The Charlson Comorbidity Index - Text Mining Techniques for Healthcare Provider Quality Determination - page 177

OUTER UNION CORR
SELECT * FROM NIS.CODE11
OUTER UNION CORR
SELECT * FROM NIS.CODE12
OUTER UNION CORR
SELECT * FROM NIS.CODE13
OUTER UNION CORR
SELECT * FROM NIS.CODE14
OUTER UNION CORR
SELECT * FROM NIS.CODE15
OUTER UNION CORR
SELECT * FROM NIS.CODE2
OUTER UNION CORR
SELECT * FROM NIS.CODE3
OUTER UNION CORR
SELECT * FROM NIS.CODE4
OUTER UNION CORR
SELECT * FROM NIS.CODE5
OUTER UNION CORR
SELECT * FROM NIS.CODE6
OUTER UNION CORR
SELECT * FROM NIS.CODE8
OUTER UNION CORR
SELECT * FROM NIS.CODE9
OUTER UNION CORR
SELECT * FROM NIS.CODE7
;
Quit ;
This next part of the code appends the 15 datasets just created; each dataset with just one of the 15
columns of diagnosis codes.
PROC FREQ DATA=WORK.SORT
ORDER=FREQ
NOPRINT
;
TABLES DX /
OUT=NIS.ONEWAYFREQOFDXINAPPEND_TABLE_000(LABEL=”Cell statistics for DX analysis of
SASUSER.APPEND_TABLE_0009”) SCORES=TABLE;
RUN ;
This last part finds a frequency count for each of the diagnosis codes in the appended dataset. Instead
of printing, we collect the values into a dataset. They are listed in descending order so that we can find
the ones with the greatest frequency. The codes that have a mortality percentage greater than 1 are col-
lected and shown in Table 17. This gives us a new list to define a new index code.