Hammersmith Infant Neurological Examination in infants born at term: Predicting outcomes other than cerebral palsy

We explored the ability of the Hammersmith Infant Neurological Examination (HINE) to identify cognitive performance delay at 2 years in a large cohort of infants born at term.

The Hammersmith Infant Neurological Examination (HINE) is well recognized as a reliable tool for early detection of motor impairments in both infants born preterm and at term. [1][2][3][4][5][6][7] When performed after the age of 5 months, it is considered the assessment with the highest predictive power for cerebral palsy (CP), together with early neuroimaging. 2 We have recently reported 8 that the HINE can also be used in the first year of live to identify infants born preterm at risk of delayed cognitive performance, identifying specific age-dependent cut-off scores to discriminate infants at risk of CP from those with a delayed performance without CP.
The relation between early neuromotor outcome and later non-motor performance in infants born at term is of particular interest as, with the advent of therapeutic hypothermia, there is a decrease in the frequency of major motor sequelae, while children remain at risk of cognitive impairments. [9][10][11][12] Earlier detection of cognitive impairment may improve possible benefits of early interventions for enhancing neuroplasticity, preventing or ameliorating neurodevelopmental impairments and enhancing parental well-being. 13 Few studies have reported the use of early neurological assessment methods to identify infants at risk not only for motor impairment (mainly CP) but also for cognitive impairment. [13][14][15][16][17][18] The assessment of general movements, that is the distinct spontaneous movement pattern evident in infants in the first months after birth, is considered a sensitive assessment for providing information on the integrity of an infant's brain function and the best predictor of CP in high-risk infants. [13][14][15] A recent review of the literature 14 suggested that abnormal general movements beyond term age, monotonous and jerky movements, as well as postural abnormalities in the first 5 months of age were associated with a higher risk of cognitive dysfunction in a population of infants born preterm. Although the association between abnormal general movements and cognitive outcome is weaker when children with CP are excluded, abnormal general movements may help clinicians in identifying infants born preterm who warrant closer follow-up of their cognitive development. 15 More recently, Hadders-Algra and colleagues [16][17][18] validated a new clinical assessment, the Standardized Infant NeuroDevelopmental Assessment, for assessing infants under 12 months of age at risk of neurodevelopment disorders. This assessment showed a good predictive power both for detecting infants at risk of CP, but also those at high risk of other non-motor developmental disorders owing to the presence of different items evaluating the quality of spontaneous movements. 19 As the HINE is widely used for the early detection of CP and has already been shown to be useful in predicting cognitive function in infants born preterm, 8 we explored its ability to identify normal and delayed non-motor performance assessed using the Mental Developmental Index (MDI) of the Bayley Scales of Infant Development, Second Edition at 2 years in a large population of infants born at term. We also explored the range and possible overlaps of HINE scores in groups of infants with normal and low MDI scores, with or without CP.

M ET HOD
All the infants in this study were part of a follow-up research project performed at the Neonatal Unit of the University of Catania and at the Fondazione Policlinico Gemelli IRCCS of Rome between January 2006 and December 2012. These are level II (specialty care) and level III (subspecialty) neonatal centres admitting, among others, infants born at term affected by perinatal asphyxia, seizures, sepsis, congenital heart disease, cerebral malformations, and those needing surgery. Infants were enrolled routinely to a 2-year follow-up research protocol.
Infants having a gestational age of at least 37 weeks were eligible for the study. The specific inclusion criteria were neurological assessments at four time-points in the first year (3, 6, 9, and 12 months) and a neurodevelopmental assessment at 2 years. Exclusion criteria were the presence of congenital disorders, transfer to another hospital, or an incomplete follow-up programme.
In this retrospective study, the charts of 496 infants born at term consecutively discharged from our neonatal units were reviewed. A total of 50 were excluded from the study: 10 because of the presence of congenital anomalies, 15 because of transfer to their local II level hospitals after stabilization of their clinical condition, and 25 because they did not complete the follow-up programme. There were no statistical differences in the patients' characteristics in this last group compared with those in the study group. A further 235 typically developing infants born at term were recruited and used as a comparison group. These infants were all discharged from the postnatal ward within 3 days from birth; none were small for gestational age, none had evidence of birth asphyxia or respiratory distress, none had neurological, metabolic, or infective concerns, or jaundice requiring phototherapy.
The study protocol was approved by the Ethics Committee of the Fondazione Policlinico Universitario A. Gemelli IRCCS, Rome, Italy, and written informed consent was obtained from the parents in all cases.

