Aim To assess whether access to a voice activated decision support system (VADSS) containing video clips demonstrating resuscitation manoeuvres was associated with increased compliance with American Heart Association Basic Life Support (AHA BLS) guidelines.

Methods This was a prospective, randomised controlled trial. Subjects with no recent clinical experience were randomised to the VADSS or control group and participated in a 5-min simulated out-of-hospital cardiopulmonary arrest with another ‘bystander’. Data on performance for predefined outcome measures based on the AHA BLS guidelines were abstracted from videos and the simulator log.

Results 31 subjects were enrolled (VADSS 16 vs control 15), with no significant differences in baseline characteristics. Study subjects in the VADSS were more likely to direct the bystander to: (1) perform compressions to ventilations at the correct ratio of 30:2 (VADSS 15/16 (94%) vs control 4/15 (27%), p=<0.001) and (2) insist the bystander switch compressor versus ventilator roles after 2 min (VADSS 12/16 (75%) vs control 2/15 (13%), p=0.001). The VADSS group took longer to initiate chest compressions than the control group: VADSS 159.5 (±53) s versus control 78.2 (±20) s, p<0.001. Mean no-flow fractions were very high in both groups: VADSS 72.2% (±0.1) versus control 75.4 (±8.0), p=0.35.

Conclusions The use of an audio and video assisted decision support system during a simulated out-of-hospital cardiopulmonary arrest prompted lay rescuers to follow cardiopulmonary resuscitation (CPR) guidelines but was also associated with an unacceptable delay to starting chest compressions. Future studies should explore: (1) if video is synergistic to audio prompts, (2) how mobile technologies may be leveraged to spread CPR decision support and (3) usability testing to avoid unintended consequences.