CohortSymmetry

1. Overview
2. Key Features
3. Installation
4. Methodological Workflow
5. Usage Example
6. Main Functions

The CohortSymmetry R package provides tools to perform Sequence Symmetry Analysis (SSA) and Prescription Sequence Symmetry Analysis (PSSA) on Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) databases.

---

Overview

Sequence Symmetry Analysis is a self-controlled study design commonly used for pharmacovigilance, post-marketing drug safety surveillance, and adverse event signal detection. By comparing the order of occurrence between an index exposure (e.g., a newly initiated drug) and a marker event (e.g., an incident diagnosis or prescription for a symptom), SSA tests whether the marker event occurs disproportionately after the index event compared to before.

Because individuals serve as their own controls, time-invariant confounding (such as genetics, socioeconomic factors, and chronic comorbidities) is intrinsically controlled for.

CohortSymmetry calculates:

• Crude Sequence Ratio (CSR): The ratio of the number of individuals who initiated the marker after the index event ($A$) versus before ($B$): $\text{CSR} = A / B$.
• Adjusted Sequence Ratio (ASR): The CSR adjusted for background prescribing trends and temporal asymmetry in the underlying population.
• Confidence Intervals: 95% confidence intervals around CSR and ASR estimates.

---

Key Features

• Standardized OMOP Integration: Operates directly on cohort tables created with CohortConstructor, DrugUtilisation, or CDMConnector.
• Temporal Symmetry Modeling: Evaluates sequence distributions across configurable time gaps and washout windows.
• Publication-Ready Outputs: Standardized summarised_result outputs formatted into tables (gt, flextable) and diagnostic plots (ggplot2).
• Signal Detection & Negative Controls: Ideal for rapid safety screening across multiple drug-outcome combinations.

---

Installation

Install CohortSymmetry from CRAN or GitHub:

# From CRAN
install.packages("CohortSymmetry")

# Development version from GitHub
# install.packages("pak")
pak::pkg_install("ohdsi/CohortSymmetry")

---

Methodological Workflow

graph TD
    subgraph "1. Cohort Instantiation"
        I["Index Cohort (e.g., Aspirin)"]
        M["Marker Cohort (e.g., Amoxicillin)"]
    end

    subgraph "2. Sequence Generation"
        S["generateSequenceCohortSet()<br/>Match Index & Marker Events per Individual<br/>(Apply timeGap, washoutWindow, daysPriorObservation)"]
    end

    subgraph "3. Sequence Ratio Calculation"
        SR["summariseSequenceRatios()<br/>Calculate Crude (CSR) & Adjusted (ASR) Ratios + 95% CI"]
        TS["summariseTemporalSymmetry()<br/>Evaluate Temporal Distribution Trends"]
    end

    subgraph "4. Visualisation & Reporting"
        T1["tableSequenceRatios()<br/>GT / Flextable"]
        P1["plotSequenceRatios()<br/>Point Estimates & CI"]
        P2["plotTemporalSymmetry()<br/>Histogram of Time Gaps"]
    end

    I & M --> S
    S --> SR
    S --> TS
    SR --> T1
    SR --> P1
    TS --> P2

---

Usage Example

The following example demonstrates conducting a Sequence Symmetry Analysis on synthetic DuckDB OMOP CDM data:

library(CDMConnector)
library(CohortConstructor)
library(CohortSymmetry)
library(dplyr)

# 0. Connect to database
con <- DBI::dbConnect(duckdb::duckdb(), eunomiaDir("GiBleed"))
cdm <- cdmFromCon(con, cdmSchema = "main", writeSchema = "main")

# 1. Instantiate Index (Celecoxib) and Marker (GI Bleed) cohorts
cdm$celecoxib <- conceptCohort(cdm, list(celecoxib = 1118084L), name = "celecoxib")
cdm$gi_bleed <- conceptCohort(cdm, list(gi_bleed = 192671L), name = "gi_bleed")

# 2. Intersect cohorts into sequence pairs
cdm <- generateSequenceCohortSet(
  cdm = cdm,
  indexTable = "celecoxib",
  markerTable = "gi_bleed",
  name = "celecoxib_gi_bleed",
  timeGap = 365,
  washoutWindow = 365,
  daysPriorObservation = 365
)

# 3. Summarise sequence ratios (CSR, ASR, 95% CI)
sr_results <- summariseSequenceRatios(
  cohort = cdm$celecoxib_gi_bleed
)

# 4. Generate formatted summary table and plot
tableSequenceRatios(result = sr_results, type = "gt")
plotSequenceRatios(result = sr_results, onlyASR = TRUE)
plotTemporalSymmetry(cdm = cdm, sequenceTable = "celecoxib_gi_bleed")

---

Main Functions

Function	Purpose
generateSequenceCohortSet()	Intersects an index cohort and a marker cohort to identify sequence pairs within a defined time window.
summariseSequenceRatios()	Calculates Crude Sequence Ratios (CSR), Adjusted Sequence Ratios (ASR), and 95% confidence intervals.
summariseTemporalSymmetry()	Extracts temporal distributions of time differences between index and marker events.
tableSequenceRatios()	Renders publication-ready sequence ratio summary tables (gt or flextable).
plotSequenceRatios()	Generates ggplot2 forest plots showing sequence ratio point estimates and confidence intervals.
plotTemporalSymmetry()	Plots the distribution of time gaps between index and marker events to visually inspect symmetry.