Introduction

Longitudinal research in the field of child psychopathology is relatively rare but there have been some large epidemiological studies in recent years. The Great Smoky Mountains Study in the US [8] provided an important longitudinal perspective on the prevalence of psychiatric disorders from 9 to 16 years with a focus on the cumulative prevalence of 36.7% over the period of 7 years in contrast to a 3 month prevalence of 13.3%, that would be suggested by cross sectional studies. The Dunedin Multidisciplinary Health and Development Study in New Zealand [18] examined the mental health of children and young adults between the ages of 11 and 26 years and provided the evidence that most psychiatric disorders of adults will first appear in childhood and adolescence. The British Child and Mental Health Survey [12] reported a rising rate of psychiatric disorders by age from 7.8% to 12.2%. The survey was based on four age clusters of 5–7, 8–10, 11–12 and 13–15 years but did not include preschool children.

Longitudinal studies on the development of child psychopathology beginning in the preschool years are particularly uncommon [23]. An important exception in this field is the study of Lavigne et al. [20], in which a clinical sample of 344 children aged between 2 and 9 years was investigated for stability of psychopathology. Over a period of almost 4 years the authors found a persistence of symptoms in 50% of the cases.

In a community sample in the Netherlands [19] consisting of 403 children at the age of 5 years the most important predictors of mood and anxiety disorders over a period of 18 months were living in a single-parent family and the occurrence of a stressful life event. Keiley et al. [16] examined externalizing and internalizing problem behaviour in yearly intervals for 8 years in a sample of 405 children between 5 and 12 years of age. There was no effect on internalizing and externalizing behaviour problems by gender. The internalizing score remained unchanged while the externalizing score declined from preschool to school age.

In the Netherlands mental health problems have been examined in a population based sample of 2,076 children between the ages of 4 and 18 years [5]. In childhood no difference between boys and girls was reported on internalizing problems. However, adolescent girls showed more internalizing problems than boys at this age. The scores of externalizing problems were reported to decline by age, with boys showing more problems than girls throughout all ages. This latter result stands in contrast to Keiley et al. [16] and a former study by Campbell [6], whose results showed no gender differences in preschool children. At primary school age more externalizing symptoms were found in boys than in girls, and in adolescence more internalizing symptoms were found in girls than in boys [25, 26].

One possible reason for inconsistencies across studies can be seen in the confounding of internalizing and externalizing symptoms. It is unclear if children with high degrees of externalizing symptoms are vulnerable to symptoms of anxiety or depression [10]. Children with externalizing symptoms may experience emotions such as sadness and fear with relative frequency because they tend to be rejected by their peers. In adolescents it is known, that anger and frustration are positively related to externalizing problem behaviour [7] and also to psychosomatic disorders [17] and depression [4], which are both internalizing types of syndromes.

In summary, the scope of literature about child mental health on the transition from preschool to primary school is unclear: there is evidence for an increasing level of mental health problems while internalizing problems seem to persist and externalizing problems even decline [5, 16]. In preschool age the findings of higher externalizing problems of boys are not consistent. There is a gap of research about preschool children because most longitudinal studies on child mental health problems do not start until the children are at school. The present study focuses on the transition from preschool to primary school. In the present second wave a population based total sample of children was examined at primary school age. In the baseline study 4 years ago the whole 1 year intake to primary school had been included [13].

Aims of the study was to explore the prevalence rate of behavioural and emotional problems at fourth year primary school, the comparison of prevalence rates between preschool and fourth year primary school, and the investigation of gender specific prevalence rates. In addition to the well established broadband scales of exclusively internalizing and externalizing symptoms the combination of those symptoms was also examined.

Method

Sample and procedure

The data of the present study were derived from a 4 year follow-up assessment in a northern German university town with a population of 260,000. The total fourth year primary school population in the town included 2,524 children. Each family of a fourth year child was given a set of questionnaires together with a covering letter. The questionnaires were completed by the parents and returned to the schools in sealed envelopes, which were collected by class teachers and forwarded to us. The participation in the study was voluntary and anonymous. Due to a 68.3% response rate and the exclusion of missing data the fourth year primary school sample consists of n = 1,481 children.

The present study was the second wave of two repeated assessments, 4 years after a baseline study at preschool age had been completed. The size of the baseline sample had been n = 1,887 children. For further details of the baseline study see elsewhere [13]. National privacy laws prohibited the allocation of personal identity codes and prevented the matching of a child’s individual measurements at preschool and primary school stage. At each wave of data collection all children of the same year of school intake were included in the assessments. The fluctuation in the period of 4 years at primary school was rated at 10–15% (Local school authority, personal communication). The overlap of the present school sample with the baseline sample was estimated by a statistical random model under consideration of the valid data at each wave and a fluctuation of 15%, indicating that between 949 and 1,259 children (64–85%) of the present sample are the same at both assessment waves.

