Performance and error analysis of automated external defibrillator use in the out-of-hospital setting

We determined whether automated external defibrillators (AEDs) can meet the American Heart Association performance criteria to detect and shock unstable cardiac rhythms (ventricular fibrillation [VF], ventricular tachycardia [VT]) in the setting of an out-of-hospital cardiac arrest.

AED performance was reviewed for cardiac arrests occurring between January 1, 1995, and December 31, 1997. After every cardiac arrest, data regarding each rhythm analyzed and subsequent response (shock or no shock) were downloaded from the AED memory module. The study paramedic and study physician independently reviewed each case and interpreted cardiac rhythms from downloaded AED data. The emergency medical services medical director resolved all discrepancies in a blinded manner. All cases of out-of-hospital cardiac arrest in which an AED was turned on and a rhythm analyzed were included. The primary objective was the correct identification and defibrillation of VF or VT. Sensitivity, specificity, and predictive values with 95% confidence intervals (CIs) were calculated. Sources of error in AED rhythm management are also described.

A total of 3,448 AED rhythms were available for interpretation. Sensitivity and specificity for appropriate AED management of a shockable (VF or VT) rhythm were 81.0% (95% CI 77.9% to 83.8%) and 99.9% (95% CI 99.7% to 100%), respectively. Positive and negative predictive values were 99.6% (95% CI 98.7% to 99.9%) and 95.5% (95% CI 94.7% to 96.2%), respectively. There were 132 errors associated with AED management. Two errors resulted in delivery of an inappropriate shock. In the remaining 130 errors, a shockable rhythm was not shocked. Fifty-five (42.3%) errors were AED dependent, 70 (53.9%) were operator dependent, and 5 (3.9%) were unclassified.

The AED had high specificity and moderately high sensitivity in detecting and shocking unstable cardiac rhythms in the out-of-hospital setting. Few cardiac rhythms were mismanaged by the AED. Elimination of operator-dependent errors could increase AED sensitivity.