Mental Health from Childhood to Early Adolescence and its impact on School Attendance Problems: A Latent Transition Analysis

doi:10.21203/rs.3.rs-5255452/v1

Despite concerted efforts among policy makers and practitioners, School Attendance Problems (SAPs) are increasing, with post covid figures indicating higher than ever rates of absenteeism. The aim of this paper was to examine how the developmental trajectory of emotional and behavioural difficulties from childhood to early adolescence might impact the frequency of chronic absenteeism and truancy at 13 years. Using a sample (N = 8570) from the Longitudinal Growing Up In Ireland Study (GUI’98), the research used Latent Class Analysis (LCA) and Latent Transition Analysis (LTA) to examine combinations of mental health symptoms at 9 and 13 years and their developmental impact on SAPs. The Strengths and Difficulty Questionnaire (SDQ) was used to measure a range of emotional and behavioural difficulties at both time points, yielding four mental health classes. Children who remained in High Risk classes, had higher odds of chronic absenteeism and/or truancy. Movement between classes significantly altered the odds of truancy, but not chronic absenteeism, highlighting the importance of differentiating between SAPs and early intervention. A secondary aim was to investigate how family, school and socio-demographic risk factors impacted those trajectories. Family factors were significantly linked to transitions into the co-morbid class, indicating that family risk factors can negatively impact the trajectory of emotional and behavioural difficulties between childhood and adolescence. This paper contributes to current knowledge on the complexities of mental health difficulties in primary school children and their impact on SAPs in early adolescence.

Mental health

School Attendance Problems

Latent Transition Analysis

School Attendance Problems (SAPs) are a complex phenomenon associated with a myriad of risk factors and outcomes [1–3]. Using the typology model, SAPs can be distinguished between school withdrawal, school exclusion, school avoidance (SA) and truancy [1]. Under this approach, school withdrawal refers to parent led absenteeism [4], school exclusion to school led absenteeism, school avoidance (SA) to a child’s difficulty in attending school, despite parents’ efforts and characterised by strong negative emotions, with truancy characterised as absence without parents’ knowledge relating to a negative attitude toward school [2, 5]. This paper focuses on the types of SAPs that relate to SA and truancy.

There has been a substantial increase in SAPs post covid [6]. In the academic year 2021/22, 40.3% of Irish primary and 26.8% of post primary students were chronically absent from school compared to an average of 11% and 15% respectively between 2014 and 2018 [7]. In Ireland, school absences are classified as explained and unexplained in accordance with the Education Welfare Act, 2000 [8]. However, whilst data is recorded, little information is available on the occurrence of SAPs and the factors that influence particular student groups [9, 10]. As SAPs can lead to a range of academic, social and emotional challenges [11–13] that increase as student’s get older [14, 15], insight into the developmental trajectory of school absenteeism is needed.

Research has also highlighted strong associations between mental health symptoms and SAPs [16], with up to 20% of young people worldwide experiencing mental health difficulties [17]. Although mental health symptoms are generally conceptualised into categories of emotional and behavioural difficulties [16, 18, 19], research indicates that there is also a co-occurrence between the two dimensions [11, 20] with several studies evidencing the association between SAPs and these symptoms [21–24]. A systematic review, for example, found a strong association between SA and emotional disorders whilst a latent class analysis exploring truancy identified a number of behavioural difficulties [25, 26]. Other studies have highlighted an association between co-occurring SA and truancy and increased rates of emotional and behavioural disorders [11]. The association is thought to be bidirectional, with mental health difficulties leading to SAPs and SAPs to mental health difficulties [27].

In recent years, a number of cross sectional studies have explored associations between mental health and SAPs. (24, 28, 29). Sharpe et al., (2023), for example, using a sample of 9 year olds assessed how rates of student absenteeism varied across classes of mental health and found that primary school children at risk of mental health difficulties had significantly higher odds of absenteeism compared to those at low-risk. They also observed significantly greater odds of family, school and demographic risk factors for the high risk symptomology classes [24]. To our knowledge, however, few longitudinal studies have examined mental health and its impact on school absenteeism. To address this gap, the authors, using the same sample as Sharpe et al., (2023), applied LTA [30] to examine the extent to which the 9 year olds remained or transitioned between classes and how the outcomes chronic absenteeism and have truanted at 13 years could be predicted by those transitions.

To this end the paper asked: i) Using the GUI (’98 cohort) datasets, can mental health classes identified at 9 years, be identified again at age 13 years ii) What is the probability of remaining or transitioning from one class to another across time (9–13 years)? iii) How are transitions affected by school, family and demographic risk factors? iv) How does remaining or transitioning between classes impact parents report of chronic absenteeism (> 20 days) and child’s report of have truanted at 13 years.

Participants

Data was collected from the first two waves of the GUI child cohort ’98, a longitudinal study examining developmental factors for children growing up in Ireland. The analyses was completed using data from GUI Researcher Microdata Files. In the first wave (2007-2008), 9-year-olds were first selected through the random sampling of 910 Irish national schools from which children were randomly selected [31]. The respondents (N=8568) were evenly distributed between females (49.8 %) and males (50.2 %). The second wave was completed in 2011/2012 age 13 (n = 7,423) [39]. Informed consent was obtained from all participants with ethical approval received from the Health Research Board’s Research Ethics Committee in Ireland [39].

