EEG Profile Associated with the Developmental Outcomes of Preschool Children Born Very Preterm

Abstract

Background

Preschool children born preterm are at risk of developmental delay and cognitive deficits. However, the neurocognitive underpinnings of the cognitive deficits in preterm children remain unclear.

Aims

This study aimed to compare quantitative EEG (qEEG) profiles and cognitive function between preterm and full-term children aged 4 to 6 years.

Method

Children's cognitive ability was assessed using the Griffiths Development Scales-Chinese (GDS-C). Resting-state EEG was recorded, and spectrum analysis was conducted. In addition to conventional band power analysis, aperiodic components (exponent and offset) and center frequency of the dominant oscillatory peak were calculated by fitting oscillations and one-over-f (FOOOF).

Results

The study recruited 28 preterm-born children (15 girls, mean age = 62.14 months, SD = 7.31) and17 age-matched full-term children (7 girls, mean age = 59.8 months, SD = 8.49) Preterm children showed significantly elevated aperiodic offset and exponent. They also exhibited significantly slower center frequency of the dominant oscillatory peak and higher delta and theta band power compared to full-term children.

Conclusion

Preterm children demonstrated elevated aperiodic offset and exponent and slower center frequency of the dominant oscillatory peak compared to full-term children, at preschool age. These findings suggest that preschool children born preterm may exhibit less efficient functional neural connections and delayed brain maturation.

Keywords

preterm children resting-state EEG 1/f power law pattern developmental assessment brain maturity EEG spectrum power

Introduction

Despite advances in perinatal care and improved survival rates,¹ preterm births remain a significant risk factor for neurodevelopmental disabilities, especially those delivered at earlier gestation.^2–4 Infants born prematurely are at high risk of neurodevelopmental disorders such as global developmental delay,⁵ cognitive deficits,⁶ and attention deficit hyperactivity disorder (ADHD).⁷ Such neurocognitive and behavioral impairments affect executive function, learning, and academic performance, which often persist well beyond middle childhood.⁸ The majority of enhanced developmental surveillance programs for preterm children advocate comprehensive developmental assessments when children are around 18-24 months old. The National Institute for Health & Care Excellence (NICE) also recommends further evaluations by multidisciplinary health professionals in preterm children at the uncorrected age of 4.⁹ However, there are no further evaluations by healthcare professionals after the age of 4. The period between ages 4 and 7 is a critical period, in which children further improve their cognitive development, and their logic and reasoning changes from symbolic thought to intuitive thought. This is represented by a greater dependence on deductive thinking rather than primitive thinking.¹⁰ Previous study found that preterm children had poorer cognitive outcomes compared with their term peers at preschool age.¹¹ However, the underlying neurological mechanism of this gap is not clear.

Few studies have investigated the neural electrophysiological differences between preterm and term children beyond the neonatal period. In adolescents, Rommel et al¹² found that preterm adolescents had increased low-frequency EEG spectral power (delta power) during the eyes closed condition compared to term adolescents.¹² Additionally, Twihaar et al¹³ found that preterm adolescents showed decreased high-frequency EEG spectral power (beta power) compared to age-matched term children.¹³ In young school-aged children, Rozhkova¹⁴ reported preterm children had increased low-frequency activities,¹⁴ whereas Poorun et al¹⁵ found preterm children had decreased high-frequency activities under anesthesia.¹⁵ The majority of the studies showed there were distinct differences in the qEEG profiles between preterm and full-term school-aged children or adolescents. Specifically, qEEG profiles of preterm children were characterized by elevated slow frequency activities (delta and theta band power) and attenuated high-frequency activities (alpha and beta band power).

Recent advances in EEG time-frequency signal processing have suggested that the canonical frequency band analysis might not show the full picture of the neural activities reflected in electrophysiological recordings.¹⁶ It is argued that the conventional method of calculating the average power of a given frequency band (absolute power) does not consider the impact of the background aperiodic component of human neural electrophysiological signals, resulting in confounding information.^16,17 This is also the case for secondary neural metrics based on absolute power, such as the ratio of power between two frequency bands or the relative power (ie, the ratio of power in one frequency band divided by the total absolute power), commonly interpreted as the relative strength of periodic oscillation. The aperiodic component shows a 1/f-like shape in the signal's power spectrum, ie, the power of a frequency band is inversely proportional to its frequency. Because of overlapping, the aperiodic component becomes entangled with the periodic oscillation pattern in canonical frequency band analysis, leading to misinterpretation of the electrophysiological neural activity in different frequency bands. For example, the theta/beta ratio, a commonly suggested biomarker for the attentional process,^18–21 would show a greater effect from the aperiodic non-oscillatory component than the hypothesized periodic oscillatory component. Furthermore, a growing body of evidence suggest that the 1/f-like aperiodic component is not background noise, but actually reflects fundamental brain functions. For instance, 1/f-like brain activity was found to be correlated with cognitive processing²² and age-related changes in the aperiodic component in adults²³ and children.²⁴ Moreover, previous research suggests that the aperiodic and periodic components of spontaneous brain activities reflect distinct aspects of cognitive functions,²⁵ and the 1/f-like brain dynamics might be associated with neuroplasticity moderated by neuromodulator systems.²⁶ Taken together, it is imperative to separate the aperiodic and periodic components in electrophysiological signals to more accurately dissect brain activities and their functional roles.

The goal of the current study was to use new analytical EEG methods to compare both aperiodic components and periodic dominant oscillations in the qEEG profiles in preschool children aged 4-6 years born preterm and full-term. This study also aimed to characterize the associations between cognitive ability and qEEG parameters between preterm and full-term children.

Methods

Participants

Participants were recruited from the preterm database at local clinics, which contains data on over 700 children born at ≤33 weeks with very low birth weight <1.5 kg who visited local clinics between 2008 and 2020. All of them received complete comprehensive developmental assessment using the GDS-C at 18-24 months. Inclusion criteria included children born preterm aged between 4 and 6 years with their first assessment showing normal development. Age-matched healthy control subjects were recruited from the community. The exclusion criteria included neurological symptoms or diseases, neurodevelopmental disorders (eg, ADHD, ASD); psychiatric disorders, diagnosis of cerebral palsy; history of structural brain abnormalities on CT/ MRI or grade III or IV IVH on either side of the brain; genetic/ syndromal disorders; or history of seizure or electroencephalogram abnormalities related to epilepsy and intellectual disability.

Parents of the participating children were invited to fill in a questionnaire to report their basic demographic details, their children's emotional, behavioral, ADHD symptoms and psychosocial well-being. Children were given a developmental assessment and an EEG.

Questionnaires and Assessments

Cognitive ability. Children's cognitive ability was assessed using the Griffiths Development Scales-Chinese (GDS-C), which has been adapted and validated for the Chinese population from the Griffiths Mental Development Scales – Extended Revised (GMDS-ER). This multidimensional assessment scale has been widely used to examine children's development.²⁷ The sub-quotient for the performance subscale (subscale E) can be used to estimate the non-verbal cognitive performance, whereas the sub-quotient for the practical reasoning subscale (subscale F) can be used to estimate children's general cognitive functioning.²⁸

Behavioral problems. The Strengths and Difficulties Questionnaire (SDQ) was used to measure children's behavior. It has 25 items scored on a 3-point scale. The total difficulties score measures emotional symptoms, conduct problems, hyperactivity/inattention, and peer relationship problems. A higher score indicates more psychosocial problems.²⁹

ADHD symptoms. Chinese SWAN is a validated instrument for assessing ADHD symptoms in Chinese children in Hong Kong.³⁰ The SWAN questionnaire was originally devised by Swanson et al in 2005 based on the criteria for ADHD in the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV).

Quality of Life. Pediatric Quality of Life Inventory (PedsQL) is a valid and widely used scale to measure core dimensions of health in healthy children and adolescents and in those with acute and chronic health conditions. Generic scales from PedsQL³¹ including psychical functioning, emotional functioning, and social functioning were used in this study.

Socioeconomic status (SES). SES was estimated based on parents’ income, education level, and average family living area (home size/number of family members) using principal component analysis (PCA).³² Z-scores were calculated to represent the SES index.

EEG Experimental Procedures

Single-channel EEG was used to assess the qEEG profiles of participants. We used three gold cup surface electrodes (Cz, A1 and A2), which were positioned on the scalp according to the international 10-20 system. The Cz electrode was placed at the midpoint between the nasion and inion, whereas A1 and A2 reference electrodes were attached on the mastoid region, the bony prominence behind the ears. The electroactivity of Cz was referenced to the average electroactivity of A1 and A2 at a sampling rate of 256 sample per second. The impedance of each EEG channel was checked and kept below 10 kΩ.

The resting-state EEG was recorded in all participants, which included two parts lasting 2.5 min each. As young preschool children likely cannot stay in a resting state with their eyes open without any focus, we used a silent cartoon movie to help children relax and stay in their seat. In the eyes open (EO) condition, participants were asked to watch a silent cartoon movie for 2.5 min. In the eyes closed (EC) condition, participants were asked to close their eyes for 2.5 min. We observed a greater variation of EEG spectrum in the EO condition, therefore we focused on the EEG results from the EC condition.

Children's movement were monitored during the experiment and manually marked in the EEG recording. The whole EEG session was recorded using Natus NeuroWorks hardware and software (Natus Medical Inc, Oakville, Canada).

EEG Data Processing

All offline EEG data processing was performed in MATLAB EEGLAB³³ using custom scripts. The data were first filtered offline through a 1-40 Hz band pass filter. The Cz channel readings were referenced to the average A1 and A2 readings. Both A1 and A2 were also cross-referenced. Any EEG segments containing artifacts were rejected based on any movement events during the experiment and by visually inspecting the data. The power spectrum density was estimated using Welch's method³⁴ on artifact-free continuous EEG data from both EO and EC conditions. The power spectrum was generated in the 1-30 Hz frequency range, using Fast Fourier transform (FFT) for each 500-ms segment with 50% overlap between adjacent segments.

Conventional Canonical Frequency Band Analysis

The absolute band power (μV²/Hz) was calculated by averaging the power spectrum in the following frequency bands: delta (1-3 Hz), theta (4-7 Hz), alpha (8 −12 Hz), beta1 (13-18 Hz), beta2 (19-21 Hz), and beta3 (22-30 Hz). The relative band power was calculated using the total band power in each frequency band divided by the total band power in the 1-30 Hz range.

Aperiodic Component Modeling

We used Fitting oscillations and one-over-f (FOOOF), an open-source python-based spectrum fitting algorithm, to parameterize the brain power spectrum to separate narrowband oscillation peaks and the 1/f-like aperiodic component. FOOOF was used to fit the parametric curve to the EEG spectrum, modeled as the summation of the Gaussian functions (for the narrowband peaks) and linear line(s) in the log-log space. This generated offset (intercept) and exponent (slope) parameters of the 1/f component, which can automatically be detected as oscillations peaks based on the model fitting. To avoid overfitting, we restricted the maximum number of peaks to three, in which the peak with maximum power was selected as the dominant peak. During preschool years, children go through an age-related shift in their dominant oscillations,²⁴ transitioning from the theta to the alpha frequency band. To account for this change, the analysis considered dominant oscillations within the frequency range of 4 to 12 Hz. We applied the fitting to each individual power spectrum estimated in both the EC and EO condition, separately. The FOOOF toolbox is available on https://github.com/fooof-tools/fooof.

Statistical Analysis

Independent T-test was used to compare the developmental sub-quotient scores, parental-reported scores (SDQ, PedsQL, SWAN), and qEEG profiles (absolute and relative band power, aperiodic parameters—exponent and offset) between preterm and term children. To control for potential confounding variables, a linear model was used for between group comparisons, with the group (preterm vs term) inputted as the fixed factor, and children's age, sex and parents’ SES inputted as the covariates. Multiple comparisons in the exploratory analyses were adjusted for using the false discovery rate (FDR) procedure (Benjamini–Hochberg), with FDR-adjusted P values (q values) reported

Correlational analysis was performed between 1) children's sub-quotient scores for performance and practical reasoning and aperiodic EEG parameters (exponent and offset); and 2) children's sub-quotient scores for performance and practical reasoning and absolute/relative EEG band power. The correlations between the focal variables and other measurements were also explored, and partial correlation was used to adjust for age, sex, and SES.

Statistical analysis was conducted using SPSS 28.0 (IBM, New York, USA) and R 4.1.³⁵ Some illustrations were plotted using MATLAB (MathWorks, MA, USA). For all statistical comparisons, a P-value < .05 was considered statistically significant. The statistical code, EEG processing code, and anonymous data are available upon request from the corresponding author.

Results

Participant Characteristics

A total of 40 preterm children aged 4-6 years at the time of the recruitment were invited to join the study, of which four declined to join or could not be reached. The remaining 36 eligible children completed the assessments. After the assessments, two children had autistic features, one had ADHD features, one had abnormal brain MRI, and one exhibited intellectual disability, and three children had excessive noise on the EEG data, and were excluded from the analysis. A total of 20 term children participated in the study and completed the assessments. Two had autistic features/ADHD features and one had excessive noise on the EEG data, and were excluded from the analysis. Overall, 28 preterm children and 17 term children were included in the final analyses. The mean age of preterm children at the time of the study was 62.14 (SD = 7.31) months and the mean age of term children was 59.8 (SD = 8.49) months. The demographic variables of the included participants are shown in Table 1. Preterm children and term children showed no significant differences in gender, age, family SES index, parent-rated SDQ, PedsQL, and SWAN scores. Preterm children had lower GSD-C E sub-quotient score (mean = 97.93, SD =26.02) and GSD-C F sub-quotient score (mean =102.28, SD =11.68) compared to term children (GSD-C E mean =114.64, SD =33.01; GSD-C F, mean =110.29, SD =16.29). However, the differences in the GSD-C sub-scores between preterm and full-term children did not reach statistical significance.

Table 1.

Comparison of Demographics and Developmental Variables Between Preterm and Full-Term Children.

	Full-Term Birth Children (n = 17)		Preterm Birth Children (n = 28)
	Count/Mean	Standard Deviation/Percentage	Count/Mean	Standard Deviation/Percentage	P-value^e
Demographic variables
Gestational age (weeks)^a	>= 37		29.65	2.79
Birth weight (kg)^a			1.15	0.36
females (%)	7	41%	15	53%	.432
age(months)	59.82	8.49	62.14	7.31	.337
SES index	−0.03	0.81	0.09	1.20	.732
Behavioral problems
SDQ^b	11.89	4.88	13.14	6.04	.474
PedsQL physical problems^c	34.21	5.39	34.04	4.83	.914
PedsQL emotional problems^c	18.79	2.86	19.32	2.67	.552
PedsQL social problems^c	20.857	3.483	21.000	2.854	.888
SWAN^d	−0.71	0.91	−0.66	1.02	.447
Cognitive ability
GDS-C: raw score E	78.71	14.03	73.93	11.81	.227
GDS-C: raw score F	81.65	11.41	79.57	10.35	.533
GDS-C: E standard Score	114.64	33.01	97.93	26.02	.063
GDS-C: F standard Score	110.29	16.29	102.28	11.68	.062

gestational age and birth weight of preterm birth children were retrieved through their clinical notes. However, such Information was not available for most term birth children.

SDQ total difficulties score, higher score indicates more parent-reported difficulties.

PedsQL subscales: higher scores indicate more parent-reported problems.

The mean SWAN score. Higher score indicates more ADHD features.

P-value as revealed by Fisher's exact test for categorical variables and independent t-test for continuous variables.

qEEG Profile

Periodic Parameters

We calculated each participant's periodic center frequency of the dominant oscillatory peak using the 1/f model, as seen in Figure 1a-c. The dominant oscillation range was defined as 4-12 Hz in the resting-state for pre-school children,^24,36 and the peak selected for the maximum power and band width was within 3 Hz of the peaks modeled by the FOOOF algorithm. The result indicated that preterm children had significantly slower resting-state center frequency of the dominant oscillatory peak compared to term children (Cohen's d = -0.80 (−1.43–-0.18), P = .005, q = .009; Figure 1d, Table 2). Notably, the center frequency in the theta range (4-7 Hz) was observed in 2 out of 17 full-term children (11.7%), in contrast to 9 out of 28 preterm-born children (32.1%). The center frequency of the dominant oscillatory peak in each participant significantly correlated with their developmental assessment results in both performance score (GDS-C E: r = 0.32, P = 0.030, q = 0.045) and practical reasoning ability (GDS-C F: r = 0.42, P = 0.005, q = 0.007), see Figure 1e-f. These correlations did not differ in each subgroup (preterm children vs term children). For correlations in each subgroup, see Table S1 and S2.

Figure 1.

1/f (aperiodic) modelling of EEG spectra in preschool children and group differences in the center frequency of the dominant oscillatory peak between preterm and full-term children. (a) Example 1/f (aperiodic) modelling of the resting-state EEG spectrum from one participant in the eyes-closed (EC) condition. Black line: original EEG power spectrum; orange line: full model fit; blue dashed line: aperiodic fit. The aperiodic component is defined as log(P) = Offset - log(f^exp), where P denotes power, f frequency, and exp exponent. (b) Group-averaged original EEG power spectra in the EC condition for children born preterm (orange) and full term (blue). (c) Group-averaged model-fitted EEG power spectra in the EC condition for children born preterm (orange) and full term (blue). (d) Group comparison of the center frequency (CF) of the dominant oscillatory peak between full-term and preterm children; error bars indicate standard deviations; P < .01. (e–f) GDS-C E and F raw scores as a function of the center frequency (CF) of the dominant oscillatory peak in preterm (orange points/line) and full-term (blue points/line) children; shaded bands indicate 95% confidence intervals.

Table 2.

Comparison of EEG Aperiodic Parameters and Frequency Band Power in Preterm and Full-Term Children in the Eyes Closed Condition.

Condition	Full-Term Children (n = 17)		Preterm Children (n = 28)		Statistical Test
	Mean	SD	Mean	SD	Raw P-value	Adjusted P-value^a	FDR q^b	Cohen's d (95% CI)
Aperiodic offset	2.26	0.16	2.39	0.23	. 047*	.028*	0.042*	0.63 (0.01-1.25)
Aperiodic exponent	1.63	0.15	1.75	0.20	.033*	.043*	0.043*	0.66 (0.04-1.27)
Peak frequency of the dominant oscillation	9.16	0.77	8.21	1.37	.005**	.003**	0.009*	−0.80 (−1.43–−0.18)
delta	1.76	0.13	1.88	0.20	.029*	.008**	0.024*	0.68 (0.06-1.30)
theta	1.23	0.15	1.44	0.25	.003**	<.001**	0.005**	0.96 (0.33-1.60)
alpha	0.94	0.15	0.93	0.17	.878	.890	0.908	−0.06 (−0.66-0.54)
beta1	0.26	0.06	0.33	0.18	.116	.154	0.308	0.48 (−0.14-1.09)
beta2	0.18	0.15	0.15	0.20	.642	.908	0.908	−0.16 (−0.77-0.44)
beta3	0.06	0.19	0.01	0.22	.465	0.732	0.908	−0.24 (−0.84-0.37)
relative delta	0.36	0.06	0.38	0.10	.569	.852	0.852	0.16 (−0.45-0.76)
relative theta	0.22	0.02	0.21	0.04	.332	.118	0.533	−0.26 (−0.87-0.34)
relative alpha	0.29	0.07	0.32	0.08	.127	.208	0.533	0.46 (−0.15-1.07)
relative beta1	0.09	0.01	0.10	0.04	.220	.266	0.533	0.31 (−0.29-0.92)
relative beta2	0.03	0.02	0.02	0.03	.249	.447	0.671	−0.33 (−0.94-0.27)
relative beta3	0.02	0.11	−0.02	0.14	.318	.570	0.685	−0.29 (−0.90-0.31)

Adjusted for covariates including age, sex, and SES. **P < .01, *P < .05.

False discovery rate (FDR) correction was applied using the Benjamini–Hochberg (BH) procedure to the adjusted P values in two prespecified families: (i) the first three outcomes (aperiodic offset, aperiodic exponent, peak central dominant frequency) and (ii) the six canonical band-power outcomes (delta, theta, alpha, beta1–beta3).

Aperiodic Parameters

Compared with full-term children, preterm children showed significantly higher aperiodic offset (raw P = .047; covariate-adjusted P = .028; FDR q = 0.042; Cohen's d = 0.63, 95% CI 0.01-1.25) and aperiodic exponent (raw P = .033; covariate-adjusted P = .043; FDR q = 0.043; Cohen's d = 0.66, 95% CI 0.04-1.27) (Table 2; Figure 2a–b). After adjusting for age, sex, and SES, group differences remained significant for both aperiodic parameters (offset adjusted P = .028; exponent adjusted P = .043) (Table 2).

Figure 2.

Aperiodic exponent and offset of EEG spectra in preterm-born and term-born children. (a–b) Group comparison of the aperiodic exponent and offset between preterm-born and term-born children; error bars indicate standard deviations; P < .05. (c–f) GDS-C E and F raw scores as a function of aperiodic exponent and offset in preterm-born (orange points/line) and full-term (blue points/line) children; shaded bands indicate 95% confidence intervals.

Preschool children's practical reasoning ability (GDS-C F) was significantly correlated with both aperiodic offset (r = −0.557, P < .001, q < 0.001) and aperiodic exponent (r = −0.351, P = .018, q = 0.018) in the combined cohort (Table S2). After adjusting for age, sex, and SES, these associations remained significant (offset r = −0.534, P = .000, q = 0.001; exponent r = −0.403, P = .008, q = 0.012) (Table S2). Preschool children's performance score (GDS-C E) was significantly correlated with aperiodic offset in the combined cohort (r = −0.352, P = .018, q = 0.045), whereas the association with aperiodic exponent was not significant (r = −0.222, P = .143, q = 0.143) (Table S1). Figure 2c–f illustrates these relationships between developmental test scores and the EEG aperiodic profile. There were no significant interactions by group (preterm vs full-term) for the associations between GDS-C E/F and aperiodic parameters. In subgroup analyses, most correlations did not survive FDR correction, consistent with limited power due to the smaller sample sizes (Tables S1–S2).

Canonical Frequency Band Power

When comparing the conventional band power, preterm children had significantly higher absolute delta and theta band power. None of the other absolute band powers and relative band power ratios showed significant differences between groups. See Table 2 for a comparison of the EEG profiles between preterm and term children. See Table S3 for the correlations between canonical frequency band power and aperiodic EEG parameters in the whole sample.

