This study was conducted in the framework of a larger population-based study designed to improve the care of newborns with perinatal asphyxia in a region of Spain with a land area of 91 217 m2.9 It was a multicentre observational cohort study that included live neonates born at or after 35 weeks of gestation with birth weights of 1800 g or greater and moderate or severe HIE between June 2011 and June 2013 in 12 hospitals (5 level III, 5 level II and 2 level I). The coordinating hospital extended the recruitment period through December 2017, applying the same study protocol.

The definition of moderate or severe HIE was fulfilment of one A criterion as well as the B criterion. The A criteria included: (1) Umbilical cord blood pH or blood pH in the first hour post birth of 7.00 or lower; (2) 5-minute Apgar score of 5 or lower and (3) need of advanced life support (intubation, cardiac massage and/or ventilation beyond 5 min post birth). The B criterion was clinical presentation compatible with moderate or severe HIE, defined as an altered level of consciousness (lethargy, stupor or coma) in the first 6 h post birth and staged with the modified Sarnat grading scale.10 To guarantee the rigour of this classification, we recorded the neurologic exams with the consent of the parents. When the evaluation was complete, 2 neonatal neurologists (A.GA, J.A.) who were blinded to the neurodevelopmental outcomes watched the video recordings to assess the severity of HIE.11

Newborns with moderate or severe HIE received servo-controlled whole-body TH (Tecotherm TSmed 200 N or CritiCool, MTRE Ltd) with a target core temperature of 33–34 °C for 72 h, according to the national guidelines in Spain.12 Patients managed with TH were monitored with amplitude integrated electroencephalography (aEEG) during cooling.

Decisions to withdraw or withhold life-sustaining treatment were based on a combination of the following findings: persistent coma, abnormal aEEG past 48 h post birth, detection of severe abnormalities in neuroimaging (cranial ultrasound and/or brain MRI); in some cases, the levels of neuron-specific enolase were measured in cerebrospinal fluid.13

We excluded neonates with polymalformative syndrome, inborn errors of metabolism, perinatal stroke, neuromuscular disease or spine lesions from the analysis.

It was performed on a 1.5T scanner. At minimum, the MRI protocol included the acquisition of axial and sagittal T1-weighted images, coronal T2-weighted images and axial diffusion-weighted images. Two blinded researchers (M.M.-B., J.A.) independently interpreted the MRI findings based on the scoring system proposed by Rutherford et al.,14 the presence and severity of lesions in the basal ganglia/thalami (BGT), white matter (WM) and cortex, and the signal of myelination in the posterior limb of internal capsule (PLIC). The total injury score (TIS) and WM × BGT were calculated as described by Thoresen et al.15 Discrepancies between the 2 researchers were resolved by consensus.

The evaluations were carried out by a single researcher (M.M.H.), a paediatric neurologist specifically trained for the purpose, who was blinded to the severity of HIE and the MRI findings. Patients underwent assessment at 36 months through a structured neurologic examination and the Bayley Scales of Infant and Toddler Development, Third Edition (Bayley-III), which yield composite scores (standard scores with a mean of 100 and a standard deviation [SD] of 15) for cognitive and motor skills.16 Language skills were assessed by means of the Peabody Picture Vocabulary Test, Third Edition (PPVT-III) Spanish version (standard scores with a mean of 100 and a SD of 15).17

In patients who could not be assessed before age 42 months, the cognitive evaluation was performed with the Spanish version of the Wechsler Preschool and Primary Scale of Intelligence III (WPPSI III)18 (standard scores with a mean of 100 and a SD of 20). In this group, motor skills were assessed with the Movement Assessment Battery for Children, Second Edition (MABC-2),19 calculating the percentile corresponding to each of the scores.20

The diagnosis of cerebral palsy (CP) was based on the criteria applied by the Surveillance of Cerebral Palsy in Europe network,21 and its severity assessed by means of the Gross Motor Function Classification System (GFMCS).22

We defined unfavourable outcome as death or adverse neurodevelopmental outcome. We defined adverse outcome as the presence of visual or hearing deficits, epilepsy and/or motor impairment (any patient with CP and/or a Bayley-III motor composite score <85 or a MABC-2 score ≤15th percentile) and/or cognitive impairment (Bayley-III cognitive composite score <85, intelligence score <85 in the PPVT-III or a full-scale intelligence quotient [IQ] score <80 in the WPPSI-III or WISC-IV).

Beyond the definition of adverse neurodevelopmental outcome, we assessed other neurodevelopmental problems, such as minimal motor impairment (MMI) and affective or behavioural problems. Minimal motor impairment was defined as absence of CP and at least 2 of the following: abnormal gait, dystonic posture, dystonic orofacial movements, intention tremor, difficulties in gross or fine motor skill coordination. Behavioural problems were assessed with the Child Behaviour Checklist (CBCL) for ages 1.5–5.23

We conducted telephone interviews with two families by means of a structured questionnaire asking about the general health and developmental milestones in the child. We also obtained follow-up health records from the hospital managing the child. The outcomes were classified as normal or adverse, without assigning scores.

We recruited an unmatched control group by convenience sampling comprising 79 term neonates who did not have perinatal asphyxia (as defined above), were not admitted to hospital in the neonatal period and with a normal physical evaluation before discharge from the maternity ward. The control group was recruited in the coordinating hospital and evaluated at age 36 months with the same protocol applied to the cases.

All the evaluations (of participants with HIE and controls) took place before the COVID-19 pandemic.

