Hypothermia after cardiac arrest: feasible but is it therapeutic?

Hay et al. [1] have added to the growing number of studies showing that hypothermia after cardiac arrest can be routinely achieved in intensive care units. They believe their audit supports the case for the wider use of a clinically effective treatment which does not require major additional critical care resources, echoing comments from an earlier survey [2] of UK practice. To someone unconvinced that deliberate hypothermia is ‘therapeutic’, the feasibility issue is irrelevant. What evidence is there to convince the clinically sceptic? Animal experiments suggested a promising effect but then so did similar studies on pharmacological interventions such as barbiturates, steroids and calcium antagonists, all of which failed to benefit human subjects with cerebral ischemia. The quality of hypothermia studies was noticeably weaker than those researching drugs, presumably due to funding differences, with the majority either underpowered or insufficiently blinded. Few studies are even available on which to base assumptions about the ‘dose’ of hypothermia needed. Even if a period of hypothermia does not appear to harm a fit rat, we then need to show that this is also true for rats that are elderly, diabetic or life-long smokers. What of the human studies? Most did not use randomised controls and we now know to discount the apparent treatment effect in studies using historical controls. The two groups are not otherwise identical and the results are generally biased in favour of the treated patients. However, such studies are easier to do and therefore outnumber randomised trials. One of the many difficulties in conducting prospective trials in intensive care is the reluctance of research ethics committees to permit research on patients unable to give consent. Another is demonstrating equipoise, which must have been almost impossible once the UK. Resuscitation Council had recommended hypothermia as standard care. Thus we are left with the two randomised studies which formed the basis of the Resuscitation Council recommendations and most subsequent review articles, editorials and meta-analyses. Both were published in the same issue of the New England Journal of Medicine [3, 4]. The primary endpoints were neurological outcome and secondarily mortality and complications. Coma persisting after successful resuscitation was the primary inclusion criterion for both studies, with sedatives and muscle relaxants given as required. One might have thought that the initial neurological findings would have been the single best predictor of outcome. A patient with a Glasgow Coma Score of three, with unreactive pupils and no respiratory effort would not be expected to do as well as one who was agitated and gagging on the tracheal tube. This information should have been presented in the results to allow the reader to confirm the effectiveness of randomisation. While much other information was given, this was not. The clinically sceptic reader could stop there and conclude the trials were invalid. However, if we attempt to discount for selection and treatment bias [5, 6] and continue to the results, what can we conclude? Hypothermia may be therapeutic for about 9% of patients presenting after out-of-hospital arrest. When designing clinical trials, getting the inclusion and exclusions criteria correct is a question of balance. Maximising the difference in outcome has to be traded against eliminating so many subjects that the study looses all validity. Extrapolating on the assumption that this benefit then exists for the other 91% in ventricular fibrillation and those in any other rhythm, is just poor science, and it is disappointingly so from a national committee. Is hypothermia actually therapeutic in any other similar conditions such as head injury or stroke? Probably not [7] and it increases the risk of infections. Although pyrexia is strongly associated with poor outcome from a range of conditions seen in intensive care, lowering the body temperature by physical or pharmacological means is not beneficial (although widely practiced still!). Could hypothermia harm patients after cardiac arrest? Our experience of peri-operative hypothermia [8] would suggest so. Brief periods of sub-normal temperature adversely influence infection rates, coagulation and nitrogen balance and we now routinely maintain normothermia with active warming in theatres and recovery. While the evidence base is not perfect, it appears consistent and is well accepted. Hay’s audit provides useful data in favour of intensive care admission for patients resuscitated from cardiac arrest and demonstrates that a positive approach can yield favourable outcomes even in those aged over 70 years. My argument is that this is equally possible without deliberate cooling and there is sufficient equipoise for another prospective trial to see if hypothermia deserves the adjective ‘therapeutic’.