In the fourth year primary school sample were 733 boys (49.5%) and 748 girls (51.5%). The age of the children was in a range of 8.8–11.9 years and a mean of 10.0 years (SD 0.48). From the total of 1,481 children 1,217 children (82.2%) lived with both parents, 244 (16.5) lived in a one parent family, 237 (16.0%) with a single mother, and 7 (0.5%) with a single father. Fourteen children (0.9%) lived with other relatives or in residential care. With regard to the housing situation there was no difference to the baseline study 4 years previously (95% confidence interval, CI). For demographic features of the present sample, the preschool sample, and the total population in town see Table 1.

Table 1 Demographic features of the samples at preschool age (PS) and fourth year primary school (S), and data of the reference population
Full size table

Measures

The parents were asked to complete the German version of the CBCL 4–18 [1, 3]. The CBCL contains a list of 120 problem items. The parents ticked the 0 if the item was “Not True” for their child, 1 if the item was “Somewhat or Sometimes True”, and 2 if it was “Very True or Often True”. By summing the scores of 1 and 2, the degree of child mental health problems for eight syndrome scales, two broadband groups of syndromes, and a total score was computed.

The standardization in T-values was done on the basis of the German norms [9]. The prevalence of behaviour and emotional problems was computed only in reference to the clinical range of the CBCL scales [1]. The cut-off point was at T = 64 in the total score (92. percentile) and at T = 71 (98. percentile) in the syndrome scales.

For those children who met the criteria of mental health problems, three subtypes of problem behaviour were defined. Children who had a T-score of 64 and more on one broadband scale—internalizing (INT) or externalizing (EXT) syndromes—and a T-score of less than 60 on the other broadband scale were considered to belong either to the INT or to the EXT syndrome group. The definitions of INT and EXT were precise implementations of the CBCL manual [1]. In addition to the CBCL manual a third grouping of co-morbidity (COM) was defined and applied to the sample. If children had a T-score of 60 and more on both broadband scales, they were considered to be in the COM group.

Due to the need for comparability with the baseline study at preschool age [13] four of the 120 items had to be dropped from the CBCL (items 73 and 96 on sexual problems, item 101 “truancy”, and item 105 “alcohol, drugs”). These items were considered not applicable at preschool and primary school age. This proceeding was in accordance to other studies of children at the age of 6 and 10 years [10, 15]. An item analysis revealed negligible frequencies in the four items of 0.00% in the items 73 and 105, 0.03% in item 96, and 0.01% in item 101 for children up to the age of 11 years [1].

Results

Prevalence

In the present study 267 (18.0%) children met the criterion of the clinical range of mental health problems. Significantly more children had mental health problems at fourth year primary school than in the preschool study, where prevalence had been 12.4%: χ²(1, n = 3,368) = 20.75, p < .001 (Table 2). The increase of prevalence was highly significant in boys (14.2–22.2%; χ²(1, n = 1,708) = 18.8, p < .001) but not in girls (10.5–13.9%; χ²(1, n = 1660) = 4.4, p = .04). The risk ratio (RR) of child mental health problems for boys versus girls was 1.60.

Table 2 Frequencies of child mental health problems at preschool age (PS) and fourth year primary school (S)
Full size table

Internalizing and externalizing symptoms

From 267 children with child mental health problems at fourth year primary school, 73 (27.3%) children showed exclusively internalizing symptoms (INT) and 16 (6.0%) externalizing symptoms (EXT). Most of the children with mental health problems, 140 (52.4%) showed a combination of internalizing and externalizing symptoms (COM). The symptoms of 38 (14.2%) children were not related to any of the broadband groups. In Table 3 the distribution of the symptom groups is broken down by gender and significant differences between boys and girls are shown: χ²(3, n = 267) = 13.82, p = .003. Post hoc comparisons revealed significantly more EXT symptoms in boys (χ²(1, n = 267) = 7.65, p = .006 and significantly more COM symptoms in girls (χ²(1, n = 267) = 8.31, p = .004).

Table 3 CBCL-Broadband Groups: symptoms of boys and girls with mental health problems at preschool age (PS) and fourth year primary school (S)
Full size table

Between preschool and primary school the frequencies of the three symptom groups did not differ significantly: χ²(3, n = 501) = 3.93; p = .27. But at fourth year primary school there was a tendency for more INT symptoms in boys and more COM symptoms in girls compared to the baseline at preschool age (Table 3).

Syndrome scales

The mean T-scores of the eight syndrome scales are shown in Fig. 1. The scores of the scale Anxious/Depressed were the highest of all CBCL syndrome scales both at baseline and at follow-up. The multivariate analysis of the eight syndrome scales was statistically significant in the main effects of time, F(8, 490) = 7.15, p < .001 and gender F(8, 490) = 2.95, p = .003. Post hoc tests revealed significant differences between boys and girls on the scales Anxious/Depressed (boys: 64.8, girls: 67.6; U(n 1 = 163, n 2 = 104) = 2.58, p = .01, d = 0.4) and Attention Problems (boys: 64.6, girls: 62.2; U(n 1 = 163, n 2 = 104) = 3.18, p = .001, d = 0.3) with effect sizes between small and medium.

