About tmm

tmm · ‎05-17-2012

PROC FREQ DATA = ReAdm.CHI ORDER = DATA; BY COMPARISON; WEIGHT COUNT; TABLES SVRTY_LVL_CD*WHO*ADMISSION/OUT=READM.FINAL1 EXPECTED OUTEXPECT; EXACT OR; RUN; The above code is not including anything that has a zero. Example data: who comparison svrty admission count 1111 1111 1 readmit 1 1111 1111 1 noreadmit 66 1111 ntl 1 readmit 955 1111 ntl 1 noreadmit 33812 1111 1111 4 readmit 1 1111 1111 4 noreadmit 0 1111 ntl 4 readmit 39 1111 ntl 4 noreadmit 219 I have a lot like this and need the above CHISQ to include in the expected output the ones with 0's too. when I run this it will only include those that have anything but a 0

tmm · ‎05-17-2012

That does not return anything other than who svrty admit 1111 3 12 It leaves out the other svrty codes which I need them to be returned. A case where it would not be returned is this: who svrty admit 2222 1 1 2222 2 3 2222 3 0 2222 4 0 None of these would be returned because the sum of the admit is not >=10 it is 4 I need all severities returned if the who has an admit sum of >=10 so in my first example all 4 severities should be returned because the sum of the TIN is 28

tmm · ‎05-17-2012

/*********************************GET TINS WITH TOTAL ADMISSIONS >=10 ONLY*********************/ PROC SQL; CREATE TABLE ReAdm.TIN_Merge3 AS (SELECT * FROM ReAdm.TIN_Merge2 WHERE ADMIT >=10 GROUP BY TIN); RUN; I need to code this to grab all my severities. Right now it will only grab those that have admit >=10 and my example table looks like this who svrty admit 1111 1 5 1111 2 4 1111 3 12 1111 4 7 If I run the code above it only returns the svrty 3 and I need it actually to return all because the admits summed together are 28. I guess I need to code it where the sum of admit >=10 so it will return all svrty items.

tmm · ‎05-16-2012

Nevermind. I was able to take a previous table I created that summed my noreadmits and readmits and give me a totaladmission column and I just created a SAS query off that which says give me the totaladmissions >=10 but leave rest of the columns for me.

tmm · ‎05-16-2012

That does not return the right results. The old code that worked for the readmit side was this: PROC SQL; CREATE TABLE ReAdm.CHI2 AS SELECT * FROM ReAdm.CHI GROUP BY COMPARISON HAVING SUM(UPCASE(ADMISSION)='READMIT' AND COUNT=0)=0; RUN; But now the goal is to return the readmit and noreadmits where the sum of the readmit and noreadmits has a count of >=10

tmm · ‎05-16-2012

And this will keep the readmit and noreadmit part of the admission column?

tmm · ‎05-16-2012

I have a table that has the following: comparison admission count 1111 noreadmit 0 1111 readmit 0 2222 noreadmit 12 2222 readmit 1 3333 noreadmit 12 3333 readmit 4 I want to return all the above data but only if there is a sum between readmit and noreadmit >=10 but I need it to also return the readmit associated. This is the code I used but it does not return correctly. /*********************************GRAB ONLY THOSE TINS THAT HAD A READMIT**************************/ /*********************************FOR CHI ALONG WITH NATIONAL OVERALL****************************/ PROC SQL; CREATE TABLE ReAdm.CHI2 AS SELECT * FROM ReAdm.CHI GROUP BY COMPARISON HAVING SUM(UPCASE(ADMISSION)='NOREADMIT' AND COUNT>=10)=0; RUN;

tmm · ‎05-09-2012

I know it does not make a whole lot of sense but it is what they want. Because when I run the 2 separate and look at it you end up with 2 expecteds. 1 for the admission and one for the readmission. I will go back to them and see if we can outst the expected because it is just not going to work the way they think. Might need just 2 separate tabs so they can see expected. Don't know. Execs just want what they want and do not care about obstacles.

tmm · ‎05-09-2012

I am just going to do both codes, mine and yours and then do a proc sql to get the expected frequency under independence into the output table I produce with my code. That will save me some manual work via Excel because I can link on the TIN and just say <> national in the who section. I just don't see in sashelp.heart where there is anyway to do an odds ratio, cmh and expected and have it output odds ratio, cmh and expected information. Kind of strange that the SAS creators would not have a code that outputs expected along with or, cmh, chisq, all or anything else. I have tried all and expected does not show up using just all. I have tried all with expected outexpect and still I do not get any OR, CMH, CHISQ nothing. I get the same output with only 7 columns no matter what I do with an expected outexpect in it. I don't want that. I want the 38 or so columns I produce with OR and CMH but I also want the expected frequency under independence in there.

tmm · ‎05-09-2012

No. The data is in the CHI2 table and final is the output table.

tmm · ‎05-09-2012

I need expected, or and cmh and I need it to output to that readm.final table so I can export it to Excel.

