Developmental changes in the efficiency of attentional networks in very preterm children during the transition from preschool to school

Abstract

This longitudinal study examined changes in the efficiency of the attentional system during the transition from preschool to school in children born very prematurely. Seventy‐one Polish children participated: 34 born before 32 weeks of gestation (47.1% boys) and 37 full‐term peers (48.6% boys). The Attention Network Task—Child Version was administered at ages 5 and 6. At age 5, very preterm children showed reduced efficiency in alerting and orienting compared to full‐term children. By age 6, alerting efficiency improved in the preterm group, and no longer differed from the control group. However, group differences in attentional orienting persisted at the follow‐up assessment at age 6. These findings suggested partial improvement in attentional functioning in very preterm children during early development, with persistent differences in attentional orienting compared to full‐term peers.

Statement of contribution.

What is already known on this subject?

• Children born very preterm are at increased risk of attentional difficulties.

• Evidence on attentional networks in preterm children is mixed and heterogeneous.

• Little is known about developmental change in attentional networks in early childhood.

What this study adds?

• Longitudinal evidence on attentional networks between ages 5 and 6 in very preterm children.

• Alerting improves during the transition from preschool to school.

• Differences in attentional orienting persist compared to full‐term peers.

INTRODUCTION

Preterm birth is associated with a range of short‐ and long‐term consequences, affecting both the child and their family. Although the incidence of severe disabilities in this population has declined over recent decades, prematurity continues to pose a significant risk for a variety of developmental challenges (e.g., problems with attention) that impact both families and professionals who support these children (e.g. educators).

Prematurity has consistently been identified as a major risk factor for attention‐related problems, particularly those characteristic of the inattentive and combined subtypes of attention‐deficit/hyperactivity disorder (ADHD) (Brogan et al., 2014; Franz et al., 2018; Jaekel et al., 2013; Johnson et al., 2010; Montagna et al., 2020). These difficulties appear to be especially pronounced among very preterm children (born before 32 weeks of gestation) (Bogičević, Pascoe, et al., 2021; de Jong et al., 2018; Montagna et al., 2020). Most available studies focus on school‐age children when attention‐related problems tend to become more apparent and are more likely to lead to a formal ADHD diagnosis. Nevertheless, identifying early indicators of attentional deficits is of critical importance, as emphasised in the literature (Montagna et al., 2020). Detecting such difficulties during the preschool years may enable early intervention and better predict challenges in learning‐related cognitive domains (Bogičević, Verhoeven, & van Baar, 2021; Jaekel et al., 2013; Johnson & Marlow, 2017; Walczak‐Kozłowska et al., 2022a).

Although some studies have reported difficulties in visual orienting in preterm children (Bijlsma et al., 2024; Pel et al., 2010), their attentional functioning appears to be heterogeneous. Some studies have identified distinct profiles of attentional functioning in both moderate‐to‐late preterm and full‐term children (Bogičević et al., 2020), as well as in longitudinal cohorts of extremely preterm children, although these profiles may change over time (Bogičević, Pascoe, et al., 2021; Bogičević, Verhoeven, & van Baar, 2021). Ginnell et al. (2021) note that current evidence remains insufficient to delineate a distinct attentional phenotype specific to this population. One major limitation of the existing research is the lack of studies that employ a unified, theoretically grounded model of attention—one that accounts for both the multidimensional nature of the construct and its developmental trajectory. Additionally, methodological inconsistencies and differences in sample characteristics (e.g., gestational age, age at assessment, task type) further hinder cross‐study comparisons and preclude clear conclusions regarding developmental trends in attention among preterm children. Moreover, the substantial heterogeneity observed in attentional functioning among children born preterm may itself limit the emergence of a single, clearly defined attentional phenotype.

A theoretical framework that effectively integrates both the complexity of attention and its developmental underpinnings is provided by Michael Posner's model of the attentional system. This model conceptualises attention as comprising three distinct but interrelated components: alerting, orienting and executive control (Posner & Fan, 2008; Posner & Petersen, 1990), each supported by specific neural networks (Attention Network Theory). Alerting supports rapid responses to stimuli and the maintenance of an alert state over time (related to sustained attention), orienting allows the shifting of focus between objects within the visual field, and executive attention facilitates the management and control of both ongoing and planned activities (De Jong et al., 2016a, 2016b; Posner & Petersen, 1990). Building upon this model, Posner and colleagues developed the Attention Network Task (ANT), which allows for separate assessment of these three attentional mechanisms (Fan et al., 2002). A child‐friendly version of the task (Child‐ANT) was later introduced to assess children as young as 4–5 years old (Rueda et al., 2004).