Fig. 1
figure 1

Profile of the CBCL syndrome scales: mean T-scores of boys and girls with mental health problems at preschool age (PS) and fourth year primary school (S)

Full size image

The comparison between baseline measurement (T1) and the present second wave (T2) indicated a significant increase on the scale Somatic Complaints from T1 = 59.3 to T2 = 63.8. (U(234, 267) = 5.66, p < .001, d = 0.5). This increase was significant for boys and girls and the effect size was medium. On the scale Social Problems a significant increase of the mean was found in girls only (T1: 62.5, T2: 65.1; U(96, 104) = 1.93, p = .05, d = 0.3). The scale Attention Problems showed a significant increase of the mean in boys only (T1: 62.6, T2: 64.6; U(138, 163) = 2.01, p = .04, d = 0.2). The decrease of the mean from preschool to primary school on the scale Aggressive Behaviour was not statistically significant (T1: 62.8, T2: 61.7; U(234, 267) = 1.36, p = .18). Also no significant differences were found on the remaining scales Withdrawn (T1: 63.0, T2: 62.8; U(234, 267) = 0.21, p = .84), Thought Problems (T1: 59.5, T2: 58.6; U(234, 267) = 0.92, p = .36) and Delinquent Behaviour (T1: 60.4, T2: 60.7; U(234, 267) = 0.73, p = .47).

Discussion

Four years after the preschool study [13] the prevalence of child mental health problems was re-examined at fourth year primary school and the prevalence had increased significantly from 12.4% to 18.0%. These results of the second wave at school age correspond with findings of earlier studies [8, 11, 16, 25]. Gender specific evaluation showed a significantly higher prevalence of child psychiatric symptoms in boys than in girls. The risk of child mental health problems for boys was 1.6 times higher than for girls. Already in the baseline study at preschool age there was a higher level of child mental health problems in boys than in girls. This difference persisted and increased at fourth year primary school.

In contrast to earlier studies [2, 14], the increase of symptoms at school age was primarily due to an increase of internalizing symptoms and the increase was only significant in boys, but not in girls. Of those children with a psychiatric disturbance at school age the rate of internalizing symptoms was more than the fourfold of externalizing symptoms (27.3% vs. 6.0%). The high rate of internalizing child psychiatric symptoms especially in boys is unexpected and requires further longitudinal and prospective studies of early child mental health.

The additional category of comorbid symptomatology (COM) has proven to be very helpful. A combined symptomatology of internalizing and externalizing symptoms was present in 52.4% of the disturbed children. Especially externalizing symptoms were often combined with internalizing symptoms but were rarely present on their own. A high correlation of externalizing and internalizing symptoms has been already highlighted by Eisenberg et al. [10].

On the level of single syndrome scales the differences between boys and girls at fourth year primary school, the significantly higher scores for girls on Anxious /Depressed and for boys on Attention Problems, were in line with previous findings [5, 8, 11, 22]. Also the significant increase between preschool and primary school on the scale Somatic Complaints in boys and girls [24], and on the scale Social Problems in girls only were in line with previous findings [5]. Contrary to previous findings [16, 21], the higher number of psychiatrically disturbed boys was primarily related to a significantly increased prevalence of internalizing symptoms and attention deficits, with at the same time unchanged prevalence of externalizing symptoms.

The present sample at fourth year primary school proved to be representative of the study population. The strict anonymity of the subjects did not allow a one-to-one matching with the baseline data. Both assessments had to be treated as two cross-sectional samples from the same population. The overlap between the subjects of the present sample and the baseline study was 64–85%.

Given the large number of subjects, the symptom checklist approach offers useful information about children at preschool and primary school. For international comparison of the results the late school entry in Germany, generally at the age of 6 years, needs to be taken into account.

A teacher perspective could add valuable information about problem behaviour at school, which might possibly result in reports of more externalizing problems. Due to the principle of identical data collection at both points of assessments and the anonymity of the subjects, it was only possible to obtain the parental report. Despite this limitation the core findings of the present study remain unaffected, which show a high number of persistent internalizing symptoms at fourth year primary school, occurring solely or in combination with externalizing symptoms.

The results point to the need for early child psychiatric research on child mental health beginning in infancy and preschool age. Especially the high prevalence of internalizing symptoms in boys demands more sensitivity towards internalizing child psychiatric symptoms not only in girls but also in boys. The gender- and age-specific vulnerability for internalizing psychopathology in boys requires the attention of all child mental health workers and demands appropriate mental health diagnostic and treatment facilities already at preschool and at primary school age.