1

ΠΑΡΑΛΛΗΛΑ ΚΕΙΜΕΝΑ

ΕΙΔΙΚΉ ΑΓΩΓΉ
& ΕΚΠΑΊΔΕΥΣΗ

Comorbidity in Child Psychopathology:

Concepts, Issues and Research Strategies

Chantal Caron and Michael Rutter

1.1
J. Child Psycho/. Psychiat. Vol. 32, No. 7, pp. 1063-1080, 1991

Printed in Great Britain.

0021-9630/91 $3.00 + 0.00

Pergamon Press plc

© 1991 Association for Child Psychology and Psychiatry

Comorbidity in Child Psychopathology:

Concepts, Issues and Research Strategies

Chantal Caron and Michael Rutter

CHANTAL CARON*† AND MICHAEL RUTTER*

*MRC Child Psychiatry Unit, Institute of Psychiatry, De Crespigny Park, Denmark Hill, London SE5 8AF, U.K.
†Present address: Centre de Recherche Laval, Robert-Giffard, 2601, Chemin de la Canardiere, Beauport,
Quebec G1J 2G3, Canada.

Accepted manuscript received 28 September 1990

Requests for reprints to: Professor Michael Rutter, MRC Child Psychiatry Unit, Institute of Psychiatry, De Crespigny Park,

Denmark Hill, London SES 8AF, U.K.

3

ABSTRACT

Epidemiological data show that the co-occurrence of two or more supposedly separate
child (and adult) psychiatric conditions far exceeds that expected by chance (clinic data
cannot be used for this determination). The importance of comorbidity is shown and it is
noted that it is not dealt with optimally in either DSM-III-R or ICD-9. Artifacts in the detection of
comorbidity are considered in terms of referral and screening/surveillance biases. Apparent
comorbidity may also arise from various nosological considerations; these include the use
of categories where dimensions might be more appropriate, overlapping diagnostic criteria,
Κeywords: artificial subdivision of syndromes, one disorder representing an early manifestation of the
other, and one disorder being part of the other. Possible explanations of true comorbidity
are discussed with respect to shared and overlapping risk factors, the comorbid pattern
Comorbidity, child psychiatric disorder, constituting a distinct meaningful syndrome, and one disorder creating an increased risk for
epidemiology, classification the other. Some possible means of investigating each of these possibilities are noted.

INTRODUCTION

Every medical student is taught that, whenever possible, a single diagnosis should be
made (Jaspers, 1963; Kendell, 1975). Patients may present with multiple diseases but, when
there is complex mixed symptomatology, an unusual presentation of a single disorder is more
likely than the simultaneous occurrence of two or more unrelated conditions (comorbidity).
Yet in child psychiatry, all epidemiological studies that have examined the issues have shown
that comorbidity is extremely common (Anderson, Williams, McGee & Silva, 1987; Kashani
et al., 1987; Flament et al., 1988; Szatmari, Boyle & Offord, 1989; Weissman et al., 1987). The
same has been found in adult psychiatry (Boyd et al., 1984). In this paper, we consider the
empirical findings, concepts, and research implications of this important issue.
4

FREQUENCY OF COMORBIDITY

The first question is whether the observed rate of comorbidity in epidemiological surveys exceeds that
expected by chance alone. The expected rate is obtained by multiplying the base rates of each of the separate
conditions involved in the comorbidity patterns studied. The epidemiological studies undertaken by Anderson et
al. (1987) and Kashani et al. (1987) both presented their findings in a manner that allows this calculation. The data
are summarized in Table 1a and 1b.

Table 1a. Base rates of disorders in three community studies that serve to
calculate the expected number of comorbid cases

Author n Age Disorders Base rates

Anderson 785 11 Oppos/Conduct d. 9.2%

et al. Anxiety d. 7.5%
(1987) Attention deficit d. 6.7%
Depressive d. 1.8%

Kashani 150 14-16 Oppos/Conduct d. 14.7%

et al. Anxiety d. 8.7%
(1987) Depressive d. 8.0%

d. = disorder.

Table 1b. Expected and observed number of comorbid cases

in two community studies

Number of Expected Observed 95%

Author disorders cases (E) cases (O) O/E C.I.

