Early Neurodevelopmental Assessments for Predicting Long-Term Outcomes in Infants at High Risk of Cerebral Palsy

Key Points Question What is the effectiveness of early neurodevelopmental assessments performed at corrected age (CA) 3 to 4 months for estimating cerebral palsy, cognitive impairment, and neurodevelopmental impairments and their severity at CA 24 to 36 months in infants at high risk of adverse neurodevelopmental outcomes? Findings This diagnostic study including 116 infants born extremely preterm or preterm with extremely low birth weight or born full term with encephalopathy and received therapeutic hypothermia found that early neurodevelopmental assessments were effective in identifying infants with cerebral palsy and predicting its severity. However, the assessments had limited accuracy in identifying cognitive impairment and its severity, as well as in detecting any neurodevelopmental impairment and its severity. Meaning These findings support the potential to identify cerebral palsy and its severity as early as CA 3 to 4 months through early neurodevelopmental assessments, but the role of these tests is limited in identifying cognitive and neurodevelopmental impairments.

However, the assessments had limited accuracy in identifying cognitive impairment and its severity, as well as in detecting any neurodevelopmental impairment and its severity.
Meaning These findings support the potential to identify cerebral palsy and its severity as early as CA 3 to 4 months through early neurodevelopmental assessments, but the role of these tests is limited in identifying cognitive and neurodevelopmental impairments.

Introduction
Prematurity and hypoxic-ischemic encephalopathy (HIE) represent critical concerns in neonatal health, posing significant risks to the well-being of newborns.Prematurity affects approximately 1 in 10 births globally, 1 whereas HIE affects 1.6 per 1000 live births in high-income countries and 12.1 per 1000 in low-or middle-income countries. 2Interventions to mitigate the risk of morbidities linked to prematurity and HIE are limited, 3 leaving affected infants at an elevated risk of neurodevelopmental complications that adversely affect motor skills and cognitive abilities.While various neurodevelopmental concerns exist, cognitive impairment and cerebral palsy stand out as primary concerns, affecting approximately 16.9% and 6.9% of very preterm infants, respectively. 4Early detection of these complications is recommended because prompt identification allows for timely intervention, promoting optimal motor and cognitive development in infants. 5veral tools for early neurodevelopmental assessments in infants are available, including the Prechtl General Movements Assessment (GMA), Hammersmith Infant Neurological Examination (HINE), and magnetic resonance imaging.While individual tools are valuable, a comprehensive approach involving these neurodevelopmental assessments, medical history, and neuroimaging is strongly recommended for early detection of cerebral palsy. 5However, several nuances with respect to early neurodevelopmental assessments must be addressed.
First, the existing data primarily originate from research studies, and there need to be more data from practical settings comparing the early neurodevelopmental assessments with long-term outcomes, which is essential for understanding the effectiveness in practical settings.Second, these early neurodevelopmental assessments have shown high predictability for identifying cerebral palsy, but their effectiveness in detecting cognitive and other neurodevelopmental impairments (NDI) is limited. 6While early neurodevelopmental assessments primarily focus on evaluating motor function and development, systematic reviews have highlighted that these assessments may be useful in identifying cognitive delay to some extent while also emphasizing the limited literature specifically addressing cognitive impairment. 6,7Third, there is a lack of sufficient evidence regarding the effectiveness of these tools in evaluating other NDI commonly used in clinical trials, especially those incorporating a sensory component alongside cognitive and motor components.To address these knowledge gaps, our study was designed to assess the broader ability of early neurodevelopmental assessments to predict cerebral palsy, cognitive impairment, and other NDI, including their severity, within a practical setting, at 24 to 36 months' corrected age (CA).

Methods
This diagnostic study was approved by the Monash Health Human Research Ethics Committee as a quality assurance project, and parental consent was not required because it was an observational study and the exposure was considered within routine care.This study is reported following the Standards for Reporting of Diagnostic Accuracy (STARD) reporting guideline.

JAMA Network Open | Pediatrics
Early Neurodevelopmental Assessments to Predict Cerebral Palsy

Design, Setting, and Participants
This diagnostic study assessed test accuracy by comparing observational data from our multidisciplinary, early neurodevelopmental clinic at CA 3 to 4 months with long-term neurodevelopmental outcomes at CA 24 to 36 months.This study included children born between January 1, 2019, and July 30, 2021.We chose this time frame to allow for neurodevelopmental outcome assessments, with the end point for these assessments set at 36 months.Infants considered at high risk of cerebral palsy or other NDI were recruited from the early neurodevelopmental clinic 8 at Monash Children's Hospital, Melbourne, Australia, a perinatal and surgical tertiary neonatal unit providing care to preterm and term infants in Victoria, Australia.These infants were born extremely preterm (born before 28 weeks) or had extremely low birth weight (<1000 g) or had moderate to severe HIE who received therapeutic hypothermia.

