Too Old for Cold? Editorial Commentary on Busch and Søreide,2011

Subgroups kill people (Peto, 1990). The slicing and dicing of clinical trials continues, as bedside clinicians seek to “tailor” therapy to the presumed unique constellation of risk factors, comorbidities, genetics, and circumstances in their patients (Caplan, 2007). The arrogance of individualizing therapy seems limitless, as clinicians seek to identify particular subgroups that either benefit more or less than the study group, or suffer harm more or less often. The search for target subgroups leads the casual reader to scrutinize inclusion/exclusion criteria, as well as post hoc subgroup analyses, hoping to find in the tea leaves an indication for tailoring therapy in individuals. In fact, subgroups—by definition—are smaller than the full study population, and thus underpowered to answer the question at hand. Subgroup analysis will lead to erroneous conclusions, leading to either overzealous treatment or withholding of treatment and can lead to spurious conclusions (Peto, 1990; Yusuf et al., 1991). There are, however, powerful methods that can be used for finding hidden effects apart from the main study results (Lu et al., 2005).

A different problem arises in extending clinical trial results to patient groups that were not included in the original definitive studies. Of the two original studies of therapeutic hypothermia (TH) for cardiac arrest, for example, the larger trial excluded patients over the age of 75 years and the other was too small to allow any conclusions about TH in the elderly (The Hypothermia After Cardiac Arrest Study Group, 2002; Bernard et al., 2002). In the absence of evidence, therefore, guidelines for TH after cardiac arrest contain no upper age limit (Nolan et al., 2003). In 2011, Busch and Soreide (2011) published a large retrospective analysis of TH in comatose survivors of out-of-hospital cardiac arrest. The trial contained 113 patients with a median age of 62 years. There were 23 patients aged 70 to 79 years and 11 patients older than 80 years. While survival with good function was lower in the older cohort, it was better than historical control data for cardiac arrest. Logistic regression analysis showed that good outcome was more likely in the young, but still present in the old. The data did not identify any age cutoff after which TH could be shown ineffective.

Certainly, the best way to determine efficacy of TH in the elderly would be to perform a randomized trial. Pending those data, however, how do we interpret both the original studies and the retrospective analysis of elderly patients? How do we make a decision in real time at the bedside? On the one hand, the bedside clinician could simply refuse to treat any patient that does not strictly fit the published inclusion/exclusion criteria. This approach might be termed the “strict evidence” approach. Alternatively, the clinician could call upon previous experience and knowledge of the individual patient's history and physical condition, then use his or her judgment to tailor a therapy. This approach might be termed the “best hunch” approach. Neither approach seems entirely satisfactory, however, so the rational clinician could approach the dilemma with a bit more thought, as follows.

First, consider whether there is any evidence of harm in the elderly due to TH. In the Intravascular Cooling Trial of Ischemic Stroke (ICTuS-L) trial, risk of pneumonia was associated with TH, but age did not seem to be a predisposing factor (Lyden et al., 2012; Hemmen et al., 2010). Other TH studies have not shown an age effect in assessing risk of harm, but to be fair, very few studies enrolled significant numbers of patients over age 80. So the answer at present appears to be no, there is no compelling evidence that using TH in the elderly is clearly associated with increased risk, but there is a paucity of data. Second, consider whether there is any compelling biological reason to think the therapy will not work in the elderly. Experimental studies in aged animals are seldom performed, but there is no molecular or cellular mechanism occurring in the young that does not exist in the elderly, although many pathological processes run more slowly or to a lesser degree in the elderly. So on balance, if there is no compelling evidence of harm, and no biological reason why TH would not work in the elderly, then it's prudent to consider using TH. In assessing risk in the elderly, the case series by Busch and Soreide is helpful in that TH did not appear to cause harm.

The gratifying response rate in this case series, however, is difficult to interpret because of secular improvements in the care of cardiac arrest patients. Improvement in streetside cardiopulmonary resuscitation (CPR) correlates with improved perfusion and better survival and presumably influenced the care of patients in this study (Adielsson et al., 2011). Deployment of automated defibrillators have also significantly improved outcomes after cardiac arrest. The authors do not comment on these secular trends, but presumably automated defibrillators and improved CPR would have been deployed in the study community as rapidly as in other European and North American countries. Nevertheless, at least a portion of the good outcomes in this case series could be due to widespread use of TH in this community.

Further trials—randomized, blinded, controlled—are needed before TH can be routinely recommended for the elderly victim of cardiac arrest. In selected cases, however, the data presented by Busch and Soreide reassure the bedside clinician that there is no obvious risk to TH in older victims of cardiac arrest.