Automated external defibrillator and operator performance in out-of-hospital cardiac arrest.

AIM: An increasing number of failing automated external defibrillators (AEDs) is reported: AEDs not giving a shock or other malfunction. We assessed to what extent AEDs are 'failing' and whether this had a device-related or operator-related cause. METHODS: We studied analysis periods from AEDs used between January 2012 and December 2014. For each analysis period we assessed the correctness of the (no)-shock advice (sensitivity/specificity) and reasons for an incorrect (no)-shock advice. If no shock was delivered after a shock advice, we assessed the reason for no-shock delivery. RESULTS: We analyzed 1114 AED recordings with 3310 analysis periods (1091 shock advices; 2219 no-shock advices). Sensitivity for coarse ventricular fibrillation was 99% and specificity for non-shockable rhythm detection 98%. The AED gave an incorrect shock advice in 4% (44/1091) of all shock advices, due to device-related (n=15) and operator-related errors (n=28) (one unknown). Of these 44 shock advices, only 2 shocks caused a rhythm change. One percent (26/2219) of all no-shock advices was incorrect due to device-related (n=20) and operator-related errors (n=6). In 5% (59/1091) of all shock advices, no shock was delivered: operator failed to deliver shock (n=33), AED was removed (n=17), operator pushed 'off' button (n=8) and other (n=1). Of the 1073 analysis periods with a shockable rhythm, 67 (6%) did not receive an AED shock. CONCLUSION: Errors associated with AED use are rare (4%) and when occurring are in 72% caused by the operator or circumstances of use. Fully automatic AEDs may prevent the majority of these errors.