Thread: Cross sectional study- how to treat multiple samples from same patient?

1. Cross sectional study- how to treat multiple samples from same patient?

Hey there,

I know nothing about stats, but am hoping there is a simple answer to this problem.

I am doing a study looking at different immune responses in patients. Some patients I have only one sample for, but most I have 2-6 samples (over a period of several years). As of now, I've grouped them all together in my analysis, but I know that can't be right. Is there a correction to do that will take into consideration that some patients are represented more than once?

I know I could analyze this data taking just one sample per patient, but now I'm not sure how I could justify which of the 2-6 samples from each patient I've chosen, as there is no really good criteria to use to choose one over the others.

Help??

2. Re: Cross sectional study- how to treat multiple samples from same patient?

Yes, there is a way. Depends on your dependent variable, is it continuous or discrete ?
If it is continuous, you need to use a linear mixed model (LMM). If it is discrete, you can use either a generalized estimating equations (GEE) model or a generalized linear mixed model (GLMM) model.

Your study is a longitudinal study if the time aspect is important, or a repeated measures study if it doesn't.

Taking only one sample per subject is bad, since you lose data. However, taking all data without taking the correlation within a subject into account is even worse, because you artificially increase your power (your sample size which was for example 10 becomes now 20-60)

3. Re: Cross sectional study- how to treat multiple samples from same patient?

The analysis required for your kind of data can be quite complicated for beginners, so unless you are mathematically oriented (mathematician, physicist, computer scientist, etc,...), you better turn to a statistician...

4. The Following User Says Thank You to WeeG For This Useful Post:

BioMel (03-25-2014)

5. Re: Cross sectional study- how to treat multiple samples from same patient?

Excellent, thank you so much for the tip. I should have known it wouldn't be a simple 'fix'- but knowing there is a way to work this data out is very reassuring.

At the very least I can do some background reading before I go for a biostats consultation... so I have at least a vague idea what we're discussing.

Thanks again- Mel

 Tweet

Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts