DOI: 10.1097/CCM.0000000000000394

,  

,  

PMID: 25029143

Issn Print: 0090-3493

Publication Date: 2014/08/01


Print

Prognostication Following Cardiac Arrest: Do We Have Our Patients’ Safety in Mind?*

Romergryko G. Geocadin; Santosh B. Murthy

    loading  Checking for direct PDF access through Ovid

Excerpt

The practice of prognostication in post cardiac arrest (PCA) patients has grave importance particularly in the era where resources are limited and cost containment is emphasized. Prior to the introduction of therapeutic hypothermia (TH) in comatose survivors of cardiac arrest, a clinical practice guideline advocates for 72 hours postarrest as an acceptable time cutoff for prognostication of neurological recovery (1). The introduction of TH shifted the paradigm as it improved outcomes and increased attention to the postarrest care. TH has also renewed the interest in prognostication (2). Hence, developing validated prognostic tools and establishing the optimal timing and predictive accuracy of tests used in response to the changes to outcomes that TH has caused is one of the most pressing research needs in postresuscitation care (3).
In this issue of Critical Care Medicine, Golan et al (4) present a meta-analysis on predicting neurological outcome following TH PCA that included 20 studies with 1,845 patients. Measures of poor neurological outcome were Glasgow outcome score (GOS) of 1–3 or Glasgow-Pittsburgh cerebral performance categories (CPC) score of 3–5. The study assessed several validated tools such as clinical tests, electroencephalogram, somatosensory-evoked potentials (SSEPs), neuroimaging, and serum biomarkers such as neuron-specific enolase and Tau. The authors concluded that bilateral absence of pupillary reflexes more than 24 hours after return of spontaneous circulation was the best predictor of outcomes with a false-positive rate (FPR) of 0.02 (95% CI, 0.01–0.06), whereas bilateral absence of corneal reflexes was the second best predictor (FPR of 0.04; 95% CI, 0.01–0.09). SSEPs accurately prognosticated neurological recovery especially when performed in the first week post arrest (FPR, 0.03; 95% CI, 0.01–0.07). Notably, the authors cautioned that the specificity of available tests improved when these were performed beyond 72 hours. Moreover, they warned that clinicians should use caution with these predictors as they carry the inherent risk of becoming self-fulfilling.
Several meta-analysis on this topic have been published (5, 6), but some of the strengths of the study by Golan et al (4) include the comprehensive inclusion of relevant observational, retrospective, and prospective studies, along with formal appraisal of the quality of evidence. One of the highlights of this study is the standardized approach with regard to timing of the tests. By contrast, the meta-analysis by Sandroni et al (5) examined the role of different tests at different time points, making its application in clinical practice less feasible. Furthermore, Golan et al (4) included predictors of neurological outcome within the first 14 days, as opposed to 7 days by Sandroni et al (5), to minimize the confounding effect of sedation and paralytics. Another meta-analysis by Kamps et al (6) that included 10 studies also reported similar findings regarding the diagnostic accuracy of pupillary reflexes and somatosensory potentials when performed within 72 hours. Despite the similarities in the sensitivity analyses, the study by Golan et al (4) had narrower CIs due to the inclusion of more studies and correspondingly a larger sample size. As we compare the three meta-analyses, all three studies provide some serious caution regarding their observations and recommendations.
Critical evaluation of published literature on neuroprognostication raises a few noteworthy questions. First, on timing, are we justified in using the 72-hour cutoff for arriving at a prognosis in patients treated with TH? A recent study of 194 patients reported that the time to awakening after PCA was variable and often longer than 72 hours (7).

Related Topics

    loading  Loading Related Articles