• The
Hammersmith Infant Neurological Examination (HINE) can be used in the first year to detect risk of delayed cognitive performance at age 2 years. • Age-dependent HINE cut-off scores identify increased risk of delayed cognitive performance in infants born at term.

Neurological examination
The HINE was used for the clinical neurological examinations. It includes 26 items that assess five subsections: cranial nerve function, posture, movements, tone, and reflexes. 1,7 Each item can be scored from 0 to 3, 3 being considered as optimal, giving a maximum total score of 78. The HINE is readily performed and accessible to all clinicians and has been shown to have good interobserver reliability. [1][2][3][4][5][6][7][8]19 In the typically developing term-born population, optimal scores at 9 and 12 months are at least 73, 7 and at 3 and 6 months at least 67 and at least 70 (median) respectively. 20 In this retrospective study, we analysed data from infants born at term, both those at high risk and typically developing, with HINE assessments at 3, 6, 9, and 12 months using both the five available subsection scores and the global score, which is the sum of the subsection scores, for each time period.

Outcomes
At 2 years all infants were assessed using a structured neurological examination 21 and the Bayley Scales of Infant Development, Second Edition. 22 The Bayley Scales of Infant Development, Second Edition comprise two outcome scores: the Psychomotor Developmental Index and the MDI with a mean of 100 and SD of 15 (range 50-150). For the present study, we only used the MDI. An MDI score of 50, representing the minimum possible score, was apportioned to those infants so severely affected that a structured cognitive assessment could not be completed. Two child neurologists (EP, CB) with experience in neurodevelopmental assessment performed the assessments with the Bayley Scales of Infant Development, Second Edition. The examiners were blind to the HINE results.
The presence of CP was determined from the neurological examination. CP was defined as a developmental disorder of movement and posture, causing activity limitation, attributed to non-progressive disturbances related to brain injury early in development. 23 In infants with CP, magnetic resonance imaging (MRI) reports were obtained from the medical records and classified on the basis of current knowledge of the predictive value of early neuroimaging in infants at risk of developing CP. 24,25 Neurodevelopmental outcomes were classified as within the normal range (Bayley MDI scores >84 and no CP), mildly delayed cognitive performance (Bayley MDI scores 70-84 without CP), and significantly delayed cognitive performance (MDI <70 without CP). Children who developed CP were classified similarly according to their MDI score.

Statistical analysis
Birthweight and gestational age are reported as mean and SD. HINE scores are reported as median and interquartile range (for data that were not normally distributed) at the four different ages, for four groups of high-risk infants (those with normal performance, mildly delayed cognitive performance, significantly delayed cognitive performance, and those with CP) and for the typically developing infants born at term. Intergroup comparisons were done, as appropriate, by parametric (one-way analysis of variance performed separately for each age group, followed by post hoc analysis for multiple comparisons using Bonferroni's method) or by nonparametric test (Kruskal-Wallis test followed by Dunn's test of multiple comparisons). Spearman's rank correlation coefficient (R s ) was used to correlate the HINE global and subsection scores obtained at 3, 6, 9, and 12 months with the MDI scores at 2 years.
For at-risk infants, the predictive values of cut-off HINE global and subsection scores were assessed using sensitivity and specificity, according to the age at assessment, for two different outcomes: significantly delayed cognitive performance and CP. Areas under the receiver operating characteristic curve were obtained and cut-off values were estimated using the Liu method, which maximizes the product of the sensitivity and specificity. 26 The level of significance was set at p < 0.05. We used Stata statistical software version 15 (StatCorp, College Station, TX, USA) for the analysis.

