identifying similar values of two columns

[riya275]

I have data that looks like -

row	episode	detail	date1	date2	detail2
1	1	BIGUANIDES	13-Feb-13	19-Jul-13	BIGUANIDES
2	4	BIGUANIDES	6-Sep-16	12-Oct-16	BIGUANIDES
3	3	BIGUANIDES	19-Oct-15	21-Jun-16	BIGUANIDES
4	2	BIGUANIDES	19-Nov-13	18-Dec-13	BIGUANIDES
5	1	BIGUANIDES+DPP-4	19-Nov-13	18-Dec-13	DPP-4 + BIGUANIDES
6	5	DPP-4+BIGUANIDES	6-Sep-16	12-Oct-16	DPP-4 + BIGUANIDES
7	4	BIGUANIDES	19-Oct-15	21-Jun-16	DPP-4 + BIGUANIDES
8	3	DPP-4	6-May-14	9-Oct-14	DPP-4 + BIGUANIDES
9	2	DPP-4	5-Apr-14	4-May-14	DPP-4 + BIGUANIDES

I only want to keep rows where the column names:  detail and detail2 are same

however do i identify cases like line5?

[Cynthia_sas]

Hi,
In addition to LINE 5, I think you also have an issue in LINE 6. You have spaces between the 2 words and the plus in DETAIL2, but do not have extra spaces in DETAIL. Would you want those to be treated the same?
Cynthia

[nicobuettner]

Like this?

data temp;
  attrib detail1 length=$30;
  attrib detail2 length=$30;
  attrib d11 length=$30;
  attrib d12 length=$30;
  attrib d21 length=$30;
  attrib d22 length=$30;
  attrib flag length=3;

  detail1 = "BIGUANIDES+DPP-4";
  detail2 = "DPP-4 + BIGUANIDES";

  d11 = scan(compress(detail1, ' '), 1, '+');
  d12 = scan(compress(detail1, ' '), 2, '+');

  d21 = scan(compress(detail2, ' '), 1, '+');
  d22 = scan(compress(detail2, ' '), 2, '+');

  if (d11 eq d21 or d11 eq d22)
    and (d12 eq d21 or d12 eq d22)
    then flag = 1;
  else flag = 0;
run;