The predictive value of early neurological examination in neonatal hypoxic–ischaemic encephalopathy and neurodevelopmental outcome at 24 months


Dr Deirdre Murray, Department of Paediatrics and Child Health, University College Cork, Ireland. E-mail:


Aim  The clinical and electrographic signs of hypoxic–ischaemic encephalopathy (HIE) evolve over the first days of life. We examined the evolution of neurological signs over the first 3 days of life, and determined whether serial administration of the Amiel-Tison Neurological Assessment at Term (ATNAT) would predict neurodevelopmental outcome at 24 months.

Method  Term (>37wks’ gestation) neonates born with suspected HIE between May 2003 and May 2005 in a Cork maternity unit were recruited prospectively. Modified Sarnat grading was assigned. The ATNAT was administered on days 1, 2, and 3 of life and a discharge neurological examination. Time to oral feeding and demographic variables were recorded. Developmental status was assessed using the revised Griffiths Mental Development Scales at 6, 12, and 24 months.

Results  Fifty-seven infants were recruited, with 51 (31 males, 20 females) included for follow-up. Neurological examination evolved and normalized over the first 3 days of life in many cases. At 24 months, 21 children had an adverse outcome, including six deaths. Examination at all time points correlated significantly with neurological outcome at 24 months. The best correlations were found to be (1) neurological examination at discharge (r=0.65, p<0.001), (2) Sarnat grading (r=0.64, p<0.001), and (3) ATNAT on day 3 (r=0.46, p<0.001). The best predictive value was seen with neurological examination at discharge (positive and negative predictive values of 86% and 72% respectively).

Interpretation  Persistence of abnormal neurological signs correlates significantly with adverse outcome. The later a neonatal neurological examination was performed, the better its predictive ability.

List of Abbreviations

Amiel-Tison Neurological Assessment at Term


Hypoxic–ischaemic encephalopathy

Hypoxic–ischaemic encephalopathy (HIE) affects three to five per 1000 live births. In its severe form, it remains one of the leading causes of neonatal death; approximately 25/100 of infants with HIE will experience severe neurological disability including cerebral palsy (CP), epilepsy, intellectual disability, and sensory loss.1 Outcome can be difficult to predict and is traditionally described using the clinical Sarnat score assigned at 24 hours after delivery.2 Although the Sarnat score is useful at the extremes (mild and severe grades), it is more difficult to predict outcome in those infants with a moderate Sarnat grade of encephalopathy. Although the Sarnat grade provides a single summary score, the neurological signs observed in HIE continue to evolve over the first 72 hours of life.

The Amiel-Tison Neurologic Assessment at Term (ATNAT)3 has been developed to provide a framework for observing the development of cortical control in infants at term and has been shown to predict the occurrence of CP after birth asphyxia.4,5 Previous studies have not examined repeated measurements over the first days of life. The central tenet of the ATNAT is that the stage of maturation of the two motor control systems (subcorticospinal and corticospinal) can be clinically evaluated in the term infant.6 Amiel-Tison recommends serial assessments for the following reasons: (1) to reduce the confounder of transient clinical signs caused by medical treatments and metabolic processes;7 and (2) to discriminate between a ‘static’ neurological profile more usually seen in infants with prenatal damage and an ‘evolving’ profile more commonly seen in infants with HIE and healthy infants.8

The aim of this study was to examine the evolution of neurological signs in term infants with HIE over the first 3 days of life, and to investigate the ability of these repeated serial assessments to predict neurodevelopmental outcome at 2 years.


This prospective study was performed in a large maternity service with approximately 6000 deliveries per year as part of a study into early continuous video–electroencephalogram (EEG) monitoring.9 Ethical approval was obtained from the Clinical Research Ethics Committee of the Cork Teaching Hospitals.

Between May 2003 and May 2005, term infants (>37wks’ gestation) with HIE were recruited if they fulfilled two or more of the following criteria: initial capillary or arterial pH <7.1, Apgar score <5 at 5 minutes, initial capillary or arterial lactate >7mmol/L, or abnormal neurology/clinical seizures. A neurological assessment was performed by dedicated research paediatricians not involved in the clinical care of the infant. The parents of infants fulfilling the criteria were approached and written informed consent was obtained. Infants were typically recruited within 6 hours of birth, with the remainder recruited at the onset of clinical seizures. Participants were excluded if they had a coexisting congenital anomaly, or significant comorbidity.

