About shellp55

shellp55 · ‎03-26-2013

Hi Reeza Thanks for helping out but I copied exactly what you have and it still isn't working...I get where A001 shows up for both the 111111 charts that have valid readmits. Where was the typo? Thanks.

shellp55 · ‎03-20-2013

Hi Reeza Changing it to i.visit=r.visit-1 creates duplicates of the visits and doing it the other way then gets the "previous acctno" to be the next acctno. Any suggestions? Thanks.

shellp55 · ‎03-19-2013

Hi Reeza I spoke too soon. With a larger database it appears that all subsequent readmits are linked to visit #1 whereas the readmit difference is still connected to the various visits. I added a visit for chartno 111111 below and you'll note that even though the 3rd visit is connected to the 2nd visit, it shows the account number of the first visit, which I don't want. I want the 3rd visit to have the account number of the 2nd visit because that is the one that is within 30 days of it. Thanks. data test_grp; input @1 chartno $6. @7 acctno $4. @11 admdate yymmdd8. @19 sepdate yymmdd8. @27 cmg $3.; format admdate yymmdd10. sepdate yymmdd10.; cards; 111111A0012012032920120331001 222222A0022012050120120515003 333333A0032012060120120607321 111111A0042012040220120410320 444444A0052012060320120615109 555555A0062012053120120603149 222222A0072012070120120713001 222222A0082012072020120725026 444444A0092012081320120817027 333333A0102012061520120621028 111111A0112012110120121130004 111111A0122012050320120517001 run; proc sort data=test_grp; by chartno admdate;run; Data readmits; set test_grp; by chartno admdate; length index_acctno $4.; length index_cmg $3.; dif=ifn(first.chartno,.,admdate-lag(sepdate)); if first.chartno then index_acctno=""; else index_acctno=lag(acctno); if first.chartno then index_cmg=""; else index_cmg=lag(cmg); format admdate ddmmyy10. sepdate ddmmyy10.; run; Data readmits; set test_grp; by chartno; dif=ifn(first.chartno,.,admdate-lag(sepdate)); format admdate ddmmyy10. sepdate ddmmyy10.; *create an episode; if first.chartno then episode=1; if dif > 30 then episode+1; *count visits within episodes; if first.chartno the visit=1; else if dif <=30 then visit+1; else visit=1; *flag a readmit; if dif<=30 and dif ne . then readmit=1; run; *merge in fields, you can add the other fields in the same manner as acctno; proc sql; create table readmits_with_index as select r.*, i.acctno as index_visit from readmits as r left join readmits as i on r.chartno=i.chartno and r.episode=i.episode and i.visit=1 and r.readmit=1 order by chartno, episode, visit; quit;

shellp55 · ‎03-19-2013

Thanks Reeza! This works great.

shellp55 · ‎03-19-2013

Hello So below is a sample of what I'd like to see i.e. when the difference between the previous separation date and the next visit admit date is <=30 days, then I want the previous account number to show up on the line of the 30 day readmit. I realized from my original code that I will need to sort by chartno but also admdate to ensure that the visits are in chronological order. I'm hoping that whatever method is shown to me, I can then also extract previous diagnosis and triage level using that same method. Thanks for any and all assistance.

shellp55 · ‎03-15-2013

Hello I am using using SAS 9.3. I have a database where I'm trying to determine readmits for hospital patients. I consider the visit that links to the readmit as an "index" case. The code below provides me with the field of "diff" and then I can select all those that match criteria of <= 30 days. However, I wish to also report on the index case information so is there any way to have a field for "previous account number" or any previous data? So what I'm asking is how to have the readmit case also show specifics from the index case all on the same line of data? Thanks. data test_grp; input @1 chartno $6. @7 acctno $3. @10 admdate yymmdd8. @19 sepdate yymmdd8.; format admdate yymmdd10. sepdate yymmdd10.; cards; 1111110012012032920120331 2222220022012050120120515 3333330032012060120120607 1111110042012040220120410 4444440052012060320120615 5555550062012053120120603 2222220072012070120120713 2222220082012072020120725 4444440092012081320120817 3333330102012061520120621 1111110112012110120121130 run; proc sort data=test_grp; by chartno admdate;run; Data readmits; set test_grp; by chartno; dif=ifn(first.chartno,.,admdate-lag(sepdate)); format admdate ddmmyy10. sepdate ddmmyy10.; run;

