In an effort to minimize the time to deliver shocks in patients in cardiac arrest, hospitals across the country have turned to automatic external defibrillators to reduce the time to first shock, thereby improving cardiac arrest outcomes. More often than not, this policy has been effective at assuring that nursing staff and even locally-available non-medical personnel can at least treat a patient as soon as possible after a cardiac arrest.
But might there be an instance where the AED gets it wrong?
Well, of course. No shock algorithm is perfect. Take a look at this strip obtained from a Medtronic's Life Pak 20 defibrillator set to default to AED mode after a telemetry alarm prompted nursing personnel to rush to be patient's room, only to see the patient lose consciousness:
Click image to enlarge
CPR is started, but no shock was delivered. Why? Because the AED considered the rhythm "NONSHOCKABLE."
A quick call to the company suggests the algorithm (which is not published anywhere to my knowledge) involves five factors to attempt to be highly sensitive and fairly specific for ventricular arrhythmias: heart rate over 120, amplitude of the signal, slope of the received EKG morphology, QRS width, and something called "flat line content." Careful review of this rhythm demonstrates that it represents a heart rate over 120, is irregular, has varying QRS widths, and probably has different "slopes" and has an unknown "flat line content," whatever that represents.
While these findings are interesting for engineers, man (and women) must be allowed to intervene when it is perceived a machine is in error and anyone in a new, wide complex rhythm that causes a team of people to initiate CPR should consider the obvious:
Shock 'em anyway (synchronized to the QRS, of course).
Oh, and how do you deactivate the automatic nature of the AED on a Lifepak 20? Push any button besides the "Shock" button on the device and you'll be placed in manual mode.