Index Test
Skilled physiotherapists, occupational therapists, and neonatologists trained in GMA conducted early neurodevelopmental assessments at CA 3 to 4 months, using the Prechtl GMA 9 and the HINE. 10 The primary predictor was early diagnosis of cerebral palsy or a high risk of cerebral palsy.Additional predictors included the absence of fidgety movements, HINE score, and early diagnosis of neurodevelopmental impairment (NDI).Early cerebral palsy or high risk of cerebral palsy was diagnosed based on absent fidgety movements, low HINE score (<57), and a medical neurological examination involving motor function, muscle tone, reflexes, and coordination.The diagnosis of early NDI included any anomalies found during initial assessments.These could be isolated anomalies, like absent fidgety movements, a low HINE score (<57), or isolated issues noted during medical neurological examinations.It could also encompass multiple anomalies, such as a combination of 2 or more anomalies or a diagnosis of early cerebral palsy or high risk of cerebral palsy.

Reference Standard
Skilled neuropsychologists certified in Bayley Scales of Infant & Toddler Development (BSID) assessments and experienced developmental pediatricians conducted long-term neurodevelopmental evaluations between CA 24 and 36 months.The evaluations encompassed medical neurodevelopmental examination, BSID assessments (III or IV edition), 11 and assessments for blindness and deafness as part of the Australian and New Zealand Neonatal Network (ANZNN) 12 data collection.Our primary outcomes were any grade of cerebral palsy diagnosed on medical neurological examination and moderate to severe cerebral palsy, defined as cerebral palsy with grade II or above based on Gross Motor Function Classification System.We assessed 4 secondary outcomes: cognitive impairment, moderate to severe cognitive impairment, any NDI, and moderate to severe NDI.Cognitive impairment was defined as cognitive score less than 1 SD from the mean based on BSID III or IV.Moderate to severe cognitive impairment was defined as cognitive score less than 2 SD from the mean based on BSID III or IV.Any NDI was defined as the presence of any the following: cerebral palsy; motor, language, or cognitive impairment (score <1 SD on BSID); blindness (vision <6/60 in the better eye), or deafness (need for hearing aid).Moderate to severe NDI was defined as the presence of any following: moderate to severe cerebral palsy, moderate to severe motor, language, or cognitive impairment (score <2 SD on BSID); blindness, or deafness.

Data Collection
The early neurodevelopmental clinic at Monash Children's Hospital prospectively collected neurodevelopmental data at CA 3 to 4 months, while demographic and long-term data were obtained from the ANZNN, a collaborative network encompassing all 29 neonatal intensive care units in Australia and New Zealand.Monash Children's Hospital contributes data to ANZNN, which systematically gathers information for preterm infants born at less than 32 weeks' gestational age admitted to any participating neonatal intensive care unit.Specifically, ANZNN collects neurodevelopmental data for infants at high risk of adverse neurodevelopmental outcomes up to ages 2 to 3 years, including infants born before 28 weeks' gestation or weighing less than 1000 g at birth, as well as infants with moderate to severe HIE.

Statistical Analyses
Diagnostic test accuracy values, along with corresponding 95% CIs, were calculated for early cerebral palsy or high risk of cerebral palsy, early NDI, and absent fidgety movements using MedCalc statistical software version 19.2.6 (MedCalc Software).Receiver operating characteristic curve analysis was conducted to obtain the area under the curve (AUC) and 95% CI for the HINE score, using SPSS software, 2021 release (IBM).Other analyses, including summarizing demographic variables (presented as the median and IQR due to nonnormal distribution) and predictive probability of cerebral palsy, were performed using Stata software version 17.0 (StataCorp).Data were analyzed from December 2023 to January 2024.

Primary Outcome
The early cerebral palsy or high risk of cerebral palsy diagnosis demonstrated an accuracy of 85% (95% CI, 77% to 91%), correctly identifying 99 of 116 infants evaluated for cerebral palsy (Table 1 and Figure 1).Among 13 infants diagnosed with cerebral palsy, the early cerebral palsy or high risk of cerebral palsy diagnosis showed a sensitivity of 92% (95% CI, 63% to 99%) by correctly identifying 12 infants.Notably, it achieved a 100% (95% CI, 59% to 100%) sensitivity (7 of 7 infants) in identifying all patients with moderate to severe cerebral palsy.Furthermore, the absence of fidgety movements exhibited a comparable accuracy of 81% (95% CI, 73% to 88%), and the HINE score displayed good discriminatory power, with an AUC of 0.88 (95% CI, 0.79 to 0.97) for predicting cerebral palsy (Figure 2).On the other hand, the early NDI predictor, while correctly identifying all infants with cerebral palsy, also yielded false positives, resulting in a lower accuracy of 48% (95% CI, 38% to 57%).
The probability of cerebral palsy in infants was calculated depending on the anomalies noted on the early assessments, with none of the infants developing cerebral palsy having a HINE score within reference range (>57) and fidgety movements.The probability was noted at 7% if an infant had either a low HINE score or absent fidgety movements.If both were present (low HINE score and absent fidgety movements), the probability was 24%, which was not significantly different compared with infants with either a low HINE score or absent fidgety movements alone (absolute difference, 17% [95% CI, −2% to 36%]).However, infants with a very low HINE score (<40) and absent fidgety movements had a significantly higher predicted probability of 67%, in contrast with infants with either a low HINE score or absent fidgety movements alone (absolute difference, 60% [95% CI, 21% to 98%]) (Figure 3).

