Conflict of Interest Declaration
The ILCOR Continuous Evidence Evaluation process is guided by a rigorous ILCOR Conflict of Interest policy. The following Task Force members and other authors were recused from the discussion as they declared a conflict of interest: (none applicable)
The following Task Force members and other authors declared an intellectual conflict of interest and this was acknowledged and managed by the Task Force Chairs and Conflict of Interest committees:
Dr Topjian was an author and investigator in pediatric post-cardiac arrest neuro-prognostication studies.
Dr Scholefield was an author and investigator in pediatric post-cardiac arrest neuro-prognostication studies and received UK NIHR funding for research into post-cardiac arrest neuro-prognostication research.
Dr Rodriguez-Nunez was an author and investigator in pediatric post-cardiac arrest neuro-prognostication studies.
Barnaby R Scholefield, Janice Tijssen, Saptharishi Lalgudi Ganesan, Mirjam Kool, Alexis Topjian, Thomaz Bittencourt Couto, Anne-Marie Guerguerian on behalf of the International Liaison Committee on Resuscitation Pediatric Life Support Task Force. Brain imaging for the prediction of survival with good neurological outcome after return of circulation following pediatric cardiac arrest Consensus on Science with Treatment Recommendations [Internet] Brussels, Belgium: International Liaison Committee on Resuscitation (ILCOR) Pediatric Life Support Task Force, 2022 XXXX. Available from: http://ilcor.org
Methodological Preamble and Link to Published Systematic Review
The continuous evidence evaluation process for the production of Consensus on Science with Treatment Recommendations (CoSTR) started with a systematic review (Scholefield, 2021, PROSPERO CRD42021279221)conducted by the members of the PLS TF with involvement of clinical content experts. Evidence for pediatric literature was sought and considered by the Pediatric Life Support Task Force. Additional scientific literature was published after the completion of the systematic review and identified by the Pediatric Task Force, and is described before the justifications and evidence to decision highlights section of this CoSTR. These data were taken into account when formulating the Treatment Recommendations.
Scholefield B et al. Brain imaging for the prediction of survival with good neurological outcome after return of circulation following pediatric cardiac arrest (in preparation)
The PICOST (Population, Intervention, Comparator, Outcome, Study Designs and Timeframe)
Population: This review is studying children (<18 years) who achieve a return of spontaneous or mechanical circulation (ROC) after resuscitation from in-hospital cardiac arrest (IHCA) and out-of-hospital (OHCA), from any cause.
Studies which include newly born infants or patients in hypoxic coma from causes other than cardiac arrest (e.g., respiratory arrest, toxidromes, drowning, hanging) will be excluded, except when a subpopulation of cardiac arrest patients can be evaluated separately.
Intervention: Index prognostic tests, recorded less than 12 hours, 12 to <24 hours, 24 to <48 hours, 48 to <72 hours, 72hrs to <7 days, and/or 7 to10 days after cardiac arrest and will include:
Imaging: Neuroimaging modalities which included head computer tomography (CT), brain magnetic resonance imaging (MRI), cranial ultrasound or trans-cranial doppler (TCD) ultrasound.
Comparators: There is no control group for intervention/exposure. However, the accuracy of the prognostic (index) test will be assessed by comparing the predicted outcome with the final outcome, which represents the comparator.
Outcomes: Primary outcome of interest is survival with good neurological outcome*.
*Good neurological outcome is defined as a Pediatric Cerebral Performance Category (PCPC) score of 1, 2 or 3, or Vineland Adaptive Behavioural scale-II ≥ 70. PCPC score ranges 1 (normal), 2 (mild disability), 3 (moderate disability), 4 (severe disability), 5 (coma), and 6 (brain death). We will also separately report studies defining good neurological outcomes with other assessment tools, or as a PCPC score 1 or 2, or change in PCPC score from baseline ≤2.
Outcome time point(s) will include hospital discharge, 30 days, 60 days, 180 days and/or 1 year.
Study Designs: Randomized controlled trials (RCTs) and non-randomized studies (non-randomized controlled trials, interrupted time series, controlled before-and-after studies, cohort studies) were eligible for inclusion. Case series were considered if greater than 5 cases reported. Unpublished studies (e.g., conference abstracts, trial protocols*) and animal studies were excluded. We selected studies where the sensitivity and false-positive rate (FPR) of the prognostic (index) test are reported.
