Thanks for the reply.
Goal of my meta analysis is to compare 2 different surgical procedures to fix
the same problem.
Data (after systematic search) are collected from studies with one or the other
procedure applied.
There are a couple of outcome variables that measure the effetivenss of the
procedures; The same outcome data was extacted from each study.
All outcome variables are categorical with number of levels 4 or 5. Levels can
be combined to make meaningful 2 levels (making binary variables), but prefer
not to do this.
My analysis plan (say for outcome variable 1):
1. Do a metanalysis for all the studies with procedure 1 (say), combining
proportions for outcome 1 (estimating the proportions).
2. Do a metanalysis for all the studies with procedure 2 (say), combining
proportions for outcome 1 (estimating the proportions).
3. Finally just discuss the proportion diffenences between the 2 procedures
(descriptive) or test the for the proportion difference between the 2
procedures. Hope this will give some insight into the problem.Thank
you.Dushanthi
> Date: Tue, 31 Jul 2012 17:42:12 +0100
> To: dushan...@bell.net; i...@aghmed.fsnet.co.uk; r-help@r-project.org
> From: i...@aghmed.fsnet.co.uk
> Subject: RE: [R] "metafor" package, proportions: single groups wrt to a
> categorical dependent variable
>
> At 01:18 31/07/2012, Dushanthi Pinnaduwage wrote:
>
> >Thanks very much for the reply Michael.
> >I guess I have to combine some levels to make categorical variable
> >to a binary variable.
>
> Without knowing more about your outcome variable (you have not even
> told us yet whether it is ordered or not) it is a bit hard to offer
> more advice. I would regard collapsing categories as a last resort
> myself as you lose information.
>
> >
> >Dushanthi
> >
> >
> > > Date: Sat, 28 Jul 2012 16:38:24 +0100
> > > To: dushan...@bell.net; r-help@r-project.org
> > > From: i...@aghmed.fsnet.co.uk
> > > Subject: RE: [R] "metafor" package, proportions: single groups
> > wrt to a categorical dependent variable
> > >
> > > At 15:47 28/07/2012, Dushanthi Pinnaduwage wrote:
> > > >
> > > >Dear Michael,
> > > >
> > > >Thanks very much for your quick response.
> > > >
> > > >My outcome is not binary. Outcome has more than 2 values (levels)
> > > >and I have counts for the levels.
> > > >I have not written a code yet.
> > >
> > > Well as the name of the key function in metafor (rma.uni) suggests
> > > metafor at present does univariate meta-analysis. If you can reduce
> > > the outcome to a single measure of effect (perhaps using proportional
> > > odds or continuation ratio) then you can enter that with its standard
> > > error using the yi and sei parameters to rma.uni. If the nature of
> > > your outcome does not allow that then you have a multivariate
> > > meta-analysis. In that case you could try mvmeta (available from CRAN)
> > >
> > > >
> > > >I would like to know if escalc() can handle categorical outcome data
> > > >(more than 2 proportions). If so, how can I input
> > > >the counts of each level of the categorical outcome to this function.
> > > >
> > > >Hope it is little clear now.
> > > >
> > > >Dushanthi
> > > >
> > > >
> > > >
> > > >You do not give us very much to go on here. When you say your outcome
> > > > >is proportions do you mean your outcome has more than two values and
> > > > > you are generating (or have been provided with) several proportions
> > > > > for each study or do you mean it is a binary variable? If the latter
> > > > > do you in fact have the numerators and denominators or just the
> > > > > proportion? It would also help if you showed us the code you have
> > > > > tried and the error message you got (if any) or told us the
> > > > > discrepancy between the output you obtained and that which
> > you expected.
> > > >
> > > >
> > > >
> > > > > Date: Sat, 28 Jul 2012 14:25:08 +0100
> > > > > To: dushan...@bell.net; r-help@r-project.org
> > > > > From: i...@aghmed.fsnet.co.uk
> > > > > Subject: Re: [R] "metafor" package, proportions: single groups
> > > > wrt to a categorical dependent variable
> > > > >
> > > > > At 01:44 28/07/2012, Dushanthi Pinnaduwage wrote:
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > >Dear all, I am using R version 2.15.0 and 'metafor' package version
> > > > > >1.6-0. Can this version of the package handle proportions from
> > > > > >a categorical dependent variable for single studies?If so how do I
> > > > > >set up my dataframe for the raw data from different studies? Also
> > > > > >how do I give inputs, specially xi, mi (or ni) to the function
> > > > > >escalc()? Thanks,Dushanthi
> > > > >
> > > > > You do not give us very much to go on here. When you say your outcome
> > > > > is proportions do you mean your outcome has more than two values and
> > > > > you are generating (or have been provided with) several proportions
> > > > > for each study or do you mean it is a binary variable? If the latter
> > > > > do you in fact have the numerators and denominators or just the
> > > > > proportion? It would also help if you showed us the code you have
> > > > > tried and the error message you got (if any) or told us the
> > > > > discrepancy between the output you obtained and that which
> > you expected.
> > > > >
> > > > > >
> > > > > > [[alternative HTML version deleted]]
> > > > >
> > > > > Michael Dewey
> > > > > i...@aghmed.fsnet.co.uk
> > > > > http://www.aghmed.fsnet.co.uk/home.html
> > > > >
> > >
> > > Michael Dewey
> > > i...@aghmed.fsnet.co.uk
> > > http://www.aghmed.fsnet.co.uk/home.html
> > >
>
> Michael Dewey
> i...@aghmed.fsnet.co.uk
> http://www.aghmed.fsnet.co.uk/home.html
>
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