Measures

Mental Health Indicators for latent classes were taken from the parents report of the Strength and Difficulty Questionnaire (SDQ) [32], a measure used to identify moderate and high-risk symptoms in children and adolescence. The current study used the twenty items from the four subscales representing adjustment difficulties: Emotional Problems, Conduct Problems, Hyperactivity–Inattention and Peer Problems.

Primary carers were asked to rate whether the statements pertaining to their child were ‘not true’, ‘certainly true’, ‘somewhat true’. Responses were dichotomized into 0 = ‘not true’ and 1 = ‘somewhat and certainly true’. Individual responses were used to create classes with combinational differences of individual SDQ items.

Outcomes

Chronic absenteeism, based on parental reports during the last 12 months was dichotomised into <20 days and >20 days, with >20 days representing chronic absenteeism as per the Education Welfare Act, 2000 [8]. Have Truanted was collected using the child’s report of whether they ever skipped classes or mitched in the last 12 months, combining responses ‘now and again’, ‘quite often’ and ‘all the time’ to equal Have Truanted.

Risk Factors

Based on previous research [26], the following risk factors were selected. Data was collected from the children, parents and teachers as part of the GUI data collection.

Socio-demographics

Gender, 0 = male, 1 = female.

Socio-economic, parents report of ‘making ends meet’. ‘Not difficult’ = 0, ‘Great Difficulty’ and ‘Difficulty’ = 1.

School

Bullying, child’s report of being bullied in the last year. No = 0, Yes = 1.

Relationship with teacher, child’s report of liking/disliking their teacher. ‘Sometimes ‘and ‘Always’ = 0, Never = 1 (Dichotomised to eliminate ambiguity on the scale response of ‘sometimes and always’).

Attitude to school, child’s report of liking school, with ‘Sometimes’ and Always’ = 0 and Never = 1. (Dichotomised to eliminate ambiguity on the scale response of ‘sometimes and always’.)

Family

Parent conflict, primary carer report of conflict between parents. No = 0 and Yes = 1.

Parent Depression, Centre for Epidemiological Studies in Depression 8-item scale (CES-D). Cronbach’s alpha was .87 [33]. Scores of seven or above indicate presence of depressive symptoms, scores below seven indicate ‘not depressed’.

Problems at home affecting school participation based on teacher response to – ‘Do any of the following limit the kind or amount of activity the study child can do at school?’ – home environment/problems at home. No= 0, Yes = 1.

Statistical Analysis

LCA and LTA with binary indicators were computed in Mplus 8.7. LCA was applied to categorise individuals with similar patterns of behaviour into latent classes, LTA was used to explore changes in those classes over time. This technique allowed for the combining of cross-sectional measurements of the categorical latent variables and longitudinal descriptions of change in the classes between time points.

Analyses were conducted using maximum likelihood estimation and MLR algorithms [34]. The number of missing cases with unknown or missing values for the LCA’s at wave 1 and 2 was 13 and 1. Missing cases with unknown or missing values for the LTA was 13, missing on one or more covariate was 1453. The sample from wave 1 reduced from 8,568 participants to 7,423. To account for missing data the maximum-likelihood based estimation default in Mplus was applied. Model data was used in the estimation of the parameters under the assumption that missing data was at random (MAR). Further, since the data was obtained at the school level, this source of non-independence was taken into account through the use of the Huber-White ‘sandwich’ estimator [35, 36].

The following approach was applied (i) measurement models explored at both time points, (ii) measurement invariance examined, (iii) structural and configural LTA model specifications observed with the final LTA model specified.

Creating Latent Classes

Mental health symptoms were examined by dichotomizing items of the SDQ with 0 = not true and 1 = somewhat and certainly true, with 5 items reversed. Dichotomisation was necessary to reduce model complexity, an approach used in a number of LCA studies using SDQ scores [24, 37, 38].

To determine the number of classes, the best model fit was selected using statistical and theoretical criteria. The estimation process started with a 2 class model increasing to 5 with the lowest indicator of Bayesian Information Criteria (BIC) and the Akaike Information Criteria (AIC) observed. Both the entropy and the Lo-Mendell Rubin likelihood ratio test (LMR-LRT) were used to examine class separation and significance. The entropy 0.714 of a four class solution at wave 1 (age 9) suggested acceptable accuracy of assignment into the correct class [39]. Given the level of accuracy within a number of the classes significant (p < 0.05) values on the LMR-LRT also indicated that the 4 class model was the significantly best fit, . Finally the theoretical interpretability of classes was considered, providing both statistical and theoretical validity. The same procedure was completed for wave 2 (Table 1), revealing a class with primarily emotional symptoms, a class with behavioural symptoms, a class with mixed emotional and behavioural symptoms and a low risk class at 9 and 13 years.

Latent Transitional Analysis

Configural and structural similarities were examined to explore longitudinal invariance between classes over time [40]. The longitudinal weight assigned for wave I was used in the model. LTA was conducted to identify changes in latent statuses over time based on responses to indicators at T2 (wave 2), conditional on membership to a latent status at T1 (wave 1) [30]. The LTA was constrained to be equal across both time points. The four class solution indicated that measurement invariances hold. Risk factors (covariates) from Time 1 were included to determine if they were predictors of the latent transition probabilities. After their inclusion the effect of latent transition patterns from T1 on outcomes at T2 were examined using Mplus 8.7 (fig.1).

