Maayke Hunfeld

231 General discussion background pattern at 12 hours post-CA are associated with a good outcome (2, 6). These EEG studies, however, are limited in several aspects. First, all were single-center studies; and most of them were retrospective studies with small sample sizes (median n = 34; IQR 28–56). Second, clinicians were not blinded for the EEG results; this knowledge may have influenced their decisions to withdraw treatment if EEG patterns were abnormal. Third, mainly short-term outcomes were presented, established with crude outcome measures such as Pediatric Cerebral Performance Category (PCPC) and Glasgow Outcome Scale (GOS). Many studies did not specify causes of death. For example, it whether the cause was refractory circulatory failure withdrawal of life-sustaining therapies (WLST) based on a poor neurological prognosis. Fourth, the studies used different EEG classifications – and, therefore, are hard to compare. Fifth, the timing of the EEG was not always specified. This should be taken into account, as studies in adults after CA suggest that EEG patterns at 12 hours after ROC have the best predictive value for good outcome, while patterns at 24 hours post-ROC have the best predictive value for poor outcome. Of note, although a normal EEG is associated with a good outcome, some studies described children with a normal EEG and poor outcome and vice versa (1, 3). The above-described review gave insight into the challenges of outcome studies after pediatric CA. Most importantly, a child’s brain – and thus the EEG patterns - are continuously developing and changing, especially at early age (7). The wide age range in pediatrics (1 month to 17 years) hampers meaningful outcome comparison. In children with brain damage, growing into deficit can occur later in life. Second, additional investigations (e.g. EEG, MRI) were often only done when clinically indicated, and clinicians were unblinded to the results. This could gave led to selection bias and self-fulfilling prophecies in relation to patient outcome (e.g. decision-making regarding WLST). Third, the majority of neuromonitoring studies included both in-hospital cardiac arrest (IHCA) and OHCA with different etiologies, resulting in heterogeneous patient groups. Finally, in the majority of studies, outcome was measured using gross outcome scales (e.g. PCPC, GOS). In chapter 5. we have shown that a normal brain MRI (without post-hypoxic injury ) on T1/T2 and diffusion weighted imaging (DWI) within 1 week after pediatric OHCA is 100% predictive for a good neurological outcome at 2 years post-OHCA. Conversely, the presence of extensive injury injury ( ≥ 50% of the cortex/white matter or in 4 or more defined brain regions (with or without involvement of deep grey matter)) on T1/T2 and DWI/apparent diffusion coefficient (ADC) is 100% predictive for a poor 8