topic Re: how to read a dataset that contains both icd-9 and icd 10 code in SAS Programming

how to read a dataset that contains both icd-9 and icd 10 code

Bal23 — Fri, 24 Jun 2016 20:43:31 GMT

I need to dvelop a code to read a dataste that contains both icd-9 and 10 codes. Any reference for me to read?

Re: how to read a dataset that contains both icd-9 and icd 10 code

Reeza — Fri, 24 Jun 2016 20:59:21 GMT

You should be reading and storing both ICD9 and ICD10 as character fields so you need to change the length of the field possibly but besides that there should be no issue.

Re: how to read a dataset that contains both icd-9 and icd 10 code

Bal23 — Wed, 06 Jul 2016 21:20:50 GMT

Re: how to read a dataset that contains both icd-9 and icd 10 code

Reeza — Fri, 24 Jun 2016 21:08:46 GMT

Without seeing anything else ... no.

Post your code, log and sample data.

Check your input statement - is it missover truncover flowover? Which is appropriate for your situation?

Re: how to read a dataset that contains both icd-9 and icd 10 code

Norman21 — Sat, 25 Jun 2016 07:18:34 GMT

This article contains information on the Lost Card note.

http://www2.sas.com/proceedings/sugi30/058-30.pdf

Re: how to read a dataset that contains both icd-9 and icd 10 code

Cynthia_sas — Sat, 25 Jun 2016 15:57:45 GMT

Hi:
Here is a paper from 2016 that discusses a SAS macro program (presumably written by the paper authors) on dealing with a file that had both ICD-9 and ICD-10 data:
https://www.nahdo.org/sites/nahdo.org/files/Resources/Other_Resources/CDC_webinar_SASmacro_Dualcoded_Feb2016.pdf

You might want to investigate whether the paper authors will share their programs. There is a contact email at the end of the paper.

cynthia

Re: how to read a dataset that contains both icd-9 and icd 10 code

Bal23 — Mon, 27 Jun 2016 13:46:19 GMT

Thanks. I have read this before.

truncover

I added truncover option. No more "loss card" but still almost all missing data, execpt the first 2-7 variavles, the first one is also blank

Should I change Irecl?

The log shows The minimum record length was 256.

Re: how to read a dataset that contains both icd-9 and icd 10 code

Bal23 — Mon, 27 Jun 2016 13:48:39 GMT

I also added firstobs=2 because I only want to test it, without waiting for it to finish the whole records,

it still reads all records

can anyone tell how to fix it

Re: how to read a dataset that contains both icd-9 and icd 10 code

Bal23 — Mon, 27 Jun 2016 13:57:18 GMT

missover truncover obs=2;

I did use this and it only read two records, but most info was missing except the first 2-7 variables

most missing data, what should I do

Re: how to read a dataset that contains both icd-9 and icd 10 code

Bal23 — Mon, 27 Jun 2016 15:03:56 GMT

I chaged the

lrecl to the maximum value but now it only reads most variables but still many values are missing. Any suggestions.

Re: how to read a dataset that contains both icd-9 and icd 10 code

Reeza — Mon, 27 Jun 2016 15:06:11 GMT

Guessing is a waste of everyone's time.

Post your code, log and sample data.

Re: how to read a dataset that contains both icd-9 and icd 10 code

Bal23 — Wed, 06 Jul 2016 22:21:13 GMT

run;

Re: how to read a dataset that contains both icd-9 and icd 10 code

Reeza — Mon, 27 Jun 2016 15:53:10 GMT

20k is the length of the record. Your last input variable starts well before that so that's why your missing variables.

Use truncover.

Fix your input statement.

This all assumes you have a fixed width file.

Use the following to subset your file so you can debug it. Extract 20 lines and make sure your input statement is correct.

https://gist.github.com/statgeek/de01cb34a61e53f2915d

Re: how to read a dataset that contains both icd-9 and icd 10 code

Bal23 — Mon, 27 Jun 2016 16:06:56 GMT

Thanks.

I have used truncover, it does not change.

I cannot open the link. doe it need to register?

I cannot understand how to fix it. I still do not understand why it is "20,000" and do not understand "Your last input variable starts well before that so that's why your missing variables"

Re: how to read a dataset that contains both icd-9 and icd 10 code

Reeza — Mon, 27 Jun 2016 16:53:46 GMT

SAS is saying a record length is 20,000 characters/columns. You've set it to the 32K characters but that doesn't matter if there's nothing beyond this.

Your last input variable starts at column 13456. I'm assuming you have a specification document, what does that say regarding length and variables?

The code was for Powershell, a script command to extract top N lines.

gc -path file_name.csv - head N > output.txt

Re: how to read a dataset that contains both icd-9 and icd 10 code

Bal23 — Mon, 27 Jun 2016 16:55:55 GMT

Thanks. But my file is not csv, it is *.dat

Re: how to read a dataset that contains both icd-9 and icd 10 code

Reeza — Mon, 27 Jun 2016 17:03:30 GMT

Seriously?

A text file is a text file. CSV means it's comma separated, DAT files don't have a fixed structure, they can be tab delimited, space delimited and/or fixed width.

Do you have a specification document that you can share?

gc -path "full path to your file.dat" - head N > "full path to output.txt"

Re: how to read a dataset that contains both icd-9 and icd 10 code

Bal23 — Mon, 27 Jun 2016 17:09:14 GMT

Thanks. What is a full statement, with data step, proc?

Re: how to read a dataset that contains both icd-9 and icd 10 code

Bal23 — Mon, 27 Jun 2016 17:10:47 GMT

I c. you want me to use powershell. Any other suggestions?

Re: how to read a dataset that contains both icd-9 and icd 10 code

Tom — Mon, 27 Jun 2016 17:24:35 GMT

You don't need to use PowerShell to look at the records in your file. Just use SAS. So to see the first 5 records run a simple data step like this and look at the log.

data _null_;
   infile 'myfilename' obs=5 ;
   input;
   list;
run;