Hello, I am trying to create a new dataset (let's call it new_table) from an old dataset (called old_table) where the diagnosis variable (dx1) is equal to one of 150 diagnosis codes I have. Problem is that many of these diagnosis codes are broad with any value accepted beyond the first 3 specified letters. for example, for epilepsy, G40.xx where xx could represent any set of letters. I tried 2 ways of achieving this and both have failed. Any tweaks to either method or a totally new method would be appreciated! Approach 1: Use wildcards within macro lists. I have all these diagnosis values readily available where I can easily code them all by condition and combine them into a general macro list. E.g. %let epilepsy = 'G40%', 'R56%'; %let nausea = 'R11%'; let% allicd10s = &epilepsy &nausea; Then, I thought of using proc sql to get my new table proc sql; create table new_table as select* from old_table where dx1 in (&allicd10s.); quit; This approach did not work because proc sql cannot read wild card function within macro lists.   Approach 2: use contains I created a new table with all of the available diagnosis values. I named the table pci and named the variable icd10 and then used proc sql again proc sql; create table new_table as select* from old_table where dx1 contains (select icd10 from pci); quit; Problem is that contains can only be used for a single string, which sounds quite a pointless feature. Are there any non-macro statement ways for me to use contains with an or function?  I am looking for a simpler solution than a macro statement only because I intend to use this for multiple variables in many ways (creating tables, creating other variables, summary counts, etc.)   Any help is appreciated! ... View more