Initial neurological assessment was performed using the ATNAT.3 Early neurological status continued to be monitored with repeated administration of the ATNAT for three consecutive days. The ATNAT consists of examination of 35 items describing cranial characteristics, alertness, behaviour and spontaneous activity, passive tone in the limbs and axis, active tone, and primary reflexes. Scores are assigned for each item as follows: 0, ‘within normal range’; 1, ‘a moderately abnormal result’; 2, ‘a definitely abnormal result’. Scores are assigned each time the ATNAT is administered. The scores are not used in a quantitative way; rather, they are designed to develop an opinion of the infant’s neurological status based on the most commonly observed scores. The final synthesis is described as normal in the absence of neurological signs, or presence of minor abnormalities (without central nervous system [CNS] depression), moderate (with CNS depression), or severe (deep CNS depression and repeated seizures for >30min) for the term infant.7

Most of the ATNAT assessments were undertaken by the research paediatrician (DMM), with the remainder completed by a designated experienced neonatal paediatric registrar (PB). In the event that some of the items could not be administered (e.g. sucking, Moro reflex, axial tone assessment) owing to endotracheal intubation, difficulties were recorded and the assessment was attempted again at a later stage.

Additional neurological assessment consisted of a designation of a modified Sarnat score assigned at 24 hours (Table I).10,11

Table I.   Modified Sarnat score
Grade I (mild)Grade II (moderate)Grade III (severe)
  1. Source: Levene et al. (1985).11

‘Hyperalert’SeizuresProlonged seizures
Mild hypotoniaMarked abnormalities of toneSevere hypotonia
Poor suckingRequires tube feedingFailure to maintain spontaneous respiration

Each infant underwent a final neurological examination by the neonatal staff at discharge from the neonatal intensive care unit. The examination was graded as ‘0’ for normal if no abnormality was seen, ‘1’ for moderate if there were any suspect findings, such as mild head lag, or ‘2’ for severe if there were significant neurological abnormalities (increased peripheral tone, poor suck, marked head lag). The date that oral feeding was fully established was also recorded.

Long-term neurodevelopmental follow-up

Neurodevelopmental follow-up was performed at 6, 12, and 24 months using the revised Griffiths Mental Development Scales – Revised12 for children aged 0 to 2 years. The scales were administered by the Griffiths’ trained paediatricians (DMM, and some by PB), either at the outpatients’ clinic, or in the participant’s home. In addition, a neurological examination assessing peripheral, axial tone, and deep tendon reflexes was performed. Maternal educational level and socio-economic status were recorded.

For this study, an abnormal outcome was defined as developmental delay (Griffiths Developmental Quotient ≤87, 100 minus 1SD), CP, or death.

Statistical analysis

Means and SDs were calculated for demographic factors, birthweight, and gestation, etc. Predictive ability of the ATNAT score was assessed using positive predictive value, negative predictive value, specificity, sensitivity of binary values, and characteristics of area under the receiver operating curve. Correlations between linear variables were assessed using a non-parametric Spearman’s rank correlation. Statistical significance was set at p<0.05.

Statistical analysis was completed using SPSS version 12.0 (SPSS Inc, Cary, NJ, USA).


In total, 57 infants were recruited to the study. Of these, four were lost to follow-up, giving a 2-year follow-up rate of 93%. In addition two infants were excluded because of uncertainty about the diagnosis of HIE. One had concurrent congenital diaphragmatic hernia repaired and one was investigated for dysmorphic features and hypotonia.

The remaining 51 infants consisted of 31 males and 20 females, with a mean birthweight of 3416g (SD 634), and a mean gestation of 40.1 weeks (SD 1.5). The mode of delivery was normal vaginal delivery in 12 (23.5/100), forceps in 5 (9.8/100), vacuum in 10 (19.6/100), combined vacuum and forceps in 5 (9.8/100), and emergency Caesarean section in 19 (37/100). Their assigned Sarnat grades were I (mild) in 24, II (moderate) in 18, and III (severe) in 9.

Early neurological examination and outcome

The results of the early neurological examinations, observed for each participant and their diagnostic and developmental outcome at 24 months, are presented in Tables SI and SII (supporting information published online).

In the first day of life, 21 out of 51 (41/100) had a ‘severely abnormal’ ATNAT score of 2. Of this group, by day 2 of life, two out of 21 had died, and of the surviving infants, 14 out of 19 continued with a score of 2, whereas 5 out of 19 had improved to a score of 1. By day 3, only 8 out of 19 continued with a score of 2, and all of these experienced a poor outcome (four deaths, three with CP, and one with global developmental delay). These findings are summarized in Fig. 1.

Figure 1.

 Evolution of a severe neurological abnormality on day 1, assessed using the Amiel-Tison Neurological Assessment at Term (ATNAT). ATNAT 1, ATNAT 2, ATNAT 3, day 1, 2, and 3 respectively. Normal, ATNAT score of 0; moderate, ATNAT score of 1; Severe, ATNAT score of 2. Discharge exam, neurological examination at discharge.

