Hi
I have retrieved data on abortions (date of abortions=d_inddto)) from a hospital discharge registry. Sometimes there are several abortions (dates) per person within a short period of time (for example 4 dates in 10 days), which I know is not right. On the other hand, one person can have multiple abortions - but there should be at least 12 weeks between each abortion. So, I want to make a rule that says that in a period of 12 weeks or 84 days, only 1 abortion (that is, only on date) should remain in my dataset per person (PNR). And it should only be the first date. How do I do that? My data looks like this:
PNR d_inddto
1 13/11/2013
1 14/11/2013
1 20/11/2013
1 22/11/2013
2 24/05/2015
3 01/09/2006
data want;
set have;
by PNR;
retain last_date;
if first.PNR
then last_date = d_inddto;
else if intck('week',last_date,d_inddto) le 12
then delete;
else last_date = d_inddto;
run;
Untested. For tested code, supply example data in a data step.
data want;
set have;
by PNR;
retain last_date;
if first.PNR
then last_date = d_inddto;
else if intck('week',last_date,d_inddto) le 12
then delete;
else last_date = d_inddto;
run;
Untested. For tested code, supply example data in a data step.
Thank you very much. It seems to work perfectly:-)
How about something like this?
data want;
set have;
by pnr;
_prevdate = lag(d_inddto);
if first.pnr or (d_inddto - _prevdate) >= 84;
drop _prevdate;
run;
This is just based on 84 days, without regard to week boundaries. It assumes the input dataset is already sorted by ascending values of pnr and d_inddto.
Josh
EDIT: syntax correction to code above
Thank you for your answer! Unfortunately, it removes all datelines except for the first one. Or maybe I haven't used it correctly..... Nevertheless, I got an answer by KurtBremser that seems to work better on my data. Again, thank you!
My mistake - I made a slight in error in the code. I've corrected it above. My apologies - I was in too much of a hurry to be the first responder and earn precious badges. 🙂
Josh
It is correct that in the very small dataset, I would expect only 3 records. However, if the dataset looked like this:
PNR d_inddto
1 13/11/2013
1 14/11/2013
1 20/11/2013
1 22/11/2013
1 03/05/2017
2 24/05/2015
3 01/09/2006
there should be 4 records left. I haven't tried it myself since I'm just trying to help a colleague. She said it didn't work on her data. But as I wrote, we may have used it incorrectly.
Ok - that makes sense:-) Thanks again!
Build your skills. Make connections. Enjoy creative freedom. Maybe change the world. Registration is now open through August 30th. Visit the SAS Hackathon homepage.
Register today!Learn how use the CAT functions in SAS to join values from multiple variables into a single value.
Find more tutorials on the SAS Users YouTube channel.
Select SAS Training centers are offering in-person courses. View upcoming courses for: