What variables in OHDSI might be used to link patient-level clinical outcomes (e.g. deaths, comorbidities, COVID rates) to nursing home quality metrics (such as number of citations)?
We are using a public nursing home dataset (Nursing Home Inspect) and a dataset we collected on assisted living facilities (like group homes) to do geospatial analysis of health disparity.
Datasets we are currently aware of in OHDSI that have nursing home information:
-
The IBM Truven MarketScan Commercial Claims & Encounters (CCAE) & Truven MarketScan Medicare (MDCR), and Truven MarketScan Multi-State Medicaid (MDCD) in the common data model described here: IBM MDCD - Janssen CDM Documentation
-
@Richard_Starr mentioned he will be converting Medicaid MAX and TAF files to the Observational Medical Outcomes Partnership common data model, but data on long term care files may not be available.
Potential variables that could help assess health disparity:
-
Geographical information is not available. @clairblacketer confirmed that while CCAE and MDCR do include state, this information is not available in the IBM CCAE/MDCR/MDCD datasets in OHDSI.
-
There is race and ethnicity information in the IBM MDCD OHDSI data. The percentage of Black population across the United States is available from the census for every ZIP code (and the address of nursing homes is available). This means the percentage of Black patients in the IBM MDCD dataset could be used to link health outcomes to nursing home metrics.
For example, we have started looking at “percentage of population that is Black” to see if there are patterns for nursing home quality metrics:
In case you might know who to ask about what other variables or datasets in OHDSI might be useful for linking to nursing home data and assisted living facility/group home data, we are all ears!
Thank you so much,
Jaan