Preschool development, temperament and genetic liability as early markers of childhood ADHD: A cohort study

Abstract Background ADHD is associated with multiple adverse outcomes and early identification is important. The present study sets out to identify early markers and developmental characteristics during the first 30 months of life that are associated with ADHD 6 years later. Methods 9201 participants from the prospective Avon Longitudinal Study of Parents and Children (ALSPAC) birth cohort were included. Outcome measures were parent‐rated ADHD symptom scores (Strengths and Difficulties Questionnaire, SDQ) and ADHD diagnosis (Development and Wellbeing Assessment, DAWBA) at age 7. Seventeen putative markers were identified from previous literature and included: pre‐ and peri‐natal risk factors, genetic liability (ADHD polygenic risk scores, PRS), early development, temperament scores and regulatory problems. Associations were examined using regression analysis. Results Univariable regression analysis showed that multiple early life factors were associated with future ADHD outcomes, even after controlling for sex and socio‐economic status. In a multivariable linear regression model; temperament activity scores (B = 0.107, CI = 0.083–0.132), vocabulary delay (B = 0.605, CI = 0.211–0.988), fine motor delay (B = 0.693, CI = 0.360–1.025) and ADHD PRS (B = 0.184, CI = 0.074–0.294) were associated with future symptoms (R 2 = 10.7%). In a multivariable logistic regression model, ADHD PRS (OR = 1.39, CI = 1.10–1.77) and temperament activity scores (OR = 1.09, CI = 1.04–1.16) showed association with ADHD diagnosis. Conclusion As well as male sex and lower socio‐economic status, high temperament activity levels and motor and speech delays in the first 30 months of life, are associated with childhood ADHD. Intriguingly, given that genetic risk scores are known to explain little of the variance of ADHD outcomes, we found that ADHD PRS added useful predictive information. Future research needs to test whether predictive models incorporating aspects of early development and genetic risk scores are useful for predicting ADHD in clinical practice.

Although ADHD is often first identified from school-age, epidemiological surveys suggest that rates in pre-schoolers are similar to those in older children and adolescents (Egger & Angold, 2006). Also, a growing body of research suggests that the manifestation of core ADHD symptoms and diagnoses may be preceded by earlier risk markers, including infant development, temperament and neural markers (Johnson et al., 2015;Shephard et al., 2022). Identifying early risk markers is important as early detection and intervention could alter adverse developmental trajectories, especially during critical periods of child development, when the brain is rapidly developing and neuroplasticity is highest (Sonuga-Barke et al., 2011).
A large body of research has focused on early temperament and infant behaviour as antecedents for ADHD (Nigg et al., 2004;Willoughby et al., 2017). Meta-analyses and prospective cohorts have shown that aspects of temperament such as multiple and persistent regulatory problems during the first months of life (e.g., excessive crying, feeding, and sleeping difficulties), are also associated with later behavioural and ADHD outcomes (Baumann et al., 2019;Hemmi et al., 2011;Schmid & Wolke, 2014).
Studies investigating acquisition of early developmental milestones in children with ADHD indicate poorer motor skills and language development scores in children with, compared to without, ADHD (Arnett et al., 2013;Havmoeller et al., 2019;Lemcke et al., 2016, Sephard et al., 2022, although a recent review (Athanasiadou et al., 2020) suggests that differences in early motor development are non-specific to ADHD.
Other early infant studies have focused on neonatal and postnatal measures of head circumference and there is some evidence to suggest that microcephaly at birth is associated with an increased risk of ADHD (Aagaard et al., 2018;Ferrer et al., 2019).
ADHD is multifactorial in origin, and early markers include known risk factors as well as infant antecedents. The most wellestablished risk factors for ADHD include preterm birth, low birth weight and genetic liability (Sucksdorff et al., 2015;Thapar et al., 2013), others include restricted foetal growth (Murray et al., 2016), younger maternal age (Galéra et al., 2011) and advanced paternal age (D'Onofrio et al., 2014).
Genetic factors contribute to ADHD (Thapar & Cooper, 2016) with genome-wide association studies (GWAS) suggesting that numerous common genetic variants contribute risk (Demontis et al., 2019). Polygenic risk scores (PRS) represent an individual's estimated total burden of common risk alleles for a particular disorder and ADHD PRS have been shown to be associated with ADHD across clinical and population samples (Stergiakouli et al., 2015).
To our knowledge, no longitudinal studies have undertaken a comprehensive examination of the simultaneous contribution of multiple infant early markers of ADHD and little research has successfully identified clinically informative early predictors. We used a UK population-based birth cohort and sought to identify early markers during the first 30 months of life associated with ADHD symptom scores or DSM-IV ADHD diagnosis. Both ADHD symptoms and diagnoses were explored to provide clinically relevant information, whilst also recognising that ADHD behaviours are dimensional, with similar risks associated across the continuum (Thapar & Cooper, 2016). Markers included pre-perinatal factors (intra-uterine growth restriction, maternal age at birth, preterm birth, APGAR score and head circumference at birth), early developmental difficulties (including speech and motor delays), infant temperament dimensions of activity and distractibility, regulatory problems (early sleeping, crying, and feeding problems), and ADHD PRS.

