Does the evidence support the use of mild hypothermia after cardiac arrest? NoBMJ 2011; 343 doi: https://doi.org/10.1136/bmj.d5889 (Published 23 September 2011) Cite this as: BMJ 2011;343:d5889
- Andrew P Walden, consultant in critical care medicine1,
- Niklas Nielsen, consultant in critical care medicine2,
- Matt P Wise, consultant in critical care medicine 3
- 1Intensive Care, Royal Berkshire Hospital, Reading, UK
- 2Department of Anaesthesiology and Intensive Care, Helsingborg Hospital, Helsingborg, Sweden
- 3Adult Critical Care, University Hospital of Wales, Cardiff CF14 4XW, UK
- Correspondence to: M P Wise
Historically, critical care physicians had a nihilistic approach towards patients who remained unconscious after a cardiac arrest outside hospital. This changed with the publication of two randomised clinical trials of mild induced hypothermia (32-34°C) that showed neuroprotection.1 2 Subsequently this treatment has been embraced by the International Liaison Committee on Resuscitation, European Resuscitation Council, American Heart Association, and, most recently, the National Institute for Health and Clinical Excellence (NICE).
Animal models of cardiac arrest showed that mild hypothermia improved neurological outcome,3 and these data were supported by small observational studies in patients. Clinical trials to determine whether this treatment benefited unconscious patients after cardiac arrest were therefore fitting. However, neither the above randomised trials1 2 nor subsequent studies4 provide sufficiently robust data to justify the conclusion that cooling to 32-34°C should be used after cardiac arrest outside hospital.
Evidence from clinical trials
A search for the terms “cardiac arrest” and “hypothermia” in PubMed identifies over 1800 publications since 2002, but they are almost all reviews, expert opinion, registries, and observational studies. Systematic review and meta-analysis, including a Cochrane review by Arrich and colleagues5 concluded that mild hypothermia should be used after out of hospital cardiac arrest. Arrich and colleagues identified five published randomised trials, including one in abstract form,1 2 4 6 7 and concluded that mild hypothermia was beneficial. However, the review did not rigorously evaluate the risks of random error, design flaws, and high risk of bias in these trials and failed to use the GRADE system8 to assess evidence quality. This may have resulted in an overestimation of the treatment effect.
The largest clinical trial to date, undertaken by the Hypothermia After Cardiac Arrest Group,1 recruited an average of just over one patient a week from nine centres over five years. Only 275 patients were randomised from 3551 screened, and this low inclusion rate of around 8% makes it difficult to generalise results to daily clinical practice. The study was discontinued because of slow recruitment and a lack of funding rather than because of defined stopping rules, and, importantly, there was no predefined power calculation. The level of coma before randomisation was not reported, and withdrawal of critical care was not standardised, introducing potential bias to the primary outcome measures of neurological outcome and death.9
The four other randomised clinical trials included in the Cochrane review also had methodological problems. Examples include quasirandomisation with odd and even dates,2 early stopping without predefined rules,4 unplanned adaptive design,2 baseline differences between groups,2 4 6 7 selective outcome reporting, and no description of sequence generation and allocation concealment6 7 or blinding.4 6 7 Reporting of adverse outcomes is also inconsistent, making it difficult to assess the harm from this treatment. Recognised adverse effects include increased risk of infection, haemodynamic instability, arrhythmias, coagulopathy, hyperglycaemia, and electrolyte abnormalities.10 11 In one prospective observational registry based study of 765 patients treated with hypothermia after cardiac arrest outside hospital adverse events were common and included pneumonia (48%), electrolyte imbalance (37%), seizures (24%), arrhythmias (14%), bleeding (6%), and sepsis (4%).11
Recently Nielsen and colleagues12 conducted a systematic review and meta-analysis of hypothermia after cardiac arrest and identified 478 patients from the same five trials1 2 4 6 7 included in the Cochrane review.5 They systematically evaluated the benefits and harms of the intervention, taking into account risk of systematic bias and random errors. Treatment effects were quantified using meta-analyses and trial sequential analysis, which reduces the risk of type I errors in cumulative meta-analysis.13 The authors concluded that there was a lack of firm evidence of a beneficial effect of mild hypothermia and that using the GRADE system the quality of evidence was low. The recent NICE guideline did not consider this systematic review, even though it was published before the evidence synthesis was conducted.
Experimental data show that mild hypothermia may be neuroprotective after cardiac arrest.3 However, the risks and benefits of hypothermia in an animal that is healthy before an experimental cardiac arrest are not the same as those in patients with vascular disease and multiple comorbidities. Adverse events relating to hypothermia have been poorly studied in people who have had a cardiac arrest and need to be examined in future clinical trials. An international, multicentre randomised controlled trial of temperature management in unconscious survivors of out of hospital cardiac arrest (randomised to 33°C or 36°C) is underway. The study overcomes many of the methodological problems of previous trials, including standardised withdrawal, blinded assessment, rigorous evaluation of adverse events, defined stopping rules, and powered for a primary outcome measure of survival.14
Data from current clinical trials are not sufficiently robust to justify the conclusion that mild hypothermia should be used routinely in unconscious survivors of out of hospital cardiac arrest. We need data from adequately designed and powered studies. Until these are available, the recommendations must be regarded as weak and should not be allowed to inhibit further research into the effects of temperature control.
Cite this as: BMJ 2011;343:d5889
Competing interests: All authors have completed the ICJME unified disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare no support from any organisation for the submitted work; APW and MPW have attended an advisory board meeting with Bard in the previous three years; APW and MPW are investigators, and NN is chief investigator for the TTM trial; they have no other relationships or activities that could appear to have influenced the submitted work.
Provenance and peer review: Commissioned; not externally peer reviewed.