Meta-regression and funnel plot for Hazard Ratios

[Mahdie Raj]

Hi Everyone,
I am conducting a meta-analysis on 11 longitudinal population-based studies. I ran my meta-analysis using metan function. However, for meta-regression analysis and/or funnel plot, I need SE associated with Effect sizes (here lnHazrd ratio). How can I get SEs associated with LnHRs? It is of note that metan function generates lower and upper limits related to each HR, not SEs.
Best,
Nader

[David Fisher]

Hi Nader,

I'm away from my computer right now, but I thought that metan did leave behind the standard error as well as the LCI/UCI, as a variable named _seloghr or similar?

If I'm mistaken, then assuming the confidence limits were generated (or may be assumed to be, given a sufficiently large sample) 95% Normal limits, you can estimate the standard error as follows (where UCI and LCI relate to the log HR):

SE(log HR) = (UCI - LCI)/(2*1.96)

Hope that helps,

David.

[Mahdie Raj]

Hi David,
Thank you so much for your valuable comment.
To running my random-effects meta-analysis, I converted HRs and corresponding limits to an "ln" format first and foremost. Then I used below "metan" function and it left lnHR, lnUL, lnLL, and weights.
Do you know what option I should add to the following "metan" function to get SEs directly?
Does your suggested equation generate lnSE corresponding to lnHR?
Thanks,
Nader

metan ln_HR ln_HR_Lower ln_HR_Upper, eform effect("Hazard Ratio") random ///
lcols(study ) astext(50) double by(study) ///
xlabel(0.2, 0.5, 1, 2, 5) textsize(100) title("XXX") null(1) force

[Mahdie Raj]

I highly appreciate your thoughts on my question!