Retrospective analysis of shock decision and delivery in a fully automated Public Access Defibrillator

Hannah Torney, Adam Harvey, Amy Kernaghan, Joann Lowry, Olibhear McAlister, Ben McCartney, David McEneaney, Jennifer Adgey

Research output: Contribution to conferenceAbstractpeer-review


A recent analysis demonstrated that 4.2% of public access defibrillator (PAD) shock decisions were incorrect and that in 5% of instances where a PAD recommended a shock, the user did not press the shock button [1]. Some PADs are fully-automated, where the devices determine if shocks are required, and administer shocks without user input. This reduces possibility that a patient does not receive therapy due to a user error/decision. Here we assess the real-world safety and performance of the fully-automated SAM 360P PAD in an out-of-hospital setting.
Patient data was collected from August 2015- April 2019. Users of SAM 360Ps were asked to download the electronic data (ECG and impedance-cardiogram trace superimposed markers denoting the algorithm decisions) from the PAD and provide demographic data. Each case was analysed for instances of motion, and the behaviour of the algorithm was assessed.
A total of 47 cases were reported. Six patients were not suffering SCA, and were excluded from analysis, leaving 41 cases for review. Mean patient age was 59+/-16 years and 85.4% were males. Twenty-six (63.4%) patients presented in ventricular fibrillation or ventricular tachycardia, and all these patients received an immediate shock, with a mean of 1.56 shocks delivered (range:1-9). Of a total of 236 cardiac rhythm analyses, 233 were correct (98.7%), and there was 1 instance of inappropriate shock delivery (0.42%). “Motion detected” warnings were sounded in 16 events (39.0%). Four (25%) warnings were due to the patient regaining consciousness. “Motion detected” warnings caused cessation of motion in 6 events (37.5%). Motion continued after the warning in 4 cases (25%).
Analysis of the fully-automatic SAM 360P demonstrates high shock decision performance, and in the absence of the rescuer pressing a button, the time to shock could improve
[1] Zijlstra et al., Resuscitation 118 (2017): p140-146.
Original languageEnglish
Publication statusPublished (in print/issue) - Jun 2019


Dive into the research topics of 'Retrospective analysis of shock decision and delivery in a fully automated Public Access Defibrillator'. Together they form a unique fingerprint.

Cite this