shellp55 · ‎02-05-2013

Hello I am using SAS 9.3. I have a dataset where one record is one hospital visit. I am running the data by health card number (unique) and I want to exclude ALL records per health card number when one of the visits indicates death. I know about using merge and "not in" but that was for a single record; this is excluding any record based on health card number. I imagine I would still create a separate data set of deaths but just not sure how to remove all versus just the record not in. Any and all assistance greatly appreciated.

shellp55 · ‎02-05-2013

Hi Because I had inpatient and ED visits all in one file, my first step was to create an ED only subset with the data elements I wanted. From there I used proc summary to get total ED visits and then a formula to rank and order them. Then I combined each subset together for one table showing health card number (this is the more unique value), each of the visits/los, and a rank and order for each element which then also allowed me to sum them. So I'm not sure it's right but it worked. Below is my code for ED visits. /*ED Visits*/ data ED_Activity; set data_final; if type = "ED" then output; run; proc summary data=ED_activity; var ed_visits; class HCN_NoVC; output out=ed_visits sum(ed_visits)=ttl_ed ; run; proc sort data=ed_visits; by ttl_ed; run; data ED; set ed_visits; /*by HCN_NoVC descending ttl_ed;*/ ed_order+1-ed_order*first.HCN_NoVC; ed_rank+ifn(ttl_ed=lag(ttl_ed),0,1)-ed_rank*first.HCN_NoVC; run;

shellp55 · ‎02-04-2013

Hi Can someone please give me a starter? Where I'm confused is because I want it to group on chart number and sum ed visits (or use the where statement of where type = ED). But I don't see how to do that in the samples provided online because their examples don't have many, many extra columns of data like I do, only the ones they will be using in their summary. Thanks.

shellp55 · ‎02-04-2013

Wow this group is fast to respond...thanks so much! VD, I reviewed that support document prior to posting and my issue is that there isn't a number field for ED or IP visit, just a field of Type which indicates IP or ED. Of course I guess I could create two number fields for these where if critiera fits then 1 so that is an option. Reeza: thanks for breaking it into steps for me, it makes sense. My only issue is that I have never used Rank or Proc Means or Proc Freq so not sure where to start on that. Thanks again everyone.

shellp55 · ‎02-04-2013

Hi I am using Base SAS 9.3 on a Windows 7 computer. I have become quite adept at creating data sets in SAS but leave them at the record level, export them and report on the data sets with other software. For this exercise, I need to group and rank the data which is totally out of my depth so hopefully someone can help. My data is where one record is one hospital visit which could be ED or inpatient. I've created a field called Type which identifies ED or IP. The unique identifier per patient is chart number. I need to come up with a single rank per chart number which will consist of the ED visits rank + inpatient visits rank + total length of stay rank. So I need to summarize the data by each of the criteria (# of ED visits, # of IP visits and # of inpatient days) and then rank each which is simply sorting in ascending order and assigning a rank based on where that chart number appears in the list for that criteria. So if chart # 12345 is 45 on the ED visits summary, 28 on the inpatient visits summary and 100 on the inpatient days summary then that chart number's total rank is 173. Ideally I'd like to end up with a list by chart number showing rank of ED visits, rank of inpatient visits and rank of inpatient days plus total rank. Is this possible? Thanks for any and all assistance.