Studies employing the Child‐ANT have provided robust evidence for the independence of the three attentional components, both in terms of cognitive functioning and neural correlates (Rueda et al., 2004). According to their findings, the efficiency of alerting and orienting networks increases up to age 6 and then stabilises. By age 6, children perform comparably to adults on tasks involving these networks. In contrast, executive attention shows a more extended developmental trajectory, continuing to improve until at least age 7 and possibly up to around age 10. These improvements manifest as fewer errors (e.g., conflict trials) and shorter response times. Similar developmental patterns were observed by Hrabok et al. (2007), who found improvements in orienting and executive attention in children aged 4–5, but no further gains in alerting.

The ANT and Child‐ANT have frequently been used to characterise attentional profiles in children diagnosed with or at risk of ADHD (see Bieleninik et al., 2023). However, few studies have applied this paradigm to very preterm children, particularly in the preschool years, and those that exist yield inconsistent results. For instance, Geldof et al. (2013) and Pizzo et al. (2010) reported reduced executive attention in very preterm children aged 5.5–6.5, with intact alerting and orienting. Pizzo et al. (2010) also found that preterm children exhibited generally slower response times across all task conditions. In contrast, our own previous findings (Walczak‐Kozlowska et al., 2022b) indicated that the majority of very preterm 5‐year‐olds demonstrate attentional functioning within normative ranges, with only approximately 26% showing reduced alerting and orienting efficiency and a tendency toward weaker executive control. Offering yet another perspective, De Kieviet et al. (2012) found no differences in any attentional domain in 7‐year‐old children born very prematurely.

As outlined above, the developmental course of attention in very preterm children—particularly during the critical transition from preschool to early formal education—remains poorly understood. It is unclear whether early attentional deficits observed at age 5 persist, resolve, or change in nature by age 6 or 7. Ginnell et al. (2021) emphasised the need for longitudinal studies that use consistent, developmentally sensitive methods to better delineate these trajectories.

Accordingly, the present study aimed to examine changes in the efficiency of attentional alerting, orienting and executive control in very preterm children across the transition from preschool to school entry (i.e., between ages 5 and 6, corresponding to the ‘class 0’ year in the Polish educational system), using the Child‐ANT. The preschool period is marked by rapid neural maturation, and it may represent a window of opportunity for catching up in certain cognitive functions. If improvement does occur—as suggested by findings in 7‐year‐olds (De Kieviet et al., 2012)—what aspects of attention are affected? Answering this question holds both theoretical and practical relevance, as attention is a critical foundation for early learning. Deficits in this domain may contribute to an unequal starting point in formal education for children born very prematurely (Posner & Rothbart, 2007; Walczak‐Kozłowska et al., 2022a).

METHOD

Design

This longitudinal study consisted of two assessment points: the first at age five and the second at age six, with an interval of approximately 12 months between assessments. At both stages, children completed a behavioural assessment of attentional system efficiency based on Posner's theoretical model. In addition, general intellectual ability was measured at the first assessment. The study received ethical approval from the Ethics Committee for Research Projects at the Institute of Psychology, University of Gdansk, Poland (approval no. 2/2017).

Assessments were conducted primarily in a dedicated testing room at the Institute of Psychology, University of Gdansk, Poland. In rare cases where parents were unable to attend, sessions were held in the child's home or kindergarten under standardized conditions: a quiet, separate room, individual testing and administration by a trained psychologist. Written informed consent was obtained from all parents or legal guardians. Each session lasted approximately 55 min; the first session was about 20 min longer due to the administration of the Stanford‐Binet Brief IQ Scale. As a token of appreciation, children received educational books after each session. During the first visit, parents completed a sociodemographic and medical questionnaire in a separate room, aided by discharge records or medical documentation. Socioeconomic status (SES) was assessed using a 5‐point Likert scale reflecting perceived standard of living (1 = ‘very bad’ to 5 = ‘very good’).

