Problems with comparing 3 variables

#1
Hi,
I need help with my data.
Data are of 9 groups of patients. Not all patients in a group received a medication and not all patients received the same number of doses.
I need to compare the groups on how many medications per se was given in a group, or in other words which group is sickest.
The example is
A1 group consist of 465 patients and 82 of them received medication and a total of 137 medications were given
A9 group consists of 319 patients and 56 patients received medication and a total of 78 medications were given.
The question is, which group was sicker?
Meaning which group had a combination of No. of patients and medications given, based on total No. of patients that had bigger "load" of medications? Or the most important question...is there is a trend from A1 to A9?
The small table is attached in excel.
Please any help about this problem What statistical procedure can "even out" the groups based on these 3 variables.
Thank you
dr. kristijan
 

Attachments

#3
http://www.talkstats.com/threads/how-to-compare-groups-of-3-variables.75464/post-221404
Please, do not post the same question into several subforums.
It might create unnecessary work for those who try to be helpful.

With kind regards

Karabiner
http://www.talkstats.com/threads/how-to-compare-groups-of-3-variables.75464/post-221404
Please, do not post the same question into several subforums.
It might create unnecessary work for those who try to be helpful.

With kind regards

Karabiner

Sorry,
I will delete one of them.
thank you
 
#5
You could perhaps consider mean number of medications in each group (e.g. 137/465 in group 1).

With kind regards

Karabiner

One mathematician suggested a coeficient (number of patients that received medication x number of medication) / number of patients in a group. Like this because first two are directly proportional and the third is inversely proportional to the "load" of medications given per group. Don't know if I can use this coeficient?
Looks logical.
 

Karabiner

TS Contributor
#6
This looks like an elegant solution. I have never seen it used in a study, though.
So I am not sure whether there might be a serious problem with it. Did you
perform a search on the web whether this coefficient was already used
somewhere?

With kind regards

Karabiner
 
#7
No, i haven't found anything specific, but cowficients are widelly ised, and principles of mathematics are true...the peoof is this computer we are using, it is nothing bit mathematics. This coeficient just potentiates the difference.
If you analyze it logically it makes sense...2 variables are directly proportional and the third is inversly proportional. Simple math that we used in primary school. Math is not wrong. The problem is that doctors don't know math.
I'm putting the coeficient along with some other statistics.


This looks like an elegant solution. I have never seen it used in a study, though.
So I am not sure whether there might be a serious problem with it. Did you
perform a search on the web whether this coefficient was already used
somewhere?

With kind regards

Karabiner