Keep track of patients across observations

[Tim Wallner]

Hello everyone,

I have a large dataset with 10 million observations of hospital admissions. I need to follow up on (or keep track of) patients who had a specific admission of interest (IndexAdmission) across all subsequent observations/admissions.

In the small example below, I need to track patients with IDs 9428683, 5779720, 3324298, 10020366, and 6020031, and flag them throughout the entire dataset.

I would greatly appreciate any help, as I am currently stuck with this task.

* Example generated by -dataex-. For more info, type help dataex
clear
input long PatientID int IndexAdmission
9428683 1
5779720 1
3324298 1
10020366 1
6020031 1
9597902 0
8664997 0
9428683 0
13746335 0
6354018 0
9428683 0
4510983 0
1164266 0
8572130 0
3324298 0
12393016 0
6164436 0
3324298 0
4764362 0
9428683 0
5701333 0
9428683 0
end

Best regards

[Clyde Schechter]

gen byte flag = inlist(PatientID, 9428683, 5779720, 3324298, 10020366, 6020031)

[Tim Wallner]

Thank you so much Clyde for your reply. It is no problem to do it by hand for the small sample here, but this method is not possible in the real dataset containing 10 million observations and hundreds of different index patients to be flagged. I am looking for a programmatical way to find/flag all those IDs who have at least one occurence of IndexAdmission = 1 in the whole dataset. Thank you very much!

[Joseph Coveney]

. . . in the real dataset containing 10 million observations and hundreds of different index patients to be flagged. I am looking for a programmatical way to find/flag all those IDs who have at least one occurence of IndexAdmission = 1 . . .

Maybe

keep if IndexAdmission == 1
contract PatientID, freq(discard)
merge 1:m PatientID using <original master dataset>, keep(match) nogenerate noreport

[Nick Cox]

FAQ . . Creating variables recording whether any or all possess some char.
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . N. J. Cox
2/03 How do I create a variable recording whether any
members of a group (or all members of a group)
possess some characteristic?
http://www.stata.com/support/faqs/data-management/
create-variable-recording/

egen wanted = max(IndexAdmission == 1), by(PatientID) 

* keep if wanted

[Nick Cox]

egen wanted = max(IndexAdmission), by(PatientID)

should work with the data example, but the code in #5 is deliberately a little more circumspect, to deal with e.g. missing values too.