Secondary Outcomes Cognitive Impairment
The accuracy of various early neurodevelopmental assessments was notably limited, ranging from 44% to 60%, in predicting cognitive impairment and its severity (Table 2 and Figure 1).Although the specificity of certain assessments showed modest values, ranging between 76% and 80%, the sensitivity and positive and negative predictive values demonstrated suboptimal performance.The area under the curve was 0.88 (95% CI, 0.79-0.97)for cerebral palsy (A) and 0.62 (95% CI, 0.51-0.73)for cognitive impairment (B).
Additionally, the HINE score displayed limited discriminatory power in predicting cognitive impairment, reflected by an AUC of 0.62 (95% CI, 0.51 to 0.73) (Figure 2).

NDI
Similar to cognitive impairment, the predictive accuracy of various early neurodevelopmental assessments was notably low, ranging from 31% to 59%, when predicting NDI and its severity (Table 2).Moreover, the sensitivity and negative predictive values showed suboptimal performance across the different assessments, while the positive predictive value and specificity were modest in predicting NDI and its severity, respectively.

Discussion
This diagnostic study, leveraging data from infants considered at high risk of adverse neurodevelopmental outcomes, assessed the effectiveness of early neurodevelopmental assessments at CA 3 to 4 months in identifying cerebral palsy, cognitive impairment, and other NDI and their severity at CA 24 to 36 months.Early assessments, including early cerebral palsy or high risk of cerebral palsy, absent fidgety movements, and HINE scores, exhibited high accuracy in identifying cerebral palsy.Moreover, these assessments achieved a sensitivity and negative predictive value of 100% for detecting moderate to severe cerebral palsy.Furthermore, the study an increase in the predicted probability of cerebral palsy with the number and severity of anomalies identified in early neurodevelopmental assessments.With respect to predicting cognitive and neurodevelopmental impairment and their severity, we found limited accuracy, sensitivity, and negative predictive value of early neurodevelopmental assessments.
While current research using observational data offers insights into the feasibility of early neurodevelopmental assessments for diagnosing cerebral palsy at an early stage, 8,[13][14][15][16][17][18][19] there exists a notable gap in studies directly comparing early data with long-term outcomes.Observational data studies are crucial, as they offer insights into the effectiveness of assessments in contextual settings.
Although systematic reviews of research studies have consistently shown high accuracy in early neurodevelopmental assessments for identifying cerebral palsy, [20][21][22] the direct applicability of these findings in clinical settings is limited.Some studies using clinical data have found lower sensitivities or accuracies in predicting cerebral palsy, [23][24][25] while others have found higher accuracy. 26This discrepancy is likely influenced by various factors, including the characteristics of the studied population, 25 the prevalence of cerebral palsy or NDI, 27 and the level of training in the assessments. 28,29Additionally, some researchers argue that continued follow-up assessments may enhance accuracy compared with 1-time assessments. 30Nevertheless, our findings support the notion that these assessments maintain high accuracy in practical settings, aligning with both previous research [20][21][22] and clinical studies. 26ing our study results in a simulated scenario of 100 infants with high risk of cerebral palsy or NDI, with an observed 11% incidence of cerebral palsy and a 6% incidence of moderate to severe cerebral palsy, the data suggest that standardized assessments with a finding of early cerebral palsy or high risk of cerebral palsy at CA 3 to 4 months would identify 10 of 11 infants with cerebral palsy and all infants with moderate to severe cerebral palsy.However, this diagnosis would also falsely identify 15 infants as having cerebral palsy.In contrast, if the early cerebral palsy or high risk of cerebral palsy diagnosis is used to identify NDI, considering a 64% incidence of NDI from this study, this diagnosis would only detect 15 of 64 infants with NDI while incorrectly identifying 20 infants with NDI.Overall, these data confirm the high sensitivity and accuracy of early standardized assessments in detecting cerebral palsy and gauging its severity, aligning with prior research [20][21][22] ; however, our findings underscore the restricted effectiveness of these assessments in identifying cognitive impairment and NDI and their severity.
Ideally, a test should be able to identify all patients with the disease, but a highly sensitive test often comes with an increased false positive rate.This phenomenon was observed in our study, where the early cerebral palsy or high risk of cerebral palsy diagnosis exhibited high sensitivity but had a 15% false positivity rate.When assessing this false positivity rate, it is worth considering that it might have been influenced by the fact that infants diagnosed with early cerebral palsy or high risk of cerebral palsy underwent early intervention, potentially positively impacting their development.Early neurodevelopmental assessments are customarily designed to focus on motor evaluation and do not inherently encompass the assessment of cognitive domains.While there is a possibility that these assessments may indirectly offer insights into cognitive domains, 5,6,17 given that disorders affecting cognition may also impact motor function, our study's findings do not demonstrate their utility in assessing cognitive impairments.Furthermore, their utility in addressing broader NDI is constrained, as this outcome is comprehensive and involves cognitive impairment within its definition.
Our study has several notable strengths.Unlike previous investigations reliant on research outputs, our study draws on observational data gathered directly from clinical settings.This distinctive approach lends credibility to our results, reflecting the practical application of early neurodevelopmental assessments.Another strength of our study is the meticulous prospective collection of data, ensuring the acquisition of reliable and valid information while minimizing the risk of recall bias.Furthermore, the assessors of long-term outcomes used standardized objective measures during assessments, which further reduces the risk of bias in the assessment process.
Additionally, the comprehensive nature of our comparisons contributes to our study's depth, thereby enhancing the robustness of our findings.

