What test should be used to compare nominal data between 5 groups

#1
Dear All,
I am a surgeon trying to investigate how indications for a device has changed over 5 years.
I have the following data set for patients;

Indications.jpg

I want to find out for each indication if there is a trend or a statistically significant relationship over the years.
What is the best statistical test that I can use?
Thanks in advance.

Best wishes,

Sobath
 
Last edited:

Karabiner

TS Contributor
#3
I want to find out for each indication over the 5 years if there is any trend to statistically significant relationship over the years.
You could calculate 6 Spearman rank correlations, one for each device,
between "year" and the respective frequencies for the respective device.

With kind regards

Karabiner
 
#5
Cochran Armitage Trend test is an option to test the linear trend of the proportions of 6 indications over 5 years. If you need to adjust for other covariates, you can use proc glimmix or proc logistic in SAS including categorical variable of 6 indications and year as continuous or categorical, depending on your research question. For sparse data in some indication groups, you may want to group them based on clinical meaningfulness.
 
#6
Dear yl2010fall, many thanks for your response. Is Cochran Armitage Trend test is appropriate only when one variable has two levels and the other variable is ordinal. I can take the years as the ordinal variable but still will it work in my case? I wanted to find out if there is a up or downward trend for each indication for the use this device.