When is it futile for ambulance personnel to initiate cardiopulmonary resuscitation?

Introduction

The scene is familiar. Ambulance staff respond to an emergency telephone call: “I think my husband has died.” They find a pulseless, apnoeic corpse, in their view clearly beyond help. His distressed widow requires professional support and care. But there are no definitive signs of death. Thus, because ambulance personnel cannot pronounce “life extinct” (they are not, after all, doctors), the charade must be acted out of instituting full resuscitation measures while removing the victim “from public gaze.” Johnson and Maggiore, reporting from New Mexico, found that most emergency medical technicians had at least once in their careers taken a decision to discontinue resuscitation without a direct physician order or protocol guidance and that nearly half were troubled by this decision.1

Since the introduction of automated external defibrillators with monitoring capability an additional tool--the electrocardiogram monitor--has become available to ambulance crews which might help in diagnosing death. Kellerman et al and Bonnin et al reported that in the United States paramedics can now identify patients who will not survive resuscitation2 3. Halting futile attempts could redirect about $1bn worth of resources to other aspects of health care.4 We reviewed ambulance resuscitation records in Scotland to see whether patients could be identified in whom resuscitation attempts are clearly futile.

Methods

Since 1988-9 the Scottish Ambulance Service has been equipped with automated external defibrillators on all front line ambulances. Details of the defibrillation protocol and outcomes of resuscitation have been reported.5 6 Ambulance personnel complete a report after each attempted resuscitation, which includes time of cardiac arrest (if known), time of the 999 call, time of arrival on the scene, and time of defibrillation. The printout from the defibrillator (Laerdal Heartstart 2000 or 3000) allows assessment of the presenting rhythm, documentation of the cardiac rhythm before and after defibrillation (when delivered), and the final rhythm achieved. Analysis of the algorithm used within the automated external defibrillator disclosed a sensitivity of 90.3% to 94.4% in identifying a “shockable” rhythm (ventricular fibrillation or rapid ventricular tachycardia) and a specificity of 99.0% to 99.7% in excluding “non-shockable” rhythms.7

The resuscitation database was examined for patients who met the following criteria: (a) more than 15 minutes between collapse and arrival of the ambulance; (b) no bystander cardiopulmonary resuscitation performed; (c) no pulse or breathing on arrival of ambulance personnel. These criteria were based on the findings of Aprahamian et al, who hypothesised that survival would be unexpected in cases of unwitnessed cardiac arrest when the onset of symptoms occurred more than 15 minutes before the arrival of the ambulance and asystole was the presenting trace on the electrocardiogram.8

At the time of this study ambulance personnel were authorised to withhold resuscitation if there were definitive signs of death such as putrefaction, decapitation, rigor mortis, or postmortem staining. They were also authorised not to initiate resuscitation in patients with known terminal diseases when the doctor responsible for the patient had given written instructions or recorded verbal instructions with ambulance control not to resuscitate. Otherwise, ambulance personnel had no discretion to withhold resuscitation. All patients fulfilling the above criteria underwent a full resuscitation attempt.

Results

At the time of analysis details of resuscitation and ultimate outcome (death or discharge alive from hospital) were available for 8337 patients in the Heartstart Scotland database. Of these patients, 414 met the study criteria.

Table I gives the outcome of the resuscitation attempt in the 414 patients according to whether defibrillation was performed. Only one patient was discharged alive. This patient had sustained a witnessed cardiac arrest, the ambulance had arrived 15 minutes later, and a shockable rhythm was present on first rhythm analysis. The time from arrest to first shock was 19 minutes and cardiopulmonary resuscitation was supported by adrenaline given by a doctor. Five other patients with a shockable rhythm were admitted alive to hospital but died after admission. There were no survivors among the 240 patients not given a defibrillator shock.

In order to consider whether the automated external defibrillator might have failed to identify a potentially treatable rhythm in the non-shocked group we carried out a detailed analysis of the rhythm strips of patients in this group (table II). Electrocardiogram printouts were retrieved for 193 (80%) of the 240 patients. Asystole was confirmed in 100 cases (52%). A possible diagnosis of ventricular fibrillation was tenable in eight patients (4%), though in all of these the amplitude was fine or definitive analysis of the rhythm was rendered impossible by artefact from ongoing cardiopulmonary resuscitation.

There were no survivors among patients who met the conditions of the hypothesis, so that the omission of treatment in these patients would not have influenced survival. We propose an algorithm (figure) to govern the actions of ambulance staff confronted with a patient in whom attempts at resuscitation may be futile.

Algorithm for use by ambulance personnel faced with patient in whom attempts at resuscitation may be futile

• Download figure
• Open in new tab
• Download powerpoint

Discussion

Studies of the apparent futility of cardiopulmonary resuscitation given certain criteria2 3 have examined only the effect of discontinuing full advanced cardiac life support performed as routine. Most reports are from North America, where full advanced life support paramedic teams begin resuscitation routinely in all cases and thereafter use protocols for discontinuation. Our study differs by addressing whether to start resuscitation given the criteria of an unwitnessed arrest of more than 15 minutes' duration and a flat trace on the monitor.

The defibrillators used in the Heartstart Scotland project are automated. Concern might be expressed that patients with a flat trace in whom no shock is advised by the automated external defibrillators may be in fine ventricular fibrillation. Weaver et al, however, showed that when the amplitude of ventricular fibrillation is less than 0.1 mV full resuscitation, including defibrillation, is rarely associated with survival (only 23 (5.8%) of 394 cases).9 We identified eight patients with possible ventricular fibrillation in the non-shocked group; none had truly “flat line” asystole, and with our algorithm a resuscitation attempt would have been initiated routinely.