Tips for using calendar date range requirement in defining entry (or attrition)? about capr

[pa-nathaniel]

We tried the method below and looks like we got this to work in Capr 2.0.7.

ch <- cohort(
  entry = entry(
    conditionOccurrence(cs0, 
                        startDate(methods::new("opAttributeDate", op = "gte", value = as.Date("2022-01-01"))),
                        endDate(methods::new("opAttributeDate", op = "lte", value = as.Date("2023-12-31"))))
  ),
  exit = exit(
    endStrategy = observationExit()
  )
)

Note this didn't work in version 2.0.5 but seems to have been addressed as of 2.0.6 according to (https://github.com/OHDSI/Capr/blob/main/NEWS.md#capr-206). Thanks for those updates Capr team!

While the results look right I wonder if there is a simpler way to do this than calling methods::new(). Curious what @mdlavallee92 thinks.

[mdlavallee92]

you shouldnt need to make a new method that is hidden behind the op functions (gt, gte, lt, lte, eq, neq, bt, nbt).

The attribute class op is built by these functions and nested within the attribute domain. For example

age <- age(gte(18))
startDate <- startDate(bt(lubridate::as_date("2022-01-01"), lubridate::as_date("2023-12-31")))

So for your cohort it would look something like this.

ch <- cohort(
  entry = entry(
    conditionOccurrence(
      cs0,
      startDate(gte(lubridate::as_date("2022-01-01"))),
      endDate(lte(lubridate::as_date("2023-12-31")))),
    observationWindow = continuousObservation(priorDays = 365)
  ),
  exit = exit(
    endStrategy = observationExit()
  )
)

You can also set a censor window in the eras:

ch <- cohort(
  entry = entry(
    conditionOccurrence(
      cs0,
    observationWindow = continuousObservation(priorDays = 365)
  ),
  exit = exit(
    endStrategy = observationExit()
  ),
  era = era(
    studyStartDate = lubridate::as_date("2022-01-01"),
    studyEndDate = lubridate::as_date("2023-12-31")
  )
)

[pa-nathaniel]

Ah makes sense, thank you!!!!