Sample
cohort of a representative sample of parents and children Fraser et al., 2013) Our sample included a total of 9201 children with information on ADHD measures at age 7 years. Where families included multiple births (twins), we included only the oldest sibling. Specific details of the sample are provided in the Supporting Information.

Ratings of ADHD at age 7 years
Strengths and Difficulties Questionnaire (SDQ) The SDQ is a brief, extensively evaluated behavioural screening questionnaire for children aged 3-16 years, with demonstrated reliability and validity (Goodman, 2001) and was used to provide a continuous, symptom score at age 7 years. The five questions in the hyperactivity/inattention subscale rated on a 3-point scale are summed to give a total score. Data were available for 8302 children.

Development and Wellbeing Assessment (DAWBA)
The DAWBA structured assessment was sent as a paper questionnaire to carers and teachers when the children were 7 years old and provided information for ADHD diagnoses. The ADHD section assesses inattention, hyperactivity, impulsivity, age of onset and impairment. In accordance with the DAWBA protocol, initially the DAWBA computer algorithm ascertained the probability of having ADHD for each participant (Goodman et al., 2011;www.dawba.info).
Child psychiatrists then reviewed the data for each participant, to determine a final definitive DSM-IV diagnosis of ADHD.
DAWBA data were available for 8121 children. Good validity of the diagnoses assigned with the DAWBA has been confirmed (Goodman et al., 2000).

Genetic data
We used previously calculated PRSs for ADHD, defined using alleles associated at p < .05 in an independent GWAS (Demontis et al., 2019). Details of how the ADHD PRS scores were derived are provided in the Supporting Information. Scores were standardised using Z-score transformation.

Parental age
These variables were dichotomised to make them clinically relevant.
Younger maternal age was defined as less than 20 years and advanced paternal age as above 45 years (D'Onofrio et al., 2014;Galéra et al., 2011).

Perinatal factors
Intra-uterine Growth Restriction (IUGR), preterm birth and APGAR score at 5 min were obtained from obstetric clinical records. Head circumference (HC) at birth was assessed by trained ALSPAC staff.
Preterm birth was defined as those born at <37 weeks of gestation based on best clinical estimate of expected date of delivery.
APGAR score is a measure of vitality of a new-born infant, routinely calculated by the attending clinician after every birth (Cnattingius et al., 2017), and IUGR is a clinical definition implying a pathological restriction of the growth potential in utero, recorded as present or absent. APGAR score at 5 min after birth was dichotomised to define healthy new-borns (score of 10) and those with some problems after birth (score of 0-9). Individuals with a head circumference ≤2 SD below the mean for their sex were considered to have microcephaly.

Concerns over vision and hearing
At age 15 months, parents were asked as part of a paper questionnaire if their child had been referred to a Hearing Assessment Centre or Eye Specialist. Where the answer was "yes" for either question children were considered to have early concerns with vision or hearing.

Motor development
Fine and gross motor development were assessed at 18 months old using an adapted version of the Denver Developmental Screening Test (DDST; Frankenburg & Dodds, 1967). Parents completed a questionnaire about their child demonstrating 16 fine motor and 12 gross motor skills rated on a three-point scale (0 = not yet started, 1 = once or twice, 2 = yes, can do well). Analysed motor scores were age adjusted z-scores. To identify a clinically relevant category of motor developmental delay, a dichotomous variable was created for each score with motor skills >1 SD z-score below the mean, indicating the presence or absence of motor delays.  Fullard et al., 1984) with questions about frequency of child behaviour at age 2 years rated on a five-point scale (1 = almost never, 2 = rarely, 3 = sometimes, 4 = often, 5 = almost always). 10 Distractibility items (effectiveness of extraneous stimuli in altering the direction of ongoing behaviour) and nine Activity items (level of the motor component in the child's functioning) were summed to give the relevant sub-scores.

Regulatory problems
Parents completed questionnaires about their child's feeding (at age 24 months), sleeping and crying (at 30 months) habits and these questions were utilised to assess regulatory problems, following a previously used scoring system. The presence of seven sleeping problems and six feeding problems were assessed on a 4-point scale used to identify the presence of a problem (1-3 vs. 0) and summed to create total scores. A total crying problem score was derived by summing the answers to five items and binary clinically relevant crying, sleeping and feeding variables were defined as >1 SD above the sample mean, in-line with previous work (Winsper & Wolke, 2014).
Details of how scores were derived prorated measures were used for SDQ Score, Denver motor score, Carey temperament score and regulatory problems score where <50% items were missing.

