Subscribe to DSC Newsletter

I am presently working with a task force on ways to reduce readmission rates at my employer, specifically readmissions that happen within 30 days of discharge from the initial admission. I want to see if I can find out any similarities between patients that are readmitted versus patients that are not readmitted. So my thought is constructing some sort of data matrix that would include things such as:

Patient Demographics: [date of birth, age, zip code, insurance, live at home, nursing home, family]

ED demographic notes: [time and date of ED arrival, arrival method, chief complaint, turn around time (total time in emergency department)]

Triage notes: [acuity level, mortality risk level, height, weight, bp, bmi, pulse, respirations]

Lab work: [basic panel of tests, ie glucose, cbc, etc]

Hospital Admission notes: [length of stay, unit (telemetry, cardiac care, etc), daily vitals (height, weight, bp, bmi, pulse, respirations), lab results, medications, tests ie (radiology, ct, mri), admitting diagnosis, discharge diagnosis]

I think this in of itself would be enough to get started. I could do this for all patients over a 6 month period getting all patients that are admitted to the hospital and I can run a report that will tell me who was readmitted within 30 days of their discharge, like I said the goal is to see why they were readmitted, do they have something in common with each other like a specific diagnosis, do they have similar ages, zip codes etc, what makes them different from those who are not readmitted?

Any idea on where to start the comparisons and how I should go about it?

Views: 494

Reply to This

Replies to This Discussion

Hi Steven - You might want to summarize some data for the floor / ward and merge it onto each readmission.  For each nurse / PA / doctor / orderly who works in the hospital, you might want to summarize data based on the rates of readmission, errors, secondary infection, death, etc. for every patient they touch or ward they work in over an x month period (3 months, 6 months).
You're going to want to determine whether readmission is related to the original reason for hospitalization or not which might be gleaned from the intake and treatment records.  Also determine whether the individuals went home with drains or other apparatus which required individuals to do specialized wound care once home (complexity / length of discharge instructions might be a good proxy).

Thank you Lynne, this is going to be a very long and hopefully eye opening process.

Lynne Mysliwiec said:

Hi Steven - You might want to summarize some data for the floor / ward and merge it onto each readmission.  For each nurse / PA / doctor / orderly who works in the hospital, you might want to summarize data based on the rates of readmission, errors, secondary infection, death, etc. for every patient they touch or ward they work in over an x month period (3 months, 6 months).
You're going to want to determine whether readmission is related to the original reason for hospitalization or not which might be gleaned from the intake and treatment records.  Also determine whether the individuals went home with drains or other apparatus which required individuals to do specialized wound care once home (complexity / length of discharge instructions might be a good proxy).

Hello, Steven---Dr. Scott Zasadil at University of Pittsburgh Medical Center accomplished such a feat and presented a webinar on the subject for us (StatSoft) in June 2012. If I remember correctly, he was able to text mine nurses' notes, in addition to compiling many of the stats like you describe, in order to reduce readmissions. You can view his webinar here http://www.statsoft.com/webcasts/Zasadil_UPMC/lib/playback.html.

Thank you Paul. I am currently working with trying to improve the predictability of a LACE Score by modifying it to based on known readmit rates by MS-DRG, becoming a little difficult for now but will keep at it. So far the LACE as setup has fair predictability which is nice, so I just have to find a way to push the ROC curve up and to the left.

Paul Hiller said:

Hello, Steven---Dr. Scott Zasadil at University of Pittsburgh Medical Center accomplished such a feat and presented a webinar on the subject for us (StatSoft) in June 2012. If I remember correctly, he was able to text mine nurses' notes, in addition to compiling many of the stats like you describe, in order to reduce readmissions. You can view his webinar here http://www.statsoft.com/webcasts/Zasadil_UPMC/lib/playback.html.

RSS

Videos

  • Add Videos
  • View All

Follow Us

© 2019   Data Science Central ®   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service