Participants

The initial sample included 100 five‐year‐old children. Of these, 71 participated in the follow‐up: 34 very preterm children (born <32 weeks gestation) and 37 full‐term peers (see Table 1 for group characteristics). Attrition was due to loss of contact, withdrawal, scheduling conflicts, or relocation abroad. Comparative analyses revealed no significant differences between participants who completed the follow‐up and those lost to follow‐up in any sociodemographic, medical, or cognitive variable (all p > .05).

TABLE 1.

Characteristics of children born very prematurely and full‐term controls, n = 71.

Characteristic	Very preterm children (n = 34)	Full‐term controls (n = 37)	t‐test	p Value	Cohen's d
Mean age, months (SD)	66.50 (4.29)	65.81 (4.25)	0.679	.499	.161
Mean gestational age, weeks (SD)	30.62 (1.48)	39.27 (1.31)	−26.20	<.001	6.23
Mean birth weight, g (SD)	1538.12 (566.44)	3387.49 (428.79)	−15.59	<.001	3.70
Mean initial rating of Apgar score, n (SD)	6.06 (2.03)	9.38 (.79)	−9.22	<.001	2.19
Mean final rating of Apgar score, n (SD)	7.53 (1.31)	9.76 (.44)	−9.79	<.001	2.33
Mean IQ score in Stanford Binet Brief IQ Scale, n (SD)	98.97 (12.08)	107.27 (10.76)	−3.06	.003	.73

Characteristic	Very preterm children (n = 34)	Full‐term controls (n = 37)	χ 2	p value	Cramer's V
Mean socioeconomic status, n (SD) a	4.00 (.78)	4.11 (.68)	1.32	.725	.137
Sex
Girls, n	18	19	.18	.893	.016
Boys, n	16	18

^a Assessed using a Likert Scale ranging from 1 (‘very bad’) to 5 (‘very good’).

Participants were of Polish nationality and recruited from the same urban region. Recruitment involved collaboration with a university hospital, parent support groups for premature children and local kindergartens in the Pomeranian and West Pomeranian Voivodeships of Poland. Exclusion criteria included cerebral palsy, significant motor impairments, uncorrected visual or hearing deficits and intellectual disability (IQ < 70). Sensory screening data were obtained through parent interviews or verified using Snellen charts when necessary. Medical histories of very preterm children are detailed in Table 2.

TABLE 2.

Characteristics of medical condition and complications among very preterm children, n = 34.

Medical characteristic	Min	Max	M	SD
Body length after birth, cm	29	56	43.31	5.49
Head circuit, cm	21.50	34.50	28.92	3.23
Initial (1st minute) Apgar score, n	1	10	6.45	2.17
Final rating of Apgar score, n	5	10	7.79	1.33
Length of stay in the intensive care unit/department for premature babies, days	0	140	35.21	29.13
Days in incubator	0	122	29.45	25.78
Days of ventilation	0	70	10.04	15.48
	Occurrence (%)
Type of incubator	74.1% – closed 3.7% – open 22.2% – both
Type of delivery	23.5% – physiological 76.5% – Caesarean section
Type of pregnancy due to its course	76.5% – endangered 14.7% – not threatened 8.8% – no data
Intrauterine growth restriction (IUGR)	26.5% 5.9% – no data
Central Nervous System (CNS) haemorrhages	29.4% 2.9% – no data
Severity of a haemorrhage	40.0% – Ist stage 50.0% – IInd stage 10.0% – no data
Extension of the lateral ventricles	8.8%
Periventricular leukomalacia (PVL)	2.9%
Perinatal asphyxia	14.7%
Retinopathy of prematurity (ROP)	29.4%
Severity of ROP	40.0% – Ist stage 50.0% – IInd stage 10.0% – IIIrd stage
Patent ductus arteriosus	20.6%
Anaemia	47.1%
Respiratory distress syndrome	38.2%
Bronchopulmonary dysplasia	23.5%
Chronic lung disease	5.9%
Perinatal apnea	50.0%
Abnormal muscle tone	67.6%
Eating disorders	35.3% 2.9% – no data
Hyperbilirubinemia	35.3%
Phototherapy	76.5%
Blood transfusion	47.1%
Oxygen therapy	73.5%
Sepsis	14.7%
Subsequent hospitalisations	41.2%

Attention network task—Child version

Attentional network efficiency was assessed using the Attention Network Task adapted for children (Child‐ANT; Rueda et al., 2004), which evaluates alerting, orienting and executive attention. The task was programmed in E‐Prime 2.0 and presented on a 15.6′ Sony Vaio laptop.

