Friday, August 04, 2006

Having Computers Decide We're Dead

Death can come easy now.

Or at least deciding when to say enough is enough. And it seems the computer will help us make that decision. And if it does, about one in three patients would actually qualify for advanced cardiac life support.

This week's New England Journal of Medicine prospectively analyzed a prediction rule using the computer algorithm contained in an automatic external defibrillator (AED) to decide if it was worth transporting a patient found "down" to an Emergency Room. Actually the prediction rule required three things: Emergency Medical Technicians (EMTs) may consider the termination of resuscitation (1) if there is no return of spontaneous circulation before a patient is transported to the emergency department, (2) if the patient received no shocks by the AED before transportation was initiated, and (3) had a cardiac arrest that was not witnessed by EMS personnel responding to the call. Using these three indicators, the prediction rule had a positive predictive value of 99.5 percent and a specificity of 90.2 percent. Among patients whose condition met these three criteria, the survival rate was 0.5 percent. (Yes, there was a very rare circumstance where an individual would survive, even when all three criteria were met). But the prediction rule would have resulted in the transportation of 37.4 percent of patients (464 of 1240), rather than the current rate of 100 percent, and likely saved countless wasted medical resources and perhaps avoided unnecessary risk to the public as the ambulance rushes to the hospital.

AEDs interpret EKG signals from the patches applied to the patient's chest. Using these signals, the device can instruct an early responder whether to apply a shock to a person's chest or not. In effect, these algorithms help decide when not to be overly aggressive at resuscitative efforts.

The authors make clear that these prediction criteria should not be absolute:
"Finally, we consider that our prediction rule offers guidance for clinicians but is not obligatory. In an editorial published more than 20 years ago, Cummins and Eisenberg (Cummins RO, Eisenberg MS. Prehospital cardiopulmonary resuscitation: is it effective? JAMA 1985;253:2408-2412) suggested that prediction rules for the termination of resuscitation efforts should remain advisory and that they should be tempered by the full clinical picture, taking into account the very small possibility of successful resuscitation when the prediction rules suggest termination."
Certainly clinicians are aware of the cerebral-preserving effects of hypothermia. But beyond this, will our ambulance crews accept the clinical decision-making responsibility in our litiginous society? Will the families accept their loss and concur with the EMT's decision in all cases?

Not always. So even with the advances of computer technology and the use of automated schemes, "dead" patients will continue to be transported to Emergency Rooms.

And that's okay.


Heading off to the hills of Wisconsin for a bit of much-needed vacation his weekend. See you Monday.

No comments: