Maayke Hunfeld

33 Review neuromonitoring A summary of four studies including more than 20 patients is given below. The other study can be found in Supplement Table 5 (Supplemental Digital Content 2, http:// links.lww.com/PCC/B352) (30). A retrospective analysis of MRIs post CA in 22 children after near-drowning concluded a strong significant correlation between high signal on T2 weighted images (cortical and basal ganglia) or the presence of edema (occipital or generalized) at day 3-4 and outcome. The PPV and negative predictive value (NPV) for abnormal MRI for a poor outcome was 100% (T2 abnormalities cortex or basal ganglia, brain stem infarction or generalized or occipital edema) (31). In a retrospective cohort of 28 patients who underwent brain MRI after CA, an association was found between signal abnormalities in multiple brain lobes on T1/ T2 or diffusion weighted imaging (DWI) and worse outcome (p<0.01 and P=0.02 respectively) (32). Another retrospective study of 20 patients after CA showed an association of DWI abnormalities in cortex, basal ganglia or cerebellum with poor outcome (p<0.05). All five patients with a normal DWI on MRI had a good outcome (p=0.05) (33). A recent study of Yacoub et al (2019) showed that in 26 pediatric patients post CA remaining comatose or with neurological deficits, specific quantitative values of DWI MRI correlate with outcome. An apparent diffusion coefficient (ADC) threshold of less than 600 x 10 -6 mm 2 /s in greater than or equal to 7% of brain volume and less than 650 x 10 -6 mm 2 /s in greater than or equal to 11% of brain volume both showed a specificity of 1.0 and a sensitivity of 0.8 for poor outcome (34). Limitations of these neuroimaging studies are multiple. All studies included small patient groups (median n=22, IQR 17-27), were single center and the majority were retrospective (31-33). The study of Dubowitz et al dates from 1998 when DWIs were not available (31). Yacoub et al used specific quantitative values of DWI MRI and tried to determine ADC thresholds in order to predict outcome. However, validation of the two found thresholds is impossible due to a small and retrospective cohort (34) and variability in these measurements between different MRI scanners. At least four studies only performed MRI when clinically indicated by the treating clinicians, this might have caused a selection bias and influenced the decision process to continue or withdraw treatment (30, 32-34). Temperature management was different in the studies: three studies did not implement TH (30, 31, 34) and in one study the initiation of TH was at the discretion of the treating clinician (32). Different gross outcome scales were used with varying follow-up intervals. When outcome was non-survival, the cause was not clarified. Long-term follow-up studies with neuropsychological tests, evaluation of quality of life and social participation are lacking. Finally, the timing of MRI was different within the cohorts (ranging from 2