Using the ODS tables, as @Reeza has suggested, is always possible. In this specific case, however, I personally would prefer the OUTPUT statement: proc freq data = PPIP_new; tables Tmt_ID_Name*PPH*CQ / trend out=trend_data; output out=catest trend; title 'trend by MTF for each cq'; run; (CATEST is just an arbitrary dataset name.)   I find the structure of this output dataset more convenient than that of the TrendTest ODS output dataset, which you could create as follows: ods output TrendTest=tt; proc freq data = PPIP_new; tables Tmt_ID_Name*PPH*CQ / trend out=trend_data; title 'trend by MTF for each cq'; run; You will see, for example, that CATEST contains variable Tmt_ID_Name, whereas dataset TT (again an arbitrary name) does not. ... View more

Indeed, it seems that the limitation to 12 colors is imposed by ODS styles (as @ballardw has pointed out in the other thread). This could be avoided by switching the GSTYLE system option off: options nogstyle;   Now, SAS would cycle through the default color list of the device used (excluding WHITE, the background color). The number of colors in this list (at least for the device WIN) has been increased to 38 since SAS 9.2, hence would be sufficient for your 20 graphs. However, the NOGSTYLE system option has side effects (e.g. on fonts) which may or may not be acceptable to you.   Alternatively, you could specify a list of (at least) 20 colors in the COLORS= option of the GOPTIONS statement (no need for multiple SYMBOL statements then). As a generic list to start with you could simply copy the color list from the device entry and then edit the list as you like (e.g. delete colors that are hardly visible or too similar to others etc.). I did this in the following code example, based on the default color list of the device WIN (obtained by double-clicking entry WIN in the contents of 'Sashelp.Devices' in the CATALOG window):   goptions reset=all; data test; array y[20]; do x=1 to 3; do i=1 to dim(y); y[i]=i; end; output; end; run; symbol v=dot i=join; goptions colors=(BLACK RED GREEN BLUE CYAN MAGENTA GRAY PINK ORANGE BROWN LIME PERU GOLD HOTPINK INDIGO KHAKI PURPLE SILVER SKYBLUE TOMATO); proc gplot data=test; plot (y:)*x / overlay; run; quit;   ... View more

Editor's Note: Since the character date does not conform to a given SAS date informat, the code provided by @FreelanceReinh is the best solution.   You can use the B8601DJ20.6 informat after removing the punctuation marks: data have(drop=date rename=(date2=date)); date="2000-12-12-00.00.00.860184"; date2=input(compress(date, '-.'), b8601dj20.6); format date2 datetime27.6; run; Instead of listing the punctuation marks explicitly in the second argument of the COMPRESS function, you can use the 'p' modifier to remove any punctuation marks (as well as blanks): date2=input(compress(date,,'p'), b8601dj20.6);       ... View more

If you provide the macro source code to the users, they will always be able (at least in principle) to cancel your restrictions. Have you looked into the SECURE option of the %MACRO statement? This would encrypt the contents of a macro (which must be stored in a stored compiled macro library). Thus, even a straightforward check of the DATE() value would be possible, I think. ... View more

You could, for example, insert the parentheses in the macro call:   %looper(CHL_PROF_DIAG1, (&TBL_B_DIAG_SEXW5A_CDWX.));   and still use the simple   and B.icd_diag_cd in &_varnme. ... View more