OHDSI Home | Forums | Wiki | Github

Social Determinants of Health Observations at Community Group Levels (e.g. zip and census tract)

I am trying to map data that has been aggregated to zip, census tract, geo coding, etc into the CDM as observations/observable entities for each patient. Is there a convention around what concepts to use? I don’t see, for example, concepts covering the use of the CDCs social vulnerability index measures as observations for a patient or other census tract level characteristics. Some specific examples include:

% of population below poverty line
% of population unemployed
% of population with no high school diploma
% of population living in multi-unit structures/group quarters
% of population with no vehicle
census tract considered a food desert

If I use the geo coding/zip+4 for the patient location data to create observable events for these patients, how do I go about mapping the concepts and what is the suggestion for how to date these observations (e.g. year of the data collection, date it was updated…)? Thank you! JB


All good questions. Currently, we have no Domain (=fully modeled clinical detail) for any of these socioeconomical characteristics. That is in part because the OMOP model tends to capture concrete person-level data, i.e. not % below poverty line in a geographic zone, but actual income or income range, and not % unemployed in population, but employment status of that person. But we would be very interested in anybody making a proposal of what these characteristics might be, how to model them and what vocabularies to use.

Do you have that?

@johnjbrusk Thout it is not exactly location- related census data analysis, to calculate the loss of life compared to the life expectancy (census data), we stored those data in the package
like this: life expectancy, number of mid-year population.
Then these census data are linked to the CDM in the R package.


I am not a member of the GIS working group, but Perhaps the goals of the GIS working group may be similar to your question: wiki and github

1 Like

Hello @johnbrusk, We are also interested in utilizing the CDM to populate those type of census characteristics. Is there any of adjunct common data model for the SDOH that could be linked to the OMOP tables?