The World Health Organisation (WHO) has acknowledged that whilst reducing mortality for newborns is the priority there is also a need to prioritise improving health, psychosocial well-being and the learning potential of children, particularly in the early years of life (1). Over the past decades improvement in NICU care and technology has led to increased survival rates of preterm infants. However, despite this success, gains in survival have not been paralleled by improvements in morbidity.

Of the 15 million preterm births that occur worldwide each year, approximately 75% to 84% occur between 32 and 36 + 6 weeks gestational age (GA) and these infants are defined as moderate to late preterm (MLP) (2, 3). Current trends have shown that the rise in the rate of preterm birth is primarily attributed to the MLP infant group (4). This may in part reflect the increased use of artificial reproductive technology and its association with multiple pregnancies and preterm delivery and consequently an increased demand for neonatal unit services (5).

In the last six weeks of pregnancy, there is a 35% to 40% increase in brain weight with a five-fold increase in brain volume (6). After birth, the MLP infant is no longer protected by the uterine environment and this critical period of accelerated development takes place externally in the NICU or postnatal wards (7). A challenging external environment may influence neural connectivity (8). Disruption of the typical process of brain development may expose preterm infants to long-term neurodevelopmental impairment. Whilst a more positive outcome is associated with the low-risk uncomplicated preterm infant (9, 10) defining “uncomplicated” may prove problematic as preterm birth itself is atypical and its aetiology may involve deteriorating fetal or maternal health, which in itself may lead to suboptimal outcomes. Maturity at birth has a direct impact on outcome; the lower the GA of the preterm infant, the greater the impact on mortality and morbidity (11–14). Low birth weight associated with preterm birth further increases risk factors for the preterm infant (15).

The availability of neurodevelopmental outcome data for healthy MLP infants is often hindered by the variation in follow-up between institutions internationally (16) and the belief that these infants are low risk, and hence require little developmental surveillance. This limits our ability to assess the prevalence of impairments in this healthy MLP infant group. This study aims to establish the natural neurodevelopmental baseline scores for a healthy homogenous group of MLP infants admitted to the neonatal unit at birth. Secondly, we aim to compare the neurodevelopmental outcome of the MLP group to that of a healthy term-born infant's group at 18 months postmenstrual age, recruited during the same period.

Cork University Maternity Hospital is the largest Maternity Hospital in the south of the Republic of Ireland, serving the local and surrounding areas, and is a referral centre for high risk pregnancies.

One hundred and one MLP infants and ninty seven term control (TC) infants were recruited at birth and enrolled in this study. The median (IQR) birth GA (weeks) was 34.0 (33.0–35.0) for the MLP group and 39.9(38.9–40.9) for the TC group. There was no difference in maternal age between groups (p = 0.785). As expected the TC group had significantly higher birthweight and larger head circumference than the MLP group (p < 0.001 for both) (Table 1). A higher percentage of infants in both groups were male (MLP 52.5% and TC 57.7%). Participants from TC and MLP groups were predominantly Irish Caucasian (89.7% and 93.1% respectively) which reflects the local population. Whilst 96.7% of mothers of TC group were employed, maternal occupation was not documented for two participants (2.2%). In the MLP group, 74.7% of mothers were employed with maternal occupation not documented in 18.7% of the group. There was no significant difference in employment category between TC and MLP infant groups (p = 0.784).

Within the MLP group, 52.5% (53/101) were from singleton pregnancies. MLP infants from a singleton pregnancy(n = 53) had significantly higher birthweight than infants from a multiple pregnancy(n = 48) (median (IQR): 2.27(1.95–2.65) kg vs. 1.97(1.78–2.14) kg, p < 0.001); however, this difference in weight between infants from a singleton(n = 36 and multiple pregnancy(n = 35 did not persist at 18 months (median(IQR): 11.29(10.80–12.49) kg vs. 11.37(10.60–12.24) kg, p = 0.666). In the MLP group, nine infants had a birth weight < 10th centile for GA i.e., small for gestational age. Cranial ultrasounds were performed on three infants (all <1,500 g), as per local guidelines, all of which were normal prior to discharge.

At 18 months (+2 weeks) PMA 74.3% (75/101) of the MLP infant group returned for follow up. The large fall out rate (n = 26) may in part have been associated with Covid 19 and some families chose to withdraw and some parents were not contactable or known to have moved overseas. Assessments were delayed for a number of MLP infants (n = 14). These infants were assessed at 22 months PMA as soon as restrictions were lifted and Griffiths scores were adjusted according to PMA. There was no significant difference in birth demographics between MLP infants who were lost to follow up and those who returned at 18 months. Follow up of the TC group was completed prior to the global pandemic and hence study retention was not impacted by same. In the TC group 94.8% (92/97) returned for a Griffiths developmental assessment at 18 months(+2 weeks) (Table 1).

