How Split

[sasuser143]

If there are number of observations in the same variable, how to separate them

data have ;

101 1-LIGHT SENSITIVITY OU_2-BURNING OU_3-PUNCTATE EPITHELIAL EROSION OU

102 1-Cellulitis of right lower extremity_2-Abscess of the right lower extremity

103 1-Burning (Forehead, Left Cheek, Right Cheek, Chin)_2-Itching (Forehead, Left Cheek, Right Cheek, Chin)_3-Stinging (Forehead, Left Cheek, Right Cheek, Chin)

104 1-LIGHT SENSITIVITY OU

run;

data want ;

101    LIGHT SENSITIVITY OU

101    BURNING OU

101    PUNCTATE EPITHELIAL EROSION OU

102    Cellulitis of right lower extremity

102     Abscess of the right lower extremity

103     Burning (Forehead, Left Cheek, Right Cheek, Chin)

103     Itching (Forehead, Left Cheek, Right Cheek, Chin)

103     Stinging (Forehead, Left Cheek, Right Cheek, Chin)

104     LIGHT SENSITIVITY OU

run;

[Kurt_Bremser]

The SCAN function accepts a third parameter which defines the delimiter. Use it in a DO loop with the underline as delimiter, and a second time with the hyphen as delimiter to remove the leading digit.

[PeterClemmensen]

If you data is representable, then try

data have;
input ID string $ 4 - 200;
datalines;
101 1-LIGHT SENSITIVITY OU_2-BURNING OU_3-PUNCTATE EPITHELIAL EROSION OU                                                                                         
102 1-Cellulitis of right lower extremity_2-Abscess of the right lower extremity                                                                                 
103 1-Burning (Forehead, Left Cheek, Right Cheek, Chin)_2-Itching (Forehead, Left Cheek, Right Cheek, Chin)_3-Stinging (Forehead, Left Cheek, Right Cheek, Chin) 
104 1-LIGHT SENSITIVITY OU                                                                                                                                       
;

data want(keep = id w);
   set have;
   by ID;
   length w $200;
   c = countw(string,, 'd');
   do i = 1 to c;
      w = compress(scan(string, i,, 'd'), '_-');
      output;
   end;
run;