ATLAS

The most popular way is to create cohorts is to use ATLAS.

ATLAS is an open source software tool for researchers to conduct scientific analyses on standardized observational data converted to the OMOP Common Data Model V5.

GitHub.com/OHDSI/Atlas

A cohort definition for Viral Sinusitis may look like this:

Cohort information

Viral Sinusitis Cohort parameters

The following text is an export of ATLAS Version 2.12.2. The bold parts of the text are parameters set in the cohort definition.

Cohort Entry Events

People with continuous observation of 365 days before and 1,095 days after event may enter the cohort when observing any of the following:

drug eras of ‘Viral Sinusistis drugs’. Limit cohort entry events to the earliest event per person.

Inclusion Criteria

1. Viral Sinusitis diagnosis Entry events having at least 1 condition occurrence of ‘[MVK] Viral Sinusitis diagnosis’, allow events outside >observation period.

2. Index year Entry events with the following event criteria: starting after December 31, 1949.

Cohort Exit

The person exits the cohort at the end of continuous observation.

Cohort Eras

Entry events will be combined into cohort eras if they are within 0 days of each other.

ATLAS allows you to export cohort definitions to SQL, JSON, or plain text (like above).

Cohorts specified in SQL may be run on your database directly with DatabaseConnector and SqlRender to generate the cohort table.

Cohorts specified as JSON may be generated using CohortGenerator or CDMConnector to generate the cohort table.

CapR

CapR is a package that allows one to programatically generate cohorts. This is especially useful if lots of cohorts need to be created.

The goal of Capr, pronounced ‘kay-pr’ like the edible flower, is to provide a language for expressing OHDSI Cohort definitions in R code. OHDSI defines a cohort as “a set of persons who satisfy one or more inclusion criteria for a duration of time” and provides a standardized approach for defining them (Circe-be). Capr exposes the standardized approach to cohort building through a programmatic interface in R which is particularly helpful when creating a large number of similar cohorts. Capr version 2 introduces a new user interface designed for readability with the goal that Capr code being a human readable description of a cohort while also being executable on an OMOP Common Data Model.

Specifying a cohort for Viral Sinusitis with CapR may look like this:

library(Capr)

viralSinusitis <- cs(
  descendants(40481087),
  name = "viralSinusitis",
  id = "Viral Sinusitis"
)

viralSinusitisCohort <- cohort(
  entry = entry(
    conditionOccurrence(viralSinusitis),
    primaryCriteriaLimit = "First"
  ),
  exit = exit(
    endStrategy = observationExit()
  )
)

cohortSet <- list(ViralSinusitis = viralSinusitisCohort)

The cohortSet may then be generated using CohortGenerator or CDMConnector.

Cohorts for TreatmentPatterns

TreatmentPatterns includes the following dummy cohorts that were defined in ATLAS, and are usable with the Eunomia package: Viral Sinusitis, Acetaminophen, Amoxicillin, Aspirin, Clavunate, Doxylamine, Penicilin V, and Death.

These cohorts are of a specific type, meaning they serve a different purpose within TreatmentPatterns. The types are: target (required), event (at least one), exit (optional). Target cohorts is a cohort of subjects that you are interested in. Event cohorts are different treatments that may be to subjects within your target cohort. Exit cohorts are similar to event cohorts, but allow you to specify an exiting event, like death.

In the table below you can see which cohort is of what type.

Given these cohorts and their types we are looking within a target cohort of subjects diagnosed with Viral Sinusitis. We are interested in generating treatment pathways with the following event cohorts: Acetaminophen, Amoxicillin, Aspirin, Clavunate, Doxylamine, Penicilin V. If subjects exit the database by dying, within the bounds of the target cohort, tag on Death.

After installing TreatmentPatterns these cohorts are accessible through the following code.

list.files(
  system.file(package = "TreatmentPatterns", "exampleCohorts"),
  full.names = TRUE
)

## [1] "/tmp/RtmpZ1fpf9/Rinst1a062441cbc/TreatmentPatterns/exampleCohorts/Acetaminophen.json" 
## [2] "/tmp/RtmpZ1fpf9/Rinst1a062441cbc/TreatmentPatterns/exampleCohorts/Amoxicillin.json"   
## [3] "/tmp/RtmpZ1fpf9/Rinst1a062441cbc/TreatmentPatterns/exampleCohorts/Aspirin.json"       
## [4] "/tmp/RtmpZ1fpf9/Rinst1a062441cbc/TreatmentPatterns/exampleCohorts/Clavulanate.json"   
## [5] "/tmp/RtmpZ1fpf9/Rinst1a062441cbc/TreatmentPatterns/exampleCohorts/Death.json"         
## [6] "/tmp/RtmpZ1fpf9/Rinst1a062441cbc/TreatmentPatterns/exampleCohorts/Doxylamine.json"    
## [7] "/tmp/RtmpZ1fpf9/Rinst1a062441cbc/TreatmentPatterns/exampleCohorts/PenicillinV.json"   
## [8] "/tmp/RtmpZ1fpf9/Rinst1a062441cbc/TreatmentPatterns/exampleCohorts/ViralSinusitis.json"