5.6Dynamic Diagnostics – With ‘Age’ As Example

Cite as: Crijnen, A. A. M. (1999). Age-influences on the taxonomy of psychopathology in children and adolescents. In H. M. Koot, A. A. M. Crijnen, & R. F. Ferdinand (Eds.), Child psychiatric epidemiology: Accomplishments and future directions (pp. 53–66). van Gorcum, the Netherlands.

Practitioners in the field of child and adolescent psychiatry generally bear in mind the age of a child when diagnosing their patients. They do this, because they know that the relations between observable behaviors and underlying psychiatric disorder may change across development at stages and transitions. It is therefore surprising that classification systems in child and adolescent psychiatry, like DSM and ICD, almost neglect the age of a child. In the classification systems of child and adolescent psychiatry, these relations are assumed to be stable, whereas in reality some of these relations appear to change.

Classification systems, like DSM-5, assume a stable relation between symptom and disorder, whereas practitioners take the dynamic nature into account

In this chapter, we regard the need and necessity to study the structure of psychiatric syndromes in relation to age-influences, and we discuss some of the models that may explain some of these changes. The different ways by which the age of children may affect our classifications will be elaborated. Since this field of study has been neglected by scientific research, the considerations are therefore limited to a theoretical exploration supported when possible by scientific evidence.

‘Age’ as an example of the dynamic structure of psychopathology

One of the first things to ask when confronted with questions concerning a child’s emotional or behavioral problems is: How old is this child? And the second question is when it is not clear from the beginning of the interview: Is this child a boy or a girl? Age and sex form a frame of reference against which the complaints of the child are positioned: they reduce the number of disorders that we consider for our first diagnostic hypotheses.

With ‘age’ and ‘gender’ practitioners reduce their ‘problem space’ enormously

They also strongly determine our interpretation of the complaints. The emphasis that we give to the age of a child in our diagnostic reasoning and treatment strongly contrasts with the virtual absence of age-related criteria in our taxonomy of disorders in child and adolescent psychiatry.

An Historic Overview

When reading about the history of child psychiatry we learn that our colleagues in the second half of the nineteenth century had an eye for the differences between a psychologically disturbed child and severely retarded children. In the same period, psychiatrists already understood that some psychiatric disorders were carried over from parents to their children and clinicians were aware of the role of genetic influences on a child’s vulnerability. Around the turn of the 19th century more elaborate theories were formulated about the relation between the development of a child and child psychiatric disorder (Costello & Angold, 1996). The theories of Sigmund Freud stimulated a discussion about the relation and influences of psychological development on psychopathology. Anna Freud subsequently elaborated her fathers’ ideas and introduced the concept of developmental lines. She examined whether the consecutive levels of development corresponded with the age of a child. She also examined the relation between impulses and ego-development and paid much attention to the interaction between both developments (1981).

Disorder Was ‘Frozen’ To Fit Current Classification Systems

In the seventies and eighties, two new classification systems were introduced that are at odds with the idea of developmental lines.

• Both the International Classification of Diseases, developed by the World Health Organization (1992), and the Diagnostic and Statistical Manual of the American Psychiatric Association (1994) are basically a-theoretical and non-developmental in their approach.
• Both systems try to formulate the characteristics that patients with psychiatric disorders have in common at any time in their development.
• Both systems also heavily rely on an unchanging notion of disease.

Classification Systems rely on an unchanging notion of disease

The question can be posed as to how suitable these static classification systems are for use in child and adolescent psychiatry. To answer this question, we have to understand how the taxonomies are constructed and how they may be influenced by the age of a child indicating the developmental stage (Werry, 1992).

How are the taxonomies of psychiatric disorder constructed?

The design of a taxonomy is essentially simple. A taxonomy consists of criteria on the one hand and categories on the other hand.

Criteria

The criteria consist of a list of behaviors or characteristics. These criteria are in one way or the other observable: we can observe the behavior of children and we can ask them questions about their feelings or thoughts.

Categories

The categories on the other hand are unobservable. Categories are always given a name or a label and by using the names we can distinguish categories from each other. Categories are unobservable, but we assume that they are entities in real life because a number of specific criteria form clusters and combinations of certain criteria appear to cluster more often than others. 

