A multicenter RCT evaluating a new oral anticoagulant versus warfarin in atrial fibrillation patients is ongoing in India. At the interim analysis (50% of planned enrollment, N = 800), the Data Safety Monitoring Board (DSMB) observes a significantly higher rate of major bleeding in the new drug arm (5.2% vs. 2.1%, p = 0.008) with no difference in stroke prevention. The DSMB must now make a recommendation. What is the most appropriate next step?

Explanation

## DSMB Interim Analysis and Trial Stopping Rules **Key Point:** A Data Safety Monitoring Board must recommend trial termination when interim data demonstrate a **clinically important and statistically significant safety signal** that outweighs efficacy benefits, even if efficacy endpoints are not yet met. ### Ethical Framework for Trial Stopping **High-Yield:** The primary duty of a DSMB is **participant safety**, not statistical significance of the primary endpoint. If harm clearly exceeds benefit, the trial must stop. ### Assessment of This Interim Finding | Finding | Interpretation | Action Implication | |---------|-----------------|-------------------| | **Major bleeding: 5.2% vs. 2.1%** | 2.5-fold higher absolute risk; clinically significant | **Safety signal present** | | **p = 0.008** | Statistically significant at conventional α = 0.05 | **Not due to chance** | | **Stroke prevention: no difference** | New drug offers no efficacy advantage | **No offsetting benefit** | | **Interim analysis at 50% enrollment** | Early detection of harm; remaining participants at risk | **Ethical imperative to stop** | ### Stopping Rules in RCTs DSMBs use pre-specified stopping rules: 1. **Efficacy stopping** — primary endpoint met with overwhelming evidence; continue to prevent harm from ongoing control arm 2. **Futility stopping** — primary endpoint unlikely to be met; continue if safety is acceptable 3. **Safety stopping** — serious adverse events exceed acceptable threshold; **stop immediately** **Mnemonic: EFS** — **E**fficacy, **F**utility, **S**afety (in order of urgency for stopping). ### Why This Scenario Mandates Stopping ```mermaid flowchart TD A[Interim Analysis: 50% enrollment]:::outcome --> B{Efficacy signal?}:::decision B -->|Yes| C[Continue to completion]:::action B -->|No| D{Safety signal?}:::decision D -->|Yes| E[STOP trial immediately]:::urgent D -->|No| F{Futility?}:::decision F -->|Yes| G[STOP for futility]:::action F -->|No| H[Continue enrollment]:::action ``` **Clinical Pearl:** In anticoagulation trials, a **2.5-fold increase in major bleeding without efficacy benefit** is unacceptable. Continuing enrollment would expose 800 additional participants to excess harm without compensatory stroke prevention. ### Informed Consent and Equipoise - **Equipoise violation** — the new drug is now inferior (more bleeding, equal stroke prevention). Continuing violates the ethical principle that participants must be in a state of genuine uncertainty about which arm is better. - **Duty to warn** — participants in the new drug arm have a right to know the interim safety signal; blinded continuation becomes ethically indefensible. **Warning:** Continuing a trial after a clear safety signal is detected is a **violation of Good Clinical Practice (GCP)** and research ethics guidelines (Declaration of Helsinki, ICH-GCP).