In two carefully categorised prospective healthy cohorts we have shown that MLP infants had significantly lower scores compared to TC infants in the Griffiths lll MDS at 18 months. This study shows that MLP infants are vulnerable to suboptimal neurodevelopment. There were no significant differences in scores found in Subscale B (Language and Communication) between the two groups (p = 0.107) however, when we controlled for sex, a significant difference was evident between the groups (p = 0.028). In a study by Putnick et al. inclusive of very preterm infants (n = 204), moderate-late preterm (n = 276) and term born infants (n = 268), language was assessed at 5 months, 20 months, 4 years, 6 years and 8 years of age (28). Very preterm children consistently performed less well than term-born children, with moderate-late preterm children also identified as being at risk for poorer language performance compared to term-born children. By comparison, Perez-Pereira et al. compared language development of low risk preterm infants (n = 150, mean GA of 32.62 weeks) and term born children (n = 49, mean GA of 39.70 weeks) up to 30 months and concluded that low risk preterm children were not at risk of delayed language development (29).

Both MLP and TC infant groups scored high in Subscale E (Gross Motor). However, statistically significant differences were evident between the groups (p < 0.001) with the TC group out-performing the MLP infant group. When the MLP group was subdivided by pregnancy type, differences were evident in Subscale B (Language and Communication) in which MLP infants from singleton pregnancies significantly outperformed MLP infants born from a multiple pregnancy (p = 0.040). A systematic review comparing the outcome of infants from twin and singleton pregnancies described differences in language and communication but proposed that the likely cause was due to the parental demands of managing two infants (30). Thorpe identified language delay in twins as mild but prevalent especially in male infants (31). It is not certain if low birth weight in multiple pregnancies may be a potential confounder which is associated with 60% of twin pregnancies (32).

Capobianco et al. compared the outcome of preterm infants (average birth GA of 32 weeks and birth weight of 2,200 g) with no neurological sequela (n = 20) and a normal Bayley's lll score(>85) at 18 and 24 months with term born infants (average birth GA of 39 weeks and birth weight of 3,500 g) (n = 20). The preterm infant group demonstrated significantly slower language acquisition but these differences were only significant at 16 months (p < 0.001) and 18 months (p < 0.005) but not at 20, 22, 24 months of age (33). However, in the Copabianco study, assessments were based on chronological age rather than corrected age. Using the Bayley's ll developmental assessment, Gould et al. described differences in scores achieved by preterm infants (n = 554) based on chronological age vs. corrected age as 17.3 points lower on the mental scale (79.5 vs. 96.8 respectively) and 11.8 points lower on the motor scale (84.8 vs. 96.6, respectively) at 18 months (34).

In the MLP group, we did not find any significant differences in outcome related to birth GA. This may be because of low numbers at each gestational age group meaning that we were underpowered to examine this effect. In a similar study investigating the neurodevelopmental outcome and social-emotional competence of MLP infants vs. a term control infant group at 2 years, Cheong et al. also found little evidence of an association with gestational age (7).

As infants in both studies had an uncomplicated neonatal stay / hospital stay on post-natal wards, factors such as suboptimal brain development and/or a high frequency of mild brain injuries (35), maternal well-being, obstetric/fetal complications warrant consideration in explaining the differences in developmental outcome between the groups. Pre-eclampsia, male sex and interventional delivery have been identified as antenatal/perinatal risk factors for poor school performance, which Johnson et al. also identified as independent risk factors for cognitive impairment (14).

Other determinants of neurobehavior and intellectual development include level of parental education, socioeconomic status, cultural background and maternal mental health conditions which may be considered additional risks for suboptimal outcome (36–41).

Socioeconomic metrics may be derived from many different aspects including maternal education (40). Maternal education is the most strongly associated dimension of a child's neurocognitive development (42). Detailed data on maternal education were not available for our MLP cohort. Of those employed similar profiles of maternal occupation were seen, with 71.4% of mothers of MLP infants identified as professionals requiring a minimum of a university degree whilst 64.0% of TC group fell into the same category. These high percentages may in part reflect the Organisation for Economic Co-operation and Development (OECD) findings that 60% of women in Ireland between 25 and 34 years and 51% of women between 25 and 64 years have a tertiary education (43). It also may reflect the local community of 210,000 persons from which the cohort was recruited which is home to world leading information technology and software development organisations, large pharmaceutical companies, the largest acute tertiary hospital outside the country's capital and a high ranking university, all of which have high employment of professionals. Finally, it may also be reflective of the professional mothers who agree to take part in birth cohort studies, which is generally higher than the general population. We have reported similar high rates of tertiary education in a previous cohort study (44).

As 18.7% of mothers in MLP group (vs. 2.2% of TC group) had an undocumented occupation this led to employment analysis of a smaller group (n = 56) of MLP infants. This may have impacted the differences in percentage of employed professionals in the MLP infant group (71.4%) versus TC group 64.0% (57/89). It is reassuring that the differences in neurodevelopmental outcome are not due to an imbalance in maternal employment between the two groups (p = 0.784).

