Alternative code with easy understading

[raja777pharma]

Hello ,

I am unable to understand how below code is working.

can any one alternative code will provide with easy understand with same logic.

data ss3_epoch_6;
   set epoch2;
   by seq_id usubjid strvard sestd seend;
   retain outfl;
   if first.seq_id then outfl = 0;
   
   if strvard >= sestd and seend > strvard and sestd ne . and strvard ne . and outfl = 0 then do;
       output;
       outfl = 1;
   end;
   * If strtvar does not fit any epoch, set it to missing and output;
   if last.seq_id and outfl = 0 then do;
   * Fix for ongoing study where the end date for the last epoch is still missing.;
   if seend = . and strvard ne . then output;
   else do;
   EPOCH = "";
   if strvard ne . then PUTLOG "(epoch) STAR_CHECK: Provided date outside all epochs in sdtm.SE for subject: " 
USUBJID= N=;
   output;
  end;

Thank you,

Rajasekhar.

[Kurt_Bremser]

For every seq_id, the first observation that meets this condition

strvard >= sestd and seend > strvard and sestd ne . and strvard ne . and

is output; if no such observation is found, the last observation is output, with EPOCH set to a missing value if another condition is met.

[ChrisNZ]

I don't see how the code could be simplified. 

Better formatting (indentation) might help you to understand the logic.

I don't know the data, so can't comment further, but the coder did a good job.

It's your opportunity for growth it seems 🙂

[raja777pharma]

Hi , Below is example data sorry i am not able to write the data here.

For below code is output two obervations.

data out;
set input;
by seq_id usubjid strvard sestd seend;
retain outfl;
   if first.seq_id then outfl = 0;
   
   if strvard >= sestd and seend > strvard and sestd ne . and strvard ne . and outfl = 0 then do;
       output;
       outfl = 1;
   end;

run;

output;

I wrote the same logic but no observations are output with below code , please help me to understand what i am missing , aslo please help to write different from above.

data output;
 set input;
by seq_id usubjid strvard sestd seend;
 if first.seq_id then do;       
	    	if sestd ^= . and strvard ^= .  then do;
	      		if sestd <= strvard and  strvard < seend then do;
	      		    flag=1;
	       			output;
	      		end;   		 
            end;
       end;
run;

Thank you,

Rajasekhar.

[Kurt_Bremser]

Post usable data, like tghis:

data have;
input seq_id epoch :$20. sestdtc :yymmdd10.;
format sestdtc yymmdd10.;
datalines;
199 SCREENING 2019-09-24
;

Add additional variables and observations as needed to illustrate your issue.

[raja777pharma]

Hi ,

Thank you very much for help.

I have try te get some sammple data here and expalined rules and output.

Plese help me code with differently previosu code.

data ce;
input usubjid :$40 ceterm :$200 cestdtc :yymmdd10.;
format cestdtc yymmdd10.;
datalines;
D169CC00001/E0201004    CV DEATH    14-09-21
D169CC00001/E0201004    HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT    04-02-20
D169CC00001/E0201004    HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT    03-09-21
D169CC00001/E0201004    HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT    04-02-20
D169CC00001/E0201007    HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT    20-05-21
D169CC00001/E0201025    HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT    04-06-20
D169CC00001/E0201025    HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT    04-06-20
;
run;

data se;
input usubjid :$40 EPOCH :$200 SESTDTC :yymmdd10. SEENDTC :yymmdd10.;
format SESTDTC SEENDTC yymmdd10.;
datalines;
USUBJID	EPOCH	SESTDTC	SEENDTC
D169CC00001/E0201004    SCREENING	        07-02-19	12-02-19
D169CC00001/E0201004	BLINDED TREATMENT	12-02-19	15-09-21
D169CC00001/E0201007	SCREENING	        01-04-19	08-04-19
D169CC00001/E0201007	BLINDED TREATMENT	08-04-19	 
D169CC00001/E0205038	SCREENING	        12-09-19	19-09-19
D169CC00001/E0205038	BLINDED TREATMENT	19-09-19	20-02-20
D169CC00001/E0205038	FOLLOW-UP	        20-02-20	 
D169CC00001/E0201025	SCREENING	        26-09-19	03-10-19
D169CC00001/E0201025	BLINDED TREATMENT	03-10-19	08-01-20
D169CC00001/E0201025	FOLLOW-UP	        08-01-20
;
run;