Identification of Latent classes

The accelerated flattening of the Log Likelihood and BIC values in the four class model, the decrease in entropy the non-significant CI for the LMT-IRT test for class 5 and the conceptual composition of the four class model suggested an appropriate solution to identifying meaningful classes of mental health symptoms, resulting in a four class model that had configural and structural similarity for both waves (Table 1).

Table 1 Model Fit

Wave 1

C	LL	Replicated	Parameters	AIC	BIC	SSABIC	LMR-LRT (p)	Entropy
2	-90182.722	YES	41	180447.445	180736.665	180606.374	(p <0.0000)	.745
3	-88503.928	YES	62	177131.857	177569.214	177372.190	(p<0.0000)	.736
4	-87566.067	YES	83	175298.633	175883.628	175619.869	(p<0.0000)	.717
5	-87250.496	YES	104	174708.993	175442.625	175113.132	(p<0.0827)	.692

Wave 2

C	LL	Replicated	Parameters	AIC	BIC	SSABIC	LMR-LRT (p)	Entropy
2	-74265.966	YES	41	148613.931	148897.886	148767.596	(p <0.0000)	.767
3	-72680.471	YES	62	145484.942	145914.336	145717.313	(p<0.0000)	.763
4	-71780.622	YES	83	143727.244	144302.079	144038.323	(p<0.0000)	.748
5	-71527.533	YES	104	143263.066	143988.341	143652.851	(p<0.6654)	.757

C = classes; LL = log-likelihood; AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; LMR-LRT = Lo– Mendell–Rubin Likelihood Ratio Test; p = p value.

Classes: Class 1 - High Risk of Emotional and Behavioural Difficulties (HEBD), had moderate to high probabilities of behavioural difficulties, namely hyperactivity, inattentiveness, disobedience, tantrums, lying and emotional difficulties, somatic illnesses, worry, nervousness, fearfulness, unhappiness, solitary playing, bullying and better adult than peer relationships.

Class 2- High Risk of Emotional Difficulties (HED), had moderate to high probabilities of somatic illnesses, worry, fear, nervousness, solitary play and moderate probabilities of tantrums.

Class 3- High Risk of Behavioural Difficulties (HBD), had moderate to high probabilities of hyperactivity, inattentiveness, disobedience and tantrums with low probabilities of emotional difficulties.

Class 4 - Low Risk of Emotional and Behavioural Difficulties (LEBD), presented with low probabilities of emotional and behavioural difficulties. (Fig.2). See Table S1a/b/c (supplementary materials) for item responses.

Latent Transition Analysis

LTA was conducted next to examine rates of transitions across classes from age 9 to 13 years (Fig.3). The LEBD class had the highest stability rate with a small proportion transitioning to the high risk classes. The HEBD class had the lowest stability rate, with transitions mostly to the HED and HBD class with a small proportion transitioning to the LEBD Class. The HED class and HBD class had moderate to high rates of stability with transitions occurring mostly to the LEBD class. (Table 2).

Table 2 Probability and Transition Matrix without covariates

Latent Classes	Wave 1 (C1)	HEBD (13yrs)	HED (13yrs)	HBD (13yrs)	LEBD (13yrs)	Wave 2 (C2)
HEBD (9yrs)	.173	0.557	0.140	0.283	0.018	.132
HED (9yrs)	.233	0.046	0.595	0.060	0.298	.187
HBD (9yrs)	.313	0.072	0.039	0.678	0.211	.293
LEBD (9yrs)	.281	0.011	0.044	0.060	0.886	.388
Bold values indicate probability of remaining in the same class, wave 1 (9 years), wave 2 (age 13 years).

Probability and Transition Matrix with covariates

Latent Classes	Time 1 (C1)	HEBD (13yrs)	HED (13yrs)	HBD (13yrs)	LEBD (13yrs)	Time 2 (C2)
EBD (9yrs)	.173	0.517	0.165	0.293	0.025	.130
ED (9yrs)	.246	0.044	0.569	0.081	0.305	.194
BD (9yrs)	.297	0.068	0.044	0.669	0.211	.293
LEBD (9yrs)	.283	0.010	0.046	0.066	0.879	.388
Bold values indicate probability of remaining in the same class, wave 1 (9 years), wave 2 (age 13 years).

Risk Factors on Transition Probabilities

Bullying, dislike of teacher, dislike of school, parent conflict and socio-economic status did not affect movement between classes whilst parental depression and problems at home that affect school participation at 9 years significantly affected transitions. Children in the HED and HBD classes with reports of parental depression had higher odds of transitioning to the HEBD class at 13 years compared to those who remained, OR = 2.568 (1.224, 5.390), OR =1.835 (1.089, 5.094),

Children who transitioned from the HEBD class to the HED, HBD or LEBD classes at 13 years had significantly lower odds of having problems at home affecting school participation compared to those who remained, OR = 0.155 (0.031, 0.775), OR = 0.525 (0.291, 0.946), OR = 0.281 (0.094, 0.842) respectively. Conversely, children who transitioned from the HED and LEBD classes to the HEBD class had significantly higher odds of having problems at home affecting school participation OR= 6.435 [1.290, 32.101], OR = 3.560 [1.188, 10.670]. These results indicate that problems at home that affect school participation and parental depression can negatively impact developmental trajectories of emotional and behavioural difficulties between 9 years and 13 years.

Females had significantly higher odds of transitioning from the HEBD and HBD class to the HED class at 13 years OR = 2.004 (1.295, 3.101), OR = 2.045 (1.402, 2.983), indicating that females are more likely than males to transition away from behavioural difficulties. There was also higher odds of females transitioning from the HED class to the LEBD class OR = 1.533 [1.142, 2.058]. (See supplementary Table S2 for transition probabilities).

