topic Re: Analysis of multiple different time points data per patient in Statistical Procedures

Analysis of multiple different time points data per patient

techplexus — Wed, 09 Oct 2024 15:59:37 GMT

I have data from patients who underwent tests at different time points. Some underwent 10 times over 1 year and some 100 times over 20 years. I want to analyse whether drop in a variable measured at those times was significant cause of an event like death.

For example one patient had blood pressure measured at 1, 3,7 weeks. Another at 2, 6,9, 66, 98 weeks. So if drop in blood pressure at time was significant cause of death.

Re: Analysis of multiple different time points data per patient

StatDave — Wed, 09 Oct 2024 16:53:19 GMT

There are several types of models that can be used to model clusters of correlated observations like you describe. One easy approach is the Generalized Estimating Equations (GEE) model which you can fit using PROC GEE. See the example for modeling a binary response (like your death response) using this model in the Getting Started section of the PROC GEE documentation. This model does not require equal numbers of observations per cluster (subject) nor that observations be at the same times, particularly if you use a correlation structure like TYPE=IND or EXCH as in that example. The predictors can vary over time within subjects. You should, however, have a large number of clusters for validity of the method.

Re: Analysis of multiple different time points data per patient

techplexus — Thu, 10 Oct 2024 04:13:44 GMT

I have about 736 patients' data. Will that be sufficient for this model?

Re: Analysis of multiple different time points data per patient

StatDave — Thu, 10 Oct 2024 14:16:08 GMT

Yes. Hundreds of clusters would generally be considered adequate for validity of the method.

Re: Analysis of multiple different time points data per patient

techplexus — Sun, 13 Oct 2024 12:22:38 GMT

Thank you for your reply. For example if my independent variables are continuous variables like systolic blood pressure , diastolic blood pressure and age. Patients are then diagnosed to be having heart disease like Myocardial infarction. Final endpoint variables are death or time to death.

I want to know how much drop in SBP or DBP before or after diagnosis of heart disease would be significant to predict high chances of death so that additional treatment can be given to those having high risk.

Additionally if variables like gender or education status can predict the death (this is secondary objective).

Re: Analysis of multiple different time points data per patient

techplexus — Sun, 13 Oct 2024 12:24:34 GMT

I am not sure if those kind of predictions can be done with GEE or not as the output tables didn't give me much information.

Re: Analysis of multiple different time points data per patient

StatDave — Mon, 14 Oct 2024 14:14:53 GMT

You can easily get a plot showing the effect of a continuous predictor on the event probability using an EFFECTPLOT statement. For example, adding this statement in the Getting Started example in the PROC GEE documentation gives a plot of the fitted model showing the effect of Age with other variables fixed.
effectplot fit(x=age)/ilink;