Decision rules

This brings us to a third characteristic of a taxonomy: criteria and categories are linked to each other through decision rules.

• A decision rule may be, for example, that a person has to meet a minimum of 6 out of 10 criteria to fall within a certain diagnostic category.
• To fulfill, for example, the criteria for a Major Depressive Disorder, one of the categories in DSM-IV, five (or more) out of 9 symptoms, the observable criteria, have to be present during a 2-week period and they need to represent a change from previous functioning.

The structure of a taxonomy can be reproduced in a model as follows:

• Observable criteria: on the left, 6 observable criteria, 1 – 6, are formulated;
• Unobservable categories: on the right, 2 unobservable categories, A and B, are shown;
• Decision rules: arrows indicate how the criteria are linked to the categories:
  • Criteria 1 – 3 are coupled to category A through the decision rule;
  • Criteria 4 – 6 are coupled to category B.

You will recognize in Figure 1 the design of DSM-IV. By increasing the number of criteria and categories, we would be able to map the entire DSM-IV. Category A could represent, for example, Attention Deficit Hyperactivity Disorder, whereas category B might be Conduct Disorder.

In day-to-day clinical practice, the taxonomy is used in two different ways:

• Clinicians want to show that their patients’ complaints and behavior fall within a certain category or fulfill the criteria of a certain disease or disorder;
• At the same time, they also want to demonstrate that the complaints do not fulfill the criteria of another disease or disorder.

By asking a great number of detailed and very specific questions, clinicians decide whether the complaints of a patient meet the requirements of the classification they are aiming at. To exclude other categories a different strategy is used. Here it is usually sufficient to pose a few global questions. 

A taxonomy translates into a diagnostic interview and drives the questions we ask and the disorders we confirm or refute

A good taxonomy is therefore very convenient: the taxonomy directs our considerations in the diagnostic process and through this process the questions we ask during the diagnostic interview. To an adolescent presenting with complaints of depressed mood and fatigue, you will ask many questions about the symptoms as described in DSM-IV and about related syndromes, like anxiety and somatic complaints, but you will check with few questions whether there are any problems regarding his/her behavior.

The underlying disorder, however, determines whether our patients suffer from some complaints, and not our taxonomy

In the taxonomy and also in clinical practice, the complaints presented by patients are the starting point for the diagnostic process and clinicians try to understand through these complaints the underlying disease or disorder their patients are dealing with. The movement is from the observable criteria to the unobservable categories.

However, to understand reality and to study the potential influence of age on the taxonomy the line of thought needs to be turned around: we assume that in reality an underlying disease or disorder exists that causes the observable behaviors and complaints of our young patients. The arrows in Figure 2 indicate the real direction of the relation: the underlying disease or disorder determines whether there is a relation and the strength of the relation with the observable criteria. The idea is that the real relations dictate the structure of a taxonomy.

‘Age’ and the taxonomy of child psychiatric disorder

Now the question can be addressed: How may the age of a child affect the taxonomy of child psychiatric disorders?  The answer to this question will be found in the scheme that was made to understand how a taxonomy works. Three different models will be proposed to gain this insight.

The first model: ‘Age’ and the relation between symptom and disorder

Figure 3 shows the two disorders A and B, as well as the 6 criteria 1 – 6, and the arrows that indicate the relation between disorders and criteria.

Two arrows depart from Age:

• The first arrow departs from Age to Disorder A;
• The second arrow from Age to the First Criterion.

‘Age’, similar as ‘gender’ or ‘stage of disorder,’ directly influences the relation between symptoms and disorder

The first arrow indicates a direct influence from Age on Disorder A. This arrow indicates that the prevalence or the level of psychiatric disorder or problem behavior in the general population changes significantly with an increase in age.