Children responded to rows of yellow fish on a blue‐green background by indicating the direction of the central fish using a mouse. Stimuli appeared either alone (neutral condition) or flanked by distractor fish pointing in the same (congruent) or opposite (incongruent) direction. Feedback included animations and auditory cues. Incorrect responses received a brief tone without reinforcement.

Each session consisted of 144 trials across three blocks with five‐minute breaks. Sessions were preceded by 24 practice trials and index card training. Immediate feedback and guidance were given during training only. During the task, children sat approximately 50 cm from the screen. Stimulus parameters followed standard specifications (e.g., 1.6° visual angle per fish; total array: 8.84°).

Cue conditions included no cue, central cue, double cue and spatial cue (see Figure 1 for details). Targets appeared approximately 1° above or below the fixation cross. Each trial began with a fixation period (400–1600 ms), followed by a 150 ms cue (when applicable) and a 450 ms fixation before stimulus onset. Invalid cues were not used.

FIGURE 1.

ANT child version. Efficiency scores were computed as follows: Alerting: RT (no cue) − RT (double cue). Orienting: RT (central cue) − RT (spatial cue). Executive attention: RT (incongruent) − RT (congruent), across all cue conditions.

A negative orienting score may reflect increased inhibition of return (IOR), typical of younger children with immature executive control (Posner & Petersen, 1990; Rueda et al., 2004).

Stanford‐Binet Intelligence Scales 5th ed. (SB5)

Cognitive abilities were assessed using the Polish adaptation of the Stanford‐Binet Intelligence Scales, 5th Edition (Sajewicz‐Radtke et al., 2017). The Brief IQ score, comprising Vocabulary and Matrices subtests, was used to measure verbal and nonverbal domains. This tool is widely used in Poland and has high reliability (Spearman‐Brown coefficient = .92).

Statistical analysis

Statistical analyses were performed using IBM SPSS Statistics 29.0. The Kolmogorov–Smirnov test was used to assess the normality of distributions. Repeated‐measures analyses of covariance (ANCOVA) were conducted to examine between‐group differences and within‐group changes in attentional efficiency. IQ was included as a covariate in the analyses because the groups differed significantly in their IQ scores. Although attentional processes are closely related to intellectual performance, including IQ as a covariate helped to ensure that the observed effects are not solely attributable to overall differences in cognitive level between the groups. As the covariate effect turned out to be insignificant, further analyses were performed without taking this variable into account. Post hoc comparisons were performed using Bonferroni‐adjusted p‐values, and paired‐samples t‐tests were applied for within‐group follow‐up analyses. Fisher's z‐tests were used to compare independent correlation coefficients between groups, followed by multiple regression analyses. Chi‐square tests were conducted to analyse frequency distributions.

RESULTS

Raw data obtained from the two assessments with the Child‐ANT were transformed into three indicators corresponding to the separate attentional networks (specific subtraction procedures are described in the Method section). The normality of these variables was then assessed using the Kolmogorov–Smirnov test (K–S test). Results indicated that the distributions of the values from the second assessment significantly deviated from normality. Consequently, scatterplots for each variable were inspected, and three observations that strongly deviated from the mean were removed (representing 4% of the second‐assessment dataset: one in the very preterm group and two in the full‐term group). Following this adjustment, re‐analysis with the K–S test confirmed that the variables met the assumption of normality. The final sample included data from 71 participants: 34 very preterm children and 37 full‐term controls.

To test the hypothesis that there are significant within‐group changes and between‐group differences in attentional network efficiency between ages five and six in very preterm and full‐term children, a repeated‐measures analysis of covariance (ANCOVA) with a group × attentional mechanism × measurement design was conducted, with IQ entered as a covariate. The effect of the covariate (IQ) was not statistically significant, F(1, 68) = 1.64, p = .204, η ² = .02, and further analyses were performed without this variable. The three‐way interaction (group × attentional mechanism × measurement) was nonsignificant, F(1, 69) = 0.62, p = .539, η ² = .01. However, significant two‐way interactions were observed for group × attentional mechanism, F(1, 69) = 3.27, p = .041, η ² = .05, and for attentional mechanism × measurement, F(1, 69) = 3.14, p = .047, η ² = .04 (see Figure 2 for details). In addition, a significant main effect of birth status was found, F(1, 68) = 7.95, p = .006, η ² = .10, indicating that full‐term children achieved higher overall scores (M = 103.78, SD = 127.33, 95% CI [85.59, 121.96]) than very preterm children (M = 66.63, SD = 153.19, 95% CI [47.66, 85.60]), regardless of attentional mechanism or measurement time.

