Skip to main content

Self-Controlled Case Series (SCCS)

The Self-Controlled Case Series (SCCS) module implements a within-person study design where each patient serves as their own control. This eliminates time-invariant confounding (genetics, chronic conditions, socioeconomic factors) by comparing outcome rates during exposed risk windows versus unexposed periods within the same individual's observation time.

SCCS is particularly suited for studying the association between transient drug exposures and acute adverse events. Parthenon provides a full design form, configurable risk windows, model selection, covariate settings, and a results view with incidence rate ratio (IRR) estimates, forest plots, and era timelines.

Module Architecture​

SccsDetailPage
 |
 |-- Header (name, description, back navigation)
 |-- Source Selector + Execute button
 |-- Delete button
 |
 |-- Tab: Design
 |    |-- SccsDesigner
 |         |-- Basic Information (name, description)
 |         |-- Exposure Cohort selector
 |         |-- Outcome Cohort selector
 |         |-- Risk Windows (multiple, configurable)
 |         |-- Model & Population settings
 |         |-- CovariateSettingsPanel
 |         |-- Save / Create button
 |
 |-- Tab: Results
 |    |-- SccsResults
 |    |    |-- SccsVerdictDashboard (summary verdict)
 |    |    |-- Population Summary (cases, outcomes, observation periods)
 |    |    |-- IRR Estimates Table (sortable, with inline forest plots)
 |    |    |-- SccsTimeline (risk window era visualization)
 |    |    |-- Execution timing
 |    |
 |    |-- Execution History table

Feature Walkthrough​

Creating an SCCS Analysis​

  1. Navigate to Analyses and click New Analysis, then select SCCS.
  2. Fill in the Basic Information section with a name and optional description.
  3. Select an Exposure Cohort (the drug or intervention under study).
  4. Select an Outcome Cohort (the adverse event to measure).
  5. Configure one or more Risk Windows.
  6. Choose a Model Type and set population parameters.
  7. Optionally configure Covariate Settings.
  8. Click Create to save the analysis.

Configuring Risk Windows​

Risk windows define the time periods relative to the exposure era where outcome risk is assessed. Each window has four parameters:

ParameterDescriptionOptions
Start DayOffset in days from the anchor pointInteger (e.g., 0)
Start AnchorReference point for the start offsetEra Start or Era End
End DayOffset in days from the anchor pointInteger (e.g., 30)
End AnchorReference point for the end offsetEra Start or Era End

You can add multiple risk windows to study different exposure periods (e.g., 0-30 days post-exposure, 31-60 days post-exposure). Each window has an editable label. At least one risk window is required.

tip

A typical pharmacovigilance study uses a risk window of 0 to 30 days anchored to the era start. For persistent effects, consider extending the end day or anchoring to era end.

Model & Population Settings​

SettingDescriptionDefault
Model TypeStatistical model for rate estimationSimple (no adjustments)
Naive PeriodMinimum observation time (days) before first exposure to exclude prevalent users365 days
First Outcome OnlyRestrict to first occurrence of the outcome per patientEnabled

Available model types:

  • Simple -- no covariate adjustments; estimates raw IRRs.
  • Age-adjusted -- adjusts for age as a time-varying covariate.
  • Season-adjusted -- adjusts for seasonal patterns in outcome rates.
  • Age + Season adjusted -- adjusts for both age and seasonality.

Covariate Settings​

The shared CovariateSettingsPanel allows toggling which baseline features are included as covariates:

CovariateDescription
DemographicsAge, gender
Condition OccurrencePrior conditions in the lookback window
Drug ExposureConcomitant medications
Procedure OccurrencePrior procedures
MeasurementLab results and vitals

A configurable time window (default: -365 to 0 days) defines the lookback period for baseline covariates.

Executing an Analysis​

  1. Select a CDM data source from the dropdown in the page header.
  2. Click Execute. The analysis is queued and runs via the R runtime (HADES SelfControlledCaseSeries package).
  3. The page automatically switches to the Results tab when execution begins.
warning

Execution requires a connected CDM data source with Achilles results. The R runtime container must be healthy and the JDBC driver must be available at /opt/jdbc/ inside the container.

Viewing Results​

The results tab displays several sections:

Verdict Dashboard -- a summary panel providing an at-a-glance assessment of the SCCS results.

Population Summary -- three metric cards showing:

  • Cases -- number of patients with both exposure and outcome
  • Outcomes -- total outcome events observed
  • Observation Periods -- number of observation periods analyzed

IRR Estimates Table -- a sortable table with columns for covariate name, incidence rate ratio (IRR), 95% confidence interval bounds, log(RR), standard error, and an inline mini forest plot. IRR values above 1.0 are highlighted in red (elevated risk), and values below 1.0 in teal (protective effect).

Risk Window Timeline -- a visual timeline rendering the SCCS eras (exposure periods, risk windows, and observation time) using the SccsTimeline component.

Execution History -- a table listing all prior executions with status, source, start time, and completion time. Click a completed execution to load its results.

API Reference​

EndpointMethodDescription
GET /api/v1/analyses/sccs/{id}GETFetch SCCS analysis with design and executions
POST /api/v1/analyses/sccsPOSTCreate a new SCCS analysis
PUT /api/v1/analyses/sccs/{id}PUTUpdate an existing SCCS analysis
DELETE /api/v1/analyses/sccs/{id}DELETEDelete an SCCS analysis
POST /api/v1/analyses/sccs/{id}/executePOSTExecute against a source (body: { source_id })
GET /api/v1/analyses/sccs/{id}/executions/{execId}GETFetch execution results