1
$\begingroup$

I'm trying to design a model to analyze hospital length of stay, and more specifically to ascertain the effect of sex on LOS in my data. My data contains demographic, comorbidity, and intra-operative variables (mostly continuous data like heart rate and medication dosage).

My outcomes of interest are all categorical, except for LOS. The data was collected prospectively but doesn't have a temporal aspect. LOS is just a continuous variable ranging from 0 to 120 days postop.

As such, I was wondering how I should go about calculating an effect size for sex on LOS. I considered a Cox PH model with death as competing risk, but wasn't sure how to set up the survival object as I only have the continuous/numeric "LOS" variable.

I'm of the mind that a linear regression (mixed model with hospital clusters) would be most appropriate, but in that case, what would be the best way to incorporate the death as competing risk? Would it be included as a covariate? Thank you!

Improve this question
$\endgroup$
1

1 Answer 1

Reset to default
1
$\begingroup$

LOS is usually best modeled as a count variable using Poisson or negative binomial regression. Check the distribution of the variable using a histogram.

If you would like to include hospital clusters as a random effect, you can use mixed-effects Poisson/NB regression (using glmer and glmer.nb functions from lme4 package).

Improve this answer
$\endgroup$
2

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.