*please note that the search for unpublished trials was limited to a comprehensive search of three clinical trial registries for unpublished completed trials.
1. International Clinical Trials Registry Platform (www.who.int/ictrp/en/)
2. US clinical trials registry (www.ClinicalTrials.gov)
3. Cochrane CENTRAL (http://www.cochranelibrary.com/about/central-landing-page.html)
Timeframe: All years and all languages were included as long as there was an English abstract; unpublished studies (e.g., conference abstracts, trial protocols) were excluded. Literature search updated to Feb 17th 2022.
PROSPERO Registration CRD42021279221
Consensus on Science
We defined good neurological outcome prediction as imprecise when the false positive rate (FPR) was above 30%. However, there is no universal consensus on what the acceptable limits for imprecision should be in prediction for infants and children after cardiac arrest.
A low false positive rate means that a low proportion of patients, predicted to have a good outcome will have a falsely optimistic prediction (test predicted a good outcome, but patient went on to have a bad outcome). The task force felt that when focused on accuracy of predicting a good outcome - a low false positive rate (eg <30%) is more desirable to avoid falsely optimistic prediction than a high sensitivity. The cut off of 30% FPR (equivalent to 70% specificity) was chosen as the consequences of false optimism were felt by the task force to be less critical than false pessimism. False optimism may result in continued life sustaining therapy in a patient who will eventually have a poor outcome. This will involve increased resources and treatment; however, may also allow more time for further prognostic evaluation. Also, reasons for not achieving a very low false positive rate may be non-neurological causes of poor outcome or death, not attributable to the index test assessment.
A high sensitivity means the majority of patients, who have a good outcome, tested positive and therefore a corresponding low proportion will have a falsely pessimistic prediction (test predicted a poor outcome, but patient went on to have a good outcome). When considering the accuracy of predicting a poor outcome (compared to predicting a good outcome), then a low rate of falsely pessimistic predictions is very important. Our cut off threshold for considering precise sensitivity was therefore higher (>95%), as the consequences of inaccurate poor outcome prediction (e.g. false pessimism) may lead to a decision to limit or withdraw life sustaining therapies in a patient who could have a good neurological outcome.
The overall quality of evidence was rated as very low for all outcomes primarily due to a very serious risk of bias, assessed using the QUIPS tool. The individual studies were all at a moderate to high risk of bias due to confounding. Because of this and a high degree of heterogeneity, no meta-analyses could be performed.
Computer Tomography (CT) Imaging
Head CT was evaluated in three studies and reported the relationship to good neurological outcome (PCPC 1 to 3) in 173 patients [Fink 2014 664, Starling 2015 542, Yang 2019 223]. The majority of CT imaging was acquired at 24 h or 48 h after the cardiac arrest. Neurological outcome was assessed on discharge from the intensive care unit or hospital in two studies and at six months in one. Reported factors from CT included presence and absence of intracranial hemorrhage, cerebral oedema or ischemia measured by the ‘reversal sign’, grey white matter (GWM) differentiation and sulcal or basal cistern effacement. Two studies described methods of estimating GWM differentiation [Starling 2015 542, Yang 2019 223] and two reported radiologists qualitative reports [Fink 2014 664, Starling 2015 542].
The presence of GWM differentiation on CT at 24 hours, had a sensitivity of 64-100%, and FPR 35-70%. Absence of CT lesions, oedema, or intracranial hemorrhage predicted good neurological outcome with a sensitivity ranging 72-100%; however, a wide range of FPR (14% to 90%) was reported.
Absence of effacement of sulci or basal cisterns predicted good neurological outcome with a high sensitivity (93-100%) with a FPR 32-73%.
Clinicians were not blinded to the CT results in any study.
Magnetic Resonance Imaging (MRI)
MRI imaging was reported in four studies, including 215 patients, to predict good neurological outcomes [Fink 2013 31, Fink 2020 185, Kirschen 2021 e719, Yacoub 2019 103]. Median time from ROC to MRI ranged 3 to 6 days across all studies with inclusion of patients MRI up to 14 days reported in three studies [Fink 2013 31, Kirschen 2021 e719, Yacoub 2019 103]. Two studies reported presence or absence of abnormalities in multiple regions of the brain in 3 sequences (Diffusion weighted imaging (DWI), T1 and T2) [Fink 2013 31, Fink 2020 185]. Another study presented a composite of presence or absence of 1 (or more) region of abnormality [Kirschen 2021 e719]. Apparent diffusion coefficient (ADC) thresholds cut off values and overall qualitative MRI reporting of evidence of hypoxic ischaemic injury was assessed by one study [Yacoub 2019 103]. Three studies ensured the neuroradiologists MRI assessment was blinded to patient clinical status. However, the MRI findings were known by the treating clinicians and neurological outcome assessment was not blinded. [Fink 2013 31, Fink 2020 185, Kirschen 2021 e719].
