Changes in emotional problems, hyperactivity and conduct problems in moderate to late preterm children and adolescents born between 1958 and 2002 in the United Kingdom

Background: Preterm birth is a risk factor for the development of emotional and behavioural problems in


INTRODUCTION
Preterm birth is a risk factor for the development of emotional and behavioural problems in childhood and adolescence (Aarnoudse-Moens et al., 2009;Bhutta et al., 2002). Preterm birth rates (<37 weeks of gestation) have increased significantly with one in ten babies being born preterm (Chawanpaiboon et al., 2019). Among preterm children, those born moderate-to-late preterm (MLPT; born between 32 nd and 36 th completed weeks of gestation) constitute the largest subgroup, accounting for 84.7% of preterm births worldwide (Chawanpaiboon et al., 2019) and 85.3% of preterm births in the UK (Euro-Peristat Project, 2018). However, our knowledge regarding the behavioural and emotional development of MLPT born children has been limited in two ways.
First, most studies on the behavioural and emotional outcomes of preterm children have focused on those born very (VPT) or extremely preterm (EPT) rather than MLPT born children. Higher rates of behavioural and emotional problems in the VPT or EPT group have been consistently reported in childhood (Adams-Chapman, 2006;Johnson & Wolke, 2013;Shapiro-Mendoza & Lackritz, 2012). To illustrate, it was shown that hyperactivity problems were more pronounced in EPT born children with ratings being 0.46 SD higher than FT children (Mathewson et al., 2017) and very recently, it was shown that children born VPT were at 5.6-fold elevated risk of meeting the criteria for any mental health disorders at 13 years of age (Yates et al., 2020).
Traditionally, it was thought that MLPT born children are only little affected in their development. However, there is increasing evidence that MLPT born children show suboptimal neurodevelopmental outcomes during the early years (Johnson, Evans, et al., 2015;Morse et al., 2009;Woythaler et al., 2011). This could be due to the fact that MLPT birth coincides with rapid brain development, it puts infants at an increased risk for brain injury, and it is related to a smaller and less mature brain size (Walsh et al., 2014). Thus, it is important to understand if emotional, hyperactivity and conduct problems are also more common in MLPT born children given that they represent the majority of the preterm births.
Existing studies on emotional problems in MLPT born children focused on the age range from preschool to 8 years (de Jong et al., 2012;McGowan et al., 2011;Potijk et al., 2012;Rogers et al., 2013;Talge et al., 2010;van Baar et al., 2009;Westrupp et al., 2012), and the studies on hyperactivity focused on early childhood (Huddy et al., 2001;Talge et al., 2010;van Baar et al., 2009). Moreover, there is less focus on conduct problems, where one study reported higher externalising problems at 4 years of age (Potijk et al., 2012), while another one reported lower levels of aggression from 4 to 15 years of age compared to the FT group (Gurka et al., 2010). Some of these studies revealed that MLPT born children have higher scores in both internalising and externalising problems than FT born children, whereas others showed that MLPT children have higher scores only in attention problems and internalising problems, along with some other studies which showed no significant differences between the MLPT and FT groups (Fitzallen et al., 2021;Gurka et al., 2010;Potijk et al., 2012;Talge et al., 2010;van Baar et al., 2009). Thus, there is a lack of studies that has investigated emotional problems, hyperactivity and conduct problems following MLPT birth during late childhood and adolescence with existing studies revealing conflicting results.
Second, no studies have investigated whether differences between MLPT and FT children in emotional problems, hyperactivity and conduct problems have changed across recent decades during which major changes in prenatal and neonatal care practices have taken place. It may be expected that any differences between MLPT and FT children may have decreased over time due to advances in obstetric and neonatological care (Dunn, 2007), such as improved nutrition, treatment of infections and, in particular for the moderately preterm infants in high-income countries, the introduction of assisted ventilation in the late 1960s, antenatal glucocorticoids in the 1970s and postnatal surfactant therapy in the 1980s (March of Dimes, 2012;Philip, 2012). In addition to the technical developments in neonatal care, the introduction of family-centred care have been shown to be beneficial for infant morbidity and psychological development (Boundy et al., 2016).

Key points
� It is unknown if the difference between moderate to late preterm and full term children has changed from 1958 to 2000-2002 with regards to emotional problems, hyperactivity, and conduct problems given the advancements in neonatal care during this period. � Teachers should be informed about the association between moderate to late preterm birth and behavioural and emotional problems in late childhood and adolescence to prevent the long-term negative outcomes associated with moderate to late preterm birth.
In the current study, we examine whether the difference in emotional problems, hyperactivity and conduct problems between MLPT and FT born individuals has changed over a 40-year period in the UK at three distinct time points in development: early childhood, late childhood and adolescence. We used mother ratings for all outcomes. In addition, we used self-reports of emotional problems in adolescence where such data was available since self and mother reports on emotional problems may differ from each other (Eg et al., 2018).
Several factors which may play a role in the development of emotional and behaviour problems such as sex, multiple birth, parity, low parental education, minority ethnicity status, neurosensory impairments, foetal growth restriction, maternal age and maternal smoking during pregnancy were considered as covariates (Grey et al., 2004).

