The overall purpose of this study was to explore mothers’ descriptions of their children (bottom-up) and compare them with the descriptions that emerged through a standardized diagnostic interview (top-down). To make this comparison possible, we used a mixed method design. When comparing problem descriptions and diagnoses between bottom-up and top-down approaches, we found both convergent and divergent results.
Convergence between bottom-up and top-down
For the ODD diagnosis as a whole, bottom-up and top-down converged. Two of the themes from the qualitative content analysis, Problematic behavior and Problematic traits, correlated well with the DSM’s construction of the ODD diagnosis in terms of behaviors (e.g., often loses temper, often argues, often blames others) and descriptions of traits (e.g., is often touchy, spiteful, angry, and resentful). The third theme, Difficulties, described regulatory difficulties many of these children had, showing comorbidity with other DSM diagnoses such as ADHD, depression, anxiety, and autistic traits, which converged with the diagnostic interviews. There was also convergence between top-down and bottom-up descriptions of the essential qualities of ODD, expressing both behavioral and emotional disturbances.
Three of the criteria that seem to be core symptoms in the ODD diagnosis overlapped very well with the categories and subcategories in the bottom-up analysis. The criterion Often actively defies or refuses to comply with adults’ requests or rules equated with the category defiant behavior, which was the most common problem mothers reported, describing children who create conflicts in everyday routine situations and who refuse to follow corrective instruction. Often loses temper was also one of the two most common criteria in DSM met by children with ODD, with good agreement with emotionally externalizing behaviors. This category was the second largest group of problems mothers mentioned, describing tantrums, outbursts, and screaming and shouting, which matched the criterion well. The third criterion with good agreement between top-down and bottom-up was Often angry and resentful. Mothers reported as a major problem that their children had aggressive traits and described them as angry, frustrated and angry, aggressive, and prone to unprovoked aggression, but no mother described their child as resentful, which is included in the diagnostic description. Both approaches, however, highlighted a group of children with ODD who have an ongoing angry mood.
There was also convergence between top-down and bottom-up approaches in the category of
Difficulties. A large group of children with ODD are immature in their cognitive, social, and emotional development [
27,
28]. When mothers in the present study shared their experiences of the largest problems, 68% highlighted the various constraints their children faced (74% of the boys, 46% of the girls). They described deficits in emotional regulation, attention regulation, and behavioral control. Some children also had limitations in intellectual capacity and/or difficulty with flexibility. Assessment on the K-SADS interview revealed that 54% met the criteria for ADHD or ADHD UNS, showing deficits in attention and behavioral regulation. When children with regulation deficits and neurological immaturity are exposed to requirements they are not yet mature enough to cope with, their defiant behavior often occurs at the intersection of demands to self-regulate and their inability to do so [
29]. Furthermore, the inflexibility in the bottom-up category of defiant traits and the children´s aggressive behaviors could also be early signs of autism. It is important not to consider the children as simply brutal and defiant; instead we should be aware that many may have developmental neurological difficulties and to a large extent depend upon adaptions and support from their environment.
Divergences between bottom-up and top-down
In contrast to the DSM, which employs a categorical system, the bottom-up process allows more multi-dimensional thinking, as seen in the case of the category
defiant behavior. Mothers in the study described three different qualities or dimensions of defiance in their children. The first dimension,
disobedience, included children’s attempts to ignore and/or act against their parents’ directives. This typically defiant behavior occurs when children disobey, do not listen, and generate conflicts during everyday routine situations. The second defiant dimension was
inflexibility. These children had difficulty adapting to the demands of their surroundings. They had their own rules, wanted to do things their own way, and had strange and fixed ideas. They expected to live on their own terms and became very frustrated when things did not go their way. They seemed to be more rigid in their defiant behavior. The third dimension of defiance was
rebelliousness. This dimension seems to be the most severe of the three. The rebellious dimension includes more active refusals and resistance against parents. Mothers described power struggles, refusals, and behaviors associated with a strong negative affect and more aggression in refusing to comply with adults’ requests or rules, which is in the line with Wakschlag et al. [
9]. Our findings show notable variations in the quality, severity, and intrinsic degrees of aggressiveness and inflexibility in all three dimensions of defiant behavior. In children with defiant behavior, further investigations are warranted to access the quality of the behavior, which could contribute to better adapted interventions for children with more serious defiant behaviors.
Second, in the diagnostic interview, 74% of mothers responded affirmatively to the item Is often touchy or easily annoyed as a problem on the clinical level. In the bottom-up description, however, no mother said spontaneously that her child was touchy or easily annoyed. Instead, they described their children as grudging (unpleasant, jealous, negative, grumpy) and displeased (frustrated, never satisfied, whiny). We called this category of answers negativistic traits. This category seemed to express emotions of dissatisfaction (displeased and grudging) rather than a touchy mood.