tmm · ‎05-09-2012

That is so true. But I did it with 10 separate statements to get it done. They don't care how it is done as long as it is done. Got to love execs. Yeah our claim system is somewhat messy. We have far too many of them as well. Too many entities acquired in such a short period of time and we have all these ODBC and DB2 connection strings to connect to the various servers and DB's and all have different data dictionaries and ways of doing things. It does create a mess for us programmers and I am kind of new to SAS.

tmm · ‎05-09-2012

I so wished I could attach what mine produces. Maybe I can just include 1 TIN and see if that is small enough for an Excel. Well no dice. Even 1 TIN is still too large. So I took and split all the columns. What I am pasting here the columnsyou can see but they are all on row 1. I have a total of 38 columns produced with my code and again it all runs across 1 row. It is from column a all the way to column cj. I even added the expected outexpect to the code in hopes of getting what I produce with my code to include the expected in there. COMPARISON SVRTY_LVL_CD N 010473372 2 41625 010473372 41625 _RROR_ L_RROR U_RROR XL_RROR XU_RROR 7.214907508 0.750132724 69.3942401 0.13735972 89.89531756 _CMHCOR_ DF_CMHCO P_CMHCOR _CMHRMS_ DF_CMHRM 4.012935382 1 0.045152474 4.012935382 1 P_CMHRMS _CMHGA_ DF_CMHGA P_CMHGA _MHOR_ 0.045152474 4.012935382 1 0.045152474 7.214907508 L_MHOR U_MHOR _LGOR_ L_LGOR U_LGOR 0.750132724 69.3942401 7.214907508 0.750132724 69.3942401 _MHRRC1_ L_MHRRC1 U_MHRRC1 _LGRRC1_ L_LGRRC1 5.661180631 1.036309133 30.92606746 5.661180631 1.036309133 U_LGRRC1 _MHRRC2_ L_MHRRC2 U_MHRRC2 _LGRRC2_ 30.92606746 0.784650479 0.445608932 1.38165178 0.784650479 L_LGRRC2 U_LGRRC2 _BDCHI_ DF_BDCHI P_BDCHI 0.445608932 1.38165178 PROC FREQ DATA = ReAdm.CHI2 ORDER = DATA; BY COMPARISON; WEIGHT COUNT; TABLES SVRTY_LVL_CD*WHO*ADMISSION/CMH EXPECTED OUTEXPECT; EXACT OR; OUTPUT OUT = ReAdm.Final OR CMH; RUN; My code above produces what I just pasted. You see there is no national nothing in there. I only want the numbers from national cuz that is the denominator. The TIN numbers are my numerator. The problem is how to incorporate my code output with your code output for expected for the TIN only. I guess if I cannot find the right SAS code to do this by the end of today I will just manually manipulate it in Excel because this project is due by end of day Friday and I am at a loss.

tmm · ‎05-09-2012

The above is based on your code and they do not want to see it like that. My code does not have any national stuff in the who. It uses the national numbers but only returns the TIN info but there is no expected returned when using my code. What the execs want is what is produced from my code but they also want an expected column in there. If I do your code to get expected I will have to manually clean the file and the file produces using your code 1864 rows that I would have to manually clean and then add to what mine outputs. I am not sure why there is not a way to do my code and have the output include the expected with it like it does from your code. If I have to manaully complete the puzzle I will. Just did not want to do that and not sure why SAS would not have thought of this. I mean it makes no sense to me as robust as everyone keeps telling me SAS is.

tmm · ‎05-08-2012

I tried it and it gave me all the claims but it did not give me the 2 digits on the end to make the claim 12 digits in length. With the 1,10 it chops off those last 2 digits that are the page numbers for the 10 digit claim. That is ok. I just did 10 like queries and 1 of the 10 digit claims per and it is on number 6 running now. I am under a time constraint. This was due an hour and a half ago so I really don't have time to research or read any books or anything else. So I did the fastest, easiest thing I knew. I will then export it all and put it in Excel and just apologize. This project was thrown on me the last minute as things usually are and they said need it by 6pm est. Always the case.

Online Status	Offline
Date Last Visited	‎09-01-2015 07:11 AM

Place quotes around column header and data

Re: Point = error

Point = error

Re: Insert into confusion

Re: Insert into confusion

Insert into confusion

Re: create a proc sql without the need for multiple proc sqls

create a proc sql without the need for multiple proc sqls

Re: how to combine 2 inner joins to proc sql

how to combine 2 inner joins to proc sql

format max date

concatenating number columns

CHI INCLUDE 0 TOO

Re: sas code help

sas code help

Re: count help

Re: count help

Re: count help

count help

Re: chi sq with expected

Re: chi sq with expected

Re: chi sq with expected

Re: chi sq with expected

Re: Like with In statement

Re: chi sq with expected

Re: chi sq with expected

Re: Like with In statement