Multiple imputation for missing data based on domain EQ5D -3L

[Rose Matthews]

I'm not sure if there are people on here who would like to share

But is there common stata syntax to share amongst us users who would like to do a Multiple imputation on domains for EQ5D-3L

For a PROM that is reportedly used worldwide, with value sets available for most countries. It seems awkward there is no common syntax to calculate the missing data

There is stata syntax to compute the total scores but none available for missing data.

Is anyone keen to share?

[Joseph Coveney]

. . . is there common stata syntax . . . to do a Multiple imputation on domains for EQ5D-3L

If you're interested specifically in multiple imputation, is there some reason why Stata's standard suite of multiple imputation commands won't do?

help mi

Beyond that, I don't know whether there is anything peculiar to Stata, but Googling EQ5D-3L multiple-imputation seems to bring up a bunch of hits that you could look at for general approaches that others use.

[Rose Matthews]

Thanks for this, really helpful. Yes I have read most articles on MI for EQ5D-3L. The main paper is by Simons et al, who recommend using multiple imputation at Domain level rather than index level which I DON'T understand what this means. They use a OLS..
However, I thought as this is a commonly used scale that perhaps this may be readily available for everyone but I suppose as one is using different outcomes perhaps that's why MI stata code for EQ5D 3L isn't readily available for everyone to use....

unlike mean imputations for a oxford hip score...