Mental health and school absenteeism in children with long-term physical conditions: A secondary analysis of the British Child and Adolescent Mental Health Surveys 2004 and 2007

Background: Children and young people (CYP) with long-term physical conditions (LTCs) are more likely to have poorer mental health and more school absenteeism compared with CYP with no LTCs. However, there is limited longitudinal research, and the extent to which these difficulties persist in CYP with LTCs is unknown. Furthermore, little is known about the relative impact of different types of LTC on mental health and absenteeism. Methods: We investigated cross-sectional and longitudinal associations of different LTCs with mental health and school absenteeism in a large ( N = 7977) nationally representative survey of CYP in Great Britain and its 3-year follow-up. Psychopathology was assessed using the parent-reported Strengths and Difficulties Questionnaire (SDQ), and diagnosis of any psychiatric disorder using the Development and Wellbeing Assessment (DAWBA). Days absent and persistent absence (missing 10% or more of school days) were reported by parents. Results: Compared with those with no LTCs, CYP with any LTC had higher SDQ total difficulties scores at baseline (adjusted mean difference 1.4, 1.1 – 1.6) and follow-up (1.1, 0.8 –


| INTRODUCTION
Children and young people (CYP) with long-term physical conditions (LTCs) are more likely to experience mental health problems compared with their peers (Glazebrook et al., 2003;Hysing et al., 2007;Lum et al., 2019;Pearce et al., 2018;Pinquart & Shen, 2011;Wolock et al., 2020). For example, 5-to 15-year-olds attending paediatric outpatient clinics in the United Kingdom were more than twice as likely to have mental health problems compared with CYP from a general population sample, and for those attending neurological clinics the prevalence was more than five times greater (Glazebrook et al., 2003).
The association between LTCs and poor mental health has been attributed to factors such as stress caused by the illness and its impact on quality of life; social stigma, difficulties maintaining peer relationships and increased susceptibility to bullying; difficulties in the family environment and impact of the LTC on family functioning; and biological pathways, including the physiological impact of LTCs on the brain, particularly with regard to neurological disorders (Davies et al., 2003;Layte & McCrory, 2013;Leeman et al., 2016;Olsson et al., 2003).
LTCs may also negatively impact attendance at school (Gottfried & Gee, 2017;Ingul et al., 2012;Lum et al., 2017Lum et al., , 2019. In a survey of 5-to 19-year-olds in Australia, those with chronic illness were nearly five times as likely to have missed at least 1 day of school in the last 2 weeks for illness-related reasons and more than twice as likely to have missed at least 1 day for non-illness reasons (Lum et al., 2019). Other research has shown that poor physical health is one of the strongest predictors of persistent absenteeism (Gottfried & Gee, 2017). Symptoms of the LTC, treatment side effects and attendance at medical appointments are likely to contribute to absenteeism, as well as difficulties with peer relationships and keeping up with schoolwork (Lum et al., 2017(Lum et al., , 2019. Given that anxiety and depression are predictors of absenteeism (Finning et al., 2020), poor mental health may be an additional pathway through which LTCs impact attendance.
School provides children with important opportunities for academic, social and emotional learning, and chronic absenteeism is associated with poor academic attainment, school dropout, poor peer relationships and adult unemployment (Attwood & Croll, 2014;Hancock et al., 2013;Heyne et al., 2019;Malcolm et al., 2003). For CYP with LTCs regular school attendance may play an important role in maintaining a sense of normality and in supporting mental wellbeing (Boonen & Petry, 2012). This impact on mental health and school attendance may contribute significantly to the burden of LTCs and interventions targeting mental health, in addition to the provision of educational support, may be important considerations for CYP with LTCs (Boonen & Petry, 2012;Lum et al., 2019;Shaw et al., 2019).
Although associations between LTCs, mental health and absenteeism are well established, most research is cross sectional and the extent to which these difficulties persist is unknown. Furthermore, less is known about the relative impact of different types of LTC (Allison & Attisha, 2019;Gottfried & Gee, 2017), and the mental health and educational needs of CYP with different conditions may not be equal.  (Green et al., 2005). Approval for this secondary analysis was granted by the University of Exeter College of Medicine and Health ethics committee (May20/D/247).

| Sample
The 2004 BCAMHS involved a representative sample of CYP aged 5-16 years living in Great Britain, sampled via the Child Benefits Register (Green et al., 2005). The Office for National Statistics

Key messages
• Children and young people (CYP) with long-term physical conditions (LTCs) had poorer mental health than those with no LTCs, both in the short term and at 3-year follow-up.
• CYP with LTCs had greater school absenteeism than those with no LTCs in the short term and, to a lesser degree, at 3-year follow-up.
• Neurodevelopmental conditions, migraines/severe headaches and atopic conditions were the conditions most strongly and consistently associated with mental health and school absenteeism.  Green et al., 2005). A diagram showing flow of participants is provided in Figure S1.

