Counting the number of patients and frequencies for other categorical variables per study arm

[PaulaC]

Hello,

I have a clinical trial dataset and am trying to determine the frequencies of a few categorical variables as well as the number of unique patients per study arm at baseline.

I had initially done proc freq, but since I have multiple rows for each patient due to multiple visits per patient I was told that for proc freq there should only be one row/person.  So, the values I got were a count of the same patient numerous times.  I was told that I could use the where statement to only get the frequency at the baseline visit, but that did not work either.  No observations were read.

My code is below:

proc freq data=xxxxl;
where visit="001";
tables race*group sex*group /norow nopercent;
run;

I received the following message in the log: NOTE: No observations were selected from data set xxx.
NOTE: There were 0 observations read from the data set xxx.
WHERE visit='001'

I am not sure what I did wrong and what I need to do to get the correct answer.  

I am expecting the output to look as follows:

Group 1    Group 2

xxx               xxx   (these numbers would the number of unique patients in each group)

Race    Group 1    Group 2

White    xxx            xxx

Black     xxx            xxx

Other     xxx            xxx

Any help would be greatly appreciated.

Thanks

[mkeintz]

Visit is a character variable?

Then do a proc freq of visit to see what values it actually has in the data step.

And also note;

where visit='001' will NOT match a right-justified 4-byte visit value of ' 001' (note leading space).  If that's the problem, then either

   where visit=' 001'

or

   where left(visit)='001'

[PaulaC]

Yes, visit is a character variable. I strip(visit)="001.00" and I got a result but it did not seem to subset the data to give the counts with only one observation per patient. The number I got was much more than it should be. Thanks.

[mkeintz]

If you get fewer "001.00" cases than expected, then perhaps the original visit coding has errors or incosistencies.  I think you have to find a patient with no visit="001.00" to see whether there are other codes indicating baseline visit.

[PGStats · Accepted solution]

Assuming your data is sorted by some patient_Id variable, try keeping the first record for each patient:

data patients;
set xxxxl;
by patient_Id;
if first.patient_Id;
run;

proc freq data=patients;
tables race*group sex*group /norow nopercent;
run;

[PaulaC]

Thank you for your suggestion. I will try that.