want to create symptom episodes

[Aryyyan]

Hi all @Ksharp ,

I want to create variable episode (symptoms episode).

Criteria to create episode are

1) first episode = start with first symptom day and end when there is a gap of 2 or more days but minimum of 15 days

2) second episode = first symptom after first episode (has to be gap of 2 or more days and after minimum of 15 days since first day of first episode

3) create as many episode there is using criteria of second episode

ADT	SUBJID	PARAM	F_SYMPDT	SYMPDAY	EPISODE
29-Sep-20	VAC31518COV3001-3000107	MUSCLE ACHES PAINS	29-Sep-20	0	1
29-Sep-20	VAC31518COV3001-3000107	SHAKING CHILLS	29-Sep-20	0	1
29-Sep-20	VAC31518COV3001-3000107	SHAKING CHILLS	29-Sep-20	0	1
8-Oct-20	VAC31518COV3001-3000107	GISYMP	29-Sep-20	9	1
2-Dec-20	VAC31518COV3001-3000107	GISYMP	29-Sep-20	64	2
2-Dec-20	VAC31518COV3001-3000107	MALAISE	29-Sep-20	64	2
2-Dec-20	VAC31518COV3001-3000107	MALAISE	29-Sep-20	64	2
2-Dec-20	VAC31518COV3001-3000107	MALAISE	29-Sep-20	64	2
2-Dec-20	VAC31518COV3001-3000107	MUSCLE ACHES PAINS	29-Sep-20	64	2
2-Dec-20	VAC31518COV3001-3000107	SHAKING CHILLS	29-Sep-20	64	2
2-Dec-20	VAC31518COV3001-3000107	SHAKING CHILLS	29-Sep-20	64	2
24-Dec-20	VAC31518COV3001-3000107	FEVER	29-Sep-20	86	3
24-Dec-20	VAC31518COV3001-3000107	GISYMP	29-Sep-20	86	3
24-Dec-20	VAC31518COV3001-3000107	HEADACHE	29-Sep-20	86	3
24-Dec-20	VAC31518COV3001-3000107	MALAISE	29-Sep-20	86	3
24-Dec-20	VAC31518COV3001-3000107	MALAISE	29-Sep-20	86	3
24-Dec-20	VAC31518COV3001-3000107	MALAISE	29-Sep-20	86	3
24-Dec-20	VAC31518COV3001-3000107	MALAISE	29-Sep-20	86	3
24-Dec-20	VAC31518COV3001-3000107	MUSCLE ACHES PAINS	29-Sep-20	86	3
24-Dec-20	VAC31518COV3001-3000107	SHAKING CHILLS	29-Sep-20	86	3
25-Dec-20	VAC31518COV3001-3000107	FEVER	29-Sep-20	87	3
25-Dec-20	VAC31518COV3001-3000107	GISYMP	29-Sep-20	87	3
25-Dec-20	VAC31518COV3001-3000107	HEADACHE	29-Sep-20	87	3
25-Dec-20	VAC31518COV3001-3000107	MALAISE	29-Sep-20	87	3
25-Dec-20	VAC31518COV3001-3000107	MALAISE	29-Sep-20	87	3
25-Dec-20	VAC31518COV3001-3000107	MALAISE	29-Sep-20	87	3
25-Dec-20	VAC31518COV3001-3000107	MUSCLE ACHES PAINS	29-Sep-20	87	3
26-Dec-20	VAC31518COV3001-3000107	GISYMP	29-Sep-20	88	3

Thank you!

[ballardw]

I think that you may want to point out observations/work through a couple examples of exactly what this means:

1) first episode = start with first symptom day and end when there is a gap of 2 or more days but minimum of 15 days

Do you actually already have values for those variables Sympday and Episode? If not you need to point that out that you actually need help calculating Sympday (and maybe a few more rules about how)

If the minimum is 15 then exactly where does the 2 days actually come into play. Doesn't make much sense to me on the surface why the "2 or more" is included.

Is this supposed to per value of Param? Not at all mentioned why Param is included.

You should include something that is "not an episode" as well.

[Aryyyan]

I think that you may want to point out observations/work through a couple examples of exactly what this means:
1) first episode = start with first symptom day and end when there is a gap of 2 or more days but minimum of 15 days
• This is a symptom dataset so first episode starts when SYMPDAY = 0 for all subjid and it will continue for 15 days doesn’t matter if what’s a gap. If we have symptoms on sympday 0, 1,2,3,5,8,10,12,13,15,16,17,20 then the first episode will be until SYMPDAY 17 and the second episode will start on SYMPDAY 20. The second episode will be for at least 15 days (20+15=35) regardless of the gap. So, if we have symptoms on day 35,36,37,44 then the second episode will be until day 37 and the third episode will start on day 44.
Do you actually already have values for those variables Sympday and Episode? If not you need to point that out that you actually need help calculating Sympday (and maybe a few more rules about how)
• Yes, I have valued for SYMPDAY but not EPISODE.
If the minimum is 15 then exactly where does the 2 days actually come into play. Doesn't make much sense to me on the surface why the "2 or more" is included.
• I explained this above.
Is this supposed to per value of Param? Not at all mentioned why Param is included.
• We can ignore param.
You should include something that is "not an episode" as well.
• All observations must have episodes because this dataset includes only observations with positive symptoms.
I hope this helps. Thank you!

[Ksharp]

/*
OK.Assuming I understood what you mean.
And SYMPDAY has been sorted by ascending.
*/
data have1;
retain id 1;
infile cards dsd;
input SYMPDAY @@;
cards;
0,1,2,3,5,8,10,12,13,15,16,17,20,35,36,37,44 
;
data have2;
retain id 2;
infile cards dsd;
input SYMPDAY @@;
cards;
0,1,2,3,5,8,10,12,13,15,16,17,20,35,36,37,44 
;
data have;
 set have1 have2;
run;


data want;
 set have;
 by id;
 retain first dif_days;

 gap=dif(SYMPDAY);

 dif_days=SYMPDAY-first;
 if first.id or (gap>2 and dif_days>15) then do;first=SYMPDAY;EPISODE+1;end;
 keep id SYMPDAY EPISODE;
run;

[Aryyyan]

Thank you! Ya, that's what I was looking for with one minor change. When ID changes, the episode should reset to 1 instead of 4.

[Ksharp · Accepted solution]

OK. Add one more line:

data want;
 set have;
 by id;
 retain first dif_days;

 gap=dif(SYMPDAY);

 dif_days=SYMPDAY-first;
 if first.id then EPISODE=0;   /*<--------*/
 if first.id or (gap>2 and dif_days>15) then do;first=SYMPDAY;EPISODE+1;end;
 keep id SYMPDAY EPISODE;
run;

[Aryyyan]

Thank you so much for your help!