| LTC (predictor)
We defined an LTC as a physical condition that is predicted to persist for at least 3 months, interferes with everyday activities, is likely to require medical care, and for which a cure is unlikely van der Lee et al., 2007). At baseline, parents reported whether their child had any condition from a list of 29 (Table S1), which we restricted to the 17 that met the criteria above. These were asthma, eczema, epilepsy, cerebral palsy, muscle disease(s), coordination problems, heart problems, kidney/urinary tract problems, migraines/severe headaches, deformities, spina bifida, chronic fatigue syndrome, cystic fibrosis, blood disorders, missing limb(s), diabetes, and cancer. We examined associations between mental health/ absenteeism and having any of the 17 conditions listed above ('Any LTC'), as well as individual types of condition where sample sizes were sufficient. These were atopic conditions (asthma, eczema), migraine/severe headaches, neurodevelopmental conditions (epilepsy, cerebral palsy, muscle disease(s), coordination problems), deformities, kidney/urinary tract problems, and heart problems.

| Mental health (outcome)
Psychopathology Both surveys used the parent-reported Strengths and Difficulties Questionnaire (SDQ), which is a validated questionnaire that screens for childhood psychopathology (Goodman, 2001). The SDQ comprises five subscales: emotional, conduct, hyperactivity-inattention, and peer problems; and prosocial behaviour. The first four are combined to produce a total difficulties score, which ranges from 0 to 40, with higher scores indicating greater difficulties. The SDQ has satisfactory testretest reliability (teacher Pearson correlation coefficient [r] = 0.84, parent r = 0.76) and is able to discriminate between clinical and community samples (Stone et al., 2010). In the current sample the SDQ had good internal consistency (Cronbach's alpha [α] = 0.80).

Psychiatric disorder
The Development and Wellbeing Assessment (DAWBA) was used to assess psychiatric disorders according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) (Goodman et al., 2000). The DAWBA is a diagnostic interview that combines structured and open-ended questions and incorporates information from parents, children and teachers. Computer-generated predictions of psychiatric disorders are produced, which are reviewed by clinical raters. In a validation study the DAWBA demonstrated excellent discrimination between clinical and community samples, and in the clinical sample there were high levels of agreement between DAWBA diagnoses and clinical case notes (van der Lee et al., 2007). We analysed a binary variable indicating diagnosis of any versus no psychiatric disorders.

Days absent
Parents reported the number of days in the previous school term that the child was absent. Seven parents reported the number of days to be well in excess of the total number of days available in a school term. Most schools in England do not exceed 70 days of teaching in a term (Department for Education, 2015), so 70 was set as the maximum and the seven observations greater than this were coded as missing.

Persistent absence
Due to its importance in education policy, we also explored associations with persistent absence based on the 10% threshold used by the Department for Education (2019) (i.e. 7 days absent or more in the previous term).

| Potential confounders
Parents reported the following variables at baseline: child's age, sex and ethnicity; whether the child had any learning difficulties; number of stressful life events experienced by the child (e.g. parental separation, death of a close relative); family structure; housing tenure; and parental academic qualifications. All of these variables were selected for the current analysis based on their theoretical plausibility as confounders and/or previous literature indicating their relevance as confounders.

| Analysis
Analyses were conducted using Stata/SE 16.1. Variables were summarized for CYP with and without LTCs using means and standard deviations for continuous and count variables and numbers and percentages for categorical variables. We examined associations for 'Any LTC', followed by specific types of LTC. For the latter, the control group (i.e. no LTC) involved CYP with none of the 17 LTCs, and the LTC group included those with only that condition. Multiple imputation was used to adjust for the bias and loss of statistical power associated with missing values, on the assumption that data were missing at random (MAR) (Rubin, 1976;Sterne et al., 2009). Fifty imputed datasets were generated using the chained equations approach (Lee & Carlin, 2010). The imputation model included LTCs, outcome variables, and background variables.
Regression models examined the impact of having an LTC at baseline on all outcomes at baseline and 3-year follow-up using linear, logistic and negative binomial regression for continuous, binary and count variables, respectively. We excluded individuals who were missing data on absence (n = 387) at baseline because otherwise sensitivity analyses would have involved different individuals across the imputed datasets. The final analytic sample, therefore, included 7590 CYP. Unadjusted models were followed by models adjusted for the background variables that were associated with at least one outcome at the 5% level. These were age, sex, ethnicity, learning difficulties, number of stressful life events, parental qualifications, housing tenure and family structure. All regression analyses were weighted to correct for unequal sampling probabilities for the children and variation across regions in the response rate and to make the sample representative of the age-sex-region structure of the population aged 5-16 years in Great Britain (Green et al., 2005). Table 1 summarizes characteristics of children with (N = 2341) and without (N = 5249) any LTC at baseline. The proportion of individuals with missing data on key variables ranged from 0% (baseline psychiatric disorder) to 49.8% (follow-up absence) (see Table S2 for full details). Several background variables were associated with missingness on the predictor or outcomes (Table S3); these variables were all included in imputation models.

