This vignette provides a number of example cohorts built with Capr from the OHDSI Phenotype Phebruary event.
Type 2 diabetes mellitus
Persons with new type 2 diabetes mellitus at first diagnosis
https://atlas-phenotype.ohdsi.org/#/cohortdefinition/88/definition
cs0 <- cs(descendants(443238, 201820, 442793),
descendants(exclude(195771, 201254, 435216, 761051, 4058243, 40484648)),
name = "Type 2 diabetes mellitus (diabetes mellitus excluding T1DM and secondary)")
ch <- cohort(
entry = entry(
conditionOccurrence(cs0),
observationWindow = continuousObservation(priorDays = 365)
),
exit = exit(
endStrategy = observationExit()
)
)
chJson <- toCohortJson(ch)Persons with new type 2 diabetes and no prior T1DM or secondary diabetes
https://atlas-phenotype.ohdsi.org/#/cohortdefinition/89/export
cs0 <- cs(descendants(443238, 201820, 442793),
descendants(exclude(195771, 201254, 435216, 761051, 4058243, 40484648)),
name = "Type 2 diabetes mellitus (diabetes mellitus excluding T1DM and secondary)")
cs1 <- cs(descendants(201254, 435216, 40484648),
name = "Type 1 diabetes mellitus")
cs2 <- cs(descendants(195771),
name = "Secondary diabetes mellitus")
ch <- cohort(
entry = entry(
conditionOccurrence(cs0),
observationWindow = continuousObservation(priorDays = 365)
),
attrition = attrition(
't1d' = withAll(
exactly(0, conditionOccurrence(cs1), duringInterval(eventStarts(-Inf, 0)))
),
'secondaryDiabetes' = withAll(
exactly(0, conditionOccurrence(cs2), duringInterval(eventStarts(-Inf, 0)))
)
),
exit = exit(
endStrategy = observationExit()
)
)
chJson <- toCohortJson(ch)Persons with new type 2 diabetes mellitus at first dx rx or lab
https://atlas-phenotype.ohdsi.org/#/cohortdefinition/90
cs0 <- cs(descendants(443238, 201820, 442793),
descendants(exclude(195771, 201254, 435216, 761051, 4058243, 40484648)),
name = "Type 2 diabetes mellitus (diabetes mellitus excluding T1DM and secondary)")
cs1 <- cs(descendants(201254, 435216, 40484648),
name = "Type 1 diabetes mellitus")
cs2 <- cs(descendants(195771),
name = "Secondary diabetes mellitus")
cs3 <- cs(descendants(4184637, 37059902),
name = "Hemoglobin A1c (HbA1c) measurements")
cs4 <- cs(descendants(21600744),
name = "Drugs for diabetes except insulin")
ch <- cohort(
entry = entry(
conditionOccurrence(cs0),
drugExposure(cs4),
measurement(cs3, valueAsNumber(bt(6.5, 30)), measurementUnit(8554L)), # percent
measurement(cs3, valueAsNumber(bt(48, 99)), measurementUnit(9579L)), # mmol/mol
observationWindow = continuousObservation(priorDays = 365)
),
attrition = attrition(
'no T1D' = withAll(
exactly(0, conditionOccurrence(cs1), duringInterval(eventStarts(-Inf, 0)))
),
'no secondary diabetes' = withAll(
exactly(0, conditionOccurrence(cs2), duringInterval(eventStarts(-Inf, 0)))
)
),
exit = exit(
endStrategy = observationExit()
)
)
chJson <- toCohortJson(ch)Type 1 diabetes mellitus
Persons with new type 1 diabetes
https://atlas-phenotype.ohdsi.org/#/cohortdefinition/92
cs0 <- cs(descendants(195771),
name = "Type 1 diabetes mellitus")
ch <- cohort(
entry = entry(
conditionOccurrence(cs0),
observationWindow = continuousObservation(priorDays = 365)
)
)
chJson <- toCohortJson(ch)Persons with new type 1 diabetes and no prior T2DM or secondary diabetes
https://atlas-phenotype.ohdsi.org/#/cohortdefinition/91
cs0 <- cs(descendants(443238, 201820, 442793),
descendants(exclude(195771, 201254, 435216, 761051, 4058243, 40484648)),
name = "Type 2 diabetes mellitus (diabetes mellitus excluding T1DM and secondary)")
cs1 <- cs(descendants(201254, 435216, 40484648),
name = "Type 1 diabetes mellitus")
cs2 <- cs(descendants(195771),
name = "Secondary diabetes mellitus")
ch <- cohort(
entry = entry(
conditionOccurrence(cs1),
observationWindow = continuousObservation(priorDays = 365)
),
attrition = attrition(
"no prior T2DM" = withAll(exactly(0, conditionOccurrence(cs0), duringInterval(eventStarts(-Inf, 0)))),