FIGURE 2.

Within‐group changes and between‐group differences in the efficiency of attentional mechanisms, measured with the Child‐ANT, among very preterm (n = 34) and full‐term children (n = 37), N = 71 (Error bars represent ± 1 SE).

Pairwise comparisons with Bonferroni correction were conducted to examine group differences. At age five, full‐term children scored significantly higher than very preterm children in Alerting (M = 105.50, SE = 23.04, 95% CI [59.55, 151.45] vs. M = 24.74, SE = 24.03, 95% CI [−23.20, 72.67], p = .018, Cohen's d = 0.58) and Orienting (M = 36.15, SE = 20.66, 95% CI [−5.07, 77.37] vs. M = −42.69, SE = 21.55, 95% CI [−85.69, 0.31], p = .010, Cohen's d = 0.63). No significant group difference was found in Executive Attention (p = .882). At age six, full‐term children demonstrated a significant advantage over very preterm children only in Orienting (M = 70.82, SE = 22.15, 95% CI [26.63, 115.00] vs. M = −8.44, SE = 23.10, 95% CI [−54.53, 37.65], p = .016, Cohen's d = 0.59), whereas differences in Alerting (p = .755) and Executive Attention (p = .324) were nonsignificant. Across both measurement points, full‐term children scored significantly higher than very preterm children in Alerting (M = 105.74, SE = 15.26, 95% CI [75.31, 136.18] vs. M = 60.98, SE = 15.92, 95% CI [29.23, 92.73], p = .046, Cohen's d = 0.48) and Orienting (M = 53.48, SE = 16.64, 95% CI [20.29, 86.68] vs. M = −25.57, SE = 17.36, 95% CI [−60.19, 9.06], p = .002, Cohen's d = 0.78), whereas no significant difference was observed in Executive Attention (M = 152.11, SE = 19.08, 95% CI [114.04, 190.17] vs. M = 164.48, SE = 19.90, 95% CI [124.77, 204.18], p = .655, Cohen's d = 0.11). Overall, these results indicate that full‐term children consistently outperformed very preterm children in Alerting and Orienting, while performance in Executive Attention was similar between groups.

Paired‐samples t‐tests revealed a significant improvement in Alerting efficiency in the very preterm group between the first and second assessments (t = 2.25, p = .016, Cohen's d = 0.386). Changes in Orienting (t = −1.15, p = .130, Cohen's d = 0.197) and Executive Attention (t = 0.33, p = .373, Cohen's d = 0.056) were not statistically significant. In the full‐term group, no significant within‐group differences were observed for Alerting (t = −0.02, p = .493, Cohen's d = 0.003), Orienting (t = −1.37, p = .090, Cohen's d = 0.225), or Executive Attention (t = 1.60, p = .059, Cohen's d = 0.263); a trend toward a decrease in mean value was observed here.

The next stage of the analysis aimed to examine whether the strength of the association between attentional mechanisms and perinatal variables (gestational week at delivery and birth weight) differed between very preterm and full‐term children. Fisher's Z tests were used to compare the two groups' independent correlation coefficients. The results are presented in Table 3.

TABLE 3.

Comparison of correlations between attentional mechanisms and perinatal variables (gestational week at delivery, birth weight) in very preterm and full‐term children, with Fisher's Z tests.

Measurement	Dependent variable	Very preterm children		Full‐term controls
		Gestational week at delivery	Birth weight	Gestational week at delivery	Birth weight
5th year	Alerting	−0.08	0.09	0.16	0.11
	Orienting	−0.01	0.22	−0.16	0.02
	Executive attention	0.20	0.08	−0.05	0.08
6th year	Alerting	0.11	−0.32*	−0.36*	0.15
	Orienting	−0.14	0.05	−0.16	0.11
	Executive attention	−0.06	0.00	0.28	0.02

^* p < .05.