R E SU LTS
Of the 446 eligible at-risk infants (Table 1), 408 did not develop CP at 2 years and, of them, 211 had MDI results classified as normal (MDI ≥85), 32 were mildly delayed (MDI 70-84), and 165 significantly delayed (MDI <70); infants with normal and mildly delayed cognitive performance had similar findings in terms of gestational age, birthweight, and HINE scores. We therefore combined these two groups for the analysis and refer to them as having normal/mildly delayed cognitive performance (n = 243).
A total of 38 infants developed CP; 10 of the 38 had a normal MDI (≥85), two were mildly delayed (MDI 70-84), and 26 had significantly delayed MDI (<70). Seventeen of the 26 infants with significantly delayed MDI were so severely affected that a structured cognitive assessment could not be completed and were given a score of 50. Infants with CP and normal or mildly delayed MDI were similar in terms of gestational age, birthweight, and HINE scores and were therefore combined for the analysis and we refer to them as infants with CP and normal/mildly delayed cognitive performance (n = 12). Infants with CP had some abnormalities consistent with their MRI scans. All the 235 typically developing infants had a normal neurological examination at 2 years; MDI scores were within the normal range (≥85) in all but 15 children, who had scores between 70 and 84; none were less than 70. Table 1 provides the clinical characteristics of the population related to outcome at 2 years. All the three groups of at-risk infants (normal/mildly delayed cognitive performance, no CP; significantly delayed cognitive performance, no CP; CP and all MDI levels) and the typically developing infants on the postnatal ward had a similar mean gestational age and birthweight. No differences were also observed for mean gestational age and birthweight in infants with CP when subdivided according to the MDI.

HINE
The typically developing infants born at term on the postnatal ward reported higher HINE scores than the at-risk groups at all ages (p < 0.001). At each time-point (Figure 1), infants with a normal/mildly delayed cognitive performance but no CP at 2 years had significantly higher (p < 0.001) HINE global scores than infants with significantly delayed cognitive performance and no CP and those with CP. Infants with significantly delayed cognitive performance had higher HINE global scores (p < 0.001) than those with CP; infants with CP and an MDI of at least 70 had significantly higher HINE global scores (p < 0.01) than those with CP (with or without low MDI scores). Infants with CP and MDI of at least 70 and those with significantly delayed cognitive performance had similar HINE global scores (p > 0.05). Numerical details of these data are given in Table 2. Table 3 provides the sensitivity and specificity of HINE global cut-off scores for detecting normal/mildly delay performance at the four assessment ages (3 months, 58 out of 78; 6 months, 64 out of 78; 9 months, 69 out of 78; 12 months, 71 out of 78). An age-dependency of the predictive values is reported with lower sensitivity and specificity at 3 months.
In high-risk infants, we found a significant (p < 0.001) correlation between MDI at 2 years and HINE global scores at 3 months (R s = 0.569), 6 months (R s = 0.611), 9 months (R s = 0.634), and 12 months corrected age (R s = 0.680) ( Figure 2). Table 4 reports the sensitivity and specificity of cut-off scores for the five HINE subsections at each assessment age for detecting significantly delayed cognitive performance, significantly delayed cognitive performance, and/or CP. The scores for the movements subsection showed the highest sensitivity and specificity in the four time-points. Moreover, the movements subsection scores had the most significant correlation at all four assessment ages (3 months, R s = 0.56; 6 months, R s = 0.58; 9 months, R s = 0.58; 12 months, R s = 0.58) with 2-year MDI scores (Table 5).