Anderson 1 — 100
et al. 2 17.5 27
(1987) 3 0.62 4
4 0.007 8
Total >1 18.43 39 2.1 1.5-2.9

Kashani 1 — 12
et al. 2 4.74 7
(1987) 3 0.17 7
Total > 1 4.91 14 2.85 1.7-4.8

95% C.I. = 95% confidence interval.

5

Thus, in the Anderson et al. (1987) survey, 7.5% of 11-year-olds showed an anxiety disorder and 6.7 % an
attention deficit disorder; the expected comorbidity between these two disorders therefore is the product of 7.5%
and 6. 7 % , namely O.5% . By summing all possible combinations, the overall expected comorbidity rate for different
numbers of conditions can be obtained (Table 1b). It is evident that the observed comorbidity rate was more than
double that expected by chance; in the Kashani et al. (1987) study the excess was even greater. The data used in
these calculations (as presented in the original papers) were based on pooled diagnostic groupings. For example,
all the various anxiety disorders were combined to form one grouping; the same applied to depressive disorders
and to oppositional/conduct disorders. This means that comorbidity within these pooled categories was not taken
into account. Accordingly, the true observed comorbidity rate must have been considerably higher, so that the
calculated two- to three-fold excess over chance expectation constitutes a gross underestimate of the frequency of
comorbidity. These two reports were chosen because they presented their findings in a fashion that made it easy to
calculate observed and expected comorbidity rates. However, other epidemiological studies are agreed in showing
a very high comorbidity rate and it may be concluded that the observed co-occurrence of supposedly separate child
psychiatric disorders far exceeds that expected by chance alone.
The fact that the pooling of diagnoses greatly underestimates the true rate of comorbidity is evident in the adult
epidemiological data presented by Boyd et al. (1984). For example, the population base rate for panic disorder was
67 out of 11, 176, 0. 6 % . However, out of 266 individuals with major depression, 30 had a panic disorder, a rate of
11%, representing a huge increase over chance expectation.
In drawing this conclusion, we have restricted our attention to epidemiological data because general population
base rates are essential for the calculation of expected comorbidity rates. Because much of the discussion of
comorbidity in the literature is based on the findings of studies of clinic samples, it is necessary to note that clinic
data alone cannot provide information on whether or not the observed comorbidity rate exceeds that expected by
chance. A simple example serves to illustrate the point. Suppose that disorders A and B are both present in 10% of
the general population and that there is a 5 % rate for ‘pure’ A, 5% for ‘pure’ B and 5 % for the comorbidity pattern
of A+ B. As the comorbidity expected by chance is 1% (i.e. 10% x 10% ), this means that the observed comorbidity
rate is five times the chance expectation.
Let us also suppose that all children with these disorders are referred to clinics without any referral bias (and
that there are no other disorders). The clinic pattern will, therefore, be the same as that in affected members of
the general population; namely 33% with disorder A, 33% with disorder Band 33% with A+ B. If the clinic base rate
(instead of the general population rate) is used to calculate the comorbidity expected by chance, the expectation
is 67 % x 67 % , i.e. 45 % . That is, instead of the true five-fold excess, the clinic data misleadingly appear to indicate
that the observed rate is less than that expected by chance. Clinic data can be used to assess comorbidity only if
the general population rates for each disorder are known and if data are available on the clinic referral rate and
biases for each disorder.
In that connection, it should be noted that, whenever less than all subjects with disorder are referred, clinic
samples will always contain a disproportionately large proportion of patients showing comorbidity (Berkson,
1946). That is because the referral likelihood for subjects with disorders A and B will be a function of the combined
likelihood of referral for each disorder separately. That is so irrespective of referral biases. However, it is known that
in practice there are various referral biases and these also need to be taken into account. For example, Shepherd,
Oppenheim and Mitchell (1971) found that children were more likely to be referred if there was also parental
psychopathology or family problems.
For all these reasons, clinic data need to be used with considerable caution in studying comorbidity. However,
provided that epidemiological data have already shown that there is a significantly raised comorbidity, and provided
that enough is known on referral influences to consider their possible effects, clinic data may be used to examine
possible explanations for comorbidity patterns.
6