Overall, no statistically significant between‐group differences in correlation strength were observed across most measures (Table 3). Effects approaching statistical significance (p < .10) were found only for Alerting in the 6th year of life, both for gestational week at delivery (Z = 1.95, p = .051) and for birth weight (Z = −1.91, p = .056). These findings suggest that, except for this specific case, the associations between attentional mechanisms and perinatal variables were similar for very preterm and full‐term children. Subsequently, separate multiple regression analyses were conducted in the group of very preterm children in order to examine whether performance in attentional mechanisms at the follow‐up point (the sixth year of life) could be predicted by corresponding scores obtained during the first measurement point (the fifth year of life) as well as by the gestational week at delivery and the birth weight, as presented in Table 4.

TABLE 4.

Multiple regression analyses predicting attentional mechanisms at the follow‐up assessment point (6th year) by the baseline performance (5th year), gestational week at delivery and birth weight in very preterm children.

Independent variable	B	SE	β	t	p	95% CI for B	F	p	$R_{\mathrm{adj}.}^2$
Dependent variable: Alerting – 6th year
Alerting – 5th year	0.14	0.08	.25	1.61	.117	[−0.02, 0.30]	5.52	.004	.291
Gestational week at delivery	45.67	13.53	.70	3.37	.002	[18.05, 73.29]
Birth weight	−0.13	0.03	−.79	−3.86	.001	[−0.19, −0.07]
Dependent variable: Orienting – 6th year
Orienting – 5th year	0.13	0.24	.10	0.53	.597	[−0.36, 0.62]	0.79	.509	.019
Gestational week at delivery	−29.06	26.70	−.29	−1.09	.285	[−83.57, 25.45]
Birth weight	0.06	0.07	.24	0.95	.348	[−0.08, 0.20]
Dependent variable: Executive attention – 6th year
Executive attention – 5th year	−0.04	0.14	−.05	−0.28	.778	[−0.33, 0.25]	0.09	.965	.090
Gestational week at delivery	−9.74	27.66	−.09	−0.35	.727	[−66.22, 46.74]
Birth weight	0.02	0.07	.06	0.24	.813	[−0.12, 0.16]

Abbreviations: 95% CI for B, 95% confidence interval for the unstandardized regression coefficient (B); β, unstandardized regression coefficient; B, unstandardized regression coefficient; F, analysis of variance result; $R_{\mathrm{adj}.}^2$, adjusted R‐squared; SE, standard error; t, Student's t‐test result.

For Alerting (see Table 4), the overall model was statistically significant, F(3, 30) = 5.52, p = .004, $R_{\mathrm{adj}.}^2$ = .291. Both gestational week at delivery (β = .70, p = .002) and birth weight (β = −.79, p = .001) were significant predictors of Alerting. This indicates that higher gestational age is associated with better Alerting, whereas higher birth weight is unexpectedly related to poorer Alerting. For Orienting, the model was not statistically significant, F(3, 30) = 0.79, p = .509, $R_{\mathrm{adj}.}^2$ = .019, and none of the predictors reached statistical significance (p > .05). Similarly, the model for Executive attention was not significant, F(3, 30) = 0.09, p = .965, $R_{\mathrm{adj}.}^2$ = .090, and no predictor showed a significant relationship with performance at the second (follow‐up) measurement point.

Subsequently, we evaluated the frequencies of values indicating the improvement in Alerting, Orienting, and Executive attention (getting a better value in the second measurement compared to the initial assessment). Although in the case of all three attentional mechanisms the values indicated better performance during the follow‐up testing in the control group (see Table 5), the Chi‐squared analyses revealed that these observations were statistically insignificant (p > .05).

TABLE 5.

Improved outcome in the Child‐ANT experiment in the second assessment (at 6th year of life) in comparison to the initial measurement (at 5th year of life), n = 71.