Stability on Outcomes

Children who remained in The HEBD and HED classes compared to those who remained in the LEBD class were three times more likely to be chronically absent at 13 years, OR = 2.670 (1.231, 5.788), OR = 2.844 (1.400, 5.775). Children who remained in the HEBD and BD classes were 5 times more likely to report have truanted at 13 years compared to those remaining in the LEBD class OR = 5.598 (3.121, 10.043), OR = 4.600 (2.565, 8.24) respectively. Children remaining in the HED class did not have significant odds of having truanted and children remaining in the HBD Class did not have greater odds of chronic absenteeism These results indicate greater odds for chronic absenteeism for children experiencing stability of emotional difficulties and greater odds of truancy for children experiencing stability of behavioural difficulties. Children experiencing stability of co-morbid symptoms had greater odds of chronic absenteeism and truancy (Table 5).

Table 3 OR for Chronic Absenteeism and Have Truanted for stability

Stability

OR Chronic Absenteeism

95% CI

Est. S.E Lower Upper

OR Have Truanted

95% CI

Est. S.E Lower Upper

Stability in HEBD compared to

stability in HED class

stability in HBD class

stability in LEBD class

0.939(.36) (0.441, 1.998)

1.580(.67) (0.684, 3.653)

2.670(1.05) (1.231, 5.788)

4.223(1.8) (1.778, 10.028)

1.217(.35) (0.685, 2.163)

5.598(1.6) (3.121, 10.043)

Stability in HED compared to

Stability in HBD class

Stability in LEBD class

1.683(0.5) (0.845, 3.354)

2.844(1.02) (1.400, 5.775)

0.288(0.124) (0.109, 0.816)

1.326 (0.5) (0.547, 3.214)

Stability in HBD compared to

Stability in LEBD class

1.682(0.6) (0.836, 3.412)

4.600 (0.05) (2.565, 8.24)

Transitions on Outcomes

Odds Ratio was also calculated to compare the odds of chronic absenteeism and/or have truanted for children who leave one group compared to those who remain. No significant OR for chronic absenteeism was found for any group leaving one class compared to those who remained suggesting that mental health at 9 years had a greater effect on chronic absenteeism at 13 years than the class status transitioned to (Table 6). Transitions did affect the outcome Have Truanted for children moving between groups. Children who transitioned from the HBD to the HED class had significantly lower odds of reporting have truanted, OR = 0.233 (0.082, 0.662). Children who remained in the HEBD group were 6 times more like to report having truanted compared to those who transitioned to the HED and 10 times more likely than those who moved to the LEBD class OR = 6.893 (2.620, 18.136), OR = 10.177 (4.639, 22,327). Children who remained in the HED class were significantly less likely to truant compared to those who moved to the HBD class, OR = 0.233 (0.082,0.662).

The results indicate that children with emotional difficulties at 9 years were more likely to be absent with their parents knowledge at 13 years whilst children with behavioural difficulties at 9 years or who later develop behavioural difficulties were at risk of truancy. The children in the HEBD had significantly high odds of chronic absenteeism and have truanted indicating that children with co-morbid symptoms have a greater risk of SAPs.

Table 4 OR for Chronic Absenteeism and Have Truanted for Class Transitions.

Transitions

OR Chronic Absenteeism

95% CI

Est. S.E Lower Upper

OR Have Truanted

95% CI

Est. S.E Lower Upper

Remaining in the HEBD class compared to

transitioning from HEBD to ED class

Remaining in the HEBD class compared to

transitioning from HEBD to BD class

Remaining in the HEBD class compared to

transitioning from HEBD to LEBD class

Transitioning from the HED to HEBD class

compared to remaining in the HED class

Remaining in the HED class compared to

transitioning from the HED to HBD class

Remaining in the HED class compared to

transitioning from the HED to LEBD class

Transitioning from the HBD to HEBD class

compared to remaining in the HBD to HBD

Transitioning from the HBD to HED class

compared to remaining in the HBD class

Remaining in the BD class compared to

transitioning from the BD to LEBD

Transitioning from the LEBD to HED class

compared to remaining in the LEBD class

Transitioning from the LEBD to HBD class

compared to remaining in the LEBD

0.901 (0.77) (0.166, 4.886)

0.825(0.49) (0.256, 2.660)

1.857(1.41) (0.418, 8.256)

0.901(0.767) (0166, 4,866)

0.915(0.515) (0.304, 2.755)

2.061(0.88) (0.892, 4.762)

0.825(0.49) (0.256, 2.660)

0.915(0.51) (0.304, 2,755)

2.522(1.01) (0.935, 5.424)

2.016(0.8) (0.892, 4.762)

2.522(1.01) (0.935, 5.424)

6.893 (3.4) (2.620, 18.136)

1.603(0.47) (0.897, 2.865)

10.177(4.80) (4.639, 22,327)

0.613 (0.24) (0,283, 1.326)

0.233 (0.12) (0.082,0.662)

1.476 (0.82) (0.495,4,407)

0.759(0.25) (0.392, 1.472)

0.233(0.12) (0.082,0.662)

6.349(2.3) (3.081,13.085)

1.476(.82) (0.495, 4.407)

6.349(2.3) (3.081,13.085)

Bold values indicate significant OR’s for Chronic Absenteeism /Truancy

The aim of the paper was to examine how the developmental trajectory of patterns of mental health symptoms from childhood to adolescence impact the frequency of chronic absenteeism and have truanted at 13 years. A secondary aim was to investigate how family, school and socio-demographic risk factors impacted those trajectories. To achieve this LCA and LTA were computed to identify classes of mental health symptoms. Analysis yielded four classes; a class with mixed emotional and behavioural symptoms, a class with patterns of emotional symptoms, a class with patterns of behavioural symptoms and a low risk class.

