Preterm birth and early life environmental factors: neuropsychological profiles at adolescence and young adulthood

To establish neuropsychological profiles after high- and low-risk preterm birth (i.e., with and without neonatal brain injury) during adolescence and young adulthood and to assess the potential role of early life environmental factors in cognition. Participants (N = 177; Mage = 20.11 years) of both sexes were evaluated when adolescent or in young adulthood. They were grouped according to their birth status: 30 high-risk preterm, 83 low-risk preterm and 64 born at full term. Significant differences were found in several cognitive domains between groups. Furthermore, familial socioeconomic status (SES) moderated the relation between the degree of maturity/immaturity at birth and cognition (F(5,171) = 11.94, p < 0.001, R2 = 0.26). The findings showed different neuropsychological profiles during adolescence and young adulthood, with the high-risk preterm sample evidencing lower cognitive values. In addition, higher scores in the familial SES score in this study seem to have a protective effect on cognition.

Preterm birth, defined as any birth before 37 weeks of gestation [1], is associated with a higher or lower risk depending on its clinical conditions.High-risk preterm newborns frequently suffer from neonatal brain damage, which can have long-term consequences [2,3].The neuropathology of preterm infant brain injury encompasses a variety of pathologies, including periventricular leukomalacia (PVL) with neuronal/axonal abnormalities and severe germinal matrix-intraventricular hemorrhage (GM-IVH), notably with periventricular hemorrhagic infarction [4].Low-risk prematurity, unassociated with neonatal brain injury, implies a gestational age (GA) of 30-36 weeks, modest neurological abnormalities, and no perinatal comorbidities [5,6].However, these children's underdeveloped neural system could account for long-term neurodevelopmental alterations [7].
Preterm-born children frequently exhibit worse cognitive scores than their full-term peers at a preschool age, with similar differences in their intelligence quotient and neuropsychological functions to those observed later in life [8][9][10].Preterm delivery can result in neuropsychological problems that last far beyond the first decade of life, implying a clear association between GA and cognition [11,12].The limited evidence gathered from preterm adolescents and young adults born at younger GA's showed structural and functional brain alterations involved in cognitive performance [13], suggesting there had been and would be no improvement over time [14,15].Although great advances in neonatal intensive care have been made, prematurity-related consequences (i.e., neuropsychological and behavioral difficulties, hypertensive illnesses and metabolic syndrome) can be expected [16] and will lead to a heavy burden of lifelong neurological morbidity [17].
Apart from the presence or absence of neonatal brain damage, it is important to consider the potential impact of certain environmental factors on neurodevelopment after a preterm delivery.Preterm-born children have a higher biological vulnerability due to greater immaturity at birth, increasing their likelihood of suffering neurological, personality, and behavioral problems [16,18].Hence, preterm birth might also lead to a greater susceptibility to early life environmental factors.For instance, an adverse influence from neonatal variables is not sufficient to explain the worsened cognitive performance, but a mother's lower educational level might also influence the outcome [19,20].The effect of socioeconomic factors appears to influence the relation between prematurity and cognitive delays during childhood [21].What is more, the quality of the home environment has also been found to have a moderate effect on early cognition in the preterm population [22].Nevertheless, the extent to which early life environmental factors may affect long-lasting cognition in preterm-born adolescents and young adults is still an area that needs to be determined.
Degrees of neonatal immaturity, neurological damage, environmental factors, and the time of their occurrence all affect the way prematurity shapes the brain [23].For this reason, the present study aims to ascertain the neuropsychological profiles of highand low-risk preterm children when they reach adolescence and/ or young adulthood.Furthermore, it also seeks to retrospectively evaluate the potential role that early life environmental factors may play in preterm-born adolescents and young adults in so far as their long-term cognitive outcome.

