Filtering cohorts

library(CohortConstructor)
library(CohortCharacteristics)
library(ggplot2)

For this example we’ll use the Eunomia synthetic data from the CDMConnector package.

con <- DBI::dbConnect(duckdb::duckdb(), dbdir = eunomia_dir())
cdm <- cdm_from_con(con, cdm_schema = "main", 
                    write_schema = c(prefix = "my_study_", schema = "main"))

Let’s start by creating two drug cohorts, one for users of diclofenac and another for users of acetaminophen.

cdm$medications <- conceptCohort(cdm = cdm, 
                                 conceptSet = list("diclofenac" = 1124300,
                                                   "acetaminophen" = 1127433), 
                                 name = "medications")
cohortCount(cdm$medications)
#> # A tibble: 2 × 3
#>   cohort_definition_id number_records number_subjects
#>                  <int>          <int>           <int>
#> 1                    1           9365            2580
#> 2                    2            830             830

We can take a sample from a cohort table using the function sampleCohort(). This allows us to specify the number of individuals in each cohort.

cdm$medications |> sampleCohorts(cohortId = NULL, n = 100)
#> # Source:   table<main.my_study_medications> [?? x 4]
#> # Database: DuckDB v1.0.0 [eburn@Windows 10 x64:R 4.4.0/C:\Users\eburn\AppData\Local\Temp\Rtmp8iVoDT\file5c2859907d63.duckdb]
#>    cohort_definition_id subject_id cohort_start_date cohort_end_date
#>                   <int>      <int> <date>            <date>         
#>  1                    1       3968 1954-01-03        1954-02-02     
#>  2                    1       3968 2019-01-27        2019-02-02     
#>  3                    1       2647 1988-01-31        1988-02-07     
#>  4                    1       5334 1970-10-25        1970-11-08     
#>  5                    2       1600 1978-01-12        1978-01-12     
#>  6                    1       3641 1979-04-01        1979-04-15     
#>  7                    1       1024 1992-02-10        1992-02-24     
#>  8                    1       3022 1991-07-18        1991-07-25     
#>  9                    1        199 2010-02-04        2010-02-18     
#> 10                    1       2580 2012-06-14        2012-06-28     
#> # ℹ more rows

cohortCount(cdm$medications)
#> # A tibble: 2 × 3
#>   cohort_definition_id number_records number_subjects
#>                  <int>          <int>           <int>
#> 1                    1            391             100
#> 2                    2            100             100

When cohortId = NULL all cohorts in the table are used. Note that this function does not reduced the number of records in each cohort, only the number of individuals.

It is also possible to only sample one cohort within cohort table, however the remaining cohorts will still remain.

cdm$medications <- cdm$medications |> sampleCohorts(cohortId = 2, n = 100)

cohortCount(cdm$medications)
#> # A tibble: 2 × 3
#>   cohort_definition_id number_records number_subjects
#>                  <int>          <int>           <int>
#> 1                    1           9365            2580
#> 2                    2            100             100

The chosen cohort (users of diclofenac) has been reduced to 100 individuals, as specified in the function, however all individuals from cohort 1 (users of acetaminophen) and their records remain.

If you want to filter the cohort table to only include individuals and records from a specified cohort, you can use the function subsetCohorts.

cdm$medications <- cdm$medications |> subsetCohorts(cohortId = 2)
cohortCount(cdm$medications)
#> # A tibble: 1 × 3
#>   cohort_definition_id number_records number_subjects
#>                  <int>          <int>           <int>
#> 1                    2            830             830

The cohort table has been filtered so it now only includes individuals and records from cohort 2. If you want to take a sample of the filtered cohort table then you can use the sampleCohorts function.

cdm$medications <- cdm$medications |> sampleCohorts(cohortId = 2, n = 100)

cohortCount(cdm$medications)
#> # A tibble: 1 × 3
#>   cohort_definition_id number_records number_subjects
#>                  <int>          <int>           <int>
#> 1                    2            100             100