DOES COMORBIDITY MATTER

Comorbidity has been largely ignored in the research literature for years and it is necessary to ask whether it
matters sufficiently for detailed attention to be paid to it now. There are two main reasons why a failure to pay
attention to comorbidity may lead researchers to draw quite misleading conclusions. First, a study of condition X
may produce findings that in fact are largely a consequence of the ignored comorbid condition Y. For example,
Anderson et al. (1987) found that of the 14 children with depressive disorders, no less than 11 had at least one
other psychiatric condition as well. Indeed eight out of the 14 children showed depression and an anxiety disorder
and a conduct disorder and an attention deficit disorder! It follows that the correlates, outcome or genetic features
reported for childhood depression could in reality be those of attention deficit or conduct disorders.
The second reason is that, when comorbidity is ignored, the implicit assumption is made that the meaning of
condition A is the same regardless of the presence or absence of condition B. As shown below, that is an unsafe
assumption and in some circumstances it appears to be mistaken.
It might be thought that the solution would be to exclude comorbid cases in order to focus on “pure” groups.
However, the extent of comorbidity is such that often this would result in the investigation of tiny atypical samples.
For example, as already noted, only three out of 14 cases of depression were “pure” in the Anderson et al. (1987)
study, and among the 12 cases of depression in the Kashani et al. (1987) investigation none was pure. Clearly, that
cannot constitute a general solution; moreover, necessarily it involves a loss of the opportunity to determine the
reasons for comorbidity, a search that could throw important light on aetiological mechanisms (see below).

DSM-III-R AND ICD-9

The two major psychiatric classification systems, the American Psychiatric Association’s (1987) DSM-III-R and the
World Health Organization’s (1978) ICD-9 (soon to be superceded by ICD-10-W.H.O., 1990), follow quite different
approaches to diagnosis (Rutter & Gould, 1985; Rutter, 1988). While the W.H.O. system allows multiple diagnoses,
it tends to discourage them by its adoption of a pattern approach to diagnosis. The clinician is expected to review
the overall clinical picture made up of history, signs, symptoms and laboratory findings; and then to match this
picture with the prototypical diagnostic pattern that fits best, using a process of pattern recognition. It follows the
long-standing medical tradition of recognizing that most disorders involve a complex admixture of specific and
nonspecific symptomatology and that, in ordinary circumstances, it is more likely in practice that a patient will have
one disease, rather than several. It provides the unifying explanation of protean symptomatology. Thus, there is
an appreciation that there are many nonspecific symptoms-such as fever, tachycardia, fatigue, headache and skin
rashes-that occur in many quite different somatic diseases. In much the same way, psychiatry has many nonspecific
symptoms such as anxiety, depressed mood, poor concentration and restlessness. Of course, a dilemma arises
from the fact that many of the nonspecific symptoms also constitute the hallmarks of specific diagnostic entities.
The problem lies in the difficulty of deciding when, say, depression is an indicator of a major depressive disorder
and when it is just an indication that psychopathologically something is the matter.
The strength of the W.H.O. approach is that the underlying concept is probably correct in many, perhaps most,
instances. Thus, in reality, it is unlikely that a high proportion of patients truly have three, four or five entirely
separate conditions. The outstanding weakness, however, is that, for many symptom patterns, the data are not
available to determine when and how to give precedence to one diagnosis over another (Rutter, 1988). In some
instances, this is dealt with by having a combination code. For example, it has long been recognized that there is a
need to code schizoaffective disorder (that applies to DSM-III-R as well, although the details of criteria are not quite
the same). More controversially, ICD-10 (W.H.O., 1989) has a code for depression combined with conduct disorder.
A further limitation of the W.H.O. approach (at least as exemplified in ICD-9; ICD-10 has come closer to DSM-III-R) is
that true patterns of comorbidity may be concealed and, therefore, neglected or that a wrong hierarchical principle
may lead to invalid diagnoses.
7

