Elsevier

Psychiatry Research

Volume 278, August 2019, Pages 228-234
Psychiatry Research

Comorbidity profile of mental disorders among adolescents: A latent class analysis

https://doi.org/10.1016/j.psychres.2019.06.007Get rights and content

Highlights

  • The three-class solution best fits the data with the classes that are characterised by adolescents with high probability of emotional disorders (Class I), high probability with behavioural disorders (Class II), and with low probability of mental disorders (Class III).

  • Adolescents in the emotional disorders comorbidity class were more likely to be girls, while as adolescents in the behavioural disorders comorbidity class were more likely to be boys.

  • Adolescents in the emotional and behavioural disorders comorbidity classes were more likely to live with their non-biological parents.

Abstract

The aim of this study was to identify the number of comorbidity profiles among adolescents. Sociodemographic factors associated with the comorbidity profiles were also examined. Latent class analysis was conducted using data from the National Comorbidity Survey Adolescent Supplement (NCS-A), a representative sample of adolescents (N = 10,123) in the United States. Latent classes were derived from 26 lifetime mental disorders which were assessed using the World Health Organization Composite International Diagnostic Instrument (CIDI). A three-class solution provided the best fit for the data, with classes labelled as comorbid emotional disorders (Class I), comorbid behavioural disorders (Class II), and normative (Class III). Class I (15.62% of the participants) included adolescents with a high probability of having anxiety, depressive, and intermittent explosive disorder. Class II (6.97%) was characterised by adolescents with a high probability of having substance use, behavioural disorders, and major depression. Class III (77.41%) was characterised by adolescents with a low probability of having any mental disorders. Characterising comorbid profile of mental disorders using person-based approach yields a higher-order classification that could have important clinical implications.

Introduction

The presence of comorbid disorders is the norm among adolescents with mental disorders (Nottelmann and Jensen, 1999). As reported in numerous studies, as high as 60% of adolescents with a mental disorder not only have one, but multiple disorders (Essau et al., 2000, Lewinsohn et al., 1997, Rohde et al., 1991). These findings have been interpreted as providing support for the latent general structure of mental disorders, indicating the presence of general psychopathology factor that underly symptoms of psychopathology (Caspi et al., 2014, Widiger and Clark, 2000). According to this view, the shared variance among all forms of psychopathology may to some extent, share a common etiology, or may represent dynamic processes in which one disorder increases the risk of developing another disorder (Laceulle et al., 2015). However, other research has suggested that comorbidity may be accounted for by an overlap of core symptoms across mental disorders (e.g., Copeland et al., 2013). Until this controversy is resolved, understanding latent organisation of comorbid disorders is important as it could provide more insight into shared aetiology of mental disorders and may explain differences in treatment responses (El-Gahalawy et al., 2013).

Studies on the structure of comorbid disorders of adolescents conducted over the last decade have largely relied on variable-centered approach (e.g., confirmatory factor analysis and latent growth curve) and have identified two factor structures of mental disorders, comprising emotional (internalising) and behavioural (externalising) disorders (Carragher et al., 2015, Cosgrove et al., 2011, Krueger, 1999: Krueger and Markon, 2006). However, variable-centered methodology may be inflexible and insufficient in providing information on that variability and heterogeneity of comorbidity profile among adolescents (Lanza and Cooper, 2016). In other words, variable-centered methods assume that all the individuals within a population show the same pattern of relationships between the variables of interest. This approach may lead to biased conclusions because it ignores the key premise of developmental psychopathology which emphasizes that the processes, functions, and development of behaviours are, in part, specific to individuals (McClelland et al., 2015, Weems, 2008).

Hence, an approach based on identifying common patterns among subgroups of adolescents might be a better alternative to accurately describe the comorbidity profiles. However, the few studies that used a person-centered approach (e.g., latent class analysis) have identified inconsistent comorbidity profiles. For example, in a study among adults, Kessler et al. (2005) examined the structure of 19 disorders using data from the National Comorbidity Survey Replication (NCS-R) and reported 7-class solution that best fit the data. These classes are called unaffected respondents, pure internalising disorders, externalising disorders, comorbid internalising disorders, comorbid internalising-externalising disorders which is dominated by comorbid social phobia and ADHD, highly comorbid major depressive episode, highly comorbid bipolar disorder. Approximately, 7% of the participants were in the classes with high comorbidity, with almost half (43.6%) of serious cases being in these classes. Correlates of pure externalising disorders were being young, male, Hispanic, not low income, and living in a rural area. Correlates of pure internalising disorder include being female, married, high education, and living in suburbs of small metropolitan areas. In the reanalaysis of the NCS-R (Vaidyanathan et al., 2011), a 5-class solution was found to best fits the data, which they labelled as fear class (all phobias and panic disorder), a distress class (depression, generalised anxiety disorder, dysthymia), and externalising class (alcohol and drug dependence and conduct disorder), a multimorbid class (highly elevated rates of all disorders), and a few-disorders class (very low probability of having any disorders). In a study by Olino et al. (2012), four classes of comorbidity profile were found: (a) one class (62.5% of the sample) which included individuals with the lowest rate of mental disorders; (b) an internalising class (16.4%) with elevated rates of internalising disorders; (c) an externalising class (16.9%) which were largely characterised by externalising disorders; and (d) a comorbid internalising and externatising disorders class (4.2%) which was characterised by both disorders. Individuals in the class with externalising disorders were likely to be men, whereas those in the class with internalising disorders were likely to be women. While informative, all these studies were focused on adult samples, and thus the findings might not be generalizable to adolescents.

