* Prostate Cancer --- Baseline data;

DATA baseline;

INFILE "C:\Documents and Settings\kmed\My Documents\PrCa_cases.txt" ;

INPUT id t_0 MMDDYY8. psa_0 tx $ ;

RUN;


* Prostate Cancer --- Post-treatment PSA values ;

DATA post_tx;

INFILE "C:\Documents and Settings\kmed\My Documents\psa_file.txt"
        linesize = 189 ;

INPUT id 1-8 +1 (t_1-t_20) (MMDDYY8. +1 ) / +9 (psa_1-psa_20) (8.1 +1)  ;

RUN;

proc sort data=baseline; by id;
proc sort data=post_tx;  by id;

* Merge the baseline and follow-up data
  and (at same time) create 24 new time variables;

data all;

merge baseline post_tx; by id;

* calculate time elpsed from t_0 ;

Array t(24)       t_1-t_24; 

array fu_time(24) fu1-fu24;  * new variables to hold the differences;

do i=1 to 24; fu_time(i) = t(i) - t_0;  end;

run;

* shape of <<all>> file is WIDE ;
* make the datafile LONG ;
 
data long; * 1 observation per PSA value;

keep id time psavalue;

array psa(24)     psa_1-psa_24;
array fu_time(24) fu1-fu24;

set all;

do i=1 to 24;   
  psavalue= psa(i);
  time = fu_time(i);
  if psavalue ne . then output;  * put the non-missing psa values into dataset;
end;

proc sort data=long; by id;
proc plot data=long; plot psavalue*time; by id;
run;