Improvement with time (n)	Group		χ 2	p	Cramér's V
	Very preterm children, n = 34	Full‐term controls, n = 37
Alerting	47% (16)	51% (19)	0.131	.819	.043
Orienting	50% (17)	57% (21)	0.325	.569	.068
Executive attention	53% (18)	57% (21)	0.104	.747	.038

Since at both the age of 5 and 6 the diminished efficiency among very preterm children was manifested only in the attentional Orienting, we decided to verify the differences in the occurrence of the increased IOR effect. Table 6 presents the obtained frequencies for each group when the mean value for attentional Orienting was below 0 (‘IOR effect’) and equal or above 0 (no ‘IOR’). Chi‐squared analysis revealed no differences in the frequencies in both the first (χ ²  = 3.281; p = .070; Cramér's V = .215) as well as the second (follow‐up) testing (χ ²  = 2.263; p = .132; Cramér's V = .179).

TABLE 6.

Increased inhibition of return (IOR) effect in the Child‐ANT experiment among very preterm and full‐term children measured at 5th and 6th year of life, n = 71.

IOR effect, % (n)	Group		χ 2	p	Cramér's V
	Very preterm children, n = 34	Full‐term controls, n = 37
5th year	65% (22)	43% (16)	3.281	.070	.215
6th year	50% (17)	32% (12)	2.263	.132	.179

DISCUSSION

This study examined developmental changes in attentional network efficiency in very preterm children during the transition from preschool to school age. Our findings indicate that significant improvement over time was observed only for the alerting network, with very preterm children reaching a level comparable to their full‐term peers by age six. In contrast, changes in orienting and executive attention were non‐significant in both groups. Importantly, the efficiency of attentional orienting remained consistently lower in the preterm group across both assessments. Additionally, correlations between attentional mechanisms and perinatal variables were generally similar across groups, with only alerting in the sixth year showing trends toward significance for gestational age and birth weight. Regression analyses further indicated that, among very preterm children, higher gestational age predicted better alerting performance, whereas no significant predictors emerged for orienting or executive attention.

The observed improvement in alerting efficiency suggests that, although very preterm preschoolers initially show reduced ability to use external cues to facilitate stimulus detection and maintain that state over time, this skill develops by school entry. This aligns with earlier findings (Kraut, 1976; Smothergill & Kraut, 1989) that children, in general, use alerting cues less effectively than adults, possibly reflecting immaturity in neural systems supporting external activation. Premature birth may be associated with an altered developmental trajectory of the neural network underlying the alerting mechanism, rather than a permanent deficit. It should be noted, however, that at later stages of development this attentional network undergoes substantial changes. As noted by Posner et al. (2012) and Rueda et al. (2004), younger children rely more on phasic alerting, whereas older children and adults depend increasingly on tonic alertness. Consequently, the developmental trajectory of this network in very preterm children warrants further longitudinal investigation, as deficits may re‐emerge during adolescence, particularly in those born at lower gestational ages (see Ginnell et al., 2021).

The reduced efficiency of the orienting mechanism observed in very preterm children may indicate relative vulnerability within this particular attentional system. Unlike the alerting network, attentional orienting appears less amenable to compensatory development during the preschool period. Its neuronal substrates include the superior parietal lobule, engaged during visual cueing and the temporal–parietal junction, activated when attention shifts to unexpected stimuli (Posner & Rothbart, 2007). Damage or atypical development in these brain regions may negatively affect the ability to shift attention effectively. Neuroimaging studies indicate that preterm children are more likely to show atypical development of temporal and parietal cortices, reduced cortical volumes in sensorimotor, premotor, parietal and occipital regions, as well as altered brain microstructure and white matter connectivity, all of which may contribute to attentional deficits (Caldinelli et al., 2017; Fuchino et al., 2013; Loe et al., 2013; Mullen et al., 2011; Murray et al., 2016; Nagy et al., 2003; Peterson et al., 2002).

Attentional orienting is also shaped by neurobiological processes in late gestation, including maternal‐fetal thyroid hormone exchange (Ruff & Rothbart, 2001). Very preterm infants miss part of these critical processes and are exposed to different postnatal stimulation, potentially affecting the development of neural networks underlying attention (de Jong et al., 2012; Mulder et al., 2009; Posner & Rothbart, 2007). Preterm infants also miss critical intrauterine periods for myelination, with a significant portion of these processes occurring in the extrauterine environment. This may influence attentional performance, though confirmation requires further research. Additionally, the third trimester represents a peak period for DHA accretion in the brain, and preterm birth may compromise this process, contributing to neurodevelopmental vulnerability. DHA supplementation has been proposed as a potential intervention to support attentional development (Hewawasam et al., 2021), though its effects have yet to be reliably demonstrated using experimental tasks such as the ANT.

