We prospectively recorded all AED activations in the Girona Territori Cardioprotegit program from June 2011 to June 2015.

This project is a public access defibrillation initiative promoted by the Girona City Council (Dipsalut) and began in 2011. The program was enabled by a change in regional law allowing AEDs to be used by any person in the event of an emergency. Although no CPR training was given to the general public, an awareness campaign was launched involving public events attended by more than 10 000 people, as well as specific school courses that reached more than 4000 students.

A total of 747 AEDs were distributed according to population density (1 per 1000 population) in either public places (577 permanent AEDs) or as part of the equipment of police, fire department, and basic life support ambulance crews (170 mobile AEDs). These professionals received an official course on AED use.20 The location of the permanent devices can be consulted on the website of the Girona Territori Cardioprotegit project.21

The AEDs distributed by the program are the PowerHeart AED G3 9300 manufactured by Cardiac Science (Bothell, Washington, United States), authorized for use in Europe by the Medical Device Safety Service. This model delivers a charge that is adapted to the patient's impedance and increases in successive defibrillations, beginning with 126 to 260J and continuing with charges between 170 and 351J.

Electrocardiographic tracings were obtained from the AEDs; we excluded those that were not actually OHCAs or had incomplete or unintelligible information.

The records extracted from the AEDs were saved in PDF and contain information on the time of device opening, the time of cardiac rhythm detection and interpretation, and the number of shocks administered, as well as the electrocardiographic tracings. Also recorded from each device activation were the type of AED used (mobile or permanent) and information on patients’ vital status after CPR maneuvers. A retrospective descriptive analysis was performed of the data collected.

The initial rhythm recorded by the AED for each patient was used for the descriptive analysis of the arrhythmias underlying the cardiac arrest and the study of the diagnostic accuracy of the algorithms. All recorded arrhythmias were used in the analysis of the therapeutic effectiveness of the device.

The cardiac rhythms of the electrocardiographic tracings obtained by the activated and used AEDs were classified according to the consensus document of the American Heart Association.14 Accordingly, VT, coarse VF (> 200μV), and fine VF (≤ 200μV) were considered shockable rhythms; sinus rhythm, atrioventricular block, atrial fibrillation, and other supraventricular rhythms were considered nonshockable. An AED shock was considered effective if it managed to reverse the shockable rhythm, regardless of the final rhythm.

The type of AED activation was classified into 3 categories: a) correct, involving a patient with sudden loss of consciousness who failed to respond to commands and who was administered basic CPR maneuvers; b) intermediate, involving a patient with loss of consciousness who quickly recovered and did not thus require CPR or a patient who did not lose consciousness but whose situation was considered serious by possible resuscitators; and c) incorrect, involving antisocial behavior.

CPR performance was indirectly evaluated via the chest compression waves visible on the electrocardiographic tracing. Similar to previous work,22 resuscitation maneuvers were considered to have been performed if this type of wave was visible and were considered adequate if they met the following characteristics: a) the AED instructions were followed; and b) the CPR was consistently performed and at a frequency of > 50 compressions/min.

To study the sensitivity and specificity of the arrhythmia detection algorithms of the AED, the reference standard was an independent reading of the electrocardiographic tracings by 4 cardiologists. Any disagreements were resolved via joint analysis of the tracing until a consensus was reached. Continuous variables are presented as mean±standard deviation and categorical variables as absolute values and percentages. Categorical variables were compared using the chi-square test and continuous variables using the Student t test. A 2-sided P < .05 was considered statistically significant.

From the initiation of the Girona Territori Cardioprotegit project in June 2011 to June 2015, there were 231 AED activations. The activation involved a mobile AED in 176 cases (76%) and a permanent device in 55 (24%).

Of these 231 activations, 210 (91%) were considered correct uses, 14 (6%) were intermediate, and 7 (3%) were incorrect. No electrocardiographic tracings were obtained from the 21 cases classified as an intermediate or incorrect use. Of the 210 cases classified as a correct use, 20 lacked an electrocardiographic recording and 2 had an unintelligible tracing. Thus, 188 activations could be analyzed (81.4% of the total); 154 were from mobile AEDs (82%) and 34 from permanent AEDs (18%).

The rhythms underlying the OHCAs are shown in Figure 1. The rhythm most frequently recorded by the AEDs was asystole, in 80 patients (42.5%). VF was present in 43 patients (22.8%), comprising 35 coarse VFs (18.6%) and 8 fine VFs (4.2%).

