regroup and count cases within string variables

My dataset of 13000 patients has four variables to describe specific treatments they have received. The treatment variables are set up as string variables. Treatment 1, treatment 2, treatment 3, treatment 4. Within each treatment, there are between 0-7 different treatment types (e.g. chemo, surgery, radiation, none, refused). I am trying to regroup the variables into three groups (surgery only, surgery plus chemotherapy and or radiation therapy, all other treatments). I need to eventually determine frequencies of the new variables based on the overall cohort and then subgroups of the cohort. I have tried multiple response sets - however, this gives me total number of patients that have had each type of treatment but doesn't give information on specific combinations. (e.g. ~11000 patients had surgery, however this includes patients that had ANY surgery, not ONLY surgery). I also tried compute and aggregate functions - however, don't think I set up the include cases "if" properly. I also tried to recode the string variables into numeric, but this didn't get me any further ahead. Any other suggestions?
Hello, I'm sure you figured this out by now. First recode all your variables so they are converted from String to Numberic. After that do a simple recode to rebin the 4 variables into 3. And if I understand your final question you want to count the combinations that were the result of you recoding the 4 variables into 3. If so, you can just create a new variables and concat all three fields. I"m assuming your data is structured properly and in wide format. I don't know which program you are using, excel, spss, SQL,...but its easy regardless. If you want your data to look nice you can use spaces and commas like so:
concat(rtrim(v1),' ',rtrim(v2) etc......)

Good luck and make sure nobody dies! :)