Only a handful of studies have been conducted in the adolescent population. Based on the Child Behaviour Checklist (CBCL), Bianchi et al. (2016) found four classes of psychopathology symptom profile: (a) the internalising problems which is characterized by a high probability of being in the clinical range for anxious/depressed scale (15.68% of participants), (b) the attention/hyperactivity class which has elevated probability of being in the clinical range for attention problems (10.19%), and (c) low problem class which included participants with a low probability for each CBCL syndrome scale (66.32%) and (d) the severe dysregulated class which included participants with an elevated probability of being in the clinical range for all CBCL scales except for somatic complaints and rule-breaking behaviour (7.82%). A major problem with Bianchi et al.’s study is that the data were based on parent report and may not represent a true report of adolescent's emotional (internalising) and behavioural (externalising) problems. As reported in several studies, there exists a lack of agreement between self- and parent-report (Cantwell et al., 1997, Gould et al., 1993), where self-report ratings of anxiety and depressive synptoms have been found to be higher than parent ratings (Essau and Petermann, 2001).

Van Lang et al. (2006) conducted a symptom-level study (based on Youth Self Report) in order to identify comorbidity patterns between anxiety and depression among early adolescents. Five distinct groups appeared to provide the best fit for the data. However, almost all their participants (99%) had comorbid symptoms, with very small number of them having only anxiety or only depression. A major problem with Van Lang et al.’s study was the narrow inclusion of age groups (10–12 years old) and that only symptoms of anxiety and depression were measured. These limitations are important to address in future studies because as argued by some authors (Lanza and Cooper, 2016), psychopathology is manifested through various developmental pathways in adolescence, leading to the emergence of different disorders at the same time in the same individual.

Studies that use person-centered methods and robust assessment protocols (i.e., diagnostic interview) and cover a wide range of mental disorders are needed to provide a more accurate picture of comorbidity profiles in adolescence. Thus, the main aim of the present study is to identify the number of comorbidity profiles of mental disorders (based on DSM-IV criteria) among adolescents. Another aim is to analyse the relationships between these comorbidity profiles and sociodemographic variables.

Section snippets

Sample

The present study used data from the National Comorbidity Survey Adolescent Supplement (NCS-A), which is a nationally representative survey of 10,123 adolescents between 13 and 18 years of age in the United States (51.07% girls, mean age = 15.18, SD = 1.51); its predecessor, the National Comorbidity Survey Replication (NCS-R), is a nationally representative survey of 9282 English-speaking household residents ages 18 years and over in the United States.

The adolescents were recruited from a

Results

The sociodemographic characteristics of all the participants and the lifetime prevalence of mental disorders are shown in Tables 1 and 2, respectively. The goodness-of-fit indices for the one-to 10-class models are shown in Table 3. The AIC and SABIC decreased for all models up to the third class solution. The 3-class solution was retained due to the low AIC and SABIC and acceptable entropy R2 value, and meaningful proportion of participants (>5%) in all the identified classes.

Table 4 shows the

Discussion

To our knowledge, the present study was amongst the first to have examined the comorbidity profile of mental disorders using as person-centered approach in a nationally representative sample of adolescents. Consistent with previous studies (Carragher et al., 2015, Widiger and Clark, 2000), the comorbidity profiles consisted of an externalising (i.e., related to behavioural problems) and an internalising (i.e., related to emotional problems) profile. However, a close inspection of each class

Funding

The National Comorbidity Survey Replication Adolescent Supplement (NCS-A) was funded by: United States Department of Health and Human Services, National Institutes of Health, National Institute of Mental Health (U01-MH60220); United States Department of Health and Human Services, National Institutes of Health, National Institute of Drug Abuse (R01-DA12058-05); United States Department of Health and Human Services, Substance Abuse and Mental Health Services Administration; Robert Wood Johnson

Human participant protection

The survey was administered by the professional staff of the Institute for Social Research at the University of Michigan. The recruitment and consent procedures were approved by the Human Subjects Committees of Harvard Medical School and the University of Michigan.

Declaration of Competing Interest

The authors declare that they have no conflicts of interest.