Absence of any region of abnormality on restricted diffusion, at a median of 4 days after ROC, predicted good neurological outcome with a sensitivity of 88% and corresponding very low FPR 2% [Kirschen 2021 e719]. ADC threshold above >600 x10 power -6 mm2/s in >93% and >650 x10 power -6mm2/s in >89% of brain volume, at a median of 4 days after ROC, predicted good neurological outcome with a sensitivity of 100% and low FPR (20%) [Yacoub 2019 103]. In the same study, a normal MRI by qualitative reporting of absence of hypoxic ischaemic injury, predicted a good neurological outcome at 6 months with a sensitivity of 81% and FPR of 10% [Yacoub 2019 103].
For individual regions of the brain, at 4-6 days post ROC, DWI MRI sequence had a sensitivity for predicting good neurological outcome ranging 67-100% although associated FPR rates were moderate to high. Absence of lesions in the Lentiform regions on T2 weighted imaging had a sensitivity of 67% and the lowest FPR (7.7%) for any single region of the brain.
Transcranial Doppler Ultrasound (TCD)
The prediction of good neurological outcome using presence of flow velocities of intracranial vessels measured on Transcranial Doppler Ultrasound (TCD) was evaluated in only one study including 17 patients who were all treated with hypothermic targeted temperature management [Lin 2015 182]. Flow patterns without any reversal (or absence of diastolic) flow, mean flow velocity and pulsatility index were assessed before, during and after hypothermia therapy. Continuous flow velocities without reversal of diastolic flow pattern had a sensitivity of 100% and FPR of 44%. Within 1 hour of the event in the pre-hypothermia phase, mean flow velocity had a sensitivity for good neurological outcome of 38% and FPR of 0%, and having a normal pulsatility index had a sensitivity of 38% and FPR of 22%. In the hypothermia phase, mean flow velocity had a sensitivity of 25% and FPR of 11%; Pulsatility index had a higher sensitivity of 100%, and FPR 22%. By 72 hours, normal pulsatility index predicted a good outcome with 88% sensitivity and 11% FPR.
Clinicians were not blinded to the TCD results in this study.
We identified no studies examining the role of cranial ultrasound and good neurological outcome after cardiac arrest in children.
All evaluated tests were used in combination with other tests by clinicians in these studies. Although the predictive accuracy of tests were evaluated individually, we recommend that no single test should be used in isolation for prediction of good neurological outcome (good practice statement).
- We suggest against using normal CT imaging at 24 to 48 hours from return of circulation for predicting good neurological outcome (weak recommendation, very-low-certainty evidence).
- We suggest using normal MRI between 72 hours and 2 weeks after return of circulation for predicting good neurological outcome (weak recommendation, low-certainty evidence).
- We cannot make a recommendation for or against the use of transcranial doppler for predicting good neurological outcome (weak recommendation, very-low-certainty evidence).
Justification and Evidence to Decision Framework Highlights
- The Task Force considered the use of individual imaging tests to help the clinician in predicting a good neurological outcome. This assessment is different to predicting a poor neurological outcome, which may involve consideration of withdrawal of life sustaining therapies. Recommendations for or against tests to predict good neurological outcomes cannot be transferred to recommendations for poor outcome prediction.
- The available scientific evidence had a high risk of bias based on high heterogeneity across studies, small number of studies and small number of patients included in addition to lack of blinding, variation in test assessment and performance, and variability in outcome measurement. Therefore, no meta-analysis was performed. Overall assessment of test performance was based on visual assessment of forest plots.
- If only one study was available (with small patient sample size) then a suggestion or recommendation could not be made.
- The low false positive rate (high specificity) for normal MRI on global assessment for predicting good neurological outcome reduces the chance of false optimism if a normal MRI predicts a good neurological outcome.