National Child Development Study
The National Child Development Study (NCDS) is a cohort of 17,415 children born in one week (3 rd -9 th March) in 1958 in England, Wales and Scotland (Fogelman, 1983;Power & Elliott, 2006). It includes assessments at birth, 7, 11, and 16 years and later time points.
More information regarding the sample is available on the website: http://www.cls.ioe.ac.uk/. In the current study, the assessments at

The Avon Longitudinal Study of Parents and Children
The Avon Longitudinal Study of Parents and Children (ALSPAC) is a UK birth cohort examining the determinants of development, health and disease. The study is described in detail elsewhere Fraser et al., 2013). ALSPAC recruited 14,541 pregnant women with expected delivery dates of 1 st April 1991-31 st December 1992. Of the initial pregnancies, there were 14,676 foetuses resulting in 14,062 live births; 13,988 children were alive at 1 year of age. Ethical approval was obtained from the ALSPAC Law and Ethics committee and the local research ethics committee.
Informed consent for the use of data collected via questionnaires and clinics was obtained from participants following the recommendations of the ALSPAC Ethics and Law Committee at the time. From the first trimester of pregnancy, parents completed postal questionnaires about themselves and the study child. The study website contains details of all data available through a fully searchable dictionary (http://www.bris.ac.uk/alspac/researchers/data-access/data-dictionary/). In the current study, assessments at 6 (N MLPT = 378; Assessments across cohorts were grouped into three points: early childhood (5-7 years), late childhood (10-11 years) and adolescence (14-16 years). Characteristics of the participants included in the study are shown in Table 1 (See Table S1 for characteristics of the participants who dropped-out). The ethical approval for the current study was obtained from the Ethics Committee of the University of Warwick (96/17-18).

Moderate to late preterm birth
Gestational full weeks were recoded into a categorical variable as follows: 0, full-term (FT: 37 th -41 st completed weeks of gestation); and 1, moderate to late preterm (MLPT; 32 nd -36 th completed weeks of gestation).

Emotional problems, hyperactivity and conduct problems
At all measurement points, we used mother ratings of emotional problems, hyperactivity and conduct problems using the Rutter behaviour questions (Rutter et al., 1970) in NCDS (at 7, 11, 16 years) and BCS70 (at 5, 10, 16 years) and the Strengths & Difficulties Questionnaire (SDQ) (Goodman, 1997) in ALSPAC (at 6, 11, 16 years) and MCS (at 7, 11, 14 years). Although SDQ was developed based on the Rutter scale and the two scales are highly correlated (Stone et al., 2010), we wanted to make sure the scores are comparable across the scales. Thus, we only used the items which are very similar across the scales following the guidance from Centre for Longitudinal Studies regarding the harmonisation of mental health measures across UK cohorts (Table S2). The harmonised answering scale ranged for all measures from one (not true) to three (certainly true).
With respect to emotional problems, we further used selfreports in adolescence in the cohorts where such data was available (BCS70, ALSPAC and MCS). Adolescents completed the 15-item BILGIN ET AL. psychological subscale of Malaise Inventory (Rutter et al., 1970) in BCS70, which measures the level of depression with a 3-point response scale (0 = rarely or never, 1 = some of the time, 2 = most of the time) and has good reliability and validity (Rodgers et al., 1999).

Covariates
The following variables were included as covariates: sex (0 = male; 1 = female), ethnic group (0 = majority; 1 = minority), birth status (0 = singleton; 1 = multiple birth), parity (0 = first born child; 1 = second born or later), neurosensory impairment (0 = no; 1 = yes), obligatory education only or lower of either parent at participants' birth (0 = A level/vocational equivalent or higher education/university degree; 1 = obligatory education only or lower), foetal growth restriction, maternal age at birth, and smoking during pregnancy (0 = no; 1 = yes). Neurosensory impairments were either extracted from medical records or parent-reported and indicated by the presence of either (a) an IQ score for fluid reasoning of 3SDs below the mean identified between 5 and 11 years, (b) parent-or schoolreported visual or hearing impairment, or (c) severe congenital malformations identified during the child's first year of life. As a measure of foetal growth, birth weight z-scores were computed via the model proposed by Hadlock (Hadlock et al., 1991).