The APA classification DSM-III-R works the other way round (although its predecessor, DSM-III, included
more exclusionary hierarchies). Diagnoses are made on algorithms based on specified symptom constellations
without regard to the presence or absence of accompanying symptomatology of a different kind (apart from a few
exceptions). The consequence, perhaps the inevitable consequence, of this convention is that when a patient has
any one diagnosis, there is usually at least one other diagnosis as well. The first obvious disadvantage of this system
is that it contravenes common sense. Indeed, Weinstein, Stone, Noam, Grimes and Schwab-Stone (1989) found
that, in a child psychiatric in-patient unit sample, strict adherence to DSM-III rules (freed of hierarchies) led to 78%
of the patients showing comorbidity, but this fell to 20 % when clinicians were allowed to use their own judgement.
However, there are two other disadvantages. Although, in theory, a statistical system could be devised to present
data on all possible patterns of comorbidity, no system has been used that way to date. Moreover, even if such data
were produced, their complexity would make it extremely difficult, if not impossible, to make sense of the huge
number of possible double, treble, and quadruple combinations.
The second disadvantage is that although DSM-III-R is supposedly free of diagnostic hierarchies that conceal
comorbidity, in fact it includes a bewildering mix of inconsistencies on which combinations are, and which are not,
allowed. For example, in adults it is not allowed to diagnose overanxious disorder in the presence of generalized
anxiety disorder, but it is allowed to diagnose both separation anxiety disorder and overanxious disorder in
children. Or again, it is not allowed to diagnose both social phobia and avoidant disorder but it is possible to
diagnose both social phobia and agoraphobia. Similarly, oppositional disorder cannot be diagnosed with conduct
disorder but it can be diagnosed with attention deficit disorder. Evidently, there are more diagnostic hierarchies in
DSM-III-R than is usually appreciated.

DETECTION ARTIFACTS

Referral factors
Before considering possible reasons for different patterns of comorbidity, it is necessary to note the variety of
ways in which it can be produced artifactually. As already discussed, there is the Berkson (1946) effect by which,
for statistical reasons separate from referral biases, the comorbidity rate in clinic samples will always be greater
than that in the general population whenever only a small proportion of the conditions making up the comorbidity
pattern are referred to clinics-the state of affairs with the majority of child psychiatric disorders other than the most
severe (Rutter, Tizard & Whitmore, 1970). In addition, of course, referral biases may further distort clinic data on
comorbidity. For example, when a clinician is known to have a special interest in a particular pattern of comorbidity
such cases are more likely to be referred to him or her. Similarly, tertiary referral centres seeing difficult and
complicated cases are likely to have a disproportionately high level of comorbidity.

Screening and surveillance factors

It is important also to recognize that any general population epidemiological study that relies on screening
or surveillance procedures is also open to possible detection artifacts. Thus, it is a common practice to use high
scores on questionnaires designed to tap a broad range of behaviour as a means of picking out subjects with a
high probability of disorder, who may then be studied individually in greater detail (Newman, Shrout & Bland,
1990; Rutter, 1989a,b). The procedure works well for many purposes but it will tend to miss monosymptomatic
disorders and oversample children whose psychopathology include symptoms of many different types. A similar
detection bias will apply whenever the diagnosis of a second disorder is dependent in part on subjects with one
diagnosis being subjected to a closer degree of surveillance. This is particularly likely to operate in longitudinal
studies investigating patterns of comorbidity over time (rather than concurrently).
These detection artifacts are all open to systematic quantified investigation by means of comparisons with total
population unscreened samples.
8

NOSOLOGICAL CONSIDERATIONS

The usual concept of comorbidity implies the co-occurrence of two independent conditions or disorders. Even
if the statistical data have not been distorted by detection artifacts, the apparent overlap between supposedly
different disorders may not represent comorbidity as usually conceptualized. This possibility arises because the
basic nosological concepts may themselves be mistaken. As Bukstein, Brent and Kamliner (1989) pointed out, one
of the major difficulties in studying psychiatric comorbidity is the lack of well validated diagnostic criteria.