DISCUS SION
The HINE, already used to predict CP in young infants, 1-7 has recently been used in a population of infants born preterm to identify infants at risk of cognitive delay. 8 The possibility of differentiating infants who will have a typical outcome from those with later atypical development is essential in clinical practice. In the present study we wished to determine whether the HINE could also be used for this purpose in infants born at term. This is also the first study, to our knowledge, reporting longitudinal HINE data from typically developing infants born at term from 3 to 12 months.
In the present study, more than 50% of high-risk infants born at term had, at 2 years, a normal/mildly delayed cognitive performance and 37% had a significantly delayed cognitive performance (37%). CP was diagnosed in 9% of cases. The low incidence of infants with CP is probably due to the characteristics of the population, that included infants with sepsis, small for gestational age, heart disease, and mild asphyxia that are not often associated with the development of CP.
A significant correlation between the 2-year MDI scores and the HINE global scores at 3, 6, 9, and 12 months was observed, with a better correlation in infants from 6 months onwards. The relatively lower correlation of the HINE with 2-year outcome at 3 months was also reported in the study in infants born preterm with the same aims. 8 This finding is because HINE was originally planned and validated from 12 months onwards, and some of the items are age-dependent. 1,3,4 When we subdivided the outcome at 2 years in our cohort of at-risk infants born at term on the basis of cognitive performance, we found that there was a gradient of HINE scores progressively decreasing in patients with increasing severity of cognitive performance. In agreement with previous studies, 1,4,5,27 higher HINE global scores during the first year were found in infants with normal/mildly delayed cognitive performance at 2 years compared with those who had a delayed cognitive performance at 2 years with or without CP. It is of interest that even in this group of infants (normal/mildly delayed cognitive performance) the HINE scores were lower than those of typically developing infants born at term reported in the present and in previously published studies. 7,20 This is probably because infants with normal/ mildly delayed cognitive performance include both low-and high-risk infants born at term with some mild neurological findings leading to slightly lower HINE scores during the first year. In addition, performance in the normal/mildly delayed range at 2 years does not exclude the possibility of later neurodevelopmental issues, for example attention, memory, and behavioural problems which cannot be detected using the Bayley Scales of Infant Development, Second Edition at an early age. 28 Infants with CP and low MDI had the lowest HINE scores, lower than those found in infants with CP and an MDI of at least 70 that were similar to those with significantly delayed cognitive performance without CP. The use of brain imaging helped to identify patients with brain lesions who were at higher risk of developing CP. These data are consistent with that we reported in a population of infants born preterm, 8 in which there was a clear gradient, as infants performing normally at 2 years had higher HINE scores than those with delayed cognitive performance but without CP, followed by those with CP but an MDI greater   term with CP (regardless of their MDI) had a higher trend in HINE scores than infants born preterm with CP. 8 Although comparing the HINE scores according to the gestational age was not a specific aim of the present study and a statistical analysis was not performed, our findings are consistent with an early pattern of neurological findings showing a gradient related both to later motor and cognitive performance in infants born preterm; but in infants born at term the MDI scores for those with CP may be affected only partly by their motor skills and therefore the MDI score could be a true reflection of their non-motor skills. The HINE has a good neuroanatomical correlation 1 with brain abnormalities and outcome; in infants born preterm with cystic periventricular leukomalacia, 1 those with severe early abnormalities on the HINE (increased neck and trunk extensor tone and flexed arms and extended legs) and CP outcome were mostly associated with lesions around the trigone and rarely with lesions more anterior to the trigone. On the other hand, in infants born at term with hypoxic-ischaemic encephalopathy, a normal neonatal MRI or moderate white matter lesions were reported in infants with optimal HINE scores and normal 2year outcome. In contrast, infants with severe basal ganglia lesions had suboptimal HINE scores and abnormalities particularly in items assessing axial and limb tone, movements, and vision, developing severe CP with no independent sitting. Therefore, it is plausible that the characteristic score distribution is somewhat different in infants born preterm and at term with CP as they are susceptible to different types of brain injury.
