Wait. Hypothermia DOES improve outcomes after cardiac arrest ... right?

Last week's analysis of the randomized and retrospective evidence base for hypothermia after cardiac arrest generated several dissenting emails and comments. Including this especially thoughtful one: “When a therapeutic intervention such as hypothermia shows a statistically significant improvement in neurological outcomes in a recent, independent meta-analysis—and when a benefit cannot be excluded in another meta-analysis—this signal gains further weight if it is consistent with findings from both retrospective human studies and animal research. In the absence of evidence indicating harm to patient-centred outcomes, particularly the capacity for independent living, there is a strong rationale to support its use while continuing to advocate for well-designed, prospective studies to strengthen the evidence base.” References: "https://pubmed.ncbi.nlm.nih.gov/38126249/" and "https://pubmed.ncbi.nlm.nih.gov/38845543/" —Dr. Athanasios Chalkias (@Thanos) Reading this comment gave me an acute pang of the anxiety I periodically get from publishing this newsletter. I recalled reading meta-analyses, but I did not include any in the post. What did they say? Did I leave out something vital from this discussion? Am I going to need to apologize again? ICYMI, a quick recap: Hypothermia as a treatment after out-of-hospital shockable cardiac arrest showed massive benefits in preserving brain in two randomized trials published in NEJM in February 2002. One (HACA, n=275) reported a 23% absolute improvement in good neurologic outcome and a 14% absolute reduction in mortality; the other (Bernard et al, n=77) found a 16% absolute improvement in good neurologic outcomes. In critical care, these are huge, almost unheard-of magnitudes of benefit. Unfortunately, these findings were not replicated in 10 subsequent larger randomized trials. (HYPERION, finding a neurologic benefit from hypothermia for nonshockable rhythms, was the outlier.) The last ongoing hypothermia trial in the U.S., ICECAP, quietly shut down for statistical futility in 2025 after enrollment of about 1,150 patients (short of the intended enrollment of 1,800). Futility was determined in both the shockable and nonshockable rhythm arms of the trial. Altogether, hypothermia was tested in randomized trials enrolling over 9,140 patients. And after 2002 (other than HYPERION), it didn’t work. But what if it did? What if it does? Bring in the Meta-Analysts Read more