Discussion

This is the first study to demonstrate significant differences in aperiodic neural activity between preschool children born preterm and full term. Preterm children had elevated aperiodic offset and exponent (1/f slope) compared to term children. Since the introduction of algorithms to parameterize the periodic and aperiodic components in the EEG spectrum and approaches for fitting to a 1/f-like model,^16,17 there has been a growing body of research on the functional relevance of the aperiodic slope (exponent) and offset (intercept).^{22,23,37–39} Several studies have shown the association of aperiodic components with cognitive abilities,^22,38 aging,²³ and ADHD status.³⁹ Cellier et al²⁴ found the aperiodic offset and exponent were inversely correlated to the age of the children from 3 to 24 years,²⁴ which suggest that the 1/f offset and exponent could reflect the developmental trajectory of brain maturation. The 1/f offset is highly correlated to low frequency EEG power in canonical frequency bands. Our band power analysis results showed excessive low frequency (delta and theta) resting-state EEG activities in preterm children, which is consistent with previous studies.^12–14 The neural mechanisms underlying the reduction in aperiodic offset throughout children's development might be related to the elimination of the over-production of synapses during early brain development.⁴⁰ This process was found to be associated with the age-related reduction of overall EEG spectral power.⁴¹ As children's brains mature, redundant inter-neuron connections are removed, resulting in more efficient and functionally related neural connections. Therefore, the reduction in aperiodic offset might represent more efficient inter-neuron connections in the maturing brain. On the contrary, the excessive aperiodic offset shown in preterm children could suggest immature brain connectivity.

We also found preterm children had elevated 1/f exponent compared to term children. The 1/f exponent is suggested to be negatively associated with cognitive abilities in adults,^22,38 and with chronological age in the childhood.^24,37 The elevated 1/f exponent may therefore indicate an abnormal state of excitation-inhibition balance,^42,43 a mechanism that has important functional relevance.⁴⁴ An abnormal excitation-inhibition balance could be linked to disrupted or immature neural circuits.⁴⁵ Recently, a magnetoencephalography (MEG) study demonstrated that the 1/f exponent could allow children with autism with below average IQ to be differentiated from those with average intellectual functioning.³⁸ This is consistent with our results that showed a significant difference in the 1/f exponent between preterm and term children, which was also associated with the children's development.

As illustrated in our results and in previous research,¹⁶ conventional canonical frequency band power is partially modulated by the 1/f function. Through separating the aperiodic components from the power spectrum, we have achieved a better estimation of periodic oscillation peak during the resting-state spontaneous neural activity. In our results, preterm children demonstrated resting-state dominant oscillation with a slower peak frequency. Such peak frequency was correlated to the cognitive abilities in our study, and was shown to be associated with the chronical age of children in previous studies.^24,46 In adults, the dominant neural oscillation is in the alpha range, which is argued to have functional importance in cognitive processing, working memory,^47,48 processing speed,²² and inhibition⁴⁹ and selective attention.^50,51 In children, we demonstrated that the dominant oscillation could be extended to the theta range, as reported in previous studies.^24,52,53 More importantly, preterm children showed a slower transition from the theta range to the alpha range, indicating slower brain maturation.

Our results did not show significant differences in beta band power between preterm and term children. Such null findings have also been reported in previous studies.^12–14 Although slower beta waves (13-21 Hz) are suggested to be associated with cognitive processing ability (eg, reaction time), and fast beta waves (22-30 Hz) are correlated with stress, anxiety, and overstimulation,⁵⁴ we did not find significant differences even in the sub-beta bands of beta1 (13-18 Hz), beta2 (19-21 Hz), and beta3 (22-30 Hz) between the preterm and term children. Nevertheless, it is important to note that our study may not be able to reveal differences in beta oscillations, as the dominant oscillations in the children were in the theta and alpha bands.

We have also provided preliminary evidence for associations among aperiodic parameters (1/f offset and exponent), periodic parameters (peak frequency in dominant oscillation), canonical frequency band power (delta wave), and the cognitive ability of preschool children. Future studies with a larger sample size will be needed to confirm such findings. Although developmental assessments and cognitive testing are the gold standards for diagnosing children with developmental disorders or cognitive deficits, qEEG could potentially be used as a screening tool to identify children at risk of cognitive deficits. As previous studies showed close associations between aperiodic parameters and cognitive ability in young children, and our findings showed significant differences in aperiodic parameters between preterm and full-term children, the aperiodic components detected in qEEG might be better electrophysiological markers of developmental delay in preterm preschool children than band power ratios such as theta/beta ratio⁵⁵ or delta/alpha ratio.⁵⁶

Our study findings have important clinical relevance. When conducting developmental assessment for children born preterm, it is standard practice to correct for gestational age up to the age of 2,⁹ assuming preterm children have caught up with their term counterparts by 2 years of age. Nevertheless, our study showed that at the age of 4-6 years, preterm children still have significant brain activity differences compared to term children according to the EEG parameters. It is plausible that the differences in the EEG parameters are related to brain immaturity and delayed formation of neuronal connections and networks in preterm children, considering the third trimester serves as a critical period for the early development of such networks.⁵⁷ Moreover, the delays in the neuronal network formation would then persist beyond preschool in preterm children. Based on our findings showing the close associations between brain activity and cognitive ability of preschool children born preterm, we recommend that follow-up for cognitive developmental assessments should continue beyond the preschool years.

The study had several limitations that could be addressed in future studies. First, one channel EEG was used due to experimental constraints in the study. Although the single channel EEG can produce meaningful research data,^58,59 multiple-channel EEG data would allow the correction and removal of artifacts using algorithms such as Independent Component Analysis (ICA).⁶⁰ Thus, single channel EEG would make the processed EEG data more likely to be affected by motor artifacts. Moreover, the use of single-channel electrodes results in poor spatial resolution, therefore information on specific cortical sources generating the electrical signal on the scalp is very limited. Second, due to the relatively small sample size, the current study may be underpowered to detect correlation between EEG parameters and developmental sub-quotients for examining differences in cognitive ability between preterm and term children. Several large-scale cohort studies^6,61 have reported differences in cognition, attention, and behavior between preterm children and full-term children. Longitudinal studies using resting-state qEEG with a larger sample size are needed to replicate the current findings and further explore the developmental trajectory as well as the neurological basis of the cognitive performance differences between preterm and full-term children. Third, our study focused on eyes-closed resting EEG rather than the eyes-open condition. It was difficult for preschool children to sit still for 3 min without a clear focus; therefore, we introduced a silent cartoon video. However, previous studies using similar procedures²⁴ showed that age-related EEG aperiodic signal differences were exhibited in posterior areas, whereas our study measured the central area, which may not reveal such spontaneous neural activities during the eyes open condition in a resting state. Fourth, a potential sampling bias may exist in this study. Parents of both preterm and full-term children might be more willing to join the study if they think their children have neurodevelopmental issues. However, as the response rate of parents of preterm children was high in this study, the effects of such sampling bias could be minimal.

