Added ability to compute aggregated statistics for multiple cohorts at once: The
cohortId argument of the
getDbCovariateData() function now accepts a vector of IDs. The aggregated statistics now contains a new field called
cohortDefinitionId that can be used to distinguish between the various cohorts.
filterByCohortDefinitionId() function to select covariates of one cohort from a
covariateData object containing multiple.
cohortId argument now also supports integers greater than 32-bits.
measurement_concept_id, the selection of which measurement value to include is now deterministic.
Adding timeRef table to CovariateData object for temporal covariates.
Throw meaningful error when
createCovariateSettings() is called without specifying any covariates.
getDbCovariateData() returns empty covariates instead of NULL covariates when no covariates are specified.
Switching from ff to Andromeda for storage of large data objects.
Adding option to createTable1 to (also) show absolute counts.
For analyses that restrict to inpatient diagnoses, the string ‘(inpatient)’ is now appended to the covariate name.
Dropping spurious ‘analysis_name’ field in temporary covariate reference table.
Fixed covariate name of age groups over 100 years old.
Fixing aggregation of covariates generated using the cohort_attribute table.
Fixed error when calling tidyCovariates without removing redundancy, when there are no infrequent covariates.
Fixed server-side computation of median and interquartile range for measurement values.
Restricting by concept ID now works for measurement values.
Removing redundant covariates now precedes removing infrequent covariates when calling tidyCovariates. Analyses where the most prevalent (redundant) covariate was removed are now exempt from removal of infrequent covariates.
Fixed some typos in the vignette