There was also a divergence between approaches in the results for antisocial behavior. The mothers reported antisocial behavior in the bottom-up approach, but the ODD criteria did not catch these behaviors. Aggressive behavior (physical violation), together with provocative behavior (seeking and initiating physical conflict), and/or norm-breaking (destroying things, running away, lying), were presented as major problems in about one quarter of the children in this study. Mothers described children who had physically attacked parents, peers, and siblings. Several studies have attempted to identify the characteristics of children with ODD who develop CD and antisocial personality [
30‐
32]. Subthreshold CD symptoms have been identified as predictive [
8], and persistent physical fighting is particularly important [
33,
34].
Clinical psychologists sometimes seem to lack support in identifying such children. However, it might be inappropriate to use the criteria for a diagnosis of CD in young children. Approximately a quarter of CD symptoms seem to be developmentally impossible in early childhood and approximately one third are developmentally improbable in preschoolers [
9]. In the present study, only one of 57 (< 2%) children with ODD also met the full criteria for CD, and this child was in the oldest age group of 8 years. The present study has shown that aggressive, norm-breaking, and provocative behaviors manifest in preschool and early school years are not caught in their entirety by the ODD diagnostic tool, nor are they always sufficiently alarming in early childhood to meet the diagnosis of CD (three criteria need to be met for a diagnosis). Furthermore, some of the children also displayed traits, similar to CU-traits and impulsive traits, which might increase the risk to develop conduct problems and later an antisocial personality [
35]. DSM diagnostic tools seem sometimes to be too blunt and not adapted to capture emerging symptoms of severe conduct problems in very young children; this study shows a gap between ODD and CD criteria and diagnoses for the youngest children. In the top-down approach the ODD criteria were found to help identify and separate commonly occurring oppositional behavior from conduct problems, but in the bottom-up approach the accepted criteria did not capture the entire range of problematic behaviors, especially aggressive, provocative, and norm-breaking behaviors that risk developing into persistent antisocial behavior. One way to verify suspicion of early antisocial behavior in preschool children would be to have the possibility to specify in the ODD diagnosis, if there is also subclinical CD (one or two criteria), in the same way as in the CD diagnosis where it is possible to retrospectively specify whether the individual has shown at least one CD symptom before the age of ten (child-hood onset type).
Comparing the content of reported problem behaviors (bottom-up), we found similarity between our descriptions of
Problematic behaviors and the dimensions described by Wakschlag and collegues [
16]. Wakschlag’s four core dimensions for disruptive behavior: Noncompliance, Temper Loss, Aggression, and Low Concern for Others correspond well with our four categories of
defiant,
emotionally externalizing,
aggressive, and
provocative behaviors
. However, in addition to the dimensions proposed by Wakschlag and colleagues [
16], and the criteria for ODD, we also found a group with norm-breaking behavior.
Associations
An important part of the mixed-method analysis is to merge the two data sets [
22]. Although the sample size was small, there appeared to be a distinction between overt and covert externalizing behavior in this study. These results are similar to those of Loeber and Burke [
32], who described three different pathways into antisocial behavior: 1. Overt Pathway, beginning with annoying others and bullying, moving on to more aggressive behaviors such as physical fighting, gang fighting, and rape; 2. Covert Pathway, beginning with lying and shoplifting, leading to vandalism, pick-pocketing, and serious delinquency (theft, burglary); and 3. Authority Conflict Pathway, beginning with stubborn behavior, then disobedience, staying out late, running away, and finally truancy and avoiding authority.
In the present study, several of the behaviors described above, such as annoying others, fighting, lying, destroying things, disobedience, running away and avoiding authority, were represented among the children and were clearly visible in the qualitative content analysis. Furthermore, children with norm-breaking behavior seemed to be less angry and resentful, while children who showed aggressive behavior toward others seemed less often to actively defy or refuse to comply with adults’ requests or rules. The data are based on small groups and these findings merit further research. However, they suggest that maybe it could be possible to trace different pathways of externalizing behavior even in children as young as 3–8 years.
Differences between girls and boys
While, relatively few studies have investigated differences between boys and girls when it comes to ODD criteria, one of our aims in this study was to examine whether there seemed to be any differences between gender and between bottom-up and top-down approaches. We found no considerable difference between boys and girls in the different ODD criteria or the 14 categories from the content analysis. Thus, ODD seems to be expressed quite similarly in boys and girls. However, some small differences were found that is needed to be mentioned.