Statistical analysis
Statistical analyses were performed with Stata version 15. Differences between MLPT and FT-born individuals in emotional problems, hyperactivity and conduct problems within cohort in early and late childhood and adolescence were tested using multivariate linear regressions with the survey command (svy), which includes the samplespecific weights to account for the survey design of MCS, which over-sampled areas of higher proportions of ethnic minorities and child poverty. Standard analytic procedures and sampling weights, developed for use with MCS (Plewis, 2007), accounted for this, and ensured results were nationally representative. All utilised outcome variables were standardised according to the scores of FT children (z-scores). All analyses were adjusted for sex, ethnic group, multiple birth, parity, neurosensory impairment, parental education at birth, foetal growth restriction, maternal age, and smoking during pregnancy (Table S3).

Power analysis
Based on the sample sizes, observed statistical power to detect a small difference between MLPT and FT of d = .20 was calculated with Gpower (version 3.1.9.4) (Faul et al., 2007), which revealed statistical power greater than .85 for all of the analyses (Table S4).

Sensitivity analyses
Three sensitivity analyses were conducted to test whether the findings were robust to changes in the analytic plan: (1)

Emotional problems in MLPT born children and adolescents
In early childhood, there were no significant differences between MLPT and FT born children in all cohorts ( Figure 1, Table 2). In late childhood and in adolescence, the MLPT group had significantly higher scores on mother-rated emotional problems in comparison to the FT group in the most recent cohort (β = .147, 95% CI: .023; .271; and β = .158, 95% CI: .048; .267, respectively) but not in the earlier cohorts. However, there were no significant differences between the MLPT and FT groups in self-reported emotional problems from 1970 to 2000-2002.

Hyperactivity in MLPT born children and adolescents
In early childhood, there were no significant differences between MLPT and FT born children in hyperactivity scores in all cohorts ( Figure 1, Table 2). In late childhood, MLPT born children had higher levels of hyperactivity than FT born children in the two most recent cohorts (β = .208, 95% CI: .081; .334 and β = .158, 95% CI: .054; .261, respectively). In adolescence, MLPT group had higher scores of hyperactivity than the FT group only in the most recent cohort (β = .199, 95% CI: .073; .325).

Conduct problems in MLPT born children and adolescents
In early childhood, the MLPT and FT groups were not significantly different in conduct problems in any of the cohorts (Figure 1,

Sensitivity analyses
Sensitivity analysis comparing the LPT group separately with the FT born group showed consistent findings with the main analyses for emotional problems and hyperactivity; however, there was no significant difference between the two groups in conduct problems in any of the assessment points across all cohorts (Table S5). Furthermore, sensitivity analysis using imputed dataset revealed consistent results with main analyses (Table S6). When the full-scales were used, findings regarding late childhood and adolescence remained consistent, and higher scores of emotional problems and hyperactivity were found in early childhood in the two most recent cohorts (ALSPAC and MCS, Table S7).

DISCUSSION
This study of four cohorts born between the late 1950s and the early 2000s in the UK showed that the difference between MLPT and FT born individuals in emotional problems and hyperactivity has increased in recent cohorts in comparison to cohorts born in 1958 and 1970. This was found in mother reports in late childhood and adolescence, whereas no significant differences were found in BILGIN ET AL. Our finding of an increase in mother-reported emotional problems and hyperactivity in the UK during late childhood and adolescence in MLPT individuals born between the late 1950s and the early 2000s is contrary to our expectation of a decrease in these problems over the years given the advances in the prenatal and neonatal care (Dunn, 2007)  Since the 1970s, the incidence of emotional problems and hyperactivity in childhood and adolescence has increased on a population level in the UK (Collishaw, 2015;Collishaw et al., 2004). It has been suggested that this increase in emotional problems and hyperactivity in the general population could be due to societal changes in recent years such as a greater emphasis on academic performance and pressure to perform (Livingstone & Smith, 2014;Sweeting, West, Young, & Der, 2010;Twenge et al., 2010). It is possible that MLPT children are more vulnerable to the increased pressure to perform at school as they have more difficulties with academic performance, particularly maths . In addition, MLPT born children represent a clinically distinct population since many of them are delivered for maternal medical complications during pregnancy (e.g., early delivery due to preeclampsia via cesarean section or induction of labour), which are consequently linked with increased rates of behavioural and emotional problems in children (Curran et al., 2016;Lahti-Pulkkinen et al., 2020;Silva et al., 2014). Potential mechanisms for such associations include changes in placental and immune system, and hypothalamic-pituitary-adrenal axis functioning during pregnancy that may lead to foetal programming (Maher et al., 2018).
Thus, there may be clinical reasons for the increase in emotional problems and hyperactivity in the recent years.
Alternatively, our findings could be explained by the increased awareness of mothers about the influence of preterm birth on the development of emotional and behavioural problems, which may be reflected in their ratings during recent years (Guralnick, 2012). Thus, our findings might actually reflect maternal perception of vulnerability of their MLPT born children and adolescents rather than actual differences in their behaviour. While being speculative, this interpretation is consistent with our finding that there are no significant differences between MLPT and FT groups in self-reported emotional problems in adolescence. The low agreement in reports of emotional and behavioural problems between parent and child reports has also been shown previously in community studies (Achenbach et al., 2005) and in preterm samples Eves et al., 2020). Moreover, this finding of no difference between MLPT and FT born adolescents in self-reported emotional problems is consistent with Festinger's theory of social comparison (Festinger, 1954), which suggests that individuals compare themselves with those they socialise most often, that is their peers in the case of adolescents. On the other hand, mothers of MLPT born adolescents are more likely to compare their adolescents to the whole birth cohort (i.e., all adolescents), which could be associated with using a different comparison level and with reporting more emotional problems than adolescents themselves (Eves et al., 2020).