Mental Health Transitions

The researchers examined the extent to which 9 year olds transitioned or remained in one mental health class before establishing if those remaining or transitioning were at greater risk of chronic absenteeism and/or having truanted at 13 years. Following the inclusion of risk factors, 27% of the sample transitioned with 73% remaining. Consistent with previous research the most prevalent and stable group was the LEBD class [41, 42], followed by The HBD, HED and HEBD classes. Stability rates for the single symptomology classes were consistent with previous studies that reported some developmental continuity of behavioural and/or emotional symptoms between childhood and adolescence [42, 43, 44]. However, transitions indicated some improvement for both classes with 21% and 30% respectively moving to the LEBD class.

Although the HEBD class had the lowest stability rate, transitions for this class occurred mainly between the high risk classes, with just 2.5% transitioning to the LEBD class. These patterns support previous research, indicating that children with co-morbid symptoms are most likely to remain in one of the symptomatic groups during early adolescence, highlighting a persistence of mental health symptoms for that group [42]. These results highlight the importance of intervening early to disrupt the emergence of emotional and behavioural problems in adolescence.

Findings also indicated that whilst bullying, dislike of teacher, dislike of school, parental conflict and socio-economics did not influence movement between classes, parental depression and problems at home that affect school participation did. Children who moved to the EBD class, had significantly higher odds of having a parent with parental depression and/or problems at home affecting participation at 9 years, supporting the notion that youths with co-morbid symptoms are more likely to experience disrupted family relationships [43].

Females had significantly higher odds of moving away from classes with behavioural symptoms, supporting research that highlights gender differences in the prevalence of internalizing problems during adolescence [20, 44].

Outcomes

The odds of chronic absenteeism and have truanted were first compared between the stable high risk classes and LEBD class. Children continuing to exhibit emotional symptoms had the highest odds of chronic absenteeism but did not have significant odds of reporting having truanted. Conversely children continuing to exhibit behavioural difficulties had significantly higher odds of reporting have truanted but did not have significant odds of chronic absenteeism. As literature has historically associated emotional symptoms with SA [25, 1] and behavioural symptoms with truancy [26], these results support research that highlights the importance of differentiating between types of SAPs, indicating that children with SA are more likely to be absent with their parents knowledge and less likely to truant [1, 26, 46]. These findings also provide insight into how the trajectory of mental health symptoms from childhood to adolescence can affect SA and truancy, highlighting the importance of identifying the developmental patterns associated with the continuous dimensions of SAPs [47]. Children remaining in the HEBD class, had high odds of both chronic absenteeism and having truanted suggesting that youth with co-morbid symptoms are at risk of both types of SAPs [11]. Consistent with previous research children with good mental health had lower odds of absenteeism.

The researchers next compared the odds of chronic absenteeism and truancy for children who left a class compared to those who remained. Surprisingly, children who moved from the high risk classes to the LEBD class did not have significantly reduced odds of chronic absenteeism compared to those who remained. This is a key finding, suggesting that mental health symptoms experienced at 9 years have a greater influence on chronic absenteeism at 13 years than the mental health profile transitioned in to. These results indicate that mental health symptoms experienced in childhood rather than the stability of those symptoms alone may impact a young person’s ability to attend school at 13 years. As previous literature has highlighted that absenteeism and emotional difficulties in early education can lead to poor academic outcomes, social difficulties, behavioural challenges and parental adaptive responses, it might be possible that these factors develop into entrenched behaviour patterns that influence chronic absenteeism at 13 years [45]. As early childhood is a crucial time for the development of emotional symptoms [20, 25] early school based interventions are advised to reduce the risk of SA.

Having truanted, on the other hand, was impacted by transitions, highlighting the importance of identifying different mental health patterns associated with different categories of SAPs. Movement between classes indicated that developing behavioural symptoms increased the odds of truancy in adolescence. Equally children with behavioural difficulties at 9 years who moved to the HED or LEBD class were less likely to truant than those children who continued to have behavioural difficulties.

These findings highlight that childhood symptoms of emotional difficulties can have long term effects on SA with behavioural difficulties increasing the risk of truancy. Children presenting with co-morbid symptoms are at increased risk of SA and truancy requiring intensive interventions during the transition to early adolescence.

Both parental depression and problems at home that affect participation at school influenced movement into the HEBD class. These findings expand on previous research highlighting the adverse effects of family risk factors on mental health in primary school students and absenteeism in adolescence. Based on these findings family based supports and policy efforts designed to prevent SAPs emerging in primary school children could reduce the number of children experiencing SAPs at post primary [24, 10]. As stability in high risk groups was also predictive of chronic absenteeism and/or truancy at 13 years, noting persistent mental health symptoms could be a useful early indicator tool to reduce the number of adolescence experiencing SAPs.

There are several limitations acknowledged. First, dichotomizing the SDQ variables may have resulted in loss of information. Although the prevalence of emotional and behavioural symptoms based on the SDQ may be reflected differently in studies using cut offs, the size of the high risk classes were consistent with LTA studies that examined emotional and behavioural profiles in children [37, 38].