MATERIAL AND METHODS Participants
A total sample size of 85 participants was estimated according to G*Power tool v.3.1;specifically, a linear multiple regression analysis was conducted with an effect size of 0.15, an error probability of 0.05, and a power of 0.8.
A total of 177 Caucasian participants participated in this study: 113 preterm adolescents and young adults (30 high-and 83 low-risk preterm) and 64 full-term born participants.The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the University of Deusto [ETK-22/17-18] and the Drug Research Ethics Committee of the Basque Country [CEIm-E, PI2018154].All participants provided prior written informed consent; in the case of under-aged participants, their consent was required as well as that of their legal tutors at the beginning of the study.
A sample of 30 preterm adolescents and young adults with a history of neonatal brain damage was recruited (henceforth the high-risk preterm group).Sixteen of these had GM-IVH, and the other 14 subjects presented a diagnosis of PVL.Among the former group, 12 of its participants (GM-IVH) had been included in a cohort that participated in a longitudinal study at the Vall d'Hebron University Hospital (Barcelona, Spain); all of the participants in the latter group (PVL) were also being followed by the Vall d'Hebron University Hospital (Barcelona, Spain) in collaboration with the University Hospital Clínic (Barcelona, Spain).The remaining four participants with GM-IVH were recruited by chain-referral sampling (i.e., a nonprobability sampling strategy whereby existing participants refer new individuals for recruitment).The characteristics of both samples over their childhood have already been published elsewhere [24][25][26].The established inclusion criteria for the high-risk preterm group were: (1) preterm birth with a GA < 37 weeks of gestation, (2) having a diagnosis of brain damage diagnosed by neonatal cranial ultrasound or magnetic resonance imaging, and (3) ranging in age at the time of evaluation from 15 to 30 years.Exclusion criteria for this group were the concurrence of other brain injuries, birth trauma, malformation, and infectious or metabolic encephalopathies.
Since isolating the specific impact of neonatal brain damage from a preterm birth on cognitive performance was also of interest, a group of preterm adolescents and young adults at lower risk of presenting major disabilities was recruited at Cruces University Hospital (Bilbao, Spain).This group consisted of 83 low-risk preterm subjects and had the following inclusion criteria: (1) GA of between 30 and 36 weeks, (2) absence of brain pathology identified by neonatal cranial ultrasound, (3) lack of substantial neonatal morbidity (i.e., congenital malformations, neurologic, cardiac or digestive, necrotizing enterocolitis or septic shock), and (4) ranging in age at evaluation from 15 to 30 years.A full-term group of 64 participants was included as controls.Inclusion criteria were: (1) GA > 37 weeks, and (2) ranging in age at evaluation from 15 to 30 years.The exclusion criteria for both the low-risk preterm and the full-term group were a history of acquired brain injury, cerebral palsy or any other neurological impairment, congenital malformations and/or chromosomal abnormalities.

Instruments
Neuropsychological assessment.Analogical reasoning was evaluated using Raven's Advanced Progressive Matrices (RPM) test [27], and receptive language was measured with the Peabody Picture Vocabulary Test III (PPVT-III) [28].Verbal fluency was assessed in minute-long trials of phonetic (P, M, R) and semantic fluency (category of animals) tests [29].
The Working memory index was assessed through the Wechsler Adult Intelligence Scale IV (WAIS-IV) and Wechsler Intelligence Scale for Children (WISC-V), using digit span (forward, backward, and increasing) and arithmetic subscales for those over 16, and digit and picture span subscales for those under that age [30,31].Cognitive flexibility was measured using The Trail Making Test (TMT) part B [32], the Modified Wisconsin Card Classification Test (M-WCST) [33], and the PC and interference measurements from the Stroop Test [34].Last, processing speed was assessed by means of TMT part A and the Stroop Test (P and C measures) [32,34].
Cognitive domains driven by motivationally and emotionally significant contexts were also assessed.Theory of mind was measured using four stories from Happé's Strange Stories Test [35].Moral competence was evaluated by means of the Moral Judgment Test (MJT) [36], and delayed gratification was assessed by the Delay-Discounting Test (DDT), also called the Monetary Choice Questionnaire [37].
Socioeconomic status (SES).The Hollingshead Index was used to consider the occupation and education domains of the participants and their parents independently [38].Thus, two different measurements were obtained: participants' current SES (i.e., participants' highest ongoing occupation and educational level) and familial SES (i.e., the average between the highest level of the parents for their occupation and education domains).For under-aged participants, data from the occupation and education domains was determined from a single parent, so familial SES was the highest score obtained for either the individual's mother or father.
Emotional-behavioral assessment.The Adult Self Report (ASR) and Child Behavior Checklist (CBCL) were employed to measure emotional-behavioral output; internalizing and externalizing problems were recorded [39,40].
Early life environmental factors.The Parental Bond (PB) was assessed by using the Parental Bonding Instrument to appraise participants' parents' behavior independently (i.e., mother's and father's care and overprotection domains separately) toward them during childhood [41].Familial SES was used as an early life environmental factor [38]. Lastly, Adverse Childhood Experiences (ACE) were evaluated by the Adverse Childhood Experiences Questionnaire for Adults and Adverse Childhood Experiences Questionnaire for Teens designed by the WHO [42].
All neuropsychological assessments during adolescence and young adulthood were prospectively conducted in the Deusto University DeustoPsych and Ramon Llull University.