Categories or dimensions?
The first possibility is that the concept of disorder (or disease) categories may itself be misconceived. Instead,
as many psychologists have argued, it could be that psychopathology is best thought of as the end product of
an admixture of extremes of personality dimensions. According to this view, disorders involve no qualitative
discontinuity between abnormality and normality but rather a pattern resulting from quantitative variations on
a range of behavioural dimensions. In so far as that is the case, apparent comorbidity is bound to arise from the
inevitability of individuals with high scores on two or more dimensions; however, the extent of such apparent
comorbidity will be much affected by the particular cut-off points used to define ‘’disorder’’ and by the extent
to which the definitions of disorder involve truncation of dimensions. The same circumstances will arise if the
behavioural dimensions operate as risk factors for disorder. For example, numerous studies have shown the strong
overlap between attention deficit (hyperactivity) and conduct disorders (Szatmari et al., 1989; Taylor, 1988). Does
this imply comorbidity between two different disorders or rather does it mean that both inattention/ overactivity
and aggressivity are risk factors for disruptive behaviour?
There has been surprisingly little research that has set out to contrast and compare the validity of dimensional
and categorical approaches. However, one test would be to determine if behavioural dimensions related to one
diagnostic category functioned as a risk factor for the second condition at levels below the diagnostic threshold.
This was the approach followed by Robins and McEvoy (1990) in examining whether conduct problems before the
age of 15 years predicted substance abuse. Their results showed that they did so at all levels of severity, indicating
that conduct problems functioned as a dimensional predictor. Thus, of those with no conduct problems, 38 %
exhibited substance abuse; of those with one conduct disorder problem, 52 % did so; of those with two conduct
problems 66 % did so; and so on. At least in this population, with these two “disorders”, a dimensional approach
seemed to account for the findings better than comorbidity between two separate disorders. This research strategy
warrants greater usage.

Overlapping diagnostic criteria

A second type of nosological confusion arrives from the fact that the same item of behaviour appears in the list
of diagnostic criteria for several different diagnostic categories-a problem highlighted in previous considerations of
comorbidity (Pfeffer & Plutchik, 1989) and by the DSM-III Child and Adolescent Psychiatry Working Party (Shaffer
et al., 1989). For example, not surprisingly, anxiety/worrying appears in some form in the lists of criteria for all the
various supposedly separate anxiety disorders; just as depressed mood forms part of the criteria for dysthymia as
well as major depressive disorder. In addition, agitation is one of the criteria for anxiety, depression, and attention
deficit/hyperactivity disorder. This is not unreasonable as so many of the behaviours that define specific disorders
are also nonspecific indicators of psychopathology. Nevertheless, the fact that this is so will lead to a degree of
artifactual comorbidity. A related problem arises from the fact that so many mood states involve mixed emotions.
Thus, it is well known that anxiety and depression very commonly occur together regardless of the diagnosis
(Grayson, Bridges, Cook & Goldberg, 1990). In so far as an increasing severity of disorder is likely often to involve
an increasing number of nonspecific indices of psychopathology (Costello et al., 1988; Bird et al., 1988; Weissman,
Warner & Fendrich, 1990), and in so far as these form part of the sets of criteria for different disorders, there is
some danger that there will be an artifactual association between severity and extent of comorbidity. That is simply
because severe disorders with many symptoms are likely to have a greater chance of fulfilling the criteria for more
than one disorder. It is possible to provide a partial check on whether any severity-comorbidity association is real
or artifactual by determining whether the association holds when severity is defined in terms of degree of social
impairment rather than number of symptoms. A partial test of the influence on comorbidity of nonspecific indices
of psychopathology is also afforded by determining whether the correlates (with respect to features such as family
history, prognosis, treatment response etc.) of the behavioural item (e.g. depression or anxiety or restlessness) are
similar when it occurs as part of a mixed clinical picture to when it occurs as part of the pure syndrome it defines.
9