On day 1, 25 out of 51 (49/100) had a ‘moderately abnormal’ ATNAT score of 1. By day 2, 20 out of 25 remained with an ATNAT score of 1, and by day 3, 17 out of 25 continued to have an ATNAT score of 1. Of this group with persistent moderately abnormal assessments, 7 out of 17 had abnormal outcomes at 24 months; four children with global developmental delay, one with hemiplegia and developmental delay, one with diplegia, and one with an isolated hemiplegia. These findings are summarized in Fig. 2.

Figure 2.

 Evolution of a moderate-grade neurological abnormality assessed by the Amiel-Tison Neurological Assessment at Term (ATNAT). ATNAT 1, ATNAT 2, ATNAT 3, day 1, 2, and 3 respectively. Normal, ATNAT score of 0; moderate, ATNAT score of 1; Severe, ATNAT score of 2. Discharge exam, neurological examination at discharge.

A small group of infants, 4 out of 51, had normal ATNAT assessments on day 1, all with normal 24-month outcomes.

In the whole series, by day 3 of life, 23 out of 49 of the surviving infants had normal ATNAT neurological assessments. Of this group, 3 out of 23 had mild developmental delay at 24 months.

Examination at discharge was performed on the final day of the infant’s stay in the neonatal intensive care unit by research or medical staff. This assessment did not use the structured ATNAT and each infant was assigned a grade of normal, moderate, or severe based on the clinical findings. The age of the infants varied from 2 to 27 days, with the sicker infants remaining hospitalized longer. Data were missing for one infant, and three infants died before discharge from the neonatal intensive care unit. Of the remaining 47, discharge neurological examination was entirely normal in 29 out of 47, and in 5 out of 47 moderate abnormalities were seen. The remaining 13 infants continued to have severely abnormal examinations on discharge from the neonatal intensive care unit.

There was a significant correlation (p<0.01) between neurological outcome at 2 years and ATNAT on days 1, 2, and 3 of life. In addition, outcome correlated with Sarnat score and examination at discharge (Table II). The best correlation was seen between outcome and examination at discharge, followed by Sarnat score assigned, and finally ATNAT on day 3. Griffiths Developmental Quotient scores at 24 months differed significantly depending on ATNAT score on day 3. The mean (SD) Griffiths Developmental Quotient scores were 106.4 (10.5) in the normal ATNAT day 3 group, 98.3 (22.9) in the moderately abnormal ATNAT day 3 group, and 61.5 (18.96) in the severely abnormal ATNAT day 3 group.

Table II.   Correlation between early neurological assessment and long-term outcome
ExaminationCorrelation with outcome (r)pCorrelation with Griffiths Developmental Quotient at 24mo (r)p AUROC (95% CI)p
  1. r value, correlation coefficient using Spearman’s rank correlation; AUROC, area under the receiver operating curve; ATNAT, Amiel-Tison Neurological Assessment at Term.

ATNAT day 10.450.0010.430.0020.652 (0.47–0.84)0.132
ATNAT day 20.420.0030.400.0040.652 (0.45–0.86)0.132
ATNAT day 30.460.0010.590.0010.754 (0.58–0.93)0.012
Sarnat grade0.640.0010.610.0010.744 (0.57–0.92)0.016
Discharge examination0.650.0010.500.0010.786 (0.64–0.96)0.005

Area under the receiver operating curves were statistically significant for ATNAT day 3, Sarnat score, and highest for discharge examination. In summary, the later that a neurological assessment was performed, the more reliable was its prediction of outcome.

Neurological outcomes at 24 months correlated with discharge examination (r=0.65, p<0.001). Of the 13 severely abnormal participants, 11 had abnormal outcomes at follow-up. Of the five moderately abnormal, two had abnormal outcomes. Of the 29 infants with normal examinations at discharge, 25 were normal at 24 months. This gave a positive predictive value of a normal examination at discharge of 86%, and a negative predictive value of 72%. Sensitivity was calculated at 77% and specificity at 83%.

Within the subgroup of 18 infants with Sarnat grade II, or moderate encephalopathy, examination at discharge was still moderate or severely abnormal in 10 out of 18 infants. A normal examination at discharge predicted a normal outcome with positive and negative predictive values of 70% (7/10) and 75% (6/8) respectively.

Oral feeding and neurological outcome

Time to establishment of oral feeding (days from birth) correlated significantly with neurological outcome (r=0.40, p=0.008). In 7 out of 51 infants, oral feeding was never established; 3 out of 7 died in the first week of life, before discharge from the neonatal intensive care unit. The four surviving infants were discharged with nasogastric feeding tubes. Of these four children, three subsequently died and the sole survivor had severe CP (spastic quadriplegia). Three further children took more than 7 days to establish feeding, and all of these infants had abnormal outcomes (one with isolated hemiplegia, two with global developmental delay).


