How to combine a range of similar variables (recordings)

[Haakon Vegrim]

Dear Statalist community,

I am currently struggling with how to combine a range of variables with similar information into one.

To elaborate on this, I work with prescription information in a wide dataset; with each individual assigned a row, and each time (day, month, year) they filled/dispensed a prescription is recorded in separate columns, with the variable named disp_date1, disp_date2, disp_date3, etc. Everyone in the dataset has filled a prescription at least one time, with some more than a thousand times.

What I would like to know is how many did fill a prescription more than let's say 400 times.

Best,
Haakon

[Øyvind Snilsberg]

perhaps,

* Example generated by -dataex-. For more info, type help dataex
clear
input float(id disp_date1 disp_date2 disp_date3 disp_date4)
1 100 75 250  .
2  25 25   0 25
3 250  0 250  0
end

egen x = rowtotal(disp_date*)
egen wanted = total(x>400)

[Haakon Vegrim]

Thanks for your reply, Øyvind!

When I try to apply "egen x = rowtotal(disp_date*)", I get the sum of the time-values, and not how many times they filled their prescription.
Not sure what "egen wanted = total(x>400)" creates.

KR,
Håkon

[Øyvind Snilsberg]

I see. if there is one disp_date per prescription, replacing -rowtotal- with -rownonmiss- should sort you out. it not, a data example would be helpful.
-total(x>400)- is the number of observations with x>400.

[Haakon Vegrim]

Perfect, rownonmiss was exactly what I looked for! Thanks (tusen takk!;-))

[Øyvind Snilsberg]

bare hyggelig!