shellp55 · ‎01-31-2013

Hi Okay, I don't know why it's working this time but I revisited all the previous information I had including the blurb from the SAS knowledgebase site. I used what they had like I did before but this time it worked. I also found the syntax to use the password function (use -s and your password right after - in the example below my password is 1234). So the following works and I thank all who tried to help me! filename testing pipe '"C:\Program Files\WinZip\wzunzip.exe" -s1234 -o -c F:\myfolder\test.zip' ; data test_import ; infile testing firstobs=10 truncover LRECL=5000; input @1 Prov $1. @2 Inst $4. @6 Fyear $4. @10 Period $2. @12 Batch $2. @14 AbsNo $3. @18 Coder $2. @20 ChartNo $10.; run; Note that as was alluded to by DBailey (and SAS knowledgebase), you have to select firstobs=10 to bypass the code from WinZip. I think another issue for me was that I wasn't including the firstobs statement first. Note that I don't have to cite the files within the zip which is great because I have lots. If I wanted to exclude some I could use the -e syntax and I would imagine I can limit by file extension versus name if I wanted to. I am so very pleased to get this darn thing working! Thanks again to anyone who responded!

shellp55 · ‎01-31-2013

Hi Thanks for sticking with me! Stig: thanks, yes I've looked at this and can't get that to work either. I read in another post where someone contacted the author of the paper and found he didn't test all of his code. DBailey: I am the only person on my computer but I have an admin and user profile. I ran the program from my admin profile just in case there were issues with my not being able to use the x command and there was the error message about no physical file existing. So, using either profile, no file shows up in the F folder suggesting the syntax for unzip is incorrect. Any other thoughts?

shellp55 · ‎01-31-2013

Hi Thanks for the responses. Henrik: there was a message of "NOTE: DATA STEP stopped due to looping" so can't it just be an if statement? DBailey: I assume that this is to unzip the file to the folder and from there I can input it? If so, that didn't work because I get the error message "ERROR: Physical file does not exist, F:\myfolder\test.txt." I got the following to work in that it shows in the log what the contents are but I want to read and extract/manipulate the data while keeping it within the zip file. Is that not possible? Even if I extract it, manipulate it and then delete the extraction (leaving the zip file as is), is that possible? Also, I haven't even gotten to password protection yet which I was hoping I could include or there is no point in this exercise. I got this to work to show the output in the log: filename testing pipe '"C:\Program Files\WinZip\wzunzip.exe" -c F:\myfolder\Test.zip test.txt'; data _null_ ; infile testing; input ; put _infile_ ; run; Thanks for any and all assistance!

shellp55 · ‎01-31-2013

Hi Thanks for the replies but still not working in that I get the message 0 files read but no error messages. Henrik according to the SAS knowledgebase, pipe and wkunzip are supposed to work together even though one of the messages showing up is "unnamed pipe access device".

Online Status	Offline
Date Last Visited	‎11-26-2023 11:40 PM

Re: Exporting SAS dataset to MS-Access

Re: Exporting SAS dataset to MS-Access

Re: Exporting SAS dataset to MS-Access

Re: Exporting SAS dataset to MS-Access

Re: Exporting SAS dataset to MS-Access

Re: Exporting SAS dataset to MS-Access

Exporting SAS dataset to MS-Access

Re: Memo Field in First Row of Excel Export?

Memo Field in First Row of Excel Export?

Re: Creating Export Name from Variables?

Re: Creating Export Name from Variables?

Re: Exporting Data to Acess - Error

Re: Exporting Data to Acess - Error

Re: Rename Macro Do Loop Where Value is 1 Less than Current Occurence

Re: ODS Excel tagattr not working for data types/formats

Re: Combining Readmit with Original Visit Data

Re: Combining Readmit with Original Visit Data

Re: Combining Readmit with Original Visit Data

Re: Combining Readmit with Original Visit Data

Re: Combining Readmit with Original Visit Data

Combining Readmit with Original Visit Data

Merge to Exclude Certain Cases

Re: Rank on Groups?

Re: Rank on Groups?

Re: Rank on Groups?

Rank on Groups?

Re: How To Correctly Use Pipe and WKUNZIP?

Re: How To Correctly Use Pipe and WKUNZIP?

Re: How To Correctly Use Pipe and WKUNZIP?

Re: How To Correctly Use Pipe and WKUNZIP?