In summary, the observed differences in attentional orienting in very preterm preschoolers warrant further investigation using neuroimaging to elucidate the functionality of its neuronal substrates. Longitudinal studies are needed to determine whether these intergroup differences persist into later childhood or adolescence, and to assess their impact on the development of executive attentional mechanisms. Lower efficiency of attentional orienting may adversely affect the normative maturation of executive networks, potentially slowing their development in very preterm children (Bijlsma et al., 2024). Our previous research (Walczak‐Kozlowska et al., 2022a) has shown that attentional orienting predicts multiple school readiness skills in very preterm children, highlighting the importance of monitoring the relationship between attentional efficiency and academic outcomes over time. Persistent differences in this mechanism may underlie broader cognitive delays and interfere with the acquisition of academic skills (Johnson & Marlow, 2017; Posner & Rothbart, 2007).

For the executive network, both groups showed a non‐significant trend toward improvement between ages five and six. This may reflect the early phase of a normative acceleration in executive function development, typically most pronounced between ages five and seven (Rueda et al., 2004). Whether preterm children show the same developmental gains as full‐term peers during this period remains to be established, as slower maturation could contribute to early school learning difficulties (Ginnell et al., 2021; Posner & Rothbart, 2007). The patterns observed in the full‐term group are consistent with previous reports that alerting and orienting networks mature rapidly in infancy and early childhood, with limited changes thereafter, while executive attention undergoes marked improvement in the final preschool period and early school years (Rueda et al., 2004). In this context, the selective catch‐up in alerting but persistence of orienting deficits in preterm children underscores the specificity of attentional vulnerabilities associated with prematurity.

From an applied perspective, the preschool period represents a critical window for supporting attentional development in very preterm children. Longitudinal data from the NOVI cohort (2024) show significant heterogeneity in the trajectories of attention problems and their clinical relevance for later ADHD development (Camerota et al., 2025). Given the established role of attention in school readiness (Posner & Rothbart, 2007, Walczak‐Kozlowska et al., 2022a) and the heightened risk for academic underachievement in this population, early identification and targeted interventions are essential (de Jong et al., 2012; Pritchard et al., 2014; Twilhaar et al., 2019). Interventions should focus on strengthening attentional and executive capacities prior to school entry, which may be particularly important for children facing additional socio‐economic disadvantages (Brumberg & Shah, 2015; Chen et al., 2014; Msall, 2014). Future studies should assess the effectiveness of such interventions on attentional network efficiency and track their impact on later academic outcomes.

Some limitations of the present study should be acknowledged. First, the sample size was relatively small. In developmental research, where effect sizes are often modest, findings based on smaller samples should be interpreted with caution. Although the longitudinal design and detailed analyses provide valuable insights into developmental changes in attentional functioning, replication in larger and more diverse samples is needed to confirm the robustness and generalizability of the observed effects. Future studies with larger cohorts may also allow for more fine‐grained analyses of individual variability in attentional development among children born very preterm.

CONCLUSIONS

This longitudinal study indicates that very preterm children display a relatively persistent lower efficiency in attentional orienting between ages five and six, whereas initial differences with alerting appear to diminish by the onset of formal schooling. These findings highlight distinct developmental trajectories across attentional networks, with orienting mechanisms potentially more vulnerable to early neurodevelopmental disruption. The observed improvement in alerting may be driven by increased environmental stimulation and task demands accompanying school readiness. Ongoing longitudinal monitoring is essential, as targeted interventions aimed at specific attentional domains may promote more optimal cognitive development and enhance later academic achievement in this group.

AUTHOR CONTRIBUTIONS

Tamara Walczak‐Kozłowska: Conceptualization; investigation; funding acquisition; writing – original draft; methodology; validation; visualization; writing – review and editing; software; formal analysis; project administration; data curation; resources. Aleksandra Mańkowska: Conceptualization; writing – original draft; methodology; writing – review and editing. Michał Harciarek: Conceptualization; investigation; funding acquisition; writing – original draft; methodology; validation; writing – review and editing; software; supervision.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT

All data, analytical code and research materials can be made available by the corresponding author upon reasonable request. The shared data will have to be anonymised and some variables will have to be removed because research data includes sensitive and confidential information of participating children (patient data and medical recordings).