The authors dichotomised the outcome have truanted based on the child’s report of whether they had skipped classes or mitched in the last 12 months. As combining the responses likely reduced the severity of the outcome, it was expected that probabilities and odds would be greater for truancy then chronic absenteeism.

Although a large sample size was utilised, the sample size for > 20 days absent was small, representing 2% of the sample at both waves. At the time of data collection Irish data recorded an average of 11% absences for > 20 days in primary schools and 16% for post primary [58], suggesting that the data for chronic absenteeism in the sample is underestimated. Although previous research highlights that parents under report their child’s absence [59] parents reports of days absent was used due to the large amount of missing data for teachers reports.

Although risk factor problems at home affecting school participation was a teachers' subjective response, previous research using the same sample found this risk factor was significantly associated with HEBD at 9 years. As a result the risk factor was included in the analysis.

The confidence intervals were high for the some of the OR’s measuring the effect of transitions on outcomes. Following inspection of the probabilities and estimates for the effect of transitions on the outcomes (supplementary files Tables 3 and 4) the authors were satisfied with the reliability of the data and included the OR’s with wide confidence intervals.

Findings indicated that the mental health status of 9 year olds significantly impacts chronic absenteeism at age 13. The study highlights the need for continued efforts at developing and implementing early intervention in early primary school children.

The complex nature of SAPs requires an interdisciplinary response that supports students showing early signs of SA and/or truancy. The authors recommend the implementation of school based mental health and wellbeing interventions that incorporate a ‘whole child’ and whole school approach to school attendance [6].

Findings also highlighted the importance of family factors, indicating early multi-faceted interventions are needed to support families of children who are absent from school. As schools can play a key role in facilitating positive school and family engagement, efforts to increase school awareness about the benefit of tiered interventions, family support and well-being are advised.

In this paper, it was not possible to determine the underlying mechanisms through which the mental health status of 9 year olds influenced chronic absenteeism at 13 years. Further research could provide additional insight into developed behaviours associated with mental health difficulties at 9 years that might later influence chronic absenteeism at 13 years.

Author Contribution

All authors contributed to the study conception and design. Data analysis was conducted by Jane Sharpe. Brendan Bunting reviewed outputs released from the Central Statistics Office (CSO). The first draft of the manuscript was written by Jane Sharpe and all authors commented on versions of the manuscript. All authors read and approved the final manuscript.

Acknowledgement

The data used in this study were obtained from the Growing Up in Ireland (GUI) study. It is managed by DCEDIY in association with the Central Statistics Office (CSO). Results in this report are based on analyses of data from Research Microdata Files provided by (CSO).

