Descriptive comparisons of those in/out of matching cohort

[pbr6cornell]

CALYPSO creates 2 cohorts: the index cohort and the matching cohort. Each of these cohorts independently can be used for subsequent analyses, including through the desriptive cohortanalysis functions available within HERACLES.

One current gap within the current OHDSI stack that may be best addressed within CALYPSO : currently, there is no place to compare the composition of the patients in the matching cohort with those patients who were in the index cohort but not the matching cohort. Being able to determine how the two groups (meet/don't meet inclusion criteria) are different in terms of demographics/conditions/drugs/procedures/measurements could be useful for a few different analytical use cases:

1. evaluating generalizability of study population to overall target population
2. understanding differences in patients who have/don't have specific event
3. stratifying an index population on a binary attribute (ex: gender)

In addition to simply having prevalence of covariates with two groups side-by-size, we could compute absolute standardized differences as a means of allowing ranking on what covariates are most different between the groups.

As we look to integrate CohortAnalysis functions into CALYPSO where the current 'Cohort Reporting' says 'coming soon',, could we also consider how to make cohort comparisons (with one of the cohorts being something that doesn't currently get persisted: index - match).