I get errors saying: expressions using equals (=) has components that are different data types ... View more

Hum. That did not work. It is still named TIN ... View more

Ok, now with astoundings fix which works great and simple. I put fake column names for sake of privacy. Not sure if I can really put the correct column names but do not see any privacy issues so going to put actual names of columns provtin          noreadm               yesreadm I used astoundings query and now it is provtin         noreadm 1111                14 1111                  2 1112              145 1112                 9 etc. It does it correctly but I want to add a column into the mix that is maybe titled Admission so my final output looks like this provtin     admission      count 1111        noreadm         14 1111        yesreadm         2 1112        noreadm         145 1112        yesreadm           9 etc ... View more

That was the ticket. Thanks ... View more

then i merge the two with this data final; set tin2 tin3; run; the output is noreadm       yesreadm 4796                                                  155 now all i have to do is figure out how to get rid of that null stuff. I tried an if in the above but it did not get rid of the blank rows. not sure if this will matter when i am doing my chisquare. i am sure there is a way to do all of this in 1 proc sql right now i have 4 proc sql's just to get to this output. ... View more

Let me start from the first query that might help /*****************************get tin level counts for re and nored*********************************/ proc sql; create table tin as (select distinct prov_tin, count(noreadm) as noreadm, count(yesreadm) as yesreadm from claims group by prov_tin); run; the out put is 4796 rows of distinct tins and in the noreadm column there is not 1 value that is 0. it takes and counts all noreadms for that tin from another table. then the yesreadm column does the same by tin and of those I have 4641 rows that have 0 and 155 that have a count that varies from 1 to 9. It will not let me do 2 counts in one query from what I have tried based on my output table above of tin. If I try it and say in a where yesreadm >0 both counts will turn up 155. If i do not put where statement both columns will return 4796. i need it to return noreadm         yesreadm 4796                  155 The only way I have gotten this to work now is by doing it like this proc sql; create table tin2 as (select distinct prov_tin count(noreadm) as noreadm from tin group by prov_tin; run; proc sql create table tin3 as (select distinct prov_tin count(yesreadm) as yesreadm from tin where yesreadm >0 group by prov_tin; run; Then I end up with 2 tables. 1 with list of distinct tins that have 4796 rows that are no readms and another table with a list of distinct tins that are 155 rows for yesreadms. but to put together to have a final output of noreadms = 4796    and yesreadms = 155 and i do not need the tins at this point does not work. this is to get me to my final phase of being able to do a chisquare ... View more

do not want sums. sums make it equal my national output and this is to count the readmissions at a TIN level so I can do a chisquare to get ratios. so if the ntl exposure is noreadm 82578 and yesreadm is 232 of that 232 I have 155 distinct TINS and of those 1 TIN might have 7 readmissions and when I do my chi once I get the data I can get their ratio based on national. i am thinking i might need 3 SAS steps to get what I want ... View more

sorry. I misspoke. I should have noreadm at 4796 and yesreadm at 155 some of the TINS have multiple yesreadm counts. I just looked at the table and found some when you said =1 in the above. so i actually see prov_tin        noreadm           yesreadm 1111            1456                     9 2222               12                     0 3333              526                    1 4444                5                     12 I know there are 4796 unique tins that have 4796 no readmissions. then there (based on looking at the table in excel in descending) that there are 155 unique tins that have a count that is no 0. I need to get a count that shows noreadm 4796        yesreadm 155 I cannot sum because then it gives me 232 and that is what the national readmission total. the national readmissions are based on 1 tick per memberid and they either had a readmission or they did not. so my national count output is noreadm             yesreadm 82578                      232 then i have to take the memberid and attach it to their actual admissions based at a TIN level. there might be 2 doctors with different TINS that saw to the 1 memberID's care on the same date therefore that equals 2 ticks of yes for that specific TIN. When I export my SAS stuff to Access and do my count it gives me what my boss says I should have as                               noreadm                          yesreadm tin exposure             4796                                   155 ntl exposure           82578                                   232 I just need to figure out how to calculate the count for TIn so it shows right in SAS so I can do a chisquare off the above. I cannot do that at this point because the code is not pulling right. I tried doing separate steps for noreadm and yesreadm at the tin and then putting in the prov_tin info but it did not work ... View more

That gave me the yesreadm as 232 and it should only be 114 ... View more