Looking at similarity across rows

[tweaked2005]

I have a dataset with 5 variables (syndrome, patientClass, datatype, location, date). It is possible that multiple records can match on everything but date. So as an example:
Dataset
=======
Obs Syndrome PatientClass DataType Location Date
1 S1 Inpatient Pediatrics Hosp1 1/1/07
2 S1 Inpatient Pediatrics Hosp1 1/5/07
3 S1 Inpatient Pediatrics Hosp1 1/30/07
4 S2 Outpatient Surgery Hosp2 1/30/07

I need to do the following:
Indicate in a comment field if records appear at the same Location, same Syndrome, PatientClass and DataType within a 1 week period. So in the above example, obs1 and obs2 should have a comment = "MultipleDays within 1 week".

I've tried doing something with first.var, but with multiple variables to evaluate I'm running into a wall. Does anyone have a good suggestion for accomplishing this task?

Thank you in advance for any assistance.

 

[DanJR]

you could use sql, e.g. something like:

proc sql;
create table blar as
select 
   t1.*
  ,case when (select 
                count(*) 
              from 
                table t2
              where
                t2.syndrome = t1.syndrome
                and t2.patientclass = t1.patientclass
                and t2.datatype = t1.datatype
                and t2.location = t1.location
                and abs(t2.date - t1.date) <= 7) > 0 
        then 'MultipleDays within 1 week'
        else ' '
        end  
from
  table t1;
quit;

you would want to ensure you have an index on your fields though. (also, other sql solutions exist)

you could also try sorting the records then use a data step, for each by group load the date variable into an array, iterate over the array and compare dates then output the results.