Kim Annink

86 Chapter 4 hyperechogenicity of the cortex within 72 hours (sensitivity 76.9%, specificity 100%). Additional lesions, not identified by CUS, were found in the brain stem, hippocampus and cerebellum with histology (21). These three items were not included in our CUS scoring system. The “four column sign” and visibility of the PLIC were included as separate items because in some infants the PLIC was visible, but there was no clear “four column sign”. The CUS conducted on day one after birth was not predictive of outcome in this study, which is in agreement with previous studies (17, 22). As mentioned above, it takes 24 to 72 hours before brain injury becomes visible as hyperechogenicity on CUS, unless the onset of the injurious process is of antenatal onset (21) . Consequently, CUS within 6 hours after birth had in a previous study a low sensitivity of 42.1% and specificity of 60% (23). Nevertheless, CUS on day one is recommended to identify antenatally acquired pathology (5, 10, 11, 32). We indeed found antenatally acquired pathology in 14 of the 83 infants in cohort I. Most of the antenatally acquired lesions, i.e. germinal layer cysts, did not influence outcome and can also be found in controls (17). However, in one infant a porencephalic cyst was found that led to a mildly asymmetrical motor outcome. It was of interest to see that the RI was not associated with adverse outcome in multivariable logistic regression analyses. The RI was highly predictive in previous studies with non-cooled infants, but appears to be less predictive in cooled infants, as reported previously (20, 21, 31, 33). Especially the positive predictive value has decreased, so the observed outcome is better than the expected outcome based on an abnormal RI. It has been hypothesized that hypothermia has a direct effect on the cerebral vessels or that hypothermia leads to a better neurodevelopmental outcome but does not lead to a normalization of the RI (33). The correlations between MRI and CUS in a study of Leijser et al. were stronger than in our cohort (0.83 versus 0.67) (26). This might be explained by the fact that they used exactly the same scoring systems for MRI and CUS and in our cohort the scoring systems differed i.e. the cerebellum was included in the MRI scoring system but not in the CUS scoring system. In cohort I there was a moderate correlation between MRI and CUS, but predictive values for MRI in the study of Weeke et al. were higher than for CUS in our study (13). The most severely affected infants were too ill to undergo