1 of 1 people found this helpful
Hi Rishab Raina
Could you possible save this as an earlier version (10.3 or 10.5)? I'm curious to see your calculations.
Still, I was able to get to your source data and would consider suggesting your change your data format from
This would allow you to filter on a common date field (EventDT) and count based on EventType. This is often how we structure our data as well. I've done similarly for all events within our Emergency Department from Arrival, Patient to Room, Patient Seen by MD, Patient Disposition Selected, Patient Discharged, etc.
Hope this helps!
Bryce -- I opened it and saved it as 10.3 for you. But it's essentially empty.
PS: Your suggestion to reshape is something Rishab should investigate. Good point on that!
SampleINCDIS_v10.3.twbx 360.8 KB
Me too, Bryce.
Let me add one more note as this is something I spent a lot of time on a couple years ago. The one thing I have an "issue" with re: Census events is that it's a snapshot in time. Similarly taking a snapshot of patients on the hour every hour. The solution I came up with requires doing all of the calculations prior to bringing into tableau.
If only tracking it on the hour is desired, you can most certainly go that route, but I would still consider summarizing your data prior to bringing it in! This would save headaches especially when there isn't a patient move each hour which is inevitable in some locations.
- Create date range table with datetimes at the start and end of every hour - :00:00 and :59:59 (Image 1)
- Identify the times a patient is admitted and discharged and create rows for every hour they were present
- Combine all census datetimes with patient event datetimes (in and out)
- Calculate MAX census for each census date + hour by identifying census at each event datetime
I felt this was most accurate way to view the flow of patients at this granularity as it ensures staffing available to handle highest influx of patients. There is probably little variation from that vs. on the hour, but still is something worth considering, especially for areas such as the ED.
In the end you have a nice dataset of datetime (to the hour) with number of patients (and beds if desired). And you could easily add # of admits and # discharges.
Hope this helps!
Also, check out: http://redheadedstepdata.io/lookup-vs-transactional/