We saw in the previous vignette how we can summarise the overlap between cohorts. In addition to this, we might also be interested in timings between cohorts. That is, the time between an individual entering one cohort and another. For this we can use the summariseCohortTiming(). In this example we’ll look at the time between entering cohorts for acetaminophen, morphine, and oxycodone using the Eunomia data.

library(CDMConnector)
library(CodelistGenerator)
library(CohortCharacteristics)
library(dplyr)
library(ggplot2)

con <- DBI::dbConnect(duckdb::duckdb(),
  dbdir = CDMConnector::eunomia_dir()
)
cdm <- CDMConnector::cdm_from_con(con,
  cdm_schem = "main",
  write_schema = "main",
  cdm_name = "Eunomia"
)

meds_cs <- getDrugIngredientCodes(
  cdm = cdm,
  name = c(
    "acetaminophen",
    "morphine",
    "warfarin"
  )
)

cdm <- generateConceptCohortSet(
  cdm = cdm,
  name = "meds",
  conceptSet = meds_cs,
  end = "event_end_date",
  limit = "all",
  overwrite = TRUE
)

settings(cdm$meds)
#> # A tibble: 3 × 6
#>   cohort_definition_id cohort_name    limit prior_observation future_observation
#>                  <int> <chr>          <chr>             <dbl>              <dbl>
#> 1                    1 11289_warfarin all                   0                  0
#> 2                    2 161_acetamino… all                   0                  0
#> 3                    3 7052_morphine  all                   0                  0
#> # ℹ 1 more variable: end <chr>
cohortCount(cdm$meds)
#> # A tibble: 3 × 3
#>   cohort_definition_id number_records number_subjects
#>                  <int>          <int>           <int>
#> 1                    1            137             137
#> 2                    2          13908            2679
#> 3                    3             35              35

Now we have our cohorts we can summarise the timing between cohort entry. Note setting restrictToFirstEntry to TRUE will mean that we only consider timing between an individual’s first record in each cohort (i.e. their first exposure to each of the medications).

meds_timing <- cdm$meds |>
  summariseCohortTiming(restrictToFirstEntry = TRUE)
meds_timing |>
  glimpse()
#> Rows: 42
#> Columns: 13
#> $ result_id        <int> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,…
#> $ cdm_name         <chr> "Eunomia", "Eunomia", "Eunomia", "Eunomia", "Eunomia"…
#> $ group_name       <chr> "cohort_name_reference &&& cohort_name_comparator", "…
#> $ group_level      <chr> "161_acetaminophen &&& 11289_warfarin", "11289_warfar…
#> $ strata_name      <chr> "overall", "overall", "overall", "overall", "overall"…
#> $ strata_level     <chr> "overall", "overall", "overall", "overall", "overall"…
#> $ variable_name    <chr> "number records", "number records", "number records",…
#> $ variable_level   <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, N…
#> $ estimate_name    <chr> "count", "count", "count", "count", "count", "count",…
#> $ estimate_type    <chr> "integer", "integer", "integer", "integer", "integer"…
#> $ estimate_value   <chr> "136", "136", "35", "35", "6", "6", "136", "136", "35…
#> $ additional_name  <chr> "overall", "overall", "overall", "overall", "overall"…
#> $ additional_level <chr> "overall", "overall", "overall", "overall", "overall"…

As with cohort overlap, we have table and plotting functions to help view our results.

tableCohortTiming(meds_timing,
  timeScale = "years",
  .options = list(decimals = c(numeric = 0))
)
CDM name Cohort name reference Cohort name comparator Variable name Estimate name Estimate value
Eunomia 11289 warfarin 161 acetaminophen Number records N 136
Number subjects N 136
7052 morphine Number records N 6
Number subjects N 6
161 acetaminophen 7052 morphine Number records N 35
Number subjects N 35
11289 warfarin 161 acetaminophen Years between cohort entries Median [Q25 - Q75] -54 [-67 - -46]
Range -92 - 3
7052 morphine Years between cohort entries Median [Q25 - Q75] -5 [-10 - 5]
Range -19 - 9
161 acetaminophen 7052 morphine Years between cohort entries Median [Q25 - Q75] 16 [5 - 34]
Range -34 - 77
plotCohortTiming(meds_timing,
  plotType = "boxplot",
  timeScale = "years"
)

If we want to see an even more granular summary of cohort timings we can make a density plot instead of a box plot. Note, for this we’ll need to set density as TRUE when getting our initial results.

meds_timing <- cdm$meds |>
  summariseCohortTiming(
    restrictToFirstEntry = TRUE,
    density = TRUE
  )
plotCohortTiming(meds_timing,
  plotType = "density",
  timeScale = "years"
)

As well as generating these estimates for cohorts overall, we can also obtain stratified estimates.

cdm$meds <- cdm$meds |>
  PatientProfiles::addAge(ageGroup = list(c(0, 49), c(50, 150))) |>
  compute(temporary = FALSE, name = "meds") |>
  newCohortTable()
meds_timing <- cdm$meds |>
  summariseCohortTiming(
    restrictToFirstEntry = TRUE,
    strata = list("age_group"),
    density = TRUE
  )
tableCohortTiming(meds_timing,
  timeScale = "years",
  .options = list(decimals = c(numeric = 0))
)
CDM name Cohort name reference Cohort name comparator Variable name Estimate name Age group
0 to 49 50 to 150 Overall
Eunomia 11289 warfarin 161 acetaminophen Number records N 8 1 9
Number subjects N 8 1 9
7052 morphine Number records N - 6 6
Number subjects N - 6 6
161 acetaminophen 7052 morphine Number records N 25 1 26
Number subjects N 25 1 26
11289 warfarin 161 acetaminophen Years between cohort entries Median [Q25 - Q75] -45 [-46 - -33] 2 [2 - 2] -44 [-46 - -32]
Range -49 - -29 2 - 2 -49 - 2
7052 morphine Years between cohort entries Median [Q25 - Q75] - -5 [-10 - 5] -5 [-10 - 5]
Range - -19 - 9 -19 - 9
161 acetaminophen 7052 morphine Years between cohort entries Median [Q25 - Q75] 9 [0 - 30] 24 [24 - 24] 10 [1 - 29]
Range -34 - 37 24 - 24 -34 - 37
plotCohortTiming(meds_timing,
  plotType = "boxplot",
  timeScale = "years",
  facet = "strata_level",
  colour = "strata_level",
  colourName = "Age group"
)