This is a bit clearer. You have explained what is compliance between refills but it is still not clear whether the patient taking the medication can be considered compliant if he/she meets the compliance criteria only 5 times in 10 refills.

From my understanding of the problem the ratio can be easily calculated with a lag function or something that is the equivalent of it. See if this what you have in mind.

data prescriptions ; input id date : mmddyy. days num ; format date mmddyys8. ; cards ; 1 2/6/06 30 0 1 3/6/06 30 1 2 2/6/06 30 0 2 3/6/06 30 1 2 4/6/06 30 2 3 2/6/06 30 0 3 4/9/06 30 1 3 7/9/06 30 2 3 8/9/06 30 3 3 10/12/06 30 4 3 12/6/06 30 5 run ;

options validvarname = v7 ; data compliance ; if not last.id then _n_ = date ; set prescriptions ; by id ; ratio = days / (date - _n_) ; if ratio > .80 then compliance = 1 ; run ;

Obs date id days num ratio compliance

1 02/06/06 1 30 0 0.00178 . 2 03/06/06 1 30 1 1.07143 1 3 02/06/06 2 30 0 0.00178 . 4 03/06/06 2 30 1 1.07143 1 5 04/06/06 2 30 2 0.96774 1 6 02/06/06 3 30 0 0.00178 . 7 04/09/06 3 30 1 0.48387 . 8 07/09/06 3 30 2 0.32967 . 9 08/09/06 3 30 3 0.96774 1 10 10/12/06 3 30 4 0.46875 . 11 12/06/06 3 30 5 0.54545 .

Venky Chakravarthy.

On Fri, 31 Mar 2006 09:57:22 -0800, sam <sghate@GMAIL.COM> wrote:

>Thanks for your prompt replies. I appreciate that. One of the reply to >my question was to give details about the ratio that I want to >calculate and would it be possible to calcualte the ratio without >converting multiple lines of data to a single line. > >To make it more clear the above database is a prescription claims >database. The variable date denotes the data the prescription was >filled and the variable days is the number of days of supply of the >medication. > >The ratio i want to calculate is to find out the compliance of patients >taking the medications. > >compliance= days supply of medication /(second fill date (date2) - >first fill date (date1) ) > >Subjects with a ratio of >= 0.80 would have high compliance.