Just a phase? Mapping the transition of behavioural problems from childhood to adolescence

Abstract

Purpose: Young people change substantially between childhood and adolescence. Yet, the current description of behavioural problems does not incorporate any reference to the developmental context. In the current analysis, we aimed to identify common transitions of behavioural problems between childhood and adolescence.

Method: We followed 6744 individuals over 6 years as they transitioned from childhood (age 10) into adolescence (age 16). At each stage, we used a data-driven hierarchical clustering method to identify common profiles of behavioural problems, map transitions between profiles and identify factors that predict specific transitions.

Results: Common profiles of behavioural problems matched known comorbidity patterns but crucially showed that the presentation of behavioural problems changes markedly between childhood and adolescence. While problems with hyperactivity/impulsivity, motor control and conduct were prominent in childhood, adolescents showed profiles of problems related to emotional control, anxiety and inattention. Transitions were associated with socio-economic status and cognitive performance in childhood CONCLUSION: We show that understanding behavioural difficulties and mental ill-health must take into account the developmental context in which the problems occur, and we establish key risk factors for specific negative transitions as children become adolescents.

Keywords: Adolescence; Childhood; Development; Education; Mental health; Nosology.

Fig. 1

CONSORT diagram of data exclusion for the childhood and adolescence survey. EFA exploratory factor analysis, HHC hybrid hierarchical clustering

Fig. 2

Items associated with each factor in the childhood data. The colour indicates the strength of the standardised factor loading and the numerical values are shown in the second column (see Supplement for the equivalent loading of the items in the adolescence data)

Fig. 3

Results of the hybrid hierarchical clustering of the childhood data. a Dendrogram of the hierarchical clustering structure. The dashed lines indicate the height cut-off to define 4 and 7 clusters. b Calinski–Harabasz index for different cluster numbers. The red line indicates the percentage change from one number of clusters to the next. c Silhouette coefficient for each group. The dashed line indicates the average silhouette coefficient across all groups. d Profiles of behavioural ratings for each cluster. The asterisks indicate the effect size (Cohen's d) relative to the whole sample: *** > 0.8, ** > 0.5, * > 0.2

Fig. 4

Results of the hybrid hierarchical clustering of the adolescence data. a Dendrogram of the hierarchical clustering structure. The dashed lines indicate the height cut-off for 4 and 6 clusters. b Calinski–Harabasz index for different cluster numbers. The red line indicates the percentage change from one number of clusters to the next. c Silhouette coefficient for each group. The dashed line indicates the average silhouette coefficient across all groups. d Profiles of behavioural ratings for each cluster. The asterisks indicate Cohen’s d relative to mu = 0 and the standard deviation across all groups: *** > 0.8, ** > 0.5, * > 0.2

Fig. 5

Overview of transitions between childhood and adolescence behavioural problem profiles. a Illustration of the transitions. The left side shows groups identified in the childhood data and the right side shows groups in adolescence data. The thickness of the line indicates the proportion of children transferring from the childhood group to the adolescent groups. Only transitions that significantly differed from the proportion expected by chance are shown. Transitions to the typical group (A1b) are omitted from the figure. The mini maps show the behavioural profile of the corresponding groups (see Figs. 3 and 4 for larger versions). b Table overview of transitions between childhood and adolescence showing the percentage of children transferring from groups identified in the childhood assessment (rows) to groups in the adolescence assessment (columns). Bold entries indicate proportions that were significantly above or below what would be expected by chance (Bonferroni corrected)