The MRI score described by Rutherford was used (Appendix 1). The separate areas of interest were scored in three categories: (1) no abnormalities (no abnormalities on T1- and T2-weighted images), (2) mild abnormalities (focal regions of abnor-mal signal intensity) and (3) moderate-to-severe abnormalities (multiple regions of abnormal signal intensity). Neurodevelopmental and neuromotor function follow-up were assessed with either the Griffiths Scale of Mental Develop-ment (GMDS), Bayley Scales of Infant Development, ed. 2, Dutch version (BSID-II-NL), or, more recently, Bayley Scales of Infant Development, ed. 3 (BSID-III). In order to enable the comparison of these different developmental tests, Z-scores were calculated for GMDS [performance developmental quotient used for neurode-velopment and the locomotor developmental quotient used for motor development], BSID-II-NL [mental developmental index and psychomotor developmental index], and BSID-III composite score outcomes. Z-scores were calculated as follows: [(score – test mean)/standard deviation (SD)] for each test. Development was classified as either mildly delayed (Z-score ≤–1) or normal (Z-score >–1). If neurodevelopmental tests were not performed, par-ents and/or general practitioners were contacted by phone to ob- tain the most recent status of neurodevelopment. Comparisons were made between groups with initial Hb con-centrations ≤3.0 mmol/l and 3.1–6.0 mmol/l, as well as between surviving and deceased patients. Comparisons were also made be-tween groups with and without MRI abnormalities. We used a χ 2 test and the Mann-Whitney U test for categorical and continuous variables, respectively. p < 0.05 was considered statistically signifi-cant. Statistical analyses were performed using SPSS Software Package version 20. Patient characteristics are presented in table 1. Hb concentrations were measured at birth or within several hours after birth and ranged from 1.0 to 5.9 mmol/l. The most frequent causes of anaemia were foetomater-nal haemorrhage (30.6%) and exsanguination (26.5%) due to vasa or placenta praevia. Less common causes were twin-to-twin transfusion syndrome (8.2%), placental ab-ruption (8.2%), bleeding due to a traumatic delivery (8.2%), umbilical cord rupture (6.1%), haemolysis (6.1%) and complications during caesarean section (2.0%). In 4.1%, the aetiology remained unknown. As hydrops was not present in any patient, the onset of the severe anaemia was most likely to be (sub)acute. Reticulocyte count data would have been useful to help determine the time of on-set of the anaemia, but these data were not available in most of the infants. Invasive mechanical ventilation was required in 85.7%, and non-invasive respiratory support (continuous posi-tive airway pressure) in 14.3%. Circulatory support with vasopressors was needed in 75.5% of the patients. There were no significant differences in the need for vital organ support or mortality between the two Hb concentration groups (table 2). Thirteen patients (31%) had at least one glucose con-centration <2.0 mmol/l. There were no differences in mortality (Fisher’s exact test, p = 0.759), morbidity [Fish-er’s exact test, need for respiratory support p = 0.186, need for circulatory support p = 0.636, and convulsions (either clinical or electrographic) p = 0.612] or neurode-velopment or motor development (χ2, p = 0.954 and p = 0.683, respectively) between normoglycaemic and hypo-glycaemic patients.Clinical or subclinical seizures were noted in 61.2% of the infants (n = 30). Of these, 43.3% of the infants (n = 13) needed one, and 56.6% (n = 17) needed multiple anti-epi-leptic drugs. Most patients were monitored with aEEG during the first days after birth (n = 42, 86%). In 28 patients, aEEG was available for assessment of the back-ground pattern. The patterns recorded and scored during the first 12 h after birth were used. Continuous normal voltage was found in 50% (n = 14), discontinuous normal voltage in 7.1% (n = 2), burst suppression in 32.1% (n = 9), continuous low voltage in 3.6% (n = 1) and flat trace in 7.1% (n = 2) of the infants. No differences in seizure activ-ity (Fisher’s exact test, p = 0.754) or aEEG background pat-tern (Fisher’s exact test, p = 0.355) were observed for the two Hb concentration groups. The aEEG background pat-terns in the deceased group were more often non-favour-able (burst suppression or flat trace; Fisher’s exact test, p = 0.010). These non-favourable patterns were associated with motor developmental delay in the surviving patients (χ2 test, p = 0.001). For neurodevelopment, non-favourable patterns showed a trend towards delay (χ2 test, p = 0.100).Eighteen patients (36.7%) did not survive the neonatal period. Thirteen patients died within 72 h after birth, mainly due to acute multiple organ failure.