The SDAC Reporting Model

The ‘Madison Firewall’ for Independent Data Monitoring

Overview of the Two-Report System

The hallmark of the Statistical Data Analysis Center (SDAC) model at UW-Madison is the absolute separation of trial conduct from trial results. This is achieved through a strict “Two-Report” system.

1. The Open Report

The Open Report is provided to everyone involved in the trial, including the Sponsor and the Steering Committee. Its primary purpose is to monitor trial integrity and progress.

  • Recruitment: Comparison of actual vs. projected accrual.
  • Baseline Characteristics: Pooled demographics to ensure the enrolled population matches the target population.
  • Data Quality: Completeness of forms, timeliness of data entry, and lost-to-follow-up rates.
  • Aggregate Safety: Serious Adverse Events (SAEs) reported for the entire study population (treatment arms combined).

2. The Closed Report

The Closed Report is the “Inner Sanctum.” It is prepared by the Independent Reporting Statistician and is seen only by the DMC.

  • Unblinded Efficacy: Primary and secondary endpoints broken down by Treatment Arm A vs. Arm B.
  • Unblinded Safety: Comparison of SAEs and specific toxicities between arms.
  • Statistical Boundaries: Presentation of interim results against stopping rules (e.g., O’Brien-Fleming or Lan-DeMets alpha-spending boundaries).

The Independent Statistical Analysis Center (ISAC)

In this model, the ISAC (often the SDAC itself) acts as the Data Firewall.

Key Responsibilities of the ISAC:

  1. Access to Randomization: The ISAC is the only group besides the clinical database manager with access to the treatment codes.
  2. Preparation of the “Closed” Pack: They generate the unblinded tables and figures.
  3. The Pre-Meeting: The Independent Statistician often meets with the DMC chair before the full meeting to walk through the “Closed” findings.
  4. No Sponsor Influence: By keeping the Sponsor blinded to the “Closed” data, the ISAC prevents the Sponsor from making changes to the trial (like increasing sample size or dropping an arm) based on a “sneak peek” at the results.

“The reporting statistician must be independent of the sponsor to ensure that the interim results do not inadvertently influence the trial’s conduct.” — Adapted from Ellenberg, Fleming, and DeMets (2002).

Why the Firewall Matters

If the Sponsor were to see the unblinded data, even without taking action, the “blinded” status of the trial would be compromised in the eyes of regulatory bodies (FDA/EMA). The SDAC model ensures that the scientific and statistical integrity of the trial remains beyond reproach until the final analysis.

Foundational References for the Independent Model

The “Potomac” Papers (The Blueprint)

  • Fisher LD, Roecker EB, and DeMets DL. (2001). The role of an independent statistical analysis center in the industry-modified NIH model. Statistics in Medicine, 20(17-18), 2593-2604.
    • Key Concept: This is the core paper defining the “Firewall.” It explains how an academic center (like SDAC) can serve as the unblinded engine for an industry trial.

Recent Reflections on Independence

  • DeMets DL and Cook TD. (2022). The role of the statistician on a data monitoring committee. Journal of Biopharmaceutical Statistics, 32(5).
    • Key Concept: DeMets revisits the model in the context of modern “Adaptive Designs.” He argues that as trials get more complex, the independence of the reporting statistician becomes more important, not less.
  • DeMets DL, et al. (2020). Data Monitoring Committees: Expect the Unexpected. New England Journal of Medicine.
    • Key Concept: Highlights that the independent statistician isn’t just a “data cruncher” but a critical advisor to the DMC when trials face external shocks (like a pandemic or a competitor’s sudden result).