Overstay Predictor Project

Overstay Predictor Project for Medicine - collection team meeting discussion

• Primary Investigator: Dr. Dan Roberts

Instructions

For every medicine patient at the Grace hospital, as soon as possible, but definitely within 48 business hrs of admission:

• find the new "Overstay" form on the chart
• for the first 6 check boxes on the form, create a record for each in tmp entries and check the checkbox if the answer is Yes. (one to one from the form, no interpretation necessary)
• (I will add a picture of the form here once we have a final version)
• Also, enter at least all the following as part of minimal dataset
  • DOB
  • Admit Date
  • ADLs
  • whether there was any CVA or Dementia as a comorbid
  • whether there was a stroke as an admit diagnosis

• Click the new “Overstay” button on the patient viewer form next to the notes field and it will turn a colour:
  • GREEN, YELLOW or RED - based on an underlying algorithm too complicated to get into here, but based on tmp entries, age, comorbids, admit dxs and ADLs
  • GRAY- if some needed data is missing
  • If the chart already has a sticker the same colour then you are done
  • if the colour in the program is different from the one on the chart put a sticker of the program colour over the sticker on the chart. Stickers will be supplied.

If you can't find data

• I have emailed Linda to please fill in who the collector should talk to if there are problems with the Overstay form or the other data required for the project in a timely manner.Ttenbergen 14:45, 2012 April 23 (CDT)

Questions

1. What does RED tag indicate? This is not really clear to me in algorithm.Trish Ostryzniuk
   • we are still working on the details, but red and yellow will lead to different priorities for staff responsible for overstay prevention; both require action. Ttenbergen 12:56, 2012 April 27 (CDT)
2. Where on the chart will this new form be and will it be a colored form?Trish Ostryzniuk 10:43, 2012 April 24 (CDT)
   • not known yet, will add details when we get them. Ttenbergen 12:56, 2012 April 27 (CDT)
3. The time we use as the time of admission will be different than the time of admission to the unit, and often significant time has passed between the 2 times so the information collected is different e.g. pt is confused at our admission time but by the time the patient goes to the unit, the patient is now A & O etc or vice versa. Will this affect the algorithm?
   • Use the data as you would have entered before as that is what the algorithm was based on. I hope that helps, since I know some collectors used to collect this much later. Pls let me know if you need more info. Ttenbergen 12:47, 2012 April 27 (CDT)
4. s\Should lacunes or lacunar infarcts that are/have been asymptomatic but are reported on a CT be included as CVAs in the comorbids?
   • No. If no physical/funtional impairment what so ever then no, don't code.--Trish Ostryzniuk 15:52, 2012 April 27 (CDT)
     • They should be reported like they always have been. We are basing the algorithm on previous data. This does not preclude us from discussing that question independently on the appropriate wiki page. Trish, I have no idea what a lacunar infarct is, could you provide the link? Ttenbergen 12:47, 2012 April 27 (CDT)
5. Should ICHs be considered CVA in comorbids?
   • Yes if it caused as stroke. If pre existing or happened in past then code 505 in comorbid.Trish Ostryzniuk 15:52, 2012 April 27 (CDT)
6. Will 502 ICH be considered a 505 CVA in the admit or acquired Dx?
   • Good question. Not sure how the algorithm is set up & if it does captures a 502 ICH & 504 BSH in admit or acquired DX as 505 stroke, it should. Certainly if a 502 ICH - intracerebral hemorrhage, or 504 BSH - brainstem hemorrhage has cause a stroke that has affected patient functional ability, they it should be coded as what it is, either 502 or 504 in admit or acquired.Trish Ostryzniuk 15:52, 2012 April 27 (CDT)
     • We are not including those other dxs/comos in generating the colour, only CVA and all its subcodes. Whether we should is a question of whether Julie provided them to Rodrigo for for building his model. If they were included in the model they should be included in the algorithm for sure. If they were not included we might invalidate the model by including them now. Trish, could you pull in Julie on this? Ttenbergen 16:03, 2012 April 27 (CDT)
   • They should be reported like they always have been. We are basing the algorithm on previous data. This does not preclude us from discussing that question independently on the appropriate wiki page. Trish, I have no idea what a ICH is, could you provide the link? Ttenbergen 12:47, 2012 April 27 (CDT)
7. Will it be necessary to document a difference in sticker colours if we need to change the colour once the algorithm has run?
   • no need. The original colour is based on the items you enter into tmp, so we can go back from that if needed. Ttenbergen 12:47, 2012 April 27 (CDT)
8. How can I test the yellow colour?-I can only get the overstay button to be red or green using multiple different combinations of the check boxes
   • yellow would be based on a combination of comorbids and ADLs and age. Make a patient a little less sick and they will become yellow. Sorry, can't give you exact instructions, the algorithm is pretty messy. Ttenbergen 12:47, 2012 April 27 (CDT)

Purpose

A small number of patients overstays for a significant amount of time after they are medically ready for discharge. Though the number of patients is small, a significant proportion of bed-days is wasted in the process. We are piloting a process at the Grace where an "overstay coordinator" will be assigned to patients at risk of overstaying. Whether a patient is at risk is determined by the ward nurse assessing risk factors on a form, and by an assessment algorithm that takes into account the values mentioned as mandatory above; the combination of these results in a chart being assigned a colour sticker, and the transition coordinators monitor these chart stickers.