Covariates
Sex and socioeconomic status (SES) were included as covariates because ADHD is more common in males and in more socially disadvantaged groups (Thapar & Cooper, 2016). SES was based on mother's occupation during pregnancy coded using the Standard Occupational Classification SOC (OPCS, 1991)

Statistical analyses
Analyses were undertaken using SPSS version 21. Linear and logistic regression analyses were conducted to investigate associations be- were subsequently included in a fully adjusted multivariable model.
We used Bonferroni correction (α = 0.05/number of significant tests) in Model 3, prior to conducting multivariate regression analysis, to correct for multiple testing. Sensitivity analyses, were conducted separately for females and males to evaluate if predictors varied by sex (Tables S2 and S3).

RESULTS
We analysed data from 9201 children (51.3% males). Descriptive statistics for all study variables can be found in Table 1.
There was no evidence of multicollinearity between early markers (see Table S1), therefore all were included in regression models (Ernst & Albers, 2017).

Associations between early markers and ADHD symptom scores
Seven early markers showed association with ADHD scores in univariable analysis when controlling for sex and SES ( ADHD PRS, fine motor delays, vocabulary delays, grammar delays, higher temperament activity, higher temperament distractibility and feeding difficulties (Bonferroni correction threshold p < .004). There was also weaker evidence of associations with younger mothers at birth, IUGR and microcephaly at birth.
In multivariable adjusted analysis (Model 4), four markers remained associated with ADHD scores: ADHD PRS, fine motor delay, vocabulary delay and higher temperament activity (see Table 2).
Stratified multivariable analysis showed that associations were generally consistent across males and females, except that associations with vocabulary delays were only seen in males (See Table S2).

Associations between early markers and ADHD diagnosis
When controlling for sex and SES, ADHD diagnosis was associated with five early markers (Table 3, Model 3): ADHD PRS, fine motor delay, high temperament activity, vocabulary delay and grammar delay (Bonferroni correction threshold p < .006). There was also weaker evidence of association with gross motor delay. In the multivariable model (Model 4), ADHD PRS and temperamental activity showed strong evidence of association (see Table 3). Due to the limited number of girls with ADHD diagnoses (N = 27), stratified analysis was carried out only for males which were generally consistent with those in the whole sample (Table S3).

DISCUSSION
In this prospective cohort, we observed that ADHD PRS, fine motor delays at 18 months, speech delays and higher temperament activity at 24 months showed independent association with ADHD traits at age seven. ADHD PRS and temperament activity were also associated with subsequent ADHD diagnosis.   of age associated with ADHD were the child not being able to fetch things on request, having speech delays or being significantly more active than average.
The results of our study also suggest that genetic liability indexed as ADHD PRS were associated with subsequent ADHD diagnosis. Many studies have shown that participants with higher ADHD PRS have an increased chance of meeting ADHD diagnostic criteria, higher ADHD symptom levels and ADHD persistence (Riglin et al., 2016;Thapar, 2020). Although PRS are not strong predictors, when combined with clinical predictors they may have clinical utility in risk assessments both in Psychiatry and Child Health as well as for physical health conditions such as cardiovascular disease or cancer (Lewis & Vassos, 2020;Murray et al., 2021).
This study is the first to our knowledge to suggest that ADHD PRS, when combined with clinical and developmental variables, is associated with later ADHD symptoms and diagnosis. Our results may therefore have important practical implications because identifying precursors and antecedents of ADHD, even when not causal, could be important in risk assessments of ADHD, for example, determining follow-up options for individuals who are at clinical risk of future illness. However, the clinical utility of these results needs evaluation especially as the variance explained by this group of early markers was limited.
As we learn more about the early manifestations of clinical psychopathology, we will also be able to suggest and evaluate preventive intervention approaches targeting some of the early predictors. Moreover, this area of research is particularly relevant for ADHD because, in contrast to Autism spectrum disorder, clinicians lack effective screening tools that help them to distinguish ADHD precursors (Carter et al., 2004).

Methodological strengths and limitations
This study has several strengths, such as a large representative sample and prospectively collected information using well-validated instruments. Although some variables are based on objective measures or obstetric data, most of the variables were collected through parental report, which may have introduced a measurement error.
However, our variables represent data available in the real world and in general, evidence suggests that parent report and directly assessed measures of language and motor skills are significantly correlated (Miller et al., 2017;Torrens & Ruiz, 2021 version of the Denver developmental scale was used, necessitating dichotomisation of the sample using sample-specific rather than standardised scores. The study also has some limitations. The ALSPAC cohort shows attrition (Taylor et al., 2018) necessitating replication.
However, we found no differences in rates of ADHD symptoms and diagnosis in those with complete and some missing data (see Table S4 and Figure S1) and although selective attrition can lead to a loss of power and underestimation of the prevalence of psychiatric disorders, it has been shown in the ALSPAC sample that this is less likely to affect the patterns of associations between disorders and risk factors (Wolke et al., 2009

CONCLUSION
The results of this study suggest that genetic liability indexed by ADHD PRS, aspects of temperament, and early developmental delays act as antecedents and represent early precursors of later ADHD. using support from 23 and Me.

CONFLICT OF INTEREST
The authors declares that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analysed in this study.