It is well known that this is the case for most psychiatric syndromes and disorders. In a meta-analysis of 12 cultures we compared information about levels of emotional and behavioral problems in children and adolescents that was obtained through the use of the Child Behavior Checklist, an empirically based paradigm for the assessment and taxonomy of psychopathology (CBCL/4-18: Achenbach, 1991a and b; 1999). For 6 out of 8 Child Behavior Checklist-syndromes significant changes in the level of parent-reported problem behavior occur between ages 6 and 17:

• Withdrawn Behavior and Somatic Complaints increase by age;
• Attention Problems and Aggressive Behavior decrease;
• Anxious/Depressed and Social Problems show an initial increase followed by a decrease at older ages.

These age-related changes appear to be universal phenomena that occur in most of the 12 cultures that were studied (Crijnen, Achenbach & Verhulst, 1997; Crijnen, Achenbach & Verhulst, 1999).

Across cultures, internalizing symptoms increase with ‘age’, whereas externalizing decrease

The increase and decrease in levels of problem behavior are mentioned here, because we have to take them into account in our considerations. Changes in mean levels of psychiatric disorder do not have an influence on our taxonomy, because the criteria, the categories, and the decision rules are not affected.

A problem

This is different for the second arrow indicating the relation between Age and the First Criterion. When age has an effect on a diagnostic characteristic this means that, when compared to the other criteria, this criterion becomes a more important or just less important characteristic of the underlying disorder by increasing age. In this figure, Criterion 1 appears to become a more important characteristic for Diagnosis A when children grow older.

Physical Attacks

Physical attacks, for example, are a less important characteristic of aggression in very young children but will become more important when children become older. The consequence of this effect is that the manifestation of the disorder changes with age. The significance for the taxonomy is that this specific criterion weighs differently in the decision rules at different ages. This has significant consequences for the diagnostic process and also for scientific research. The diagnostic process could be sharpened when we know what the most important criteria are for a disorder at a certain age. The implication for scientific research is that children’s scores on problem scales at different ages cannot be compared straight forward.

Aggressive Behaviour

In child and adolescent psychiatry, a number of studies have shown that the characteristics of aggressive, but also depressive disorders changes slightly within the course of childhood and adolescence.

Cairns and Cairns (1989) followed a cohort of around 220 youngsters from ages 10 until 16 and they studied the changing properties of aggressive behavior. They compared factor-analyses at different ages and concluded that with an increase in age a rearrangement in measures of aggression occurred. At the age of 10 years, aggression in boys and girls was characterized by direct confrontations, like fighting. At age 16, this is still the case for boys, but in 16-year-old girls’ aggression is expressed especially in social interactions mainly by hushing up other girls or by gossiping about them.

Depressive Syndromes

A similar change in the expression of psychopathology was found for depressive syndromes. Goodyer and Cooper (1993; Cooper and Goodyer, 1993) scrutinized symptom profiles of more than 1000 11- through 16-year-old girls. They compared the symptomatology of girls with major depression for 3 consecutive age-categories: 11- through 12-year-old girls with a major depression were characterized by the symptoms hopelessness and decreased speech: 13- through 14-year old girls by depressive mood, lack of concentration, poor functioning at school, feelings of guilt, and indecisiveness, whereas 15- through 16-year-old girls were characterized by depressed mood and agitation.

For both conduct disorder and depressive disorder, the relation between symptoms and underlying disorder changes over time

So, for aggression and depression some evidence exists that the nature of the problems or disorder is indeed changing with a child’s age.

The second model: ‘age’ and comorbidity

In Figure 4, the two disorders A and B, the six criteria 1 – 6, and the six arrows indicating the relation between disorder and complaints, are shown. What is added is an arrow between both disorders. The arrow indicates an association between the disorders, for example because one disorder has an influence on the other or because both disorders are influenced by a common factor. This situation occurs quite often in daily practice and leads to a tangle of disorders, usually expressed as ‘comorbidity.’ Think about the tangle of Attention-Deficit Hyperactivity Disorder (ADHD), Oppositional Defiant Disorder (ODD), and Conduct Disorder (CD), or about anxiety disorders and depression (Nottelman & Jensen: 1995).

Very confusing is the situation when the strength of this relation between both disorders changes with an increase of age. This situation is indicated with the third arrow starting from Age to the arrow connecting both disorders. As indicated above, an important goal of the taxonomy is to distinguish two disorders. No problems are encountered when the disorders are strongly different.