Artificial subdivision of syndromes

A somewhat similar problem may arise from the tendency to subdivide disorders defined in terms of one
main symptom complex into various subcategories according to particular elements or facets of that complex. For
example, anxiety disorders are subdivided into some dozen different syndromes characterized by the generality
of the anxiety (e.g. “overanxious disorder” or “generalized anxiety disorder”) or its specific focus (e.g. “separation
anxiety disorder” or “social phobia”) or the presence/absence of some particular feature (e.g. agoraphobia with or
without panic), or its association with some stresses (e.g. “post-traumatic stress disorder” or “adjustment disorder
with anxious mood”). So far as children are concerned, there is the additional problem that sometimes there are
two disorders defined in clearly similar terms, one of which is intended to apply to all age groups and one of which is
supposedly particular to childhood, but which are not mutually exclusive under the age of 18 years (e.g. “overanxious
disorder” and “generalized anxiety disorder”). Not surprisingly, numerous studies (see Barlow, 1988) have shown
extensive comorbidity between these various anxiety disorders. Most published investigations have concerned
adult patients but the same situation seems to apply in childhood (Last, Hersen, Kazdin, Finkelstein & Strauss,
1987). Thus, in the study cited, half the children with separation anxiety disorder also had overanxious disorder and
95% of this comorbid group had at least one other diagnosis as well! There have been some attempts to test the
discriminant validity of these different anxiety disorders (see Barlow, 1988) with some modest success with respect
to at least some of the differentiations. For example, Mannuzza, Fyer, Liebowitz and Klein (1990) have reviewed the
evidence suggesting that social phobia is meaningfully distinctive from panic disorder and agoraphobia in adults.
However, they also point out that the meaningful interpretation of symptoms requires that they can be considered
in the individual context (something that is difficult with a symptom check list approach to diagnosis). Less is known
about the validity of differentiations among anxiety disorders in childhood. Last et al. (1987) found that children
with separation anxiety disorders tended to be somewhat younger than those with overanxious disorders but
this could reflect age effects on patterns of manifestation rather than a difference between two distinct disorders.
Nothing is known on whether there is any meaningful distinction between ‘’ overanxious disorder’’ and ‘’ generalized
anxiety disorder’’ and the comorbidity between them seems to have received no attention, perhaps because child
psychiatrists tend to choose the category in the children’s section of DSM-III-R without checking to see whether the
all-ages diagnosis might equally apply.
However, the issue of comorbidity cannot be dealt with merely by testing the discriminant validity of the
separate anxiety categories; it is necessary to go on to examine the characteristics of the comorbid groups as they
relate to the “pure” diagnoses. For example, it could be that it is useful to separate simple phobias that arise in
the absence of generalized anxiety from generalized anxiety disorder but, equally, it might well be the case that
high levels of general anxiety tend also to lead to various focused phobias as well. At least, the possibility needs
testing. The very high level of comorbidity between different anxiety disorders suggests that some of it represents
nosological confusion.
It may well be clinically useful to be able to note that an anxiety disorder has several different facets but, if these
do indeed represent varied aspects of the same basic disorder, it seems misleading to view them as examples of
comorbidity. Clearly, further research is needed to bring better order into the confusing nosological territory of
anxiety disorders.

One disorder represents an early manifestation of the other

A further possibility is that one disorder constitutes an early manifestation of the other. When that is the
explanation, it may be desirable to code these manifestations separately because the stage of the disorder has
important clinical implications. But it would make no sense to regard transitional phases with both early and later
manifestations as representing comorbidity. There are well known examples in medicine where distinct stages of a
disease are recognized; for example, primary, secondary and tertiary syphilis or the differentiation between cervical
carcinoma and precancerous cervical dyskaryosis or dysplasia. In child psychiatry, there are several disorders in
which it has been suggested that they represent early manifestations of some other diagnosis. Thus, oppositional
disorder is a syndrome mainly diagnosed in younger children and which often seems to be a precursor of conduct
disorder (Loeber & Lahey, 1989). Similarly, separation anxiety tends to be diagnosed in younger children and it has
been suggested that it may represent an early manifestation of overanxious disorder (Last et al., 1987). Or again,
conduct disorder in childhood is an established precursor of antisocial personality disorder in adult life (Robins,
1978).
10

To test this type of disorder progression hypothesis it is crucial to have longitudinal data (Loeber, 1990). Thus,
if condition A is a precursor of B, it must be the case that the presence of A at time 1 increases the likelihood of B
at time 2; and that B never precedes A. However, equally, it is to be expected that only some cases of A will develop
into Band, if there is more than one precursor ofB, there may be instances of B that have not been preceded by A.
To date, although there are some data suggesting the plausibility of hypotheses on progression, decisive testing
has yet to be undertaken.