Heyne, D., Gren-Landell, M., Melvin, G. A., & Gentle-Genitty, C. (2019). Differentiation Between School Attendance Problems: Why and How? Cognitive and Behavioral Practice, 26(1), 8–34. https://doi.org/10.1016/j.cbpra.2018.03.006
Kearney, C. A., & Graczyk, P. A. (2020). A Multidimensional, Multi-tiered System of Supports Model to Promote School Attendance and Address School Absenteeism. Clinical Child and Family Psychology Review, 23(3), 316–337. https://doi.org/10.1007/s10567-020-00317-1
Melvin, G. A., Heyne, D., Gray, K. M., Hastings, R. P., Totsika, V., Tonge, B. J., & Freeman, M. M. (2019). The Kids and Teens at School (KiTeS) Framework: An Inclusive Bioecological Systems Approach to Understanding School Absenteeism and School Attendance Problems. Frontiers in Education, 4. https://doi.org/10.3389/feduc.2019.00061
Thambirajah, M. S., Grandison, K. J., & De-Hayes, L. (2008). Understanding school refusal: A handbook for professionals in education, health and social care. Jessica Kingsley Publishers
Melvin, G., Freeman, M. M., Ashford, L., Hastings, R. P., Heyne, D., Tonge, B. J., Bailey, T., Totsika, V., & Gray, K. M. (2023). Types and correlates of school absenteeism among students with intellectual disability. JIDR. Journal of Intellectual Disability Research (Print), 67(4), 375–386. https://doi.org/10.1111/jir.13011
Graczyk, P. A., & Kearney, C. A. (2023). Roadmap for implementing a multi-tiered system of supports framework to improve school attendance. Current Psychology. https://doi.org/10.1007/s12144-023-05478-0
Tusla Education Support Service [TESS]. (2023). School Attendance Data Primary and Post-Primary Schools And Student Absence Reports Primary and Post-Primary Schools 2019-2022. In https://www.tusla.ie/uploads/content/AAR_SAR_2019_22.pdf. TUSLA
Government of Ireland. (2000). Education Welfare Act, Dublin: The Stationery Office.
Gurhy, A. M., Perry, G., & Farrell, M. P. (2018). Factors Affecting Non-Attendance in Irish Primary Schools and Reasons for Differences between Urban and Rural levels of Non-attendance. In https://arrow.tudublin.ie/tfschcafrep/13/. Dublin Institute of technology. https://doi.org/10.21427/d75f4r
Thornton, M., Darmody, M., & McCoy, S. (2013). Persistent absenteeism among Irish primary school pupils. Educational Review, 65(4), 488–501. https://doi.org/10.1080/00131911.2013.768599
Egger, H. L., Costello, J., & Angold, A. (2003). School Refusal and Psychiatric Disorders: A Community Study. Journal of the American Academy of Child and Adolescent Psychiatry, 42(7), 797–807. https://doi.org/10.1097/01.chi.0000046865.56865.79
Gottfried, M. A. (2014). Chronic absenteeism and its effects on students’ academic and socioemotional outcomes. Journal of Education for Students Placed at Risk, 19(2), 53–75. https://doi.org/10.1080/10824669.2014.962696
Niemi, S., Lagerström, M., & Alanko, K. (2022). School attendance problems in adolescent with attention deficit hyperactivity disorder. Frontiers in Psychology, 13. https://doi.org/10.3389/fpsyg.2022.1017619
Gottfried, M. A. (2010). Evaluating the Relationship between Student Attendance and Achievement in Urban Elementary and Middle Schools: An Instrumental Variables Approach. Society for Research on Educational Effectiveness. http://files.eric.ed.gov/fulltext/ED525312.pdf
Simon, O., Nylund‐Gibson, K., Gottfried, M. A., & Mireles‐Rios, R. (2020). Elementary absenteeism over time: A latent class growth analysis predicting fifth and eighth grade outcomes. Learning and Individual Differences, 78, 101822. https://doi.org/10.1016/j.lindif.2020.101822
Finning, K., Ford, T., Moore, D. J., & Ukoumunne, O. C. (2020). Emotional disorder and absence from school: findings from the 2004 British Child and Adolescent Mental Health Survey. European Child & Adolescent Psychiatry, 29(2), 187–198. https://doi.org/10.1007/s00787-019-01342-4
Kieling, C., Baker‐Henningham, H., Belfer, M. L., Conti, G., Ertem, İ. Ö., Omigbodun, O., Rohde, L. A., Srinath, S., Ülküer, N., & Rahman, A. (2011). Child and adolescent mental health worldwide: evidence for action. Lancet, 378(9801), 1515–1525. https://doi.org/10.1016/s0140-6736(11)60827-1
Havik, T., Bru, E., & Ertesvåg, S. K. (2015). Assessing Reasons for School Non-attendance. Scandinavian Journal of Educational Research, 59(3), 316–336. https://doi.org/10.1080/00313831.2014.904424
Knollmann, M., Knoll, S., Reissner, V., Metzelaars, J., & Hebebrand, J. (2010). School Avoidance From the Point of View of Child and Adolescent Psychiatry. Deutsches Arzteblatt International. https://doi.org/10.3238/arztebl.2010.0043
Göbel, K., Ortelbach, N., Cohrdes, C., Baumgarten, F., Meyrose, A., Ravens‐Sieberer, U., & Scheithauer, H. (2022). Co-occurrence, stability and manifestation of child and adolescent mental health problems: a latent transition analysis. BMC Psychology, 10(1). https://doi.org/10.1186/s40359-022-00969-4
Ingul, J. M., Klöckner, C. A., Silverman, W. K., & Nordahl, H. M. (2012). Adolescent school absenteeism: modelling social and individual risk factors. Child and Adolescent Mental Health, 17(2), 93–100. https://doi.org/10.1111/j.1475-3588.2011.00615.x
Kearney, C. A., & Albano, A. M. (2004). The functional profiles of school refusal behavior. Behavior Modification, 28(1), 147–161. https://doi.org/10.1177/0145445503259263
McShane, G., Walter, G., & Rey, J. M. (2001). Characteristics of Adolescents with School Refusal. Australian and New Zealand Journal of Psychiatry, 35(6), 822–826. https://doi.org/10.1046/j.1440-1614.2001.00955.x
Sharpe, J., Bunting, B., & Heary, C. (2023). A Latent Class Analysis of Mental Health Symptoms in Primary School Children: Exploring Associations with School Attendance Problems. School Mental Health, 15(4), 1128–1144. https://doi.org/10.1007/s12310-023-09610-0
Finning, K., Ukoumunne, O. C., Ford, T., Danielson‐Waters, E., Shaw, L., Romero De Jager, I., & Moore, D. A. (2019). Review: The association between anxiety and poor attendance at school. Child and adolescent mental health 24(3). doi:10.1111/camh.12322