In the top-down approach, boys actively defied or refused to comply with adult’s requests or rules more often than girls. In the bottom-up analysis, it was exactly the opposite. Defiant behavior and defiant traits were more often reported as major problems among girls than boys. These contradictory results might be explained by greater social acceptance of defiance in boys than in girls, even when boys are more often and more strongly defiant. If defiant behavior and traits in girls are less socially accepted and considered more problematic, disobedience in girls is probably more noticeable and more likely to be reported by mothers as a major problem, even though these behaviors are more frequent among boys. These results, however, do not fully correspond with Wright and colleagues, who found that only fathers (not mothers) had less tolerance for daughters’ than for sons’ DBD behaviors [
36]. However, due to the small sample seize in the present study the results should be interpreted cautiously.
Another difference between boys and girls was found for the criterion
Often deliberately annoys others. The mothers´ bottom-up descriptions didn´t tally with their answers on the diagnostic question about annoying others. In the diagnostic interview, girls deliberately annoyed others significantly more often than boys. Studies have shown that girls are significantly more relationally aggressive than boys, and this sex difference is apparent as early as the preschool years [
27]. Girls tend to engage in higher levels of both proactive and reactive relational aggression compered to physical actions [
37]. This indicates that even in early childhood, relational aggression appears to be the modal type of aggression for girls. Comparing results from the bottom-up and top-down analyses, there seem to be several levels of
annoying severity; teasing and annoying others are on the border of more normative behavior levels, while seeking and initiating conflict (for boys) and relational aggression (for girls) might be on a more clinical level. The differences between boys’ and girls’ disruptive behaviors might be most visible on this criterion. The different provocative behaviors, boys’ and girls’ distinct expressions need to be further explored and perhaps more clearly defined based on diagnostic criteria.
Differences between ages
In the bottom-up description, we found no significant differences in symptoms between ages 3–5 and 6–8. However, using the top-down approach 3- to 5-year-olds were more likely to often lose temper, and 6- to 8-year-olds were touchy or easily annoyed significantly more often than children 3- to 5-year-olds. Interestingly, negative emotional traits were described in the bottom-up approach as grudging (unpleasant, jealous, negative, grumpy) or displeased (frustrated, never satisfied, whiny), while no mother spontaneously described their children as touchy, easily annoyed or resentful. This might indicate that the diagnostic criteria that describe irritability (touchy or easily annoyed and angry and resentful) are more appropriate to describe older children, while grudging and displeased are a quality of descriptions more appropriate for younger children. This might be something to explore further in future studies.
There is also frequency criterion of the diagnosis in the DSM-5, providing guidance on minimum symptom frequencies for different age groups. For children younger than 5 years, the behaviors should occur on most days for a period of at least 6 months, and for those aged 5 years or older, the behavior should occur at least once peer week [
7]. Our results from top-down and bottom-up approaches raise questions about the frequency criteria in DSM-5. Are there a risk that we over-diagnose children from 5 years with ODD? According to Wakschlag and colleagues, temper tantrums seem to be more common in preschool-aged children [
38]. Our findings is in line with this. The frequency of other symptoms appears to be more constant between the age groups in the present study. It may be, that differences in frequency of disruptive behavior symptoms between younger and older children are greater in community samples than in clinical samples, where the rates of externalizing behavior continue to be high for this group of children.
Limitations
This study had several limitations. Present study has a cross-sectional design and highlights convergences and divergences between top-down and bottom-up approaches concerning symptoms of ODD diagnosis. To really know if the children who exhibited more severe conduct problem in this study later develop conduct disorder (CD), a longitudinal study would be needed. Another important limitation was the small sample, regarding comparisons between boys and girls. In the comparison of themes and categories, our calculation of statistical significance using qualitative data might be considered problematic. However, although the qualitative questions were open-ended, they structured to limit the frames of interpretation. Another limitation was the use of K-SADS as measure for children younger than 6 years (35% of the sample). The diagnostic questions in K-SADS are designed for children aged 6 to 17 years, but the probes and scoring criteria were adapted for children 3–5 years as necessary. This adaptation relied on the three psychologists' experiences in developmental psychology. Participants in the study were mothers only; multiple participants for each child would have provided a fuller description of the children’s functioning. However, by solely interviewing mothers, we got a more homogeneous group. Even if interesting, it was beyond the scope of present study to examine effects of comorbidity on mothers´ description of the children. Furthermore, in an analysis of the most common form of comorbidity in our sample (ODD + ADHD) it did not seem to influence the bottom-up descriptions of the children.
A strength of the study was the interviewer-based diagnostics, in which the clinician decides whether a symptom is present or absent. Many studies use rating scales instead, and respondents (especially parents) usually have limited exposure to the full range of normative behaviors. Another strength was the problem-formulated scales, in which mothers described the most troublesome problems they had with their children, giving an entirely different weight to the problems described. A further strength was the vivid picture of the different behaviors and personality traits hidden behind the diagnostic criteria that were revealed in the qualitative content analysis.