"no prior secondary T1DM" = withAll(exactly(0, conditionOccurrence(cs2), duringInterval(eventStarts(-Inf, 0))))
)
)
chJson <- toCohortJson(ch)Atrial Fibrillation
Persons with atrial fibrillation per Wharton et al 2021
https://atlas-phenotype.ohdsi.org/#/cohortdefinition/93
cs0 <- cs(descendants(313217),
name = "Atrial fibrillation")
ch <- cohort(conditionOccurrence(cs0))
chJson <- toCohortJson(ch)Persons with atrial fibrillation per Subramanya et al 2021
https://atlas-phenotype.ohdsi.org/#/cohortdefinition/94
afib <- cs(descendants(313217),
name = "Atrial fibrillation")
ip <- cs(descendants(262, 9201),
name = "Inpatient or inpatient ER visit")
op <- cs(descendants(9202, 9203),
name = "Outpatient or ER visit")
ch <- cohort(
entry = entry(
conditionOccurrence(
conceptSet = afib,
nestedWithAll(
# an inpatient afib
atLeast(
x = 1,
aperture = duringInterval(eventStarts(-Inf, 0), eventEnds(0, Inf)),
query = visit(conceptSet = ip)
)
)
),
conditionOccurrence(
conceptSet = afib,
nestedWithAll(
# 2 afib in outpatient
atLeast(
x = 1,
aperture = duringInterval(eventStarts(-Inf, 0), eventEnds(0, Inf)),
query = visit(
conceptSet = op,
nestedWithAll(
atLeast(
x = 1,
aperture = duringInterval(eventStarts(7, 365)),
query = conditionOccurrence(
conceptSet = afib,
nestedWithAll(
atLeast(
x = 1,
aperture = duringInterval(eventStarts(-Inf, 0), eventEnds(0, Inf)),
query = visit(conceptSet = op)
)
)
)
)
)
)
)
)
)
)
)
chJson <- toCohortJson(ch)Source Concept Example
lf <- cs(descendants(4197819), name = "lung fibrosis")
lfSource <- cs(
725357,35208074,45543276,45548125,45552899,45562465,45567273,45581864,45586679,45586680,45601135,
name = "Lung Fibrosis Source"
)
# Want cohort where entry is lung fibrosis but only using source ICD10CM codes
cd <- cohort(
entry = entry(
conditionOccurrence(
conceptSet = lf,
conditionSourceConcept(lfSource) # add source concept as attribute
),
observationWindow = continuousObservation(365L, 0L),
primaryCriteriaLimit = "First"
)
)
cdJson <- toCohortJson(cd)Primary Criteria Limit: First vs All
Regardless of entry criteria complexity:
-
Era cohort ->
primaryCriteriaLimit = "First",expressionLimit = "First", no era collapse -
Occ cohort ->
primaryCriteriaLimit = "All",expressionLimit = "All",era = era(eraDays = 99999L)
The two patterns below show different ways to get one row per person:
- Era version: take the first qualifying event and follow until end of continuous observation.
- Occ version: take all qualifying events and stitch them into one long era from first to last event.
Example 1: Nested criteria with prior diagnosis requirement
1a. Era version
# Placeholder concept IDs for disease of interest
disease_concept_set <- c(201826L)
# Index event: first diagnosis with a prior diagnosis at least 365 days earlier
diagnosisEraCohort <- cohort(
entry = entry(
conditionOccurrence(
conceptSet = cs(disease_concept_set, name = "Disease"),
nestedWithAll(
atLeast(
1,
conditionOccurrence(conceptSet = cs(disease_concept_set, name = "Disease")),
aperture = duringInterval(eventStarts(-Inf, -365))
)
)
),
primaryCriteriaLimit = "First"
),
attrition = attrition(
expressionLimit = "First"
),
exit = exit(
endStrategy = observationExit()
)
)
diagnosisEraJson <- toCohortJson(diagnosisEraCohort)1b. Occ version (same cohort logic, different type)
# Entry at each diagnosis with a prior diagnosis at least 365 days earlier
diagnosisOccCohort <- cohort(
entry = entry(
conditionOccurrence(
conceptSet = cs(disease_concept_set, name = "Disease"),
nestedWithAll(
atLeast(
1,
conditionOccurrence(conceptSet = cs(disease_concept_set, name = "Disease")),
aperture = duringInterval(eventStarts(-Inf, -365))
)
)
),
primaryCriteriaLimit = "All"
),
attrition = attrition(
expressionLimit = "All"
),
exit = exit(
endStrategy = fixedExit(offsetDays = 0L)
),
era = era(eraDays = 99999L)
)
diagnosisOccJson <- toCohortJson(diagnosisOccCohort)