This may only be a challenge for me, as it's been awhile. I am comfortable with regular recoding, but here's the challenge I'm facing.
- I have a data set of ~10,000 order entries for a medication, with individual medical record numbers as unique identifier, date of order, and where it was ordered. Each order is entered as an individual case. Therefore, for MR 123456 who had 17 orders for the medication, there are 17 individual entries.
- I would like to know the following: by MR number (individual patient), how often was the medication ordered, and were the orders in sequential dates?
My challenge has been in trying to create a variable called "frequency" and then have the system read the MR# and assign a unique identifier to each MR (so there would be 2000 "individuals" with a total frequency of the 10000 orders). the challenge is that in usual recoding your giving a value (say 1-4 out of 100) and creating a new value of 1, and this doesn't work that way. If I have the "frequency" dummy variable than I can sort and analyze as I need to start mining the data.
I am working with SPSS 21, Windows based, and with data pulled for a QI project from a large academic medical center.
All help appreciated.
Last edited by marthak
on Sun Dec 15, 2013 3:01 pm, edited 1 time in total.