forest(negLR.DSL)
negLR.MH <- madauni(AuditC, type = "negLR", method = "MH")
summary(negLR.MH)
forest(negLR.MH)
posLR.DSL <- madauni(AuditC, type = "posLR", method = "DSL")
forest(negLR.MH)
summary(posLR.DSL)
forest(posLR.DSL)
posLR.MH <- madauni(AuditC, type = "posLR", method = "MH")
summary(posLR.MH)
forest(posLR.MH)

# bivariate meta-analysis
fit <- reitsma(AuditC)
mcmc_sum <- SummaryPts(fit, n.iter = 10^6)
summary(mcmc_sum)
$\endgroup$
6
  • 1
    $\begingroup$ The OP did say s/he was using a bivariate model so I suspect SummaryPts is the function s/he needs. $\endgroup$
    – mdewey
    May 3, 2017 at 15:30
  • $\begingroup$ @mdewey Thanks, I had misunderstood his request, I have amended the code and now supply a more comprehensive code $\endgroup$ May 3, 2017 at 16:13
  • 1
    $\begingroup$ No I meant SummaryPts not summary. $\endgroup$
    – mdewey
    May 3, 2017 at 16:22
  • $\begingroup$ @mdewey I don't follow now. Isn't the updated code correct? $\endgroup$ May 3, 2017 at 16:24
  • 1
    $\begingroup$ I think your edit had not made its way through when I made my comment, I may have been seeing a cached version. $\endgroup$
    – mdewey
    May 3, 2017 at 17:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.