Help with collapsing rows by identification number.

[blacklucas1]

Hi all - been struggling with this for a little bit and it seems incredibly simple. I've tried a simple left join but it didn't give me the results I was hoping for.

I have a dataset of hospitalizations containing patient identification numbers and various other characteristics.

in the dataset, we have index events hospitalizations for a certain procedure + readmission hospitalizations. Thus, a specific patient ID number has 2 rows of data if they were readmitted (with the latter having readmission=1 and readmitdiagnosis. Values for these variables are null in the index hospitalization row.

How can I collapse these rows so each patient just has one row containing all of their data + readmission, readmitdiagnosis from the latter line?

Thank you in advance for the help.

[ed_sas_member]

Hi @blacklucas1 

Welcome to the community!

Could you please post sample data using datalines as well as the desired output?

Thank you,

[ballardw]

---

@blacklucas1 wrote:

Hi all - been struggling with this for a little bit and it seems incredibly simple. I've tried a simple left join but it didn't give me the results I was hoping for.

I have a dataset of hospitalizations containing patient identification numbers and various other characteristics.

in the dataset, we have index events hospitalizations for a certain procedure + readmission hospitalizations. Thus, a specific patient ID number has 2 rows of data if they were readmitted (with the latter having readmission=1 and readmitdiagnosis. Values for these variables are null in the index hospitalization row.

How can I collapse these rows so each patient just has one row containing all of their data + readmission, readmitdiagnosis from the latter line?

Thank you in advance for the help.

---

Before going too far down this path you should consider definition of "readmission" and carefully share that. One concern is that a patient may have multiple readmissions and how to collapse 2 records, as described, will likely have a very poor result for someone with 2, 3, 4 or more readmissions.

Which is run reason to provide example data and what the resulting output should be as @ed_sas_member suggests.

Another bit might be describe what you expect to use that "collapsed" data for. It may well be possible to use your current data without bodging things together.