The results of our study are consistent with those of Johnson et al. (n = 638 MLP infants and n = 765 TC infants) (14) describing female infants outperforming male infants in the MLP group with significantly higher scores in overall general development (p < 0.001). However, the Johnson study was based on parental questionnaires in comparison to our study which used a well-validated neurodevelopmental assessment tool.

A gender disparity is evident in the literature with male infants having a greater mortality and respiratory morbidity than female infants which is not explained by neonatal factors (45). In a review of outcomes of male and female preterm infants, O’Driscoll et al. described an increased vulnerability of male preterm infants which they suggest is likely to be multifactorial and perhaps includes hormonal and genetic factors and immunological differences between males and females (46). Romeo et al. described significant gender differences in low-risk preterm infant groups (n = 69 very preterm and n = 71 late preterm infants) and no gender differences amongst term born infants (n = 48) based on the scores achieved in the Bayley's scales of Infant Development, 2nd edition at two years (47). However no gender differences were reported by Romeo et al. at 12 and 24 months using the Griffiths ll standardised assessment tool which the authors suggested was attributed to a higher ratio of female to male participants (F:M = 1.6:1) (48).

The risk of preterm birth and infant mortality rates are higher in multiple pregnancies (49, 50). Almost half of the MLP infants in this study were from a multiple pregnancy (49.3%). General development DQ scores of MLP infants from singleton (n = 38) and multiple pregnancies (n = 37) at 18 months were comparable (p = 0.319). Restricting the analysis to singletons in both groups did not change our findings (Table 4).

Overall cognitive performance of children born moderately preterm (MPT) and MLP have been described as consistently lower in comparison to the term-born group (51). A study by Cheong et al. (7) of MLP and term-born infants concluded that the greatest difference between the groups was in the language domain at 2 years (−11.4 (95% CI, −15.3 to −7.5). However, our study has not shown a statistically significant difference between the MLP and TC groups in the domain of language and communication (p = 0.107). These differences may in part be explained by the differences in the timing of the assessment and the assessment tool used; Cheong et al. used the Bayley Scales of Infant Development assessment (Third Edition) at 2 years corrected age.

Fernald and Marchman cite individual differences in language proficiency which can accelerate and abate at different time points for each infant (52). Our single assessment provides a snapshot of language development at one-time point at 18 months, which may be too early to detect significant language delay and the clinical relevance of our findings on long term outcomes has yet to be established. We accept that neurodevelopmental deficits may be quite subtle and frequently not become evident until preschool or school-age years and may change positively or negatively at an older age (53). All impairments, irrespective of degree, have the potential to limit professional options, life opportunities and overall quality of life.

In this study, MLP infants had significantly lower scores in the personal, social and emotional subscale of development (p < 0.001) with a moderate effect size (Cohen's d = 0.590) than the TC group. Research has shown an increased risk of delayed social competence, with social behavioural issues higher amongst MLP infants in comparison to term-born infants (54). MLP infants also have a predisposition to more challenging behavioural issues with an increased risk of hyperactivity and attention deficit hyperactivity disorder (ADHD) (51). However, the association between prematurity and hyperactivity has not been firmly established (55). Children with learning, social, motor and sensory impairments that challenge their performance prior to or during school-going years may be considered to have special educational needs (SEN). Increased SEN are well described in MLP -born groups (56). Odd et al. identified MLP infants as being 50% more likely to have SEN requiring support in comparison to healthy term-born infants (57).

A strong correlation has been shown between prematurity and a diagnosis of cerebral palsy (CP) with the absolute risk of CP increasing with decreasing birth GA. Pierrat et al. reported that 1.0% of infants born between 32 and 34 weeks GA weeks (n = 1,187) had a diagnosis of CP at 2 years of age (58). Although more mature preterm infants generally have better outcomes, MLP infants have a six times higher prevalence of CP than children born at term (7 per 1,000 vs. 1.1 per 1,000 live births) (59). This may be related to the incidence of brain lesions in MLP infants described previously on cranial ultrasound (35) the clinical relevance of which warrants further study including long-term follow-up.

Our study has shown that the greatest difference between the MLP and TC groups was in the area of gross motor development (p < 0.001, Cohen's d = 0.665) and eye-hand co-ordination (p < 0.001, Cohen's d = 0.648) with a moderate effect size for each subscale. In this study no MLP infants had a diagnosis of CP at 18 months based on neurological exam nor were there any concerns about motor impairment. However, mild motor impairment may only become apparent beyond 18 months. Williams et al. suggest that motor impairment (either CP or non-CP e.g. Developmental Coordination Disorder) may be subtle and not evident until early childhood years and may be the most common form of impairment in the preterm population (60).

MLP infants are at higher risk of suboptimal outcome at 18 months in comparison to term-born infants. This study highlights the effect of moderate to late preterm birth on neurodevelopmental outcome. Our findings support the need for a broader effort for longitudinal surveillance of the MLP infant group including those who appear healthy.