Specific HINE cut-off scores have mainly been used for predicting CP during the first year after birth; 1-7 in the present study, we also identified age-specific HINE cut-off scores for infants who do not develop CP but have delayed cognitive performance at 2 years. These scores may help to distinguish infants likely to have normal or only mildly delayed outcomes from those with significant neurodevelopmental impairments (CP and/or with significant delay) using cut-offs of 58 at 3 months, 64 at 6 months, 69 at 9 months, and 71 at 12 months. Interestingly these cut-offs are the same as those we found for infants born preterm. 8 Therefore, the HINE may be considered a potential tool not only for early detection of infants at risk of CP but also for detecting risk of other developmental disorders whether born preterm or at term. However, these cut-off scores should be used with caution as they reflect differences between groups and may not hold true in individual cases. These findings should always be interpreted as part of a comprehensive clinical evaluation together with brain imaging data. 8 As reported in the preterm population, 8 the HINE subsection showing the best predictive power for significantly delayed cognitive performance and/or with CP was 'movements'. This subsection at all assessment ages showed the highest correlation with the 2-year MDI scores, although not as good as we found in the population of preterm infants (0.58 vs 0.69). Both general movements and the Standardized Infant NeuroDevelopmental Assessment have been used as predictive tools mainly in infants born preterm at risk of CP 29 and, more recently, for detecting other developmental disorders such as intellectual disability and autism spectrum disorder. [15][16][17][18][19]30,31 Both these tools rely entirely (general movements) or heavily (Standardized Infant NeuroDevelopmental Assessment)  on the assessment of the quality of spontaneous movements that in part reflect the development of cortical networks; when suboptimal, there is an increased risk of CP and developmental delay with or without CP. The HINE is a structured tool including many aspects of neurological function, such as posture, tone, motility, and reflexes. This may increase the sensitivity for predicting cognitive outcome as these combined aspects of the exam may better reflect the maturation of the basal ganglia and thalami and cerebellum that are also important for the development of cognitive aspects. 10,32,33 A limitation of this study is the relatively short follow-up period of 24 months when considering cognitive outcome, although with the advantage of good compliance and a high follow-up rate. Some infants with significantly delayed cognitive performance at this age may improve later and some with normal scores may lag behind their peers in later school years, especially those with hypoxic-ischaemic encephalopathy. 11 Follow-up assessment at older ages would probably result in a more accurate measure of a more stable neurodevelopmental trajectory. 33 Furthermore, in the present study the neurodevelopmental outcome was performed using the Bayley Scales of Infant Development, Second Edition which has now been replaced by an updated version; however, at the time of our study, the Italian norms of the Bayley Scales of Infant and Toddler Development, Third Edition were not available. 34 The use of the Third Edition could have provided more refined information than the Second Edition by separating the original MDI into cognitive and language scales.
In conclusion, our results from a large population of infants born at term suggest that the HINE can be used to detect infants at risk not only of CP but also delayed cognitive performance at 2 years of age. The age-specific HINE cut-off scores should be considered part of a comprehensive clinical evaluation to potentially differentiate infants at risk of CP from other disabilities. 8 As the frequency of major motor sequelae has decreased in infants born at term with perinatal asphyxia, 10-12 more attention should be given to the early identification of other outcomes such as cognitive impairment. Therefore the results of the present study should be useful for clinicians involved in the follow-up and treatment of high-risk infants in designing interventions targeting improvements in the motor repertoire and in other activities designed to enhance non-motor skills and having some impact on later cognition.
The HINE is easy to learn and is a readily accessible tool that should be useful in early intervention assessments between 3 and 12 months.
Longitudinal and longer follow-up data in this and other cohorts are needed to confirm our results and to allow a more detailed correlation between early findings and preschool and school age outcomes using a more indepth assessment of cognitive functions (Bayley Scales of Infant and Toddler Development, Fourth Edition, Wechsler Scales).

DATA AVA IL A BILIT Y STATE M E N T
Data available on request from the authors.

AC K NOW L E D GE M E N T S
This research study did not receive any specific grants from funding agencies in the public, commercial, or notfor-profit sectors. The authors have stated that they had no interests that might be perceived as posing a conflict of interest or bias in setting up the study or interpreting the data.