Figure 1.

Distribution of recorded cardiac rhythms. Values are numbers and percentages of the total. AF, atrial fibrillation; AVB, atrioventricular block; Coarse VF, coarse (> 200mV) ventricular fibrillation; Fine VF, fine (≤ 200mV) ventricular fibrillation; SB, sinus bradycardia; VT, ventricular tachycardia.

(0.11MB).

The remaining tracings in order of frequency were sinus rhythm (28 patients, 14.9%), second- and third-degree atrioventricular block (22; 11.7%), atrial fibrillation (6; 3.2%), sinus bradycardia (5; 2.6%), and VT (4; 2.1%). Shockable rhythms comprised 25% of all tracings.

Of all of the activations analyzed, the AED determined that 39 tracings contained an initial shockable rhythm (20.7%). There were no false positives. A tracing of a successfully treated shockable rhythm is shown in Figure 2.

Figure 2.

Example of a successfully treated shockable rhythm. After lead placement, a coarse ventricular fibrillation was detected, which was followed by an effective shock that reversed the abnormal ventricular rhythm. The tracing shows the diagnostic, charging, and shock phases of the device. CPR, cardiopulmonary resuscitation.

(1.11MB).

AEDs determined 149 of the tracings (79.2%) to be nonshockable. Eight false negatives were identified (4 fine VFs and 4 coarse VFs) (Figure 3).

Figure 3.

Examples of false negatives. A: Fine ventricular fibrillation and the time at which the defibrillator advises against the use of a shock. B: Coarse ventricular fibrillation and the time at which the defibrillator advises against the use of a shock.

(0.74MB).

Based on these data, the sensitivity of the AED algorithms for detecting shockable rhythms was calculated at 82.9%, with a specificity of 100%. Their sensitivity for the detection of the different types of shockable arrhythmias was 89% for coarse VF, 33% for fine VF, and 100% for VT.

The time required by the AED to identify the rhythm was 7.0±2.0seconds; it was much shorter for shockable rhythms than nonshockable rhythms (4.0±1.0seconds vs 8.0±1.8seconds; P < .001).

In the 39 cases determined by the AED to have an initial shockable rhythm, a therapy was indicated. A total of 98 shocks were administered, with a mean of 2.4 shocks (1-11) per patient. In 85.4% of the administered shocks, the device managed to reverse the arrhythmia.

Comparisons between time of device use, percentage of shockable rhythms, number of shocks, number of effective shocks, performance of CPR, and rate of return of spontaneous circulation between permanent and mobile devices are summarized in Table. There were no significant differences between the 2 systems for any of the above variables, except for the number of shocks, which was significantly higher with mobile devices (2.63 vs 1.75; P=.019). Permanent devices had a nonsignificantly higher rate of return of spontaneous circulation (29.4% vs 23.4%; P=.459).

Based on analysis of the waves produced in the tracings by the CPR maneuvers (Figure 4), chest compressions were performed in 152 patients (80.1%), but only 79 (51.9%) were considered adequate according to CPR quality criteria. There were no differences in maneuver performance (79% vs 81%; P=.8) or quality (45% vs 42%; P=.7) between fixed and permanent devices.

Figure 4.

Chest compression waves visible in the tracings. The arrows indicate the waves recorded in the tracings that are due to the performance of CPR maneuvers. Two cycles of 30 compressions and 1 pause can be seen, presumably because the standard 30:2 compression-to-ventilation ratio was being followed. CPR, cardiopulmonary resuscitation.

(0.58MB).

Data were available on the return of spontaneous circulation for 176 of the 188 analyzed patients (94%); 46 (26.1%) survived to resuscitation. In the case of the 47 shockable rhythms, the rate of return of spontaneous circulation was 49% (23 patients), whereas the rate was 18% for the 129 nonshockable rhythms (P < .001). There were no accidents related to device use.

The main limitation of our work is that it is solely based on electrocardiographic recordings and the times recorded by the defibrillator. We thus have no information on any other variables crucial to the study of the role of AEDs in the treatment of OHCA, such as mid- and long-term mortality and the presence of neurological sequelae. We also have no data on the time from collapse to AED application or on OHCAs from the province of Girona that were not treated with a device forming part of the Girona Territori Cardioprotegit program. In addition, the information on CPR maneuvers was indirectly obtained via the chest compression waves recorded in the electrocardiographic tracings and should be interpreted with caution.