@xxartpopxx wrote: I'd want to look at the distribution of values so the second code block may work.    How would I add a flag in the proc freq to say if the count_diff is more than 100? There is no PROC FREQ mentioned in this thread. Only PROC SQL. In PROC SQL, you perform a test to see if the difference is greater than 100 using the CASE WHEN statement in PROC SQL. If it is greater than 100, you assign the flag variable a value of 1, otherwise you assign the flag variable a value of zero. ... View more

@xxartpopxx If you provide representative sample data via a working SAS data step that we just can copy/paste to create the data, show us the desired outcome and explain the logic you have in mind to get from have to want then there is a good chance that someone will answer your question by providing working SAS code. ... View more

It looked to me like it was an invalid SAS name. start date. Needs to be start_date ... View more

Some example data would be nice. Input and what you expect as result for that. An actual appearance is helpful as a verbal description can be interpreted multiple ways. Do you have a SAS data set yet? Is your "date" and actual SAS date value or something else?   SAS has functions and formats that use SAS date values to return information about week depending on a couple of different rules depending on first day of week as Sunday or Monday and how to indicate week for end/beginning of year when a week starts or ends in a different year.   Given an incomplete data description this is something I would start with: proc freq data=dash; tables admin_dt_1 provider_Type; format admin_dt_1 yymon. ; run The format YYMON will display dates as year and three-letter month abbreviation. If that throws an error or has values that don't make sense then like the values of the date variable have to be addressed. That will give us a starting point for getting your details.   Suggestion: it would be quite a bit easier to show just the first day of the week then trying to show a range   ... View more

Once again you are not providing any example of the data that you have that contains, as a minimum, your mentioned Admin_dt_1. So we have no idea what you actually have. Or what actual date you want to be the "eligible" date for any given record.   Your title "cumulative" implies there is something to accumulate. What is it?   Without a concrete example of starting data and what output you expect for that starting data this is not a very complete description. Showing code that does not do what you want is not very helpful unless it getting close to what you want. ... View more

Since you did not post usable example data as requested, I have to assume that admin_dt_1 is a SAS date and not some funny-looking string. So you first need to expand the dates: data want; set have; do elig_date = admin_dt_1 to admin_dt_1 + 29; output; end; format elig_date yymmdd10.; drop admin_dt_1; run; and then you can run PROC FREQ on elig_date.   ... View more

\ ... View more

If these are truly 'paired', i.e., you have a measurement on subjects at the 5 timepoints presented, then you ought to consider doing a repeated measures ANOVA.  See example https://documentation.sas.com/?cdcId=pgmsascdc&cdcVersion=9.4_3.4&docsetId=statug&docsetTarget=statug_mixed_examples02.htm&locale=en .   if your data do not have normally distributed independent errors, you may need to try something like this.    SteveDenham ... View more