/* From Above two data set are left joined by below code. */
/*before left join created one unique id in CE data */
/*for final output records same as first CE data */

data ce;
  set ce;
  seq_id=_n_;
run;

proc sql;
 create table ce_and_se as
 select a.*,b.epoch,b.sestdtc,b.seendtc
 from ce as a left join se as before on a.usubjid=b.usubjid;
quit;

Output data set as below:
1. Output data set have same observation from first CE data.
2. EPOCH values will be assinged based on date when SESTDTC <= cestdtc < SEENDTC.
3. if cestdtc is missing then epoch will missing values.
4. if  SESTDTC <= cestdtc < SEENDTC this logic not fall then print those values to log (only when cestdtc and SESTDTC and SEENDTC not missing)

/*Final output like below:*/
/*data have */
USUBJID	ceterm	cestdtc	EPOCH
D169CC00001/E0201004	CV DEATH	14-09-21	
D169CC00001/E0201004	HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT	04-02-20	BLINDED TREATMENT
D169CC00001/E0201004	HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT	03-09-21	BLINDED TREATMENT
D169CC00001/E0201004	HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT	04-02-20	BLINDED TREATMENT
D169CC00001/E0201007	HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT	20-05-21	BLINDED TREATMENT
D169CC00001/E0201025	HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT	04-06-20	FOLLOW-UP
D169CC00001/E0201025	HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT	04-06-20	FOLLOW-UP

Thank you.

Rajasekhar.

[Kurt_Bremser]

Please review your codes.

Make sure that your DATA steps work without ERRORs or invalid data messages in the log.

Are you sure you want to use a YYMMDD informat and not DDMMYY?

Why do you expect those FOLLOW-UP observations to match when their SEENDTC is missing?

Also, ALWAYS (as in ALWAYS) use 4-digit years. After the Y2K scare this should be an unquestioned given in computing.

[raja777pharma]

Hi ,

Format of the date variibles would be fine YYMMDD10.

For followup record below condition is met, so assinging the values , if SEENDTC is value then will check the cestdtc < SEENDTC 

SESTDTC <= cestdtc

Thank you,

Raja

[Kurt_Bremser]

This clarifies exactly NOTHING.

Please re-post your codes, tested, with 4-digit years. Make sure to use proper delimiters or the & informat modifier so that the strings with blanks are read correctly. I will not answer further until you do that.

[raja777pharma]

Hello ,

Thank you for help.

I have exceuted the below code and data sets were genrated without error.

Also date year is displayed 4 digits in data set.

data ce;
input usubjid $1-21 ceterm $22-78 cestdtc yymmdd10.;
format cestdtc yymmdd10.;
datalines;
D169CC00001/E0201004 CV DEATH                                                 14-09-21
D169CC00001/E0201004 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-02-20
D169CC00001/E0201004 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 03-09-21
D169CC00001/E0201004 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-02-20
D169CC00001/E0201007 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 20-05-21
D169CC00001/E0201025 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-06-20
D169CC00001/E0201025 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-06-20
;
run;

data se;
input usubjid $1-21 EPOCH $22-39 SESTDTC yyyymmdd10. SEENDTC yyyymmdd10.;
format SESTDTC SEENDTC yyyymmdd10.;
datalines;
D169CC00001/E0201004 SCREENING         07-02-19 12-02-19
D169CC00001/E0201004 BLINDED TREATMENT 12-02-19 15-09-21
D169CC00001/E0201007 SCREENING         01-04-19 08-04-19
D169CC00001/E0201007 BLINDED TREATMENT 08-04-19  
D169CC00001/E0205038 SCREENING         12-09-19 19-09-19
D169CC00001/E0205038 BLINDED TREATMENT 19-09-19 20-02-20
D169CC00001/E0205038 FOLLOW-UP         20-02-20  
D169CC00001/E0201025 SCREENING         26-09-19 03-10-19
D169CC00001/E0201025 BLINDED TREATMENT 03-10-19 08-01-20
D169CC00001/E0201025 FOLLOW-UP         08-01-20
;
run;

data ce;
  set ce;
  seq_id=_n_;
run;

proc sql;
 create table ce_and_se as
 select a.*,b.epoch,b.sestdtc,b.seendtc
 from ce as a left join se as b on a.usubjid=b.usubjid;
quit;

Thank you.

Raja

[Kurt_Bremser]

Where do you see a 4 digit year here:

14-09-21

Is this 2014-09-21, or 2021-09-14? Since you use the YYMMDD informat, it will be read as the former. 

[raja777pharma]

Hi ,

Thank you very much helping on to get knowldge more in SAS data step.

Yes , we need to consider ddmmyy10. 

