Table 2.

Key lessons from the AIRWAYS-2 trial

• When the randomisation of individual patients is not possible, a cluster randomised design may be considered. A large number of small clusters will reduce intra-class correlation.

• Automatic enrolment of all eligible patients minimises the risk of bias in circumstances where those responsible for participant enrolment are not blinded to treatment allocation.

• For research in emergency situations it may be necessary to establish ways of collecting outcome data that do not rely on patient consent. In England, approval from the Confidentiality Advisory Group (CAG) can facilitate this.

• Strong patient and public involvement (PPI) in study design helps to address challenging ethical and patient consent issues.

• Hospital teams may not be used to collaborate in emergency medical (ambulance) services-led research.

• Once an emergency medical (ambulance) service begins patient recruitment, any of the hospitals in the area covered by that service could potentially receive a patient. Therefore, a phased approach to opening hospital sites is not possible and alternative approaches are required.

• Early identification of key stakeholders at participating healthcare organisations, and the provision of concise study documents, facilitates effective engagement and raises awareness of the study.

• A study-wide communications strategy, approved by an ethics committee, enables local communities and healthcare organisations to be informed about the study, and minimises the risk of unwilling patients being enrolled.

• An electronic system can support effective and efficient study management; it facilitates online training for site staff, provides an up-to-date document repository, and enables printing of patient-specific document packs.

• For patients recruited to clinical trials in emergency situations a flexible approach to follow-up will maximise participation.