Statistical analysis
The normal distribution of the data was assessed using the Kolmogorov-Smirnov test (K-S).Tests were run on mean scores for the normally distributed data and on the ranked dependent variables for the nonnormally distributed data.Missing values for GA (1 low-risk preterm and 12 full-term participants); birth weight (BW) (1 low-risk preterm and 4 full-term participants); familial SES (3 low-risk preterm and 1 full-term participants); care (1 high-risk and 2 low-risk preterm participants); overprotection (1 high-risk and 2 low-risk preterm participants); and ACE variables (1 highrisk preterm, 16 low-risk preterm and 3 full-term participants) were imputed using the expectation maximization algorithm (i.e., the percentage of missing values was of 0.94%).All tests were standardized in order to assess cognitive performance using a composite score (henceforth called general cognitive functioning score) obtained from: RPM analogical reasoning and PPVT-III receptive language total scores, M-WCST category and perseverative error scores, WAIS-IV and WISC-V digit span and working memory index, phonetic and semantic fluencies, Stroop Test P, C, PC and interference measures, TMT parts A and B, Happé's Strange Stories Test total score, MJT moral competence final score and the DDT delay gratification ratio.Cronbach's alpha (reliability) coefficient was 0.84.
The Kruskal-Wallis test was used to analyze differences in non-normally distributed data such as neonatal data (GA and BW), age at evaluation, current SES, and emotional-behavioral measurements.Early life environmental factors (i.e., care measurement, familial SES, and ACEs) were also analyzed using this test.Further, the Chi-squared test was employed to assess differences in two qualitative sociodemographic characteristics: gender and handedness.In addition, a univariate analysis of variance was run to compare the overprotection measurement between the three Fig. 1 Moderation Models between the Degree of Maturity/Immaturity at Birth and General Cognitive Functioning Score.Hypothesized moderation models specifying paths between the degree of maturity/immaturity at birth and general cognitive functioning score at adolescence and young adulthood, moderated by care and overprotection measures, familial SES and/or ACEs.An optimal care and overprotection measures can soften the impact of the degree of maturity/immaturity at birth in the long-term general cognitive functioning score (M 1 and M 2 ).General cognitive functioning score alterations related to the degree of maturity/immaturity at birth during adolescence and young adulthood can also be attenuated through appropriate SES at childhood stage (M 3 ).Finally, elevated ACEs can exacerbate the negative effect that the degree of maturity/immaturity at birth may have on general cognitive functioning score during adolescence and young adulthood (M 4 ).Available data for emotional-behavioral assessment: 30 high-risk preterm, 79 low-risk preterm and 64 full-term.groups.A multivariate analysis of covariance was also used to compare different cognitive domains and general cognitive functioning score (adjusting for gender and age at evaluation).Bonferroni's post-hoc test was employed to assess differences between groups, and the Bonferroni corrected p-value for significance was calculated for the different cognitive tests assessed (p = 0.05/17 = 0.003).Partial eta squared was used to calculate the effect sizes of cognitive and general cognitive functioning score comparisons.To interpret this value, around 0.01 is considered a small effect size, 0.06 medium, and higher than 0.14 large.
Last of all, moderation analysis is a type of regression that examines how a third variable influences the connection between the predictor and the outcome variable.In this case, the moderating effect of early life environmental factors (i.e., care measurement, overprotection measurement, familial SES, and ACEs) was independently analyzed in the relation between the degree of maturity/immaturity at birth and the general cognitive functioning score.To assess the moderation effect, four moderation analyses (see Fig. 1) adjusted for gender and age at evaluation were executed using macro PROCESS 3.5 script for SPSS (released 1 May 2020).Before doing the moderation analyses, outlier analyses assessed for rupture of linearity, normality, multicollinearity, and homoscedasticity; the Mahalanobis and Cook distances as well as Leverage parameters were used to detect possible outliers.The scatterplot and histogram check showed no sign of any outliers in the study sample.The Johnson-Neyman outcome was tested to determine the degree to which early life environmental factors had a significant conditional effect on prematurity in the prediction of the general cognitive functioning score.For all preceding analyses, IBM SPSS version 26.0 (SPSS Inc., Chicago, USA) was used and the significance level was set at 0.05.

RESULTS
Neonatal, sociodemographic, emotional-behavioral variables as well as early life environmental factors are detailed in Table 1.Statistically significant differences were found in gender and age at evaluation between groups; thus, the following comparison and moderation analyses were adjusted for both.

