DrugUtilisation

Overview

The DrugUtilisation package is a standardized toolset designed to conduct such studies using data mapped to the OMOP Common Data Model.

The package provides the tools to answer critical questions about how medicines are used in the real world:

Who is being prescribed a certain drug?
For what indications is it being used?
How long do patients stay on treatment?
What are the patterns of switching to other therapies?

This package offers a structured, reproducible workflow. It begins by creating analysis-ready “drug user cohorts” and then applies a series of standardized functions to characterize them, similar to a series of SAS macros designed for a specific reporting pipeline.

The package is designed around a logical, sequential workflow:

Generate Drug Cohorts: Create cohorts of patients based on their exposure to specific drugs, using ingredients, ATC codes, or custom concept sets.
Refine Cohorts: Apply inclusion and exclusion criteria to define a precise study population (e.g., new users only, specific observation windows).
Summarise & Analyse: Execute a range of standardized analyses to describe drug use patterns, indications, treatment persistence, and more.

Core Concepts

The `gapEra` Parameter

A fundamental concept in drug utilisation studies is consolidating multiple drug exposures into a single, continuous “episode” or “era” of treatment. The gapEra parameter controls this by defining the maximum number of days allowed between the end of one exposure record and the start of the next for them to be considered part of the same treatment episode.

Consider an individual with four prescription records for the same drug:

Record 1: Day 1 -> Day 20
Record 2: Day 15 -> Day 30
Record 3: Day 31 -> Day 38
Record 4: Day 45 -> Day 55

If gapEra = 0, records must overlap to be joined. This would result in 3 distinct episodes.
If gapEra is between 1 and 6, the small gap between records 3 and 4 is bridged. This results in 2 episodes.
If gapEra >= 7, the larger gap between the first two joined records and the last two is also bridged, resulting in 1 single, continuous episode.

Choosing an appropriate gapEra is a critical study design decision that should be clinically informed.

The Analysis Workflow

The package follows a clear, sequential process. You start with a broad cohort of drug users and progressively refine it before performing the final analysis.

graph LR
    subgraph "1. Cohort Generation"
        A[Drug Concepts<br/>Ingredients, ATC] --> B(generateCohortSet);
        C[gapEra] --> B;
    end

    subgraph "2. Cohort Refinement"
        D[requirePriorDrugWashout<br/>requireIsFirstDrugEntry<br/>requireObservationBeforeDrug<br/>requireDrugInDateRange];
    end

    subgraph "3. Analysis"
        G[summariseDrugUtilisation]
        H[summariseIndication]
        I[summariseProportionOfPatientsCovered]
        J[summariseDrugRestart]
    end

    B --> D;
    D --> G;
    D --> H;
    D --> I;
    D --> J;

Crucially, the order of the refinement steps matters. Applying a date range filter before identifying the first drug entry will yield a different result than the other way around. The recommended order is:

Washout / First Use: requirePriorDrugWashout() or requireIsFirstDrugEntry().
Prior Observation: requireObservationBeforeDrug().
Date Range: requireDrugInDateRange().

Washout Period (requirePriorDrugWashout): This is how we identify “new users” of a drug. A washout period is a specified duration (e.g., 365 days) before the start of a drug exposure during which the patient must not have had any prior exposure to the same drug. This helps ensure we are studying incident (new) treatment episodes, not prevalent (ongoing) ones. It’s crucial for avoiding biases in studies looking at the effects of starting a new therapy.
Prior Observation (requireObservationBeforeDrug): This ensures that we have enough medical history for a patient before they start the drug. A requirement of 365 days of prior observation means the patient has been enrolled in the database for at least a year before their first prescription. This is vital for assessing baseline characteristics and comorbidities and ensuring data completeness.
Date Range (requireDrugInDateRange): This restricts the study to a specific calendar period (e.g., January 1, 2015, to December 31, 2020). This is important for defining a study period that might correspond to a specific version of a clinical guideline, the availability of a drug on the market, or a particular public health event.

Getting Started: A Complete Example

This example demonstrates a typical workflow, from creating a cohort to analysing its characteristics.

1. Setup and Cohort Generation

First, we create a mock CDM reference and generate a cohort of acetaminophen users, collapsing any records separated by 30 days or less.

library(DrugUtilisation)
library(CDMConnector)
library(dplyr)

# Create a mock CDM for demonstration
cdm <- mockDrugUtilisation(numberIndividuals = 100, seed = 1)

# Generate a cohort of acetaminophen users with a 30-day gap era
cdm <- generateIngredientCohortSet(
  cdm = cdm,
  name = "acetaminophen_users",
  ingredient = "acetaminophen",
  gapEra = 30
)

2. Refine the Cohort

Next, we apply inclusion criteria to define our study population as new users with at least 365 days of prior observation.

# Refine the cohort
cdm$acetaminophen_users <- cdm$acetaminophen_users %>%
  requireIsFirstDrugEntry() %>%
  requireObservationBeforeDrug(days = 365)

3. Perform Analyses

Now we can run our desired analyses on the refined cohort.

# a) Summarise drug utilisation metrics
drug_summary <- cdm$acetaminophen_users %>%
  summariseDrugUtilisation(
    ingredientConceptId = 1125315 # Concept ID for acetaminophen
  )

# b) Summarise indications (requires pre-defined indication cohorts)
# First, create indication cohorts for headache and influenza
indications <- list(headache = 378253, influenza = 4266367)
cdm <- generateConceptCohortSet(cdm, name = "indication_cohorts", conceptSet = indications)

# Now, summarise indications within a 30-day window before the drug start date
indication_summary <- cdm$acetaminophen_users %>%
  summariseIndication(
    indicationCohortName = "indication_cohorts",
    indicationWindow = list(c(-30, 0))
  )

# c) Summarise treatment persistence over 1 year
persistence_summary <- cdm$acetaminophen_users %>%
  summariseProportionOfPatientsCovered(followUpDays = 365)

4. View and Report Results

The package provides easy-to-use functions for tabulating and plotting the results.

# View the drug utilisation summary table
tableDrugUtilisation(drug_summary)

# Plot the indication summary
plotIndication(indication_summary)

# Plot treatment persistence
plotProportionOfPatientsCovered(persistence_summary)

Key Analyses in Detail