In conclusion, our study found significant differences in the aperiodic parameters and slower wave band power within the EEG spectrum between preterm and full-term children. The EEG aperiodic offset and center frequency of the dominant oscillatory peak were associated with children's practical reasoning ability, indicating there are functional implications between preterm and full-term children in preschool. Future studies should consider the impact of the 1/f-like aperiodic component of the EEG spectrum, particularly on aperiodic neuronal activities in children. Moreover, monitoring for cognitive and higher executive function should continue beyond preschool for children born preterm.

Supplemental Material

sj-docx-1-eeg-10.1177_15500594261454408 - Supplemental material for EEG Profile Associated with the Developmental Outcomes of Preschool Children Born Very Preterm

Supplemental material, sj-docx-1-eeg-10.1177_15500594261454408 for EEG Profile Associated with the Developmental Outcomes of Preschool Children Born Very Preterm by Yuliang Wang, Ouyang Guang, Tatia M.C. Lee, Cuixin Li, Lok Kan Leung, Shiting Liang and Winnie W.Y. Tso in Clinical EEG and Neuroscience

Footnotes

ORCID iD

Winnie W.Y. Tso

Ethical Considerations

This research was approved by the institutional review board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster: UW 20-455. The signed inform consent sheet were obtained for all the participants recruited in this study.

Author Contributions

Conceptualization: YW, WWYT, TMCL;

Original Draft: YW, WWYT, OG;

Methodology: YW, OG, WWYT;

Data Acquisition: YW, CL,LKL, WWYT;

Formal Analysis: YW;

Supervision : WWYT, TMCL.

Funding: WWYT.

Interpretation of data: All authors

Critical revision of the manuscript: All authors.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by The Society for the Relief of Disabled Children (Hong Kong).

Declaration of Conflicting Interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: I declare that I participated in the design, execution, and analysis of the paper by Wang et al entitled “EEG profile associated with the developmental outcomes of preschool children born very preterm” that I have seen and approved the final version and that it has neither been published nor submitted elsewhere. I also declare that I have no conflict of interest.

Data Availability Statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

Supplemental Material

Supplemental material for this article is available online.

References

Stoll

Hansen

Bell

, et al. Trends in care practices, morbidity, and mortality of extremely preterm neonates, 1993-2012. Jama. 2015;314(10):1039–1051. doi:10.1001/jama.2015.10244

Eryigit Madzwamuse

Baumann

Jaekel

Bartmann

Wolke

. Neuro-cognitive performance of very preterm or very low birth weight adults at 26 years. J Child Psychol Psychiatry. 2015;56(8):857–864. doi:10.1111/jcpp.12358

Serenius

Ewald

Farooqi

, et al. Neurodevelopmental outcomes among extremely preterm infants 6.5 years after active perinatal care in Sweden. JAMA Pediatr. 2016;170(10):954–963. https://doi.org/10.1001/jamapediatrics.2016.1210

Serenius

Källén

Blennow

, et al. Neurodevelopmental outcome in extremely preterm infants at 2.5 years after active perinatal care in Sweden. Jama. 2013;309(17):1810–1820. doi:10.1001/jama.2013.3786

Saigal

Doyle

. An overview of mortality and sequelae of preterm birth from infancy to adulthood. Lancet. 2008;371(9608):261–269. doi:10.1016/S0140-6736(08)60136-1

Anderson

Doyle

. Neurobehavioral outcomes of school-age children born extremely low birth weight or very preterm in the 1990s. Jama. 2003;289(24):3264–3272. https://doi.org/10.1001/jama.289.24.3264

Tso

WW-Y

FKW

Coghill

, et al. Preterm postnatal complications and risk of attention-deficit/hyperactivity disorder. Dev Med Child Neurol. 2023;65(n/a):358–366. https://doi.org/10.1111/dmcn.15401

Mulder

Pitchford

Hagger

Marlow

. Development of executive function and attention in preterm children: A systematic review. Dev Neuropsychol. 2009;34(4):393–421. doi:10.1080/87565640902964524

National Institute for Health and Care Excellence [NICE]. Developmental follow-up of children and young people born preterm, https://www.nice.org.uk/guidance/ng72 (2017).

10.

Hipsky

. Piaget's developmental stages. In: Encyclopedia of Information Technology Curriculum Integration. IGI Global; 2008:713–715.

11.

Dall’oglio

Rossiello

Coletti

, et al. Do healthy preterm children need neuropsychological follow-up? Preschool outcomes compared with term peers. Dev Med Child Neurol. 2010;52(10):955–961. doi:10.1111/j.1469-8749.2010.03730.x

12.

Rommel

A-S

James

S-N

McLoughlin

, et al. Altered EEG spectral power during rest and cognitive performance: A comparison of preterm-born adolescents to adolescents with ADHD. Eur Child Adolesc Psychiatry. 2017;26(12):1511–1522. doi:10.1007/s00787-017-1010-2

13.

Twilhaar

Janssen

de Kieviet

van Elburg

Oosterlaan

. EEG Profiles and associated neurodevelopmental outcomes after very preterm birth. Clin Neurophysiol. 2019;130(7):1166–1171. doi:10.1016/j.clinph.2019.05.002

14.