Aim 1: Differences in neuropsychological profiles
Regarding cognitive performance, as shown in Table 2, the following statistically significant differences between groups remained significant after the Bonferroni correction was applied for multiple comparisons (p = 0.003): (a) worse performance in phonetic fluency and cognitive flexibility as measured by M-WCST in both preterm groups in comparison with the full-term sample; and (b) worse performance in processing speed as measured by the Stroop Test P and C measurements, cognitive flexibility assessed using TMT Part B and Stroop Test PC measure, and working memory index in high-risk preterm adolescents and young adults compared to the low-risk preterm and full-term groups.All differences showed medium to large effect sizes.Available data for MJT: 25 high-risk preterm, 80 low-risk preterm and 64 full-term adolescents and young adults.
In addition, both preterm groups showed worse performance in the general cognitive functioning score compared to the full-term group with a large effect size (see Table 2).Further, high-risk preterm-born adolescents and young adults showed even worse scores in relation to the low-risk preterm sample.
Aim 2: Moderating effect of early life environmental factors on the general cognitive functioning score Concerning four moderation analyses (see Table 3), for familial SES as a moderator, 26% of the variance was explained by the three factors (i.e., the degree of maturity/immaturity at birth, familial SES and interaction of both) in the overall model (F (5171) = 11.94,p < 0.001, R 2 = 0.26).More specifically, for every unit increase in familial SES as a moderator variable there was also an enhancement in the general cognitive functioning score (β = 0.03, t (171) = 3.17, p = 0.002), especially in those with a higher degree of neonatal immaturity (i.e., high-risk preterm group).In terms of different backgrounds of familial SES (i.e., low familial SES = 23.00;medium familial SES = 40.50;high familial SES = 57.76),for either low or medium familial SES there was a significant relation between the degree of maturity/immaturity at birth and general cognitive functioning score (see Fig. 2).Nevertheless, having a high familial SES (i.e., higher than 55.31) did not further moderate the relation between the degree of maturity/immaturity at birth and cognition over adolescence and young adulthood.
In addition, care, overprotection and ACEs early life environmental factors did not moderate the relationship between the degree of maturity/immaturity at birth and the general cognitive functioning score.

DISCUSSION
This study, with high-risk and low-risk preterm groups, established different neuropsychological profiles in preterm-born adolescents and young adults, reporting a greater number of worse cognitive domains in the high-risk preterm sample.Similarly, young adults born with very low BW exhibited worse scores in the intelligence quotient and academic achievement [43].Furthermore, disparities in SES backgrounds during childhood moderated the effect of prematurity on cognitive performance during adolescence and young adulthood, and more markedly so in those with a greater degree of neonatal immaturity.

Neuropsychological profiles
Our findings indicated a worse performance in phonetic fluency and cognitive flexibility as measured by M-WCST in both preterm  groups compared to their full-term peers at adolescence and young adulthood.Even though cognitive performance was within the normal range, as was previously reported by Hack [44], our results are also in line with the fact that prematurity in the absence of neonatal brain injury still leads to worse general cognitive functioning in the adolescence and young adult stage.That is, any degree of preterm birth affects cognition resulting in altered neurodevelopment over time [11].
The high-risk preterm-born group, on the other hand, revealed a larger amount of worsened cognitive domains than low-risk preterm adolescents and young adults, reinforcing the concept that the lower the GA, the higher the probability of presenting neurodevelopmental alterations (i.e., increased incidences of brain injury, cerebral palsy, and cognitive deficits) [45].Processing speed and working memory were the cognitive domains that differentiated the two preterm groups.In addition, cognitive flexibility, regardless of whether it was measured with a motor component or not was also found to differ between high-and low-risk preterm adolescents and young adults.In adolescents who had had younger GAs, white matter microstructure has now been related to both working memory and cognitive flexibility, with reduced microstructure being associated with worse performance [46].In fact, brain structural alterations, mainly detected after highrisk prematurity, commonly overlap those areas implicated in a worsened cognitive function [13].
Moderating effect of early life environmental factors According to Sansavini [23], preterm birth results in atypical developmental trajectories, which may vary because of the complex interaction of biological and environmental factors, underlining the non-linear process of development.Despite biological factors having a moderate effect on cognitive development at an early stage, these factors lose their effect over time [22].On the other hand, environmental factors have been shown to be increasingly important during infancy [47], and their impact on later cognitive development seems to remain [22].Nevertheless, in the present study, parental bond and ACE did not have a moderating effect on the association between the degree of maturity/immaturity at birth and the general cognitive functioning score.
Consistent with Wolke [48], factors beyond initial neonatal care may improve long-lasting cognitive outcomes for the preterm population.For example, preterm birth and low SES, separately, have been established as risk factors with widespread effects on developmental delay during childhood [49].Children who were born preterm and also lived in low SES conditions presented worse cognition [50].Our findings revealed that differences in SES environments during childhood can modulate brain reorganization after central nervous system disorders, thereby affecting cognitive performance during adolescence and young adulthood, markedly in those with neonatal brain damage (i.e., the high-risk preterm group).The current results are consistent with a previous study which suggested that the SES impact on cognition persists until the middle age range [51].Furthermore, the association of brain injury with worse cognitive performance has been found to be attenuated in children from mothers with higher educational levels [52,53].Likewise, in our study, the highest familial SES scores seem to have prevented the impact of prematurity on cognition during adolescence and young adulthood.