Tangles of disorder not only hamper diagnostics, but bring confusion about treatment

However, confusion arises when we are dealing with tangles of disorders, like the ones shown in Figure 4. In clinical practice, children are observed that fulfill the criteria for one specific disorder, but that display a number of complaints fitting another disorder. Depending on the number and level of complaints related to the second disorder, a discussion ensues about the precedence between both disorders and the disorder that should be treated. One of the possible solutions is to delay the diagnostic process and to repeat it after a while. This practice is based on clinical experience showing that the presentation of psychopathology might be very diffuse in the beginning and crystallizes later on into a more elaborated clinical picture.

Scientific inquiry is also hampered by strong and changing correlations between two disorders, because the correlation leads to unstable analyses. At younger ages for example, criteria appear to cluster around one disorder whereas at older ages they are associated more strongly with another disorder. We have to come into the situation that the criteria characteristic for a disorder are known for each age-period and that the decision-rules are elaborated to distinguish best between children with and without the disorder.

The degree by which the correlation between disorders A and B changes with age is hard to tell, because only limited scientific evidence is available. McGee, Feehan and Williams (1992) studied the comorbidity of anxiety disorders in childhood and adolescence and found decreased associations between symptom scores at older ages when compared to younger ages.

The detailed descriptions by Loeber et al. (1993) about the developmental pathways of externalizing problem behavior give us some clues that those changes in the relations also exist for externalizing disorder. Loeber et al. studied 3 types of antisocial problem behavior: l . Authority-conflict that seems equivalent to the Oppositional Deviant Disorder; 2. Covert behavior which is characterized by secretly performed offenses, like stealing or damage to property; and 3. Overt behavior, i.e. clearly visible aggressive behavior. It appeared that within each developmental path, problem behaviors developed from mild to more severe forms and that these changes were also related to the age of the child. This supports the existence of a direct effect of Age on the criteria, indicated by arrow 2. Furthermore, Loeber et al. observed that especially boys who went along a developmental path of aggression displayed also increasingly more problem behavior related to other pathways. This shows that a relation develops between one disorder influencing another disorder. This is a direct effect of Age on the relation between both disorders, indicated in our models by Arrow 3.

The basic problem in this model is constituted by the correlation between both disorders leading to a tangle of hard to unravel disorders. Even small differences in the strength of the correlations between behaviors at different ages lead to clusters of symptoms that are inconsistent across age-groups.

The third model: one symptom and two disorders

Again, two disorders and six criteria are shown in Figure 5.

• General Disorder (G): Now, Disorder A has an influence on all criteria and therefore Disorder A is called a General Disorder.
• Specific Disorder (S): Three criteria are also influenced by a second disorder which is now called the Specific Disorder.

This model probably is far more often applicable than we are aware of and it is easy to understand that the model provides many opportunities for confusion. The confusion is created because one type of complaint or behavior is influenced by two different disorders which is for example the case in the sixth criterion.

• What about ‘sleep’, a symptom of the General Disorder Depression, but at the same time a clinical problem on its own right, a Specific Disorder

When a child displays this type of behavior and the behavior is used as a criterion in our taxonomy indicating for example the Specific Disorder, we are not aware whether the score is an indication of the specific disorder or the general disorder. This type of models has been more extensively studied in the area of intellectual functioning and school achievement (Gustafsson & Balke, 1993)

This model makes us aware of our conceptions of psychiatric disorder.

• First of all, the boundaries between disorders should be well defined at the level of the disorder and at the level of observable symptoms. It might well be that the high levels of comorbidity reported in child and adolescent psychiatry are – at least to some degree – the result of ill-defined disorders at the level of observed behavior (Nottelman & Jensen, 1995).
• Secondly, changes in level of disorder over time across a number of disorders might well be the result of changes in developmental trends rather than changes in the individual disorder. The trend involves a number of observable behaviors at the same time and is seen as a continuum rather than as a distinct disorder characterized by a cluster of symptoms. For consistency in our models, we speak about General Disorder, but you might as well think about it as a General Trend.