Acknowledgements

The authors thank the staff of the Inter-university Consortium for Political and Social Research (ICPSR) of the Institute for Social Research at the University of Michigan for access to the National Comorbidity Survey-Adolescent Supplement (NCS-A) database.

References (48)

  • H.Z. Reinherz et al.

    Prevalence of psychiatric disorders in a community population of older adolescents

    J. Am. Acad. Child Adolesc. Psychiatry

    (1993)
  • L. Su et al.

    Reliability and validity of the screen for child anxiety related emotional disorders (SCARED) in Chinese children

    J. Anxiety Disord.

    (2008)
  • U. Vaidyanathan et al.

    Patterns of comorbidity among mental disorders: a person-centered approach

    Compr. Psychiatry

    (2011)
  • N.D.J. Van Lang et al.

    Latent class analysis of anxiety and depressive symptoms of the Youth Self-Report in a general population sample of young adolescents

    Behav. Res. Ther.

    (2006)
  • C.F. Weems

    Developmental trajectories of childhood anxiety: identifying continuity and change in anxious emotion

    Dev. Rev.

    (2008)
  • Diagnostic and Statistical Manual of Mental Disorders. 4

    (1994)
  • V. Bianchi et al.

    Latent classes of emotional and behavioural problems in epidemiological and referred samples and their relations to DSM-IV diagnoses

    Eur. Child Adolesc. Psychiatry

    (2016)
  • L. Boeschoten et al.

    Estimating Classification Error Under Edit Restrictions in Combined Survey-Register Data

    (2016)
  • N. Carragher et al.

    Disorders without borders: current and future directions in the meta-structure of mental disorders

    Soc. Psychiatry Psychiatr. Epidemiol.

    (2015)
  • A. Caspi et al.

    The p factor: one general psychopathology factor in the structure of psychiatric disorders?

    Clin. Psychol. Sci.

    (2014)
  • W.E. Copeland et al.

    Indirect comorbidity in childhood and adolescence

    Front. Psychiatry

    (2013)
  • V.E. Cosgrove et al.

    . Structure and etiology of co-occurring internalizing and externalizing disorders in adolescents

    J. Abnorm. Child Psychol.

    (2011)
  • D. Eisenhower et al.

    Recall error: sources and bias reduction

  • Cited by (35)

    • Development of a teacher-delivered transdiagnostic universal prevention program for adolescents and examining its usability by teachers

      2022, Mental Health and Prevention
      Citation Excerpt :

      Internalizing problems are defined as having depression, anxiety, and/or withdrawal problems, and externalizing problems are defined as having aggressive behaviors, rule-breaking behaviors, and/or attention problems. Moreover, both the problems and disorders are frequently comorbid in adolescents (Angold, Costello, & Erkanli, 1999; Essau & Torre-Luque, 2019; Kishida, Tsuda, Takahashi, & Ishikawa, 2022). In recent years, children and adolescents worldwide have been experiencing worsening of both internalizing and externalizing problems because of the COVID-19 pandemic (Berger, Jamshidi, Reupert, Jobson, & Miko, 2021; Kishida, Tsuda, Waite, Creswell, & Ishikawa, 2021; Panda et al., 2021; Samji et al., 2022).

    • Anticipation of racially motivated police brutality and youth mental health

      2022, Journal of Criminal Justice
      Citation Excerpt :

      Finally, the mental health impacts of anticipated police brutality examined have largely been limited to depression and anxiety, overlooking additional mental health conditions (e.g., PTSD) and outcomes (e.g., hope for the future) in addition to comorbidities between these conditions. Comorbidity of mental health symptoms is common among adolescents, and is associated with worsened outcomes in adulthood (Essau & de la Torre-Luque, 2019; Teplin, Abram, McClelland, & Dulcan, 2003). Thus, exploration of potential risk factors for comorbidity is key to informing potential policy and prevention programming.

    • Subgrouping school-aged children on the autism spectrum based on co-occurring psychiatric symptoms

      2022, Research in Autism Spectrum Disorders
      Citation Excerpt :

      The Low Psychiatric Symptoms subgroup (41.02%) was the largest subgroup in our sample, suggesting that slightly less than half of school-aged autistic children are within the “normative” range for most, if not all, psychiatric symptoms measured by the CBCL/6–18. This rate is similar to rates of “no psychiatric diagnosis” in other ASD samples (37.2%; Simonoff et al., 2008) and is substantially lower than rates of belonging to the equivalent subgroup in general-population and clinically-referred samples, which are presumably majority nonautistic, of past studies that have derived subgroups based on psychiatric symptoms: 62.5% (Olino et al., 2012), 66.3% (Bianchi et al., 2017), 77.4% (Essau & de la Torre-Luque, 2019), 79.2% (Parviainen et al., 2020). Thus, this finding supports the idea of elevated, clinically-relevant psychiatric symptoms in autistic children.

    View all citing articles on Scopus
    View full text