One thing you may consider later is that you don't always need to create a separate variable to create disjoint value groups (categories that don't overlap), and often multiple variables, as in this case adds to complexity.
Custom formats can be used to create groups that are honored by almost all the analysis, reporting and graphing procedures. Example using the SASHELP.CLASS data set that should be installed in your system:
proc format library=work;
value classage
11-13 = 'Young'
14-16 = 'Older'
;
run;
proc freq data=sashelp.class;
tables age;
format age classage.;
tables age*sex/chisq;
run;
Note that the format applied to the age variable works both to create the simple frequency table and to create a group used in the chisq test with sex. This is not a good set for testing chisq as the cell counts are small but for a syntax example it is small enough to run quickly and not produce a lot of output.
A big advantage of not having to add variables is that the exact same data set can be used for somewhat different analysis just by changing the format. I have data sets that use as many as 10 different age grouping formats: teen/not teen; 5 year age categories, 10 year age categories, categories that may correspond to other important ages such as minimum recommended age for a medical test or program service model.
By having the data in a one record per person per clinic you have lots of granularity for summaries. If you had summarized the data by age (not group) to get count per clinic and age then an actual mean age would be pretty trivial for each clinic:
Proc means data=clinicagesummary mean std;
class clinic;
var age;
freq count; /* count of a single age at the clinic*/
run;
A separate distribution could be done by using age in the class statement not var and applying an age grouping format. Or in proc freq with the chisq;