But when i used this one  for SEENDTC ddmmyy10. getting missing values , i am not sure what is the issue.

Below one is updated one.

data ce;
input usubjid $1-21 ceterm $22-78 cestdtc ddmmyy10.;
format cestdtc ddmmyy10.;
datalines;
D169CC00001/E0201004 CV DEATH                                                 14-09-2021
D169CC00001/E0201004 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-02-2020
D169CC00001/E0201004 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 03-09-2021
D169CC00001/E0201004 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-02-2020
D169CC00001/E0201007 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 20-05-2021
D169CC00001/E0201025 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-06-2020
D169CC00001/E0201025 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-06-2020
;
run;

data se;
input usubjid $1-21 EPOCH $22-39 SESTDTC ddmmyy10. SEENDTC ddmmyy10.;
format SESTDTC SEENDTC ddmmyy10.;
datalines;
D169CC00001/E0201004 SCREENING         07-02-2019 12-02-2019
D169CC00001/E0201004 BLINDED TREATMENT 12-02-2019 15-09-2019
D169CC00001/E0201007 SCREENING         01-04-2019 08-04-2019
D169CC00001/E0201007 BLINDED TREATMENT 08-04-2019
D169CC00001/E0205038 SCREENING         12-09-2019 19-09-2019
D169CC00001/E0205038 BLINDED TREATMENT 19-09-2019 20-02-2019
D169CC00001/E0205038 FOLLOW-UP         20-02-2019
D169CC00001/E0201025 SCREENING         26-09-2019 03-10-2019
D169CC00001/E0201025 BLINDED TREATMENT 03-10-2019 08-01-2019
D169CC00001/E0201025 FOLLOW-UP         08-01-2019
;
run;

Thank you,

Raja

[Kurt_Bremser]

Use the colon (:) modifier for the date formats, or INPUT will not honor the delimiter.

[raja777pharma]

Hi ,

Thank you now seendtc have values.

Please find the below code 

data ce;
input usubjid $1-21 ceterm $22-78 cestdtc ddmmyy10.;
format cestdtc ddmmyy10.;
datalines;
D169CC00001/E0201004 CV DEATH                                                 14-09-2021
D169CC00001/E0201004 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-02-2020
D169CC00001/E0201004 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 03-09-2021
D169CC00001/E0201004 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-02-2020
D169CC00001/E0201007 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 20-05-2021
D169CC00001/E0201025 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-06-2020
D169CC00001/E0201025 HEART FAILURE HOSPITALIZATION/URGENT HEART FAILURE VISIT 04-06-2020
;
run;

data se;
input usubjid $1-21 EPOCH $22-39 SESTDTC ddmmyy10. SEENDTC :ddmmyy10.;
format SESTDTC SEENDTC ddmmyy10.;
datalines;
D169CC00001/E0201004 SCREENING         07-02-2019 12-02-2019
D169CC00001/E0201004 BLINDED TREATMENT 12-02-2019 15-09-2019
D169CC00001/E0201007 SCREENING         01-04-2019 08-04-2019
D169CC00001/E0201007 BLINDED TREATMENT 08-04-2019
D169CC00001/E0205038 SCREENING         12-09-2019 19-09-2019
D169CC00001/E0205038 BLINDED TREATMENT 19-09-2019 20-02-2019
D169CC00001/E0205038 FOLLOW-UP         20-02-2019
D169CC00001/E0201025 SCREENING         26-09-2019 03-10-2019
D169CC00001/E0201025 BLINDED TREATMENT 03-10-2019 08-01-2019
D169CC00001/E0201025 FOLLOW-UP         08-01-2019
;
run;

data ce;
  set ce;
  seq_id=_n_;
run;

proc sql;
 create table ce_and_se as
 select a.*,b.epoch,b.sestdtc,b.seendtc
 from ce as a left join se as b on a.usubjid=b.usubjid;
quit;

Thank you,

Raja.

[ChrisNZ]

---

data output;
 set input;
by seq_id usubjid strvard sestd seend;
 if first.seq_id then do;       
	    	if sestd ^= . and strvard ^= .  then do;
	      		if sestd <= strvard and  strvard < seend then do;
	      		    flag=1;
	       			output;
	      		end;   		 
            end;
       end;
run;

---

Since you force the output statement, your code will only output records that match:

 first.seq_id and sestd ^= . and strvard ^= . and sestd <= strvard and strvard < seend 

Note that

sestd <= strvard and strvard < seend

can be written

sestd <= strvard < seend