This situation occurs for example in pre-school children and the transition to primary education. At this age, mild levels of attention deficit and hyperactivity are rather the norm than the exception. Restlessness is a temporary phenomenon that disappears in school-aged children and we can consider it to be a developmental phenomenon. You may think about restlessness as a General Disorder. However, besides the General Disorder we are well aware of the existence of Specific Disorders. 

The question is how to distinguish the General Disorder from a Specific Disorder. Stated differently, how can we distinguish the energetic, but immature young child from a child with ADHD or ODD? The authors of the DSM-IV realized the problem but passed the solution on to the clinicians. In the DSM manual you can come upon the following instruction: Consider a criterion met only if the behavior occurs more frequently than is typically observed in individuals of comparable age and developmental level.

It is better not to bring clinicians into this situation, because a substantial degree of subjectivity is introduced. A better solution is at hand.

• First of all, the strength of the relation between the Specific Disorder and the criteria that determine this disorder should be as strong as possible, which means in our model that the relation indicated by the arrow between disorder S and the 6th criterion should be strong.
• Secondly, those criteria should be selected that are as little as possible related to the General Disorder, i.e. the arrow between Disorder G and the 6th criterion should be as weak as possible. Epidemiological studies can yield the information that is needed for the selection of these criteria.

Reconsidering the taxonomy for disorder in child and adolescent psychiatry

The models described in this chapter invite us to reconsider the taxonomy of child and adolescent psychiatry. The relation between disorder and observable criteria in DSM-IV and ICD-IO classification systems is now primarily based on experts’ opinions, but should ideally be based on empirical evidence (cf. Research Diagnostic Criteria). Epidemiological studies on psychiatric disorders in the general population could yield the statistical information for the empirical support of these classification systems (Canino, Bird, Rubio-Stipec & Bravo, 1995). Some epidemiological studies use DSM-IV and ICD-IO classifications. However, these have little value for the study of psychiatric taxonomy.

A serious drawback of the instruments by which DSM-IV and ICD-IO classifications are made in epidemiological studies is their use of gating-items.

Gating-items in epidemiological studies pose a severe threat to our understanding of psychopathology

The answers on a few questions determine whether all questions within a certain diagnostic category are asked or not. As a result, many questions about symptoms are not asked to those subjects who don’t reach the threshold in the gating-items leading to datasets with many missing data. In the empirically based assessment and taxonomy approaches to the epidemiology of psychopathology, for example the Child Behavior Checklist, all questions about emotional and behavioral problems are asked to the parents of all children (Achenbach, 1999). Datasets obtained within this tradition are therefore very useful for studies on the relation between disorder, observable criteria and the influence of age, because complete datasets are available. Since we assume, on the basis of our models and our experience as clinicians, but contrary to the DSM-IV and ICD-IO classification systems, that it is unlikely that all children will have the same set of parameter values, these data-sets need to be examined and the impact of age on the relation between criteria and categories examined. Technically spoken, the amount of population heterogeneity needs to be determined through the use of multiple-indicators-multiple-causes-analyses (Muthen, 1989; Gallo, Anthony, & Muthen, 1994).

Reconsider our taxonomy of psychiatric disorder

The models formulated above, all indicate that age may influence the relation between disorder and the criteria for that disorder.

Reconsider our expert-driven taxonomy of psychiatric disorder and include the results of data-driven empirical studies

The quality of our taxonomy of disorder in child and adolescent psychiatry might be improved through the inclusion of the factor Age in the decision rules. To achieve this, each of the different models should be studied and empirically supported by data collected in epidemiological studies of children in the general population. This type of research, however, requires large numbers of children for whom similar data are available, but fortunately these data are on hand. Recently, CBCL-data of around 15,000 children from general populations studies in 12 different cultures were brought together as part of the 12-culture study (Crijnen et al., 1997, 1999). These data will enable thorough analyses of the different models described above. In a couple of years, we hope to provide an empirical basis for the inclusion of ‘age’ in our taxonomies of child and adolescent psychiatry.