topic Re: sas programming in SAS Programming

sas programming

mounikag — Wed, 14 Apr 2021 05:24:41 GMT

Hi Team,

data x;
set y;
subj recrd dvact
101 1 action taken
101 2 withdraw from study
101 2 other
101 3 IRB given
101 4 Site reinstructed
101 5 subject counselled
101 5 other;
run;

i have raw data as above with same subject and different record numbers with different Dv actions taken .
So , i want to display that if a subject has same record numbers, then DV action should be displayed as 'MULTIPLE'(highlighted as bold letters in dataset)
as below output dataset SDTM.DV. Please look at the below output:

SDTM.DV dataset:
data DV;
set x;
subj recrd dvact
101 1 action taken
101 2 MULTIPLE
101 3 IRB given
101 4 Site reinstructed
101 5 MULTIPLE
;
run;

Multiple records in SDTM.Dv domain should be displayed in SUPP domain with their Dv actions as below output:

SDTM.SUPP_DV:

data SUPPDV;
set DV;
subj recrd QVAL qlabel
101 2 withdraw from study Multiple actions 2
101 2 other Multiple actions 2
101 5 subject counselled Multiple actions 3
101 5 other Multiple actions 3
;
run;

Could you please provide code to generate DV and SUPPDV output;

Thanks in Advance

Re: sas programming

andreas_lds — Wed, 14 Apr 2021 05:54:28 GMT

Please change the title of you topic to something more meaningful and provide data in usable form.

Re: sas programming

japelin — Wed, 14 Apr 2021 07:06:08 GMT

When submitting, please prepare the data steps written in datalines so that we can easily process the data.

The following is an example of processing.

data have;
  infile datalines dsd;
  length subj recrd 8 dvact $20;
  input subj recrd dvact $;
datalines;
101,1,action taken
101,2,withdraw from study
101,2,other
101,3,IRB given
101,4,Site reinstructed
101,5,subject counselled
101,5,other
;
run;
proc sql;
  create table dv as
    select * , count(*) as cnt
      from have
      group by  subj, recrd
  ;
  update dv set dvact='MULTIPLE' where cnt>1;

  create table supp_dv as
    select * 
      from have
      group by  subj, recrd
      having count(*)>1
  ;
quit;
proc sort data=dv(drop=cnt) nodupkey;
  by subj recrd;
run;
data supp_dv;
  set supp_dv;
  length qlabel $40;
  rename dvact=qval;
  qlabel='Multiple actions X';/*what's 2 or 3 means you describe?*/
run;

The above is handled in sql mainly, but if you are not good at sql, you can count the number of records in the data step using the retain statement, first. variable.
Once you know the number of records, all you have to do is create two datasets.
Do a little research on your own.

Re: sas programming

Kurt_Bremser — Wed, 14 Apr 2021 07:11:27 GMT

The best approach for this is a DATA step with a double DO loop. In both DO loops, read your dataset with

do until (last.recd);
  set have;
  by subj recd;

In the first loop, count the number of observations.

In the second loop, if the count is > 1, do the OUTPUTt only at LAST.RECD, setting dvact to MULTIPLE; otherwise, just OUTPUT.

For code, especially tested code, provide data in usable form, as shown here in one of your previous threads.

Re: sas programming

mounikag — Wed, 14 Apr 2021 17:40:36 GMT

Thanks will do that as suggested and the answered helped me. Thanks a lot

Re: sas programming

mounikag — Wed, 14 Apr 2021 17:42:02 GMT

From next i will try to provide data in usable form . Thanks your answered help me .