Rozhkova

. EEG Spectral power of young schoolchildren with perinatal pathology of the CNS. Hum Physiol. 2008;34(1):22–31. doi:10.1134/S0362119708010039

15.

Poorun

Hartley

Goksan

, et al. Electroencephalography during general anaesthesia differs between term-born and premature-born children. Clin Neurophysiol. 2016;127(2):1216–1222. doi:10.1016/j.clinph.2015.10.041

16.

Donoghue

Haller

Peterson

, et al. Parameterizing neural power spectra into periodic and aperiodic components. Nat Neurosci. 2020;23(12):1655–1665. doi:10.1038/s41593-020-00744-x

17.

Gerster

Waterstraat

Litvak

, et al. Separating neural oscillations from aperiodic 1/f activity: Challenges and recommendations. Neuroinformatics. 2022;20 (4):991–1012. 10.1007/s12021-022-09581-8

18.

Angelidis

van der Does

Schakel

Putman

. Frontal EEG theta/beta ratio as an electrophysiological marker for attentional control and its test-retest reliability. Biol Psychol. 2016;121(Pt A):49–52. doi:10.1016/j.biopsycho.2016.09.008

19.

Barry

Clarke

Johnstone

. A review of electrophysiology in attention-deficit/hyperactivity disorder: I. Qualitative and quantitative electroencephalography. Clin Neurophysiol. 2003;114(2):171–183. https://doi.org/10.1016/s1388-2457(02)00362-0

20.

Barry

Johnstone

Clarke

. A review of electrophysiology in attention-deficit/hyperactivity disorder: II. Event-related potentials. Clin Neurophysiol. 2003;114(2):184–198. https://doi.org/10.1016/s1388-2457(02)00363-2

21.

van Son

De Blasio

Fogarty

Angelidis

Barry

Putman

. Frontal EEG theta/beta ratio during mind wandering episodes. Biol Psychol. 2019;140:19–27. doi:10.1016/j.biopsycho.2018.11.003

22.

Ouyang

Hildebrandt

Schmitz

Herrmann

. Decomposing alpha and 1/f brain activities reveals their differential associations with cognitive processing speed. Neuroimage. 2020;205:116304. doi:10.1016/j.neuroimage.2019.116304

23.

Voytek

Kramer

Case

, et al. Age-Related changes in 1/f neural electrophysiological noise. J Neurosci. 2015;35(38):13257–13265. https://doi.org/10.1523/jneurosci.2332-14.2015

24.

Cellier

Riddle

Petersen

Hwang

. The development of theta and alpha neural oscillations from ages 3 to 24 years. Dev Cogn Neurosci. 2021;50:100969. https://doi.org/10.1016/j.dcn.2021.100969

25.

Thuwal

Banerjee

Roy

. Aperiodic and periodic components of ongoing oscillatory brain dynamics link distinct functional aspects of cognition across adult lifespan. eNeuro. 2021;8(5):ENEURO.0224-0221. https://doi.org/10.1523/eneuro.0224-21.2021

26.

S-C

Brehmer

Shing

Werkle-Bergner

Lindenberger

. Neuromodulation of associative and organizational plasticity across the life span: Empirical evidence and neurocomputational modeling. Neurosci Biobehav Rev. 2006;30(6):775–790. doi:10.1016/j.neubiorev.2006.06.004

27.

Tso

WWY

Wong

VCN

Xia

, et al. The griffiths development scales-Chinese (GDS-C): A cross-cultural comparison of developmental trajectories between Chinese and British children. Child Care Health Dev. 2018;44(3):378–383. https://doi.org/10.1111/cch.12548

28.

Pino

Donne

Vagnetti

Tiberti

Valenti

Mazza

. Using the Griffiths mental development scales to evaluate a developmental profile of children with autism Spectrum disorder and their symptomatologic severity. Child Psychiatry Hum Dev. 2022;55(1):117–126. 10.1007/s10578-022-01390-z

29.

Lai

Luk

Leung

Wong

Law

. Validation of the Chinese version of the strengths and difficulties questionnaire in Hong Kong. Soc Psychiatry Psychiatr Epidemiol. 2010;45(12):1179–1186. doi:10.1007/s00127-009-0152-z

30.

Lai

Leung

Luk

Wong

Law

. Validation of the Chinese strengths and weaknesses of ADHD-symptoms and normal-behaviors questionnaire in Hong Kong. J Atten Disord. 2013;17(3):194–202. https://doi.org/10.1177/1087054711430711

31.

Varni

Burwinkle

Seid

Skarr

. The PedsQL™* 4.0 as a pediatric population health measure: Feasibility, reliability, and validity. Ambul Pediatr. 2003;3(6):329–341. doi:10.1367/1539-4409(2003)003<0329:TPAAPP>2.0.CO;2

32.

Vyas

Kumaranayake

. Constructing socio-economic status indices: How to use principal components analysis. Health Policy Plan. 2006;21(6):459–468. https://doi.org/10.1093/heapol/czl029

33.

Delorme

Makeig

. EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods. 2004;134(1):9–21. doi:10.1016/j.jneumeth.2003.10.009

34.

Welch

. The use of fast Fourier transform for the estimation of power spectra: A method based on time averaging over short, modified periodograms. IEEE Trans Audio Electroacoust. 1967;15(2):70–73. doi:10.1109/TAU.1967.1161901

35.

Team

. R: A language and environment for statistical computing. R Foundation for Statistical Computing; 2013. http://www.R-project. org/ .

36.

Chiang

Rennie

Robinson

van Albada

Kerr

. Age trends and sex differences of alpha rhythms including split alpha peaks. Clin Neurophysiol. 2011;122(8):1505–1517. https://doi.org/10.1016/j.clinph.2011.01.040

37.

Hill

Clark

Bigelow

Lum

JAG

Enticott

. Periodic and aperiodic neural activity displays age-dependent changes across early-to-middle childhood. Dev Cogn Neurosci. 2022;54:101076. https://doi.org/10.1016/j.dcn.2022.101076

38.

