Based on existing evidence and expert consensus, a multidisciplinary writing group organized by the American Heart Association (AHA) Emergency Cardiovascular Care Science Subcommittee released recommendations to improve the quality of adult and pediatric neurologic prognostication studies for survivors of cardiac arrest. This report was published in Circulation.
The writing group, which was composed of experts from adult and pediatric neurology, cardiology, emergency medicine, intensive care medicine, and nursing, examined relevant studies and existing practices of neurologic prognostication and withdrawal of life-sustaining treatment (WLST). WLST, often based on poor neurologic prognosis, is a major cause of death for patients resuscitated after cardiac arrest.
Recommendations for Designing Neurologic Prognostication Studies
Most neurologic prognostication studies reported in the cardiac arrest literature used indirect measures of brain injury severity following cardiac arrest. The AHA recommends neurologic prognostication be approached as index tests based on neurologic functions directly related to functional outcomes as well as quality of life.
The literature further reveals that the relationships between predictive variables and outcomes are not linear and may not be restricted to a single approach for optimizing the prediction of outcome; rather, combined findings of multiple predictive modalities should contribute to neurologic prognostication and the likelihood of survival.
Recommendations for Measuring Neurologic Outcomes
Current measures of neurologic function after cardiac arrest include the modified Rankin Scale and the Cerebral Performance Categories. In children, these measures include the Pediatric Cerebral Performance Categories, the King’s Outcome Scale for Childhood Injury, and the Pediatric Stroke Outcome Measure.
The writing group recommends the modified Rankin Scale version for adults as it can better discriminate between mild and moderate post-anoxic brain impairments as well as identify dependency issues related to severe cognitive impairment rather than just locomotor problems. The Pediatric Stroke Outcome Measure is widely used in pediatric studies, but King’s Outcome Scale for Childhood Injury is more sensitive to outcomes in children less than 2 years old and may be used as a complementary measure to Pediatric Stroke Outcome Measure.
Recommendations for Characterizing of Neurologic Outcomes
Dichotomizing neurologic outcomes as “good” or “poor” (based on the thresholds of common outcome measures) results in a loss of granularity and prevents measuring the evolution of individual outcomes over time. This also prevents pooling in meta-analyses due to the heterogeneity of outcome thresholds.
Recommendations for Reporting Cause of Death
Death is used as a neurologic outcome in common outcome measures, but these do not specify the actual cause of death. The AHA recommends the clarification of death in relation to neurologic prognostication by providing the mode of death (brain death or somatic/cardiac death), the extent of medical support with or without DNAR directives, and defining WLST as due to perceived neurologic futility, medicinal futility, or both.
Recommendations for Measuring Quality of Life
Although certain outcome measures are generally reliable to assess health-related quality of life after cardiac arrest, further studies are needed to understand the correlation between quality of life and neurologic function following cardiac arrest. Therefore, quality-of-life measures, including patient-reported and physician-reported outcomes, are recommended in prognostic studies for both adult and pediatric populations.
Recommendations for Timing of Outcome Assessment
In prognostication studies, the appropriate timing to assess the impact of the index test on neurologic outcomes should ensure that neurologic deficit has stabilized and that no associated comorbidities have occurred. Although 30 days after cardiac arrest is the minimum timing for measuring neurologic outcomes, recommended follow-up times longer than 30 days are desirable.
The AHA cautions against collecting neurologic outcomes at hospital discharge because some social functions and complex activities are difficult to assess within the hospital setting. Quality-of-life assessments are recommended at a minimum 3 months after cardiac rest, and again at 6 months and 1 year. Longer follow-up times are often necessary for assessing the youngest patients, especially as their brains continue to develop.
Recommendations for Avoiding Bias
Two major sources of bias identified were self-fulfilling prophecy and sedation or medications. To avoid the former, the treating team should ideally be blinded to results; however, this is not generally feasible, and comatose resuscitated patients should be maintained with full medical support. Establishing a strict protocol for WLST is also recommended, including a description of cause of death in all patients. Sedatives and neuromuscular blocking drugs may interfere with neurologic prognostication; to avoid bias, the use of short-acting drugs is recommended.
Recommendations for Reporting Standards
To achieve substantial quality of evidence, the AHA recommends avoiding indirectness (for example, providing a description of the cause of death), inconsistency (by adopting standardized definitions and terminology), imprecision, and incomplete reporting. Furthermore, adhering to highly suggested reporting standards like Standards for Reporting Diagnostic accuracy studies or Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis is recommended.
Recommendations for Improving Study Outcomes
Future outcome assessments in cardiac arrest survivors should include neurocognitive testing, patient-reported quality-of-life measures, and psychological assessment; in particular, these should be included in studies measuring delayed outcomes (testing times recommended at 90 days or later). Furthermore, there is a need for validated tools, such as those used in stroke or traumatic brain injury studies, which embrace common data elements and standardized outcome testing.
Recommendations for Neurologic Prognostication
Neurologic prognostication of a patient after cardiac arrest requires frequent evaluation and reevaluation. The timing of the index test should adequately predict neurologic prognosis and should ensure adequate time is given for the patient awakening and recovering before decisions to limit care or WLST.
Patient characteristics and clinical factors before cardiac arrest are limited in their ability to establish neurologic prognostication, but variables including age, comorbidities, and lifestyle may impact the study design and interpretation of results. Similarly, the association of intra-arrest factors, such as duration of cardiac arrest, presentation of rhythm, and duration and quality of CPR, with overall survival still remain uncertain.
Recommendations for Post-arrest Evaluation
Assessing the extent of neurologic injury is crucial in neurologic prognostication and requires significant accuracy in all testing, including bedside examination, neurophysiological testing, neuroimaging, and biomarker testing. The AHA recommends testing be performed by well-trained and experienced examiners, and their role and qualification should be described in the study.
Assessment of consciousness can be subject to bias. To avoid this, examiners should use validated scales to quantify cognitive abilities. Brain stem integrity can be evidenced by heart rate variability as well as in the presence or absence of reflexes (such as cough, gag, pupillary, and corneal reflexes). Due to their association with poor prognosis, seizures reported as part of neurologic prognostication studies require the use of electroencephalography to confirm their activity and response to treatment.
Recommendations for Biochemical Markers
Biochemical markers can provide evidence of injury and can be obtained from cerebrospinal fluid (CSF) samples or routine blood sampling. In reporting of blood and CSF biomarkers, investigators must include clear definitions of normal ranges by age. Serial testing, rather than relying on any individual values, is highly recommended as it reflects progression of injury over time.
Recommendations for Neuroimaging
Most neuroimaging modalities, even the most sensitive, are best performed several days after cardiac arrest as there is typically a delay in the imaging appearance of hypoxic-ischemic changes in the brain. Neuroimaging in neurologic prognostication studies should be reported as an objective quantification (volume of injury) or qualitative assessment (based on injury site and related to functional outcomes). Although it is not yet validated for use in neurologic prognostication, it is recommended that neuroimaging tests be performed multiple times to capture and quantify the evolution of brain injury.
Recommendations for Post-arrest Variables
The impact of medications and therapies used in resuscitation and periresuscitation management should be accounted for in neurologic prognostication studies. The extent of organ failure in patients should also be considered in the context of neurologic outcomes, as organ dysfunction is extremely common in survivors of cardiac arrest. The use of extracorporeal membrane oxygenation, which may improve survival in a subpopulation of patients, should establish clearly defined patient selection criteria.
As a special consideration, the AHA recommends that informed consent from a legally authorized representative must be included in the design of neurologic prognostication studies, especially in clinical trials. Families should be adequately prepared to help make care decisions, including WLST, with the clear communication and guidance of healthcare professionals.
In a statement from Romergryko G. Geocadin, MD, Professor of Neurology at Johns Hopkins Hospital and study author notes, “[a]t the current state of affairs, we have to acknowledge the limitations in our practices in this area because we don’t have high-quality science to back our decision-making.” He goes on to remark that, “[w]e owe it to patients and families to ensure we are doing the best to both not prolong unnecessary suffering while balancing that with not withdrawing care too soon if the person has the potential to recover with a reasonably good quality of life.”2
1. Geocadin RG, Callaway CW, Fink EL, et al. Standards for studies of neurologic prognostication in comatose survivors of cardiac arrest: a scientific statement from the American Heart Association [published online July 11, 2019]. Circulation. doi:10.1161/CIR.0000000000000702
2. Better science needed to support clinical predictors that link cardiac arrest, brain injury, and death: a statement from the American Heart Association [news release]. Dallas, TX: American Heart Association; July 11, 2019. https://newsroom.heart.org/news/better-science-needed-to-support-clinical-predictors-that-link-cardiac-arrest-brain-injury-and-death-a-statement-from-the-american-heart-association.