• Public health

This year is the 50th anniversary of the introduction of modern resuscitation from cardiac arrest, made possible by the combination of closed chest compressions with external defibrillation and effective artificial ventilation.1 Inevitably this was restricted initially to hospitals, but within a few years the need to counter sudden death in the community led to the development of cardiac ambulances. The appreciation that lethal cardiac arrhythmias are not only due to acute myocardial infarction but can also occur unpredictably from a myriad of causes led to more complex responses. In most developed countries we now have public education on the need for rapid access to help, widespread training in cardiopulmonary resuscitation (CPR), means of early defibrillation where relevant and skilled aftercare—the so-called ‘chain of survival’.2 But daunting problems markedly limit success, irrespective of knowledge and training within the community. Even when death strikes suddenly and prematurely, many cases are complicated by severe underlying pathology that is not always amenable to prompt treatment. Even more importantly, only a very few minutes are available for effective resuscitation before apparently irreversible cerebral and cardiac changes make recovery impossible. Survival from out-of-hospital cardiac arrest (OOHCA) is therefore achieved only in a small minority, even of those ‘too young to die’. Investigating the predictors of success can help to prioritise efforts to improve results that are currently so dire. They have also been used as a guide for recognising futility, with the aim of curtailing resuscitation attempts that may have no chance of success.

Many studies have been published on the predictors of success for resuscitation of out-of-hospital cardiac arrest (OOHCA), including a recent review.3 As with all data relating to survival from major prehospital events, this topic is bedevilled by difficulties that may lead to inaccurate or misleading data and also to discrepancies that may be more apparent than real. Accurate record keeping in prehospital care of emergencies is challenging; even the identity of victims is often not known initially. Collation of data from emergency services with those from hospitals in order to ascertain discharge status can be very difficult, especially in the UK because of confidentiality rules. Some well-organised groups—particularly in Sweden4 and North America5—have largely overcome such problems and have been able to contribute greatly to our knowledge. But, criteria for inclusion of data vary widely between reports, ranging from all cases in which a resuscitation effort has been made to identifiable subgroups chosen for comparator purposes, designed to eliminate as far as possible variables that cannot be influenced by emergency services. The international Utstein group recommended in 19916 the use for this purpose of only bystander-witnessed adult arrests of presumed cardiac origin in which ventricular fibrillation or pulseless ventricular tachycardia (VF/VT) was the first recorded rhythm. A later review from the same source placed more emphasis on less restrictive data that are of more value in terms of epidemiology.7 The purpose for which data are collected can lead to appreciable differences in inclusion criteria that will have some influence on predictors of a successful outcome.

Most predictors of success are widely agreed, however, and are valid for most prehospital cardiac arrest data irrespective of the inclusion criteria. The response interval of the emergency service is an obvious one, although a recent publication highlighted the non-linear effect of delay8; the penalty of time lost in the first 4 min is slight because all short delays are favourable, but the additional penalty when delays are long is also slight because they are all unfavourable. Herlitz et al9 found that the first recordable rhythm scored even more highly than the response interval; VF is favourable because it can usually be reversed, it tends to occur where the underlying pathology is not inevitably fatal, and it also acts as a surrogate for response interval since asystole ensues in all cases within minutes. The same authors listed other factors achieving high statistical significance for success: cardiac arrest outside the home; witness by bystander; CPR given before the arrival of the ambulance; and age. Both the site of the arrest and the availability of a witness also relate to delay to the onset of treatment, so are not totally independent. Age is important principally as a marker of likely comorbidity. As an independent predictor, it seems relatively unimportant when allowance is made for the lower incidence of VF as the first recorded rhythm.10 Several accepted predictors are thus very interdependent but fundamentally reflect the times to effective first aid (CPR) or definitive treatment, together with comorbidity and underlying pathophysiology for which first observed rhythm is a surrogate.

Other potential predictors of success depend on observations available only during the resuscitation attempt. Complex analysis of the fibrillatory waveform11 can reveal characteristics that have a strong relationship to the chances of a return of spontaneous circulation (ROSC); the possibility has been explored of its use to determine the appropriate timing of an automated defibrillator shock. More importantly, the occurrence of ROSC during an attempted resuscitation is a variable of clinical significance. If ROSC cannot be achieved at least transiently, then the likelihood of eventual recovery is low; failure in cases presenting with VF has been suggested as an indication for terminating resuscitation efforts. In Ontario12 a guideline designed for terminating an attempt by those qualified to practise only basic life support and defibrillation was found also to be suitable for those who had advanced life support skills. This depended on no ROSC prior to transport, no shock having been required, the arrest unwitnessed by emergency medical services personnel or bystander and no CPR. Many might be uneasy with such guidance even after a careful study of its validity.

The Swedish group have made major contributions over at least 12 years to our understanding of the predictors of a successful outcome for resuscitation after cardiac arrest. This issue of the journal contains their latest study13 that differs from others in that it focuses on the survivors of OOHCA rather than on the totality of victims (see page 1826). This new perspective has some important implications that merit attention. The first is the need to be less pessimistic about patients with non-shockable rhythms at first contact; they comprise 20% of this series of nearly 2200, and two-thirds never received a shock. The attitude of professional rescuers has not been investigated as a predictor of success, but its importance should not be doubted. Second and less encouragingly, the outcome in terms of cerebral function tended to be less good for asystolic arrests. This may have been due to longer resuscitation attempts, but one might also speculate on possible effects of adrenaline which has been shown in an animal model to have important adverse effects on cerebral capillary flow.14 Also important is the finding that a third of all survivors had arrests witnessed by ambulance staff; this highlights powerfully the continuing requirement to educate the public in the need to call promptly for unaccustomed chest pain.

Less surprisingly, in these new data 79% of survivors had a cardiac aetiology and 90% were witnessed. Women accounted for only 28% of survivors; they were more likely to be at home, less likely than men to have VF and had less CPR. The underlying reasons are understandable and will be hard to counter.

We now understand most of the predictors of success in the treatment of OOHCA, but one important lesson is never to equate a lower chance of survival with no chance. This can have a powerful demotivating effect on management, both pre- and in-hospital, with the result that we have fewer successes than our present knowledge base can justify.