Date:         Wed, 21 Nov 2007 12:17:26 -0500
Reply-To:     George Joseph <gjman@HOTMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         George Joseph <gjman@HOTMAIL.COM>
Subject:      Re: Tricky Retain Step

The data is at the rx level, I have not yet merged with the visit data. So
every obs is an rx (and we have millions of them).

I was thinking more in the lines of retaining the first rx for a given
Patient (N=1) and then when N=3 (for the patient) comparing 3 to 1 and if
thats the same then retain the previous Rx (N=2) and comparing it to N=4
(for the patient) if both these are EQ then that the end of that.

The data is sorted BY Patient start_date stop_date;

So simplistically, it is set in:
by patient;
   if first.patient then do; Reset all the flags (repeat_rx,rx_change,
dose_change) to zero. Set Prev_rx, Prev_strength to whatever is on
first.patient values are.

Else do;
       if rx NE prev_rx then rx_change=1;
       if rx EQ prev_rx and strength NE prev_strenght then dose_change=1;
      end;

Obviously in the original example for Patient 1, _N_=1 and _N_=3 & _N_=2 and
_N_=4 are the same and hence repeating. Just donno how to check for that.