Maynard, B. R., Salas-Wright, C. P., Vaughn, M. G., & Peters, K. E. (2012). Who Are Truant Youth? Examining Distinctive Profiles of Truant Youth Using Latent Profile Analysis. Journal of Youth and Adolescence, 41(12), 1671–1684. https://doi.org/10.1007/s10964-012-9788-1
Wood, J. J., Lynne-Landsman, S. D., Langer, D. J., Wood, P. A., Clark, S. L., Eddy, J. M., & Ialongo, N. (2012). School Attendance Problems and Youth Psychopathology: Structural Cross-Lagged Regression Models in Three Longitudinal Data Sets. Child Development, 83(1), 351–366. https://doi.org/10.1111/j.1467-8624.2011.01677.x
Gallé-Tessonneau, M., Johnsen, D. B., & Keppens, G. A. (2019). The relationship between mental health and school absenteeism in a community sample of French secondary school students: four profiles derived from cluster analysis. European Journal of Education and Psychology, 12(1), 77.
Gonzálvez, C., Sanmartín, R., Vicent, M., & García-Fernández, J. M. (2021). Exploring different types of school refusers through latent profile analysis and school-related stress associations. Science Progress, 104(3), 003685042110294. https://doi.org/10.1177/00368504211029464
Lanza, S. T., & Collins, L. M. (2008). A new SAS procedure for latent transition analysis: Transitions in dating and sexual risk behavior. Developmental Psychology, 44(2), 446–456. https://doi.org/10.1037/0012-1649.44.2.446
Williams, J., Green, S., Doyle, E., Layte, R., McCoy, S., McCrory, C., Murray, A., Nixon, E., O’Dowd, T., O’Moore, M., Quail, A., Smith, E., Sword, L., & Thornton, M. (2009). Growing up in Ireland: National longitudinal study of children. The Lives of 9 year olds. Child Cohort Research Report No. 1. Government Publications
Goodman, R. (1997). The Strengths and Difficulties Questionnaire: A Research Note. Journal of Child Psychology and Psychiatry, 38(5), 581–586. https://doi.org/10.1111/j.1469-7610.1997.tb01545.x
Nixon, E. (2012). Growing up in Ireland: National longitudinal study of children. How families matter for social and emotional outcomes of 9-year-old children. Government Publications.
Muthén, L. K., & Muthén, B. (1998-2017). Mplus User’s Guide. Los Angeles, CA: Muthén & Muthén.
Huber, P. J. (1967). The behaviour of Maximum Likelihood Estiamtes Under Nonstandard Conditions. In Proceedings of the Fifth Berkeley Symposium on Mathemtical Statisitcs and Probability. Berkeley, CA: University of California Press, vol. 1, 221-233.
White, H. (1980). A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity. Econometrica, 48(4), 817. https://doi.org/10.2307/1912934
Morales, A., Melero, S., Tomczyk, S., Espada, J. P., & Orgilés, M. (2021). Subtyping of Strengths and Difficulties in a Spanish Children Sample: A Latent Class Analysis. Journal of Affective Disorders, 280, 272–278. https://doi.org/10.1016/j.jad.2020.11.047
Petersen, K. S., Humphrey, N., & Qualter, P. (2020). Latent Class Analysis of Mental Health in Middle Childhood: Evidence for the Dual-Factor Model. School Mental Health, 12(4), 786–800. https://doi.org/10.1007/s12310-020-09384-9
Fonseca, J. R. S., & Cardoso, M. G. M. S. (2007). Mixture-model cluster analysis using information theoretical criteria. Intelligent Data Analysis, 11(2), 155–173. https://doi.org/10.3233/ida-2007-11204
Morin, A. J. S., & Litalien, D. (2017). Webnote: Longitudinal Tests of Profile Similarity and Latent Transition Analyses. Montreal, QC: Substantive Methodological Synergy Research Laboratory.
Moore, S., Dowdy, E., Nylund‐Gibson, K., & Furlong, M. J. (2019). A latent transition analysis of the longitudinal stability of dual-factor mental health in adolescence. Journal of School Psychology, 73, 56–73. https://doi.org/10.1016/j.jsp.2019.03.003
Petersen, K. S., Humphrey, N., & Qualter, P. (2021). Dual-Factor Mental Health from Childhood to Early Adolescence and Associated Factors: A Latent Transition Analysis. Journal of Youth and Adolescence, 51(6), 1118–1133. https://doi.org/10.1007/s10964-021-01550-9
McElroy, E., Shevlin, M., & Murphy, J. (2017). Internalizing and externalizing disorders in childhood and adolescence: A latent transition analysis using ALSPAC data. Comprehensive Psychiatry, 75, 75–84. https://doi.org/10.1016/j.comppsych.2017.03.003
Briggs-Gowan, M. J., Carter, A. S., Bosson-Heenan, J., Guyer, A. E., & Horwitz, S. M. (2006). Are Infant-Toddler Social-Emotional and behavioral problems transient? Journal of the American Academy of Child and Adolescent Psychiatry, 45(7), 849–858. https://doi.org/10.1097/01.chi.0000220849.48650.59
Connell, A. M., Bullock, B. M., Dishion, T. J., Shaw, D. S., Wilson, M. N., & Gardner, F. (2008). Family intervention effects on co-occurring early childhood behavioral and emotional problems: A Latent Transition Analysis approach. Journal of Abnormal Child Psychology, 36(8), 1211–1225. https://doi.org/10.1007/s10802-008-9244-6
Berg, I. (1997). School refusal and truancy. Archives of Disease in Childhood, 76(2), 90-91.
Leduc, K., Tougas, A. M., Robert, V., & Boulanger, C. (2022). School Refusal in Youth: A Systematic Review of Ecological Factors. Child Psychiatry Human Development 24, 1–19. https://doi.org/10.1007/s10578-022-01469-7
Millar, D. (2011). Analysis of School Attendance Data in Primary and Post-Primary Schools, 2006/7 and 200.7/8. Report to the National Educational Welfare Board. Dublin: Educational Research Centre
Rogers, T., & Feller, A. (2018). Reducing student absences at scale by targeting parents’ misbeliefs. Nature Human Behaviour, 2(5), 335–342. https://doi.org/10.1038/s41562-018-0328-1

No competing interests reported.

Supplementaryfiles.docx

Mental Health from Childhood to Early Adolescence and its impact on School Attendance Problems: A Latent Transition Analysis

Status:

Version 1

Abstract

Figures

Introduction

Methods

Results

Discussion

Mental Health Transitions

Outcomes

Conclusion

Declarations

Author Contribution

Acknowledgement

References

Additional Declarations

Supplementary Files

Status:

Version 1