| LTC and school absenteeism
At baseline children with any LTC had more days absent than those with no LTCs (adjusted incidence rate ratio (aIRR) 1.47 [1.31-1.64], P < 0.001) (

| DISCUSSION
This study investigated associations of LTCs with mental health and school absenteeism in a large representative survey of CYP in the United Kingdom and its 3-year follow-up. In line with previous studies (Hysing et al., 2007;Lum et al., 2019), we found evidence for crosssectional and longitudinal associations between LTCs and poor mental health assessed via both dimensional and diagnostic measures. As also previously reported (Gottfried & Gee, 2017;Ingul et al., 2012;Lum et al., 2017), children with LTCs had more school absences and were more likely to be persistently absent at baseline and follow-up compared with those with no LTCs, although evidence of a relationship was weaker at follow-up.
The conditions most strongly associated with poor mental health were also largely those most strongly associated with absenteeism, specifically neurodevelopmental conditions, migraines/severe headaches, and atopic conditions. Previous research suggests that CYP with neurological disorders are at particularly high risk of mental health problems (Davies et al., 2003;Glazebrook et al., 2003), which has been attributed to 'direct and powerful brain-behaviour links' (Glazebrook et al., 2003, p. 146). Social and psychological pathways may also be important, and neurological disorders are often associated with cognitive difficulties that impede learning, which makes sufficient support at school for these children imperative.
Our findings show that mental health problems and absenteeism are common in CYP with LTCs and may contribute significantly to the burden and impact of these conditions, particularly in the case of neurodevelopmental disorders, migraines and atopic conditions. Clinicians should routinely enquire about mental health in CYP with LTCs and should consider the use of screening tools such as the SDQ to identify those who require further mental health assessment (Hysing et al., 2007), particularly as poor mental health is associated with lower quality of life, poorer physical functioning and more frequent hospital admissions in this population (Ding et al., 2008;Johnson et al., 2004;Zima et al., 2016). Effective mental health interventions for children are available, although access to and implementation of such interventions in CYP with LTCs is poor (Lum et al., 2019;Ott et al., 2003;Welch et al., 2018). Interventions tailored to the needs of those with LTCs may be beneficial, and a recent systematic review concluded that high-quality research to assess the effectiveness of psychological interventions for CYP with LTCs is needed .
Our findings support a recent review that identified school absence as a key marker of functional impairment in childhood chronic illness, suggesting that clinicians should routinely enquire T A B L E 5 Associations between long-term physical conditions and persistent absence from school  (Lum et al., 2017).
It is unknown whether mental health problems and absenteeism are concurrent outcomes of LTCs or whether one precedes or exacerbates the other; future research should investigate this using multiple data points and techniques such as cross-lagged regression. Given the high rates of psychiatric disorder in CYP with LTCs, mental health may be a key predictor of later absence (Finning et al., 2020).
However, absenteeism may also adversely affect CYP's mental health due to missed social, emotional and academic opportunities. Future research should also explore why neurodevelopmental disorders, atopic conditions and migraines are particularly strongly associated with poor mental health and school absenteeism, and may identify potential therapeutic targets.

| Strengths and limitations
The use of a large, nationally representative population survey alongside follow-up data was a strength of this study and supports the generalizability of our findings. The use of validated measures for the assessment of mental health was an additional strength. However, LTCs were reported by parents based on single yes/no questions, were not corroborated with medical reports, and did also not account for severity of the LTC. Parent-reported absences may be affected by reporting and recall biases, and although our threshold of 7 days for persistent absence was based on average school term lengths, future research should aim to make use of administrative educational data from the National Pupil Database.

| Conclusions
In a large, nationally representative survey of 5-to 16-year-olds in the United Kingdom, CYP with LTCs had poorer mental health and more school absence compared to those with no LTCs.
Neurodevelopmental disorders, migraines and atopic conditions were particularly strongly associated with mental health and absenteeism.
Clinicians should routinely enquire about mental health and school attendance in CYP with LTCs and should collaborate with families and schools to ensure these children are provided with sufficient mental health and educational support.