Limitations
It is important to acknowledge some limitations.While our study comprehensively compared early neurodevelopmental assessments with long-term outcomes in a clinical setting, the study population was relatively small, as evidenced by wide CIs in some diagnostic test accuracy measures.
Additionally, the data were sourced from a single center, which may restrict the generalizability of our findings.Furthermore, the inclusion of infants with high risk, particularly with most infants being born extremely preterm, means that our findings may not be directly applicable to a more diverse range of infant populations.

Conclusions
In this diagnostic study of infants at high risk of cerebral palsy and NDI, our findings affirm the effectiveness of early standardized assessments in detecting cerebral palsy and determining its severity as early as CA 3 to 4 months, even in clinical settings.This emphasizes the crucial role of integrating these assessments into clinical practice, enabling early interventions with the potential to influence the progression of the disease.Furthermore, our findings indicate that these assessments have limited utility in detecting cognitive impairment and other NDI, emphasizing the ongoing necessity for longer-term assessments to accurately identify impairments beyond cerebral palsy.
early neurodevelopmental assessments performed at corrected age (CA) 3 to 4 months for estimating cerebral palsy, cognitive impairment, and neurodevelopmental impairments and their severity at CA 24 to 36 months in infants at high risk of adverse neurodevelopmental outcomes?Findings This diagnostic study including 116 infants born extremely preterm or preterm with extremely low birth weight or born full term with encephalopathy and received therapeutic hypothermia found that early neurodevelopmental assessments were effective in identifying infants with cerebral palsy and predicting its severity.

Figure
Figure 1.Accuracy of Early Neurodevelopmental Assessments and Absence of Fidgety Movements at Corrected Age 3 to 4 Months Compared With Neurodevelopmental Outcomes Assessed at Corrected Age 24 to 36 Months

Figure 2 .
Figure 2. Receiver Operating Characteristic Curve for Hammersmith Infant Neurological Examination Scores in Predicting Cerebral Palsy and Cognitive Impairment

Figure 3 .
Figure 3. Probability of Cerebral Palsy by Anomalies Identified on the Early Neurodevelopmental Assessments

Table 1 .
Diagnostic Test Accuracy Values of Early Neurodevelopmental Outcomes for Predicting Cerebral Palsy Abbreviations: AUC, area under the curve; eCP, early diagnosis of cerebral palsy or high risk of cerebral palsy; eNDI, early diagnosis of neurodevelopmental impairment; HINE, Hammersmith Infant Neurological Examination; NA, not applicable; NPV, negative predictive value; PPV, positive predictive value.
1. Accuracy of Early Neurodevelopmental Assessments and Absence of Fidgety Movements at Corrected Age 3 to 4 Months Compared With Neurodevelopmental Outcomes Assessed at Corrected Age 24 to 36 Months

Table 2 .
Diagnostic Test Accuracy Values Early Neurodevelopmental Outcomes for Predicting Cognitive Impairment and NDI Abbreviations: AUC, area under the curve; eCP, early diagnosis of cerebral palsy or high risk of cerebral palsy; eNDI, early diagnosis of neurodevelopmental impairment; HINE, Hammersmith infant neurological examination; NDI, neurodevelopmental impairment; NPV, negative predictive value; PPV, positive predictive value.