Overlapping time periods

[smackerz1988]

Hi,

Okay this is not my strong point but can someone help me with this specification for ONTRTFL

If any of the period between ASTDT and AENDT (inclusive) overlaps with the period between ADSL.TRTSDT and ADSL.TRTEDT (inclusive) then set to "Y".

I guess it's just the overlapping that is throwing me off but I'm sure it's something simple that I'm not considering. ONTRTFL is On Treatment Flag for an ADCM dataset. So if someone is given concomitant medication and it coincides with when they were also being given the investigational drug then it should be flagged as Y. For example,  subject 1001 ASTDT was 11JUN2022 and AEENDT was 14JUN2022  then ONTRTFL would = Y as the period between ASTDT and AEENDT (11JUN2022 - 14JUN2022) overlaps with the period between TRSTDT AND TRTEDT (07JUN2022 -16JUN2022). The same applies to subject 1004 but not for subjects 1002,1003 or 1005 as the ASTDT/AEENDT time period doesn't overlap with TRTSDT/TRTEDT.  

SUBJID	TRTSDT	TRTEDT	ASTDT	AENDT	ONTRTFL
1001	07Jun2022	16Jun2022	11Jun2022	14Jun2022	Y
1002	27May2022	20Jul2022	01Jan2006	01Jan2006
1003	27May2022	20Jul2022	01Jan2020	20Jul2022
1004	15Jun2022	14Jul2022	18Jun2022	03Jul2022	Y
1005	15Jun2022	14Jul2022	01Jan1970	14Jul2022

[PaigeMiller]

Please provide context and explanation. Provide portions of sample data sets.

[smackerz1988]

Here is a small portion of the relevant data. ONTRTFL is On Treatment Flag for an ADCM dataset. So if someone is given concomitant medication and it coincides with when they were also being given the investigational drug then it should be flagged as Y

SUBJID	TRTSDT	TRTEDT	ASTDT	AENDT
1001	27Apr2022	21Jul2022	01Jan2021
1002	27Apr2022	20Jul2022	01Jan2018
1003	07Jun2022	16Jun2022	01Jan2000
1004	27May2022	20Jul2022	01Jan2005
1004	27May2022	20Jul2022	01Jan2006	01Jan2006

[PaigeMiller]

So am I correct in understanding that these time period overlap for the first 4 records but not for the fifth?

[smackerz1988]

Ah no I believe none of the records overlap. If say ASTDT was 01MAY2022 and AEENDT was 02MAY2022 for subject 1001 then ONTRTFL would = Y as the period between ASTDT and AEENDT (01MAY2022 -02MAY2022) overlaps with the period between TRSTDT AND TRTEDT (27APR2022 -21JUL2022). This is interim data atm but all columns should be populated

[PaigeMiller]

What is the proper thing to do if AENDT is missing? What do you mean by "This is interim data atm but all columns should be populated"? Can you show us data where all the fields are populated (even if it is made up) to illustrate the different possibilities of overlapping time periods or not overlapping time periods?

I think it would be helpful if you started over and explained all of this completely and clearly and completely and clearly rather than me having to piece together various different statements of yours which I have been confused by.

[smackerz1988 · Accepted solution]

I've updated the original post for clarity and after thinking about it this seems like the solution.

* ONTRTFL *;
 if cmiss(ASTDT,AENDT,TRTSDT,TRTEDT) = 0 then do;
   OVERLAP = max(0, min(TRTEDT,AENDT)- max(TRTSDT,ASTDT)+1);
 end;

If OVERLAP > 1 then ONTRTFL = 'Y';