This study demonstrates the pattern of neurological evolution in infants with HIE over the first 3 days of life. Neurological examination normalized in many infants, with 47/100 having a normal examination by day 3. A normal early assessment predicted a normal outcome. Those infants with persistent neurological abnormality on day 3 were more likely to progress to neurodevelopmental disability. This risk was 100% in those with a severely abnormal ATNAT rating, and 41/100 in those with a moderately abnormal rating.

Several methods to assess neonatal neurological status have been described.13–15 The ATNAT was chosen because it is quickly and easily administered, examines forebrain function, and has been previously shown to correlate with outcome at 12 to 15 months, in particular to outcome after perinatal asphyxia.5,8 A further strength of the ATNAT is the inclusion of items that assess early observations of infant behaviour such as sucking. Neonatal examination used in combination with a modified Sarnat score at 6 hours of age with amplitude-integrated EEG (aEEG) has been found to enhance aEEG’s ability to predict neurological outcome at 1 year.16 Amess et al. 5 also found that neonatal neurological assessment was a useful adjunct to magnetic resonance spectroscopy, with late examinations (>7d) having superior predictive sensitivity of 12-month outcome than early examinations (<48h).

The current study found that the later a neurological examination was performed, the greater the correlation with neurological status at 24 months. An abnormal examination at the time of discharge placed the infant in a high-risk group, with 72% progressing to developmental delay. In the moderate subgroup, the discharge examination was particularly useful. In infants with moderate Sarnat scores, the overall rate of abnormal outcome was 50%. However, this was 25/100 in those with a normal discharge examination, and 70% in those with an abnormal discharge examination. Thus a clinician may find the neurological status at discharge useful in counselling parents and guiding long-term follow-up. Sarnat scoring, or ATNAT assessment at 24 hours combined with a detailed examination at discharge, could provide outcome indicators to physicians who do not have access to more sophisticated methods of prognostication. The utility of detailed neonatal neurological assessment in these situations has been previously demonstrated in the developing world.17,18 Our data confirm the importance of a structured neonatal neurological examination and its ability to aid prognostication.

Why does the discharge examination appear to be more reliable in predicting outcome? First, several factors may interfere with accurate neurological assessment in the first few days of life. Sicker infants may be intubated, or sedated, or both. Hypotension, shock, and multi-organ failure may lead to abnormalities of tone and movement.19 Many of our infants received anticonvulsants and sedatives. One infant who had meconium aspiration syndrome was ventilated and received neuromuscular blockers for the first 2 days of life, making neurological assessment impossible. By discharge, most of these confounders will have abated, helping to explain the increased sensitivity (83%) of the abnormal examination at discharge compared with earlier examinations. Notably, one limitation was that the discharge examination was performed and recorded by different medical staff. These examinations were also performed at different ages. This highlights the importance of a detailed, consistent discharge examination of the neurological status of these high-risk infants.

A limitation of administering the ATNAT exclusively in the neonatal period is that it precludes access to later-elicited responses (e.g. postural reactions) that may possess superior predictive qualities.20 Further analysis of the individual items of the ATNAT may reveal markers for later outcome, because it includes behavioural and communication items, which share construct validity with later developmental assessments.21 There have been promising observations in relation to high-risk and preterm infants and clusters of ATNAT signs such as the ‘Amiel-Tison triad’ (imbalance of passive axial tone, phasic stretch reflex in triceps surae, and squamous ridges), that may predict subtle neurological outcome.22,23 Detailed analysis of early clinical signs is dependent on what can be elicited in a neurological examination. At birth, this is limited by the level of maturation of the infant at that time, and what their nervous system can display.

Like the argument made for the importance of detailed follow-up in preterm infants,24 the clinical criteria for neurodevelopmental follow-up in the HIE population merit attention. Earlier studies have demonstrated neuropsychological deficits in older children,25,26 and our cohort are due to be followed up into their school years. This study has highlighted eight children at age 2 with developmental delay with no motor disability. It is this group of children, if not appropriately screened, who are at risk of not being identified for early interventions. Fortunately, interesting models of serial developmental screening in young children are being developed, for example the ‘developmental pathways study’,27 that may go some way towards resolving this issue.

In conclusion, we have shown that neurological signs evolve rapidly over the first 3 days after perinatal asphyxia. Persistence of abnormal neurological signs correlates significantly with adverse outcome. The later a neonatal neurological examination is performed, the better its predictive ability, with examination at discharge having positive and negative predictive values of 86% and 72% respectively.


This research was supported by a grant from the Health Research Board of Ireland, reference number RP/2008/238. Dr Murray received the Denis O’Sullivan Research Fellowship from University College Cork.