AIRCHILL DSMB Charter
Independent oversight · development charter v0.1
AIRCHILL Data and Safety Monitoring Board Charter
This charter defines the governance model for independent unblinded safety review. Numeric stopping boundaries remain provisional until the current device hazard analysis, early clinical event rates and final statistical design are available.
1. Mandate
The DSMB protects participants and trial integrity by reviewing accumulating unblinded safety, conduct and, where prospectively authorised, efficacy/futility information. It advises the sponsor to continue, modify, pause or terminate the investigation.
2. Membership
- Independent chair with acute-care trial experience.
- Critical care/resuscitation clinician.
- Stroke/neurointervention specialist for STROKE reviews.
- Pulmonary/airway or anaesthesia expert.
- Independent biostatistician.
- Optional medical-device safety/engineering expert without sponsor conflicts.
Members sign confidentiality and conflict-of-interest declarations; material conflicts disqualify participation in the affected review.
3. Meetings
Organisational meeting: before first enrolment. Early-feasibility reviews: after predefined initial cohorts and whenever a hold trigger occurs. Routine reviews: by information fraction and at least at prespecified enrolment milestones. Formal Phase-III interim: approximately 50% information, if retained in final design.
4. Open and closed sessions
Open session covers recruitment, deviations, data completeness, device logistics and blinded aggregate safety. Closed session contains treatment-coded data and is limited to DSMB members and the independent statistician. Closed minutes are stored separately from sponsor operational files until unblinding is permitted.
5. Core safety review
- All-cause and cause-specific mortality.
- SAE and serious adverse device effects.
- Device deficiencies that caused or could have caused serious harm.
- Ventilation interruption, airway pressure/resistance, bronchospasm, gas-exchange deterioration.
- Arrhythmia, hypotension, vasopressor escalation and other haemodynamic signals.
- Pneumonia/respiratory complications and airway injury signals.
- Stroke: symptomatic intracranial haemorrhage, malignant oedema, procedural complications and study-attributable reperfusion delay.
- Cardiac arrest: seizure burden, neuroprognostication/WLST timing and imbalance in downstream care.
6. Hold / stop decision domains
A rapid review is triggered by unexpected serious device-related harm, recurring clinically significant ventilation compromise, a plausible treatment-associated mortality imbalance, a new airway injury pattern, or—during stroke enrolment—material repeat delays to arterial puncture/reperfusion attributable to AIRCHILL. Exact numerical rules belong in the final signed charter and SAP.
7. Efficacy and futility
If the confirmatory study uses a formal interim, the DSMB receives the precomputed alpha-spending boundary for overwhelming efficacy and a non-binding futility criterion. The sponsor remains blinded. Stopping for efficacy requires consideration of safety, data maturity, consistency and clinical plausibility, not only a nominal boundary crossing.
8. Recommendations
Permitted recommendations: continue unchanged; continue with enhanced monitoring; modify eligibility/procedures; temporarily pause; stop a dose/cohort; stop the study; request additional analysis. The recommendation letter states the action but does not reveal treatment-specific data unless necessary for participant protection.
9. Emergency unblinding
Individual emergency unblinding is available only when knowledge of assignment is necessary for immediate clinical management. It is logged and does not automatically expose the sponsor or blinded endpoint team.
10. Documentation
Charter, agendas, open/closed reports, attendance, COI declarations, datasets, statistical programs and recommendations are retained in the Trial Master File according to applicable retention requirements.