Emotional Problems
Perception of child vulnerability among mothers of preterm children could be due to preterm birth being an emotionally challenging event as it is often unexpected and associated with a higher risk of medical complications and neonatal difficulties (Horwitz et al., 2015). It has been suggested that mothers might continue to perceive their premature born child as vulnerable beyond the infancy period, which could result in overprotective parenting (Horwitz et al., 2015). Although there is evidence that maternal sensitive parenting is not influenced by preterm birth (Bilgin & Wolke, 2015), overprotective and controlling parenting behaviours in mothers of MLPT born children have been recently documented (Toscano et al., 2020) and were found to be linked with hyperactivityimpulsivity levels from 4 to 8 years of age (Faleschini, Matte-Gagné, Luu, et al., 2020). Nevertheless, it is unclear if the influence of overprotective parenting in mothers of MLPT born children on hyperactivity-impulsivity levels has a longer-lasting influence into late childhood and adolescence.
Our findings regarding mother-reported conduct problems were less consistent than those relating to mother-reported emotional problems and hyperactivity. MLPT born adolescents had lower scores in conduct problems in comparison to FT born adolescents in the ALSPAC cohort, which is in line with 'preterm behavioural and personality phenotype' which suggests that those born preterm are socially withdrawn and disinclined towards risk taking (Johnson & Wolke, 2013;Montagna & Nosarti, 2016 Despite the fact that MLPT children experience fewer neonatal complications and have a more mature brain size than VPT children, they are still born immature in comparison to FT children (Walsh et al., 2014). Considering the increasing number of MLPT birth survivors (Chawanpaiboon et al., 2019), our findings suggest that the teachers should be made aware of this risk and the particular increase in emotional problems and hyperactivity in late childhood and adolescence following MLPT birth in the recent years . Behavioural and emotional sequelae associated with MLPT birth is particularly important for employment and educational achievement during adulthood (Bilgin et al., 2018); thus, identification of problems during the school years would be helpful in improving the long-term negative consequences on behavioural and emotional development.
The current study has several strengths including large cohort studies with participants who were born between the late 1950s and the early 2000s, measurement of the outcomes at three assessment points from childhood to adolescence and adjustment for several covariates. Moreover, we explored the consistency of our findings through several sensitivity analyses. Nevertheless, there are also limitations of the current study. First, the drop-out rate was high, which is a common problem in longitudinal studies, where those at risk of medical and social problems are likely to drop-out (Wolke et al., 2009). Nonetheless, the findings of the analysis with imputed data revealed a similar pattern of results, which suggests that the study attrition may not have influenced our findings. Second, mother reports were based on different scales in the older and recent cohorts (Rutter Scale and SDQ). Nonetheless, the SDQ was developed based on the Rutter Scale and they show high correlation (Goodman, 1997;Stone et al., 2010). In addition, we used only the most similar items from these two questionnaires. This approach of harmonisation of questionnaires was previously used to document trends in mental health across the UK cohorts (Collishaw et al., 2004). Third, self-report measures of emotional problems were available only in three cohorts, where the scales were different between the oldest and the two newest cohorts. Last, the reliability scores of the harmonised scales were lower in comparison to the full sub-scales, particularly in hyperactivity and conduct problems.
However, sensitivity analyses using the complete scales revealed similar findings.
In conclusion, mother-reported emotional problems and hyper-

CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.

ETHICAL STATEMENT
Ethics Committee of the University of Warwick provided ethical approval for the current study and informed consent was obtained from all participants.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.