- The sensitivity of a normal MRI or CT to predict a good neurological outcome is moderate to high, but up to 30% may be falsely categorized and a falsely pessimistic prediction made. Therefore, with the very-low certainty of evidence, we cannot make a recommendation for or against the use of normal or abnormal MRI or CT for predicting poor neurological outcomes. However, we encourage further research in this area as these modalities appear promising.
- The precision of MRI and CT is affected by the timing of the investigation and is at risk of pseudonormalization.
- The definition of a presence DWI or cut off values for ADC level on MRI, or GWR on CT was inconsistent in the included studies.
- MRI and CT are both expensive tests and require specialist equipment, training, interpretation and most often, patient transport to obtain the information. This may be prohibitive in physiologically unstable patients, or some health care settings.
- Neuro-imaging for prognostication after cardiac arrest appears promising but more research is required in infants and children.
- A standardization of definitions and assessment of optimal thresholds for GWR calculation on CT, and DWI, ADC thresholds on MRI is needed.
- The optimal timing for prognostication using CT and MRI after cardiac arrest is still unknown. Studies assessing serial imaging after cardiac arrest are desirable.
- The role of assessing regional areas of the brain for predicting outcome, or the use of Magnetic Resonance Spectroscopy requires further research.
- Economic cost evaluation and cost-effectiveness studies are required as CT and MRI is an expensive diagnostic and prognostic modality.
- Further work on multi-modal prognostication, timing, definitions of testing, accurate outcome timing and definition.
- We encourage wider research and consultation with patients, children, parents, guardians and caregivers, health care professionals and members of the wider society on understanding survivorship after pediatric cardiac arrest to inform correct definitions and framework of good neurological outcome for prediction research.
FINK, E. L., PANIGRAHY, A., CLARK, R. S. B., FITZ, C. R., SITTEL, D., KOCHANEK, P. M. & ZUCCOLI, G. 2013. Regional brain injury on conventional and diffusion weighted MRI is associated with outcome after pediatric cardiac arrest. Neurocritical Care, 19, 31-40.
FINK, E. L., BERGER, R. P., CLARK, R. S. B., WATSON, R. S., ANGUS, D. C., RICHICHI, R., PANIGRAHY, A., CALLAWAY, C. W., BELL, M. J. & KOCHANEK, P. M. 2014. Serum biomarkers of brain injury to classify outcome after pediatric Cardiac Arrest*. Critical Care Medicine, 42, 664-674.
FINK, E. L., WISNOWSKI, J., CLARK, R., BERGER, R. P., FABIO, A., FURTADO, A., NARAYAN, S., ANGUS, D. C., WATSON, R. S., WANG, C., CALLAWAY, C. W., BELL, M. J., KOCHANEK, P. M., BLUML, S. & PANIGRAHY, A. 2020. Brain MR imaging and spectroscopy for outcome prognostication after pediatric cardiac arrest. Resuscitation, 157, 185-194.
KIRSCHEN, M. P., LICHT, D. J., FAERBER, J., MONDAL, A., GRAHAM, K., WINTERS, M., BALU, R., DIAZ-ARRASTIA, R., BERG, R. A., TOPJIAN, A. & VOSSOUGH, A. 2021. Association of MRI brain injury with outcome after pediatric out-of-hospital cardiac arrest. Neurology, 96, e719-e731.
LIN, J. J., HSIA, S. H., WANG, H. S., CHIANG, M. C. & LIN, K. L. 2015. Transcranial Doppler ultrasound in therapeutic hypothermia for children after resuscitation. Resuscitation, 89, 182-7.
STARLING, R. M., SHEKDAR, K., LICHT, D., NADKARNI, V. M., BERG, R. A. & TOPJIAN, A. A. 2015. Early head CT findings are associated with outcomes after pediatric out-of-hospital cardiac arrest. Pediatric Critical Care Medicine, 16, 542-548.
YACOUB, M., BIRCHANSKY, B., MLYNASH, M., BERG, M., KNIGHT, L., HIRSCH, K. G., SU, F. & REVIVE INITIATIVE AT STANFORD CHILDREN'S, H. 2019. The prognostic value of quantitative diffusion-weighted MRI after pediatric cardiopulmonary arrest. Resuscitation, 135, 103-109.
YANG, D., RYOO, E. & KIM, H. J. 2019. Combination of early EEG, brain CT, and ammonia level is useful to predict neurologic outcome in children resuscitated from cardiac arrest. Frontiers in Pediatrics, 7, 223.