We have summarised qualitative data as absolute and relative frequencies and quantitative data as median and interquartile range (IQR) or mean and standard deviation SD. We compared continuous variables by means of the Mann–Whitney U test or the Kruskal-Wallis test, as applicable. For categorical variables, we used the χ2 test or the Fisher exact test. We calculated the Spearman correlation coefficient (rs) to assess the correlation between quantitative variables. To determine the optimal MRI score cut-off point for the prediction of adverse outcomes, we analysed receiver operating characteristic (ROC) curves and calculated the area under the curve (AUC), sensitivity (Sen), specificity (Spe) positive predictive value (PPV) and negative predictive value (NPV). We also developed predictive models for the rest of neonatal variables. We considered P values of less than 0.05 statistically significant, and the analysis was performed with the statistical package SPSS version 20 (IBM, Armonk, NY, USA).

We obtained the signed consent of parents, and the study was approved by the Clinical Research Ethics Committee (file 1243).

The 42 survivors, with the exception of 2, were evaluated at a median age of 39.5 months (IQR, 38–42.3): 32 with the Bayley-III and PPVT-3, 8 with the WPPSI-III or WISC-IV and the MABC-2, and 2 through a telephone interview (Fig. 1). The 79 controls were assessed with the Bayley-III and PPVT-3 at a median age of 39 months (IQR, 37–41).

Ten of the 42 survivors of moderate or severe HIE (24%) had adverse outcomes (1/8 [12%] with severe HIE and 9/34 [26%] with moderate HIE; P = .655) (Fig. 2 and Appendix B, Supplemental Table S2), as did one control, who had language delay.

Figure 2.

Neurodevelopmental outcomes of the overall cohort.

Deceased, male.

Alive, male.

Deceased, female.

Alive, female.

No follow-up.

Adverse outcome.

No adverse outcomes.

Cerebral palsy.

Minor motor impairment.

Abnormal results in cognitive assessment (Bayley-III cognitive composite score <85 and/or PPVT-3 <85 or WPPSI-III full-scale IQ <80 or WICS-IV full-scale IQ <80).

Abnormal results in motor assessment (Bayley-III motor composite score <85 or MABC-2 ≤15th

percentile).

Clinically significant scores in one or more areas of the CBCL (percentile >97th).

Normal results in all developmental and behavioural tests.

Patient with severe hearing loss that could not be explained by any other cause.

*Patient with severe HIE.

(0.36MB).

Three of the 10 children with adverse outcomes had CP, all of whom could walk independently: 2 had GMFCS level I motor impairment and the remaining patient GMFCS level II impairment (Table 1).

Another 3 children with adverse outcomes received a diagnosis of MMI. In addition, 2 of them had motor delay and 1 language delay, and 2 had abnormal scores in some area of the affective and behavioural assessment (Table 1).

Of the 32 survivors who did not meet the criteria for adverse outcome, 3 received a diagnosis of MMI (Table 1). None of the controls had MMI.

The mean scores in the neurodevelopmental tests were significantly lower in patients with HIE compared to controls (Table 2).

Although the difference in the overall prevalence of clinically significant behavioural and emotional problems (score >97th percentile) between children with HIE and controls was not statistically significant, the separate analysis of each domain revealed more frequent introversion (P = .04) in patients with HIE. In addition, when we considered scores at the upper limit of normal (93rd to 97th percentiles), patients with HIE had depression and anxiety features more frequently than controls (P < .05) (Table 2). In addition, patients with HIE and favourable outcomes had more depression problems compared to controls (8/30 [27%] vs. 6/77 [8%]; P = .021) and more anxiety problems (10/30 [33%] vs. 9/77 [12%]; P = .008). We did not find differences in behavioural variables within the HIE group between children with adverse neurodevelopmental outcomes and children with favourable outcomes (P = .693), nor between children with and without MMI (P = 1).

We assessed whether outcomes in the group of 9 patients with moderate HIE who did not receive TH differed from the outcomes in the group of 29 patients with HIE who were cooled, and found no differences in either the frequency of adverse neurodevelopmental outcomes (P = .403) or the combined unfavourable outcome of adverse outcome or death (P = .388). We also found no differences in the frequency of cognitive impairment (P = 1), CP or MMI (P = .160) or behavioural problems (P = .372). These findings should be interpreted with caution given the small number of patients who were not cooled.

The TIS and the WM × BGT were significantly higher in both children with MMI or MMI/CP compared to children with normal motor outcomes, and in children with unfavourable outcomes (death or adverse neurodevelopmental outcome) compared to survivors with normal development (Tables 3 and 4).

The TIS was a good predictor of MMI/CP (AUC: 0.8127) and unfavourable outcome (death or adverse neurodevelopmental outcome, AUC: 0.7191). A TIS of 7 or higher predicted death or adverse outcome with a PPV of 100%, while a TIS of less than 3 had an NPV of 84%. A TIS of 5 or greater had an NPV of 89% to predict the absence of motor impairment.

We did not find significant correlations between the neuroimaging scores and neurodevelopmental outcomes in the areas of cognition or language; however, there was a correlation between higher WM × BGT values and lower scores in the Bayley-III (rs = −0.465; P = .014).

We analysed the usefulness of three neonatal variables in the prediction of neurodevelopmental outcomes: the presence of clinical and/or electrical neonatal seizures, an abnormal aEEG tracing (low voltage, burst suppression or flat patterns) and the variable “reaching the target core temperature after 3 h post birth”. The NPVs were adequate—greater than 80%—for most variables (Appendix B, Supplemental Table S3).