Limitations
This study was carried out in homogeneous high-and low-risk preterm samples in order to assess their neuropsychological profiles during adolescence and young adulthood.In other words, unlike prior studies that used heterogeneous preterm samples, this study used very stringent inclusion and exclusion criteria for neonatal data to obtain comparable groups, hence confirming the generalizability of results.In addition, the recruited groups did not differ in emotional-behavioral and early life environmental measurements.However, there was no longitudinal follow-up from early infancy to young adulthood that could have ascertained whether neurodevelopmental outcomes persisted, worsened or improved over time.Another limitation is sample size, which was not suitable for studying whether there might be different results regarding cognitive domains depending on which neonatal brain injury they had suffered from as well as within the PVL and GM-IVH group classifications as proposed by Flodmark et al. [54] (i.e., radiological classification of PVL) and Papile et al. [55] (i.e., GM-IVH grades detected through cranial ultrasound).Finally, the disparities found in our study concerning current SES have not been further explored; however, preterm birth has already been related to lower educational qualifications and a reduced rate of employment in the adult stage [56].Indeed, preterm-born children's performance in mathematics is critical for adult educational attainment [57].

CONCLUSIONS
In our opinion, establishing different neuropsychological profiles and SES background impacts after preterm delivery could support the development of specific intervention programs (e.g., in educational and social level), if necessary, during childhood.The supply of appropriate resources to the preterm population, mainly those with a greater degree of neonatal immaturity, might improve their later life neuropsychological profile.Nevertheless, more research is needed in high-risk and low-risk preterm populations in order to understand the implication of SES during childhood and determine their clinical impact on cognitive performance, since interventions may be especially important for low SES preterm children.

Fig. 2
Fig. 2 Moderating Effect of Familial SES between the Degree of Maturity/Immaturity at Birth and General Cognitive Functioning Score.Familial SES familial socioeconomic status.

Table 1 .
Neonatal, sociodemographic, emotional-behavioral variables and early life environmental factors.Alphanumeric superscripts are found in the reference group whose scores are higher than the letter represented, which are a: lower values of the high-risk preterm group; b: lower values of the low-risk preterm group; and c: lower values of the full-term group.In bold the domains that were significantly different among groups.GA gestational age, wks weeks, BW birth weight, g grams, yrs years, SES socioeconomic status, ACEs adverse childhood experiences, SD standard deviation, x mean, x ͂ median, IQR interquartile range, H Kruskal-Wallis test, F Snedecor's F distribution, and X 2 Chi-square test.

Table 2 .
Cognitive domains' and general cognitive functioning score differences.
The cognitive domains in bold are those that remained significant after Bonferroni correction was applied for multiple comparisons (p = 0.003).Alphanumeric superscripts are found in the reference group whose scores are higher than the letter represented, which are a: lower values of the high-risk preterm group; b: lower values of the low-risk preterm group; and c: lower values of the full-term group.SD standard deviation, F Snedecor's F distribution, PPVT-III Peabody picture vocabulary test-III, RPM Raven's Progressive Matrices, MWCST Modified Wisconsin Card Sorting Test, WMI working memory index, WAIS-IV Wechsler Adult Intelligence Scale IV, WISC V Wechsler Intelligence Scale for Children V, TMT Trail Making Test, MJT Moral Judgement Test, DDT Delay-Discounting Task, score; and η 2 p : partial eta squared.aAvailable data for semantic fluency: 30 high-risk preterm, 83 low-risk preterm and 63 full-term adolescents and young adults.bAvailable data for RPM: 25 high-risk preterm, 82 low-risk preterm and 64 full-term adolescents and young adults.c

Table 3 .
Moderating effect of early life environmental factors on general cognitive functioning score.
The significant moderation analysis is in bold.SES socioeconomic status, ACEs adverse childhood experiences.