Which statistical test should be used to compare categories

#1
Hello everyone,

I'm trying to figure out what the best statistical test is for analyzing and distinguishing between following treatment groups: A, B and C.
I have endpoint: Incidence of hospitalization - total and by reasons (non-cardiac,cardiac,Heart Failure (HF))

Can I use simple chi-squared test or Fisher exact test to compare treatment groups by each reason (if I will be considered each reason as independent event)?
Note that one patient may have several reasons of hospitalisation.

Thanks in advance!
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
You need to better describe your data. So what is the source sample and how big is it? And are there only 3 treatments and how were they assigned? Provide more information and I would be happy to help.
 
#3
Hello and welcome
Just thinking aloud:
-incidence is a numeric variable, so I think Factorial ANOVA is a reasonable choice.
-Can competing risk analysis be considered?

Best wishes..
 

Karabiner

TS Contributor
#4
-incidence is a numeric variable, so I think Factorial ANOVA is a reasonable choice.
Occurence of hospitalization is a binary variable (yes/no). Therefore, Chi² sound reasonable.
But as @hlsmith already noticed, there is too little information.

Only the aggregated variable (% hospitalizations, or total number of hospitalizations vs. non-hospitalizations, respectively) is numeric.

With kind regards

Karabiner
 
#5
You need to better describe your data. So what is the source sample and how big is it? And are there only 3 treatments and how were they assigned? Provide more information and I would be happy to help.
Thank you for your support!

Results look like this:

1668582653480.png

Each treatment group has around 30-33 subjects. And the subjects with events is less than 10. I repeat that one subject may be presence like in non-cardiac category, as in cardiac category.

Thanks in advance!
 

hlsmith

Less is more. Stay pure. Stay poor.
#6
What do you expect the model to generate exactly? Also, all patients are hospitalized? This is still confusing. Describe what is going on. I am interpreting this as, all patients hospitalized, what were there hospitalized for and were they on a treatment a-c.

How were treatments assigned? I feel like once you better articulate what the sample is - we will eventually have questions about what the true denominator is.