Manyukhina

Prokofyev

Galuta

, et al. Globally elevated excitation–inhibition ratio in children with autism spectrum disorder and below-average intelligence. Mol Autism. 2022;13(1):1–14. doi:10.1186/s13229-022-00498-2

39.

Robertson

Furlong

Voytek

Donoghue

Boettiger

Sheridan

. EEG Power spectral slope differs by ADHD status and stimulant medication exposure in early childhood. J Neurophysiol. 2019;122(6):2427–2437. doi:10.1152/jn.00388.2019

40.

Purves

Lichtman

. Elimination of synapses in the developing nervous system. Science. 1980;210(4466):153–157. doi:10.1126/science.7414326

41.

Whitford

Rennie

Grieve

Clark

Gordon

Williams

. Brain maturation in adolescence: Concurrent changes in neuroanatomy and neurophysiology. Hum Brain Mapp. 2007;28(3):228–237. https://doi.org/10.1002/hbm.20273

42.

Colombo

Napolitani

Boly

, et al. The spectral exponent of the resting EEG indexes the presence of consciousness during unresponsiveness induced by propofol, xenon, and ketamine. NeuroImage. 2019;189:631–644. https://doi.org/10.1016/j.neuroimage.2019.01.024

43.

Gao

Peterson

Voytek

. Inferring synaptic excitation/inhibition balance from field potentials. NeuroImage. 2017;158:70–78. https://doi.org/10.1016/j.neuroimage.2017.06.078

44.

Okun

Lampl

. Instantaneous correlation of excitation and inhibition during ongoing and sensory-evoked activities. Nat Neurosci. 2008;11(5):535–537. https://doi.org/10.1038/nn.2105

45.

Golden

Buxbaum

De Rubeis

. Disrupted circuits in mouse models of autism spectrum disorder and intellectual disability. Curr Opin Neurobiol. 2018;48:106–112. doi:10.1016/j.conb.2017.11.006

46.

Marcuse

Schneider

Mortati

Donnelly

Arnedo

Grant

. Quantitative analysis of the EEG posterior-dominant rhythm in healthy adolescents. Clin Neurophysiol. 2008;119(8):1778–1781. https://doi.org/10.1016/j.clinph.2008.02.023

47.

Bonnefond

Jensen

. Alpha oscillations serve to protect working memory maintenance against anticipated distracters. Curr Biol. 2012;22(20):1969–1974. doi:10.1016/j.cub.2012.08.029

48.

Jensen

Gelfand

Kounios

Lisman

. Oscillations in the alpha band (9–12Hz) increase with memory load during retention in a short-term memory task. Cereb Cortex. 2002;12(8):877–882. doi:10.1093/cercor/12.8.877

49.

Klimesch

Sauseng

Hanslmayr

. EEG Alpha oscillations: The inhibition–timing hypothesis. Brain Res Rev. 2007;53(1):63–88. doi:10.1016/j.brainresrev.2006.06.003

50.

Forschack

Nierhaus

Müller

Villringer

. Alpha-band brain oscillations shape the processing of perceptible as well as imperceptible somatosensory stimuli during selective attention. J Neurosci. 2017;37(29):6983–6994. doi:10.1523/JNEUROSCI.2582-16.2017

51.

Foxe

Snyder

. The role of alpha-band brain oscillations as a sensory suppression mechanism during selective attention. Front Psychol. 2011;2:154. doi:10.3389/fpsyg.2011.00154

52.

Stroganova

Orekhova

. EEG And infant states. Infant EEG and Event-Related Potentials. 2007;251:280.

53.

Stroganova

Orekhova

Posikera

. Externally and internally controlled attention in infants: An EEG study. Int J Psychophysiol. 1998;30(3):339–351. https://doi.org/10.1016/S0167-8760(98)00026-9

54.

Díaz

Cid

Otárola

Rojas

Alarcón

Cañete

. EEG Beta band frequency domain evaluation for assessing stress and anxiety in resting, eyes closed, basal conditions. Procedia Comput Sci. 2019;162:974–981. doi:10.1016/j.procs.2019.12.075

55.

Arns

Conners

Kraemer

. A decade of EEG theta/beta ratio research in ADHD: A meta-analysis. J Atten Disord. 2013;17(5):374–383. doi:10.1177/1087054712460087

56.

Schleiger

Sheikh

Rowland

Wong

Read

Finnigan

. Frontal EEG delta/alpha ratio and screening for post-stroke cognitive deficits: The power of four electrodes. Int J Psychophysiol. 2014;94(1):19–24. doi:10.1016/j.ijpsycho.2014.06.012

57.

Van Den Heuvel

M. P.

Kersbergen

K. J.

, et al. (2015). The neonatal connectome during preterm brain development. Cereb Cortex., 25(9), 3000–3013. https://doi.org/10.1093/cercor/bhu095

58.

Hosseini

Shalbaf

Nasrabadi

. Extracting a seizure intensity index from one-channel EEG signal using bispectral and detrended fluctuation analysis. J Biomed Sci Eng. 2010;3(3):253–261. doi:10.4236/jbise.2010.33034

59.

L-W

Lai

W-K

Liang

W-G

, et al. Single channel wireless EEG device for real-time fatigue level detection. 2015 International Joint Conference on Neural Networks (IJCNN); 2015.

60.

Lee

T-W

Girolami

Sejnowski

. Independent component analysis using an extended infomax algorithm for mixed subgaussian and supergaussian sources. Neural Comput. 1999;11(2):417–441. doi:10.1162/08997669930001x6719

61.

Hack

Taylor

Drotar

, et al. Chronic conditions, functional limitations, and special health care needs of school-aged children born with extremely low-birth-weight in the 1990s. Jama. 2005;294(3):318–325. https://doi.org/10.1001/jama.294.3.318

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB