I've built readmissions dashboards, unfortunately due to confidentiality and PHI I can't share them, so I'll just share what I know:
Computing readmissions (and measures like return to emergency with 24 hours) in Tableau is generally not feasible nor advisable. While there are some simple cases that are computable in Tableau, the only way to really identify readmissions in Tableau with any accuracy is with table calculations which would require having every encounter ID as a dimension in the view (which would slow performance) and based on how table calculations work the kind of nested calculations that we'd need to use to then compute readmission rates are much more difficult, plus tasks like filtering and generating dashboards are more challenging. So it's generally best to have information about whether a given encounter is a readmission or not computed prior to Tableau and be available as a record level value.
Sometimes people will ask about using Tableau's Level of Detail (LOD) expressions to track readmissions since they are aggregate functions that can return record level values. LOD expressions don't have some features that are really useful for readmissions analysis and therefore are only usable in very limited cases.
My recommendation is to pre-process the data via a stored procedure or query (or a data prep tool such as Alteryx or Trifacta) that can generate or update a table identifying index and readmit encounters with associated info (such as days to readmit) based on your desired logic***, then you can join that data with your encounter data to get the desired results and have a readmit rate calculation like SUM(IF [Has Readmit Flag] THEN 1 ELSE 0 END)/SUM([Number of Records]), this example assumes that each record represents 1 encounter. If you're using SQL you can do a web search for something like 'SQL readmissions' and find code examples for a variety of databases that you can adjust to your needs.
Another factor here is that in the USA there are at least 10 readmissions measures that CMS tracks, each with their own inclusion & exclusion criteria, and then there ways that healthcare systems can want to track readmission beyond what CMS requires & reports on. You'll need to decide on what metrics and/or algorithms to use.
Here are some examples of ways I've helped track readmissions:
- rates on monthly, quarterly, annual, rolling 12 month intervals
- comparison to system targets as well as different payor targets
- calculating "number to treat" i.e. how many readmissions needed to be prevented to meet targets
- identifying various cohorts for targeting, e.g. same day readmissions, readmissions w/in 3 or 7 days, readmissions for same/similar diagnosis or DRG
- day of week/time of day analysis for discharges and readmits
- readmission rates broken down by hospital unit, original discharge disposition, payor, service line, DRG, discharging provider, etc. and combinations thereof.
- high utilizer readmissions
- analysis of process measures like the 8P Risk Assessment