One disorder is part of the other

It may also be suggested that one disorder is part of or a secondary manifestation of the other conditions. For
example, DSM-III-R specifies that if there is a pervasive developmental disorder, neither attention-deficit hyperactivity
disorder, nor pica, nor overanxious disorder can be diagnosed as well. Presumably this is on the basis of the fact
that the symptoms characteristic of these other disorders are so very frequently part of the symptomatology of
autism. A similar exclusion applies to generalized anxiety that occurs only during the course of a mood disorder.
There are several ways in which this hypothesis may be tested. For example, if the comorbid disorders are episodic
in nature, it is possible to determine whether the remission and recurrence of the one are associated in time
with those of the other and, especially, if the administration of a treatment that is specific to the one leads to the
loss of the symptoms of the other disorder. This approach has been followed in the case of the co-occurrence of
nocturnal enuresis and other child psychiatric disorders-with largely negative findings (Shaffer, 1973, 1985). It has
also been adopted to examine the comorbidity of depression and conduct disorder (Puig-Antich, 1982; Puig-Antich,
Lukens, Davies, Goetz & Todak, 1985), with inconclusive mixed findings. Unfortunately, this strategy is weakened
by the weakness and/or nonspecifity of so many treatments; thus, antidepressant medication is not particularly
effective in the treatment of depression in children and adolescents (Puig-Antich et al., 1987; Ryan et al., 1986) and
tricyclic drugs have a wide range of actions that extend far beyond those on mood. A further limitation is that the
two constellations of symptoms may represent alternative manifestations of one disorder; variable expression
is a well recognized feature of many genetic disorders (as illustrated, for example, by neurofibromatosis). This
possibility may be examined through the use of genetic research strategies. There are many examples of the
use of family studies for this purpose; for example to examine the association between anorexia nervosa and
depression (Strober, Lampert, Morrell, Burroughs & Jacobs, in press) or that between alcoholism and depression
(Merikangas et al., 1985, 1988)-in both cases with evidence suggesting independent transmission. However, on
their own, family data cannot make a satisfactory differentiation between genetic and environmental mechanisms.
Adoptee and twin designs of various kinds (Rutter et al., 1990) are effective for this purpose and may be used to
examine comorbidity. For example, Holland, Hall, Murray, Russell and Crisp (1984) found no tendency for affective
disorders to occur in the cotwins of subjects with anorexia nervosa and Cadoret, Troughton, Moreno and Whitters
(1990) found no direct genetic connection between alcohol and antisocial problems in biological relatives and
depressive symptomatology in adopted-away subjects. It would be helpful to make greater use of genetic research
strategies in the study of comorbidity in child psychiatry.
11

POSSIBLE EXPLANATIONS OF TRUE COMORBIDITY

The explanations considered this far all involve a mechanism by which the apparent comorbidity represents
some type of artifact so that the real situation does not involve the co-occurrence of two or more truly separate
and independent conditions. As discussed, that does not mean that the findings should be dismissed; quite often
elucidation of the meaning of the apparent comorbidity should throw light on the nature of the disorders involved.
However, that is even more the case when there is true comorbidity. Several rather different underlying processes
need to be considered.

Shared risk factors

One possible reason for overlap between two disorders is that they share the same risk factor or factors. This
possibility arises from the fact that many psychiatric disorders are multifactorial in origin and that many causal
factors are not diagnosis-specific. For example, there are various extreme temperamental traits or constellations
of traits that are associated with a range of psychiatric disorders as well as with learning difficulties (Kohnstamm,
Bates & Rothbart, 1990). Thus, it has been suggested that temperamental variables such as overactivity, short
attention span and impulsivity might account for the well established comorbidity between conduct disorders and
reading difficulties (Yule & Rutter, 1985). Family adversity might also operate in the same way (Offord, Poushinksy
& Sullivan 1978; Richman, Stevenson & Graham, 1982), because large family size and social disadvantage carry an
increased risk for both conditions (Rutter & Giller, 1983; Sturge, 1982). However, it is not known whether the shared
risk factors mechanism does in fact account for this (or any other) pattern of comorbidity. What are needed are
investigations in which comorbidity is examined before and after partialling out, or statistically taking account of,
shared risk factors.

Overlap between risk factors

A variant of this mechanism is provided by the possibility that, even when the risk factors for two disorders are
distinct and different, there may be comorbidity because the risk factors themselves are associated. When that is
the case, the individual may be at risk for two separate conditions with the risk mechanisms for each independent,
but co-occurring. For example, parental depression constitutes a risk factor for a range of different child psychiatric
disorders (Rutter, 1989b). Particular attention has been focused on major depressive disorders in the offspring with
the suggestion that they are genetically mediated (Weissman et al., 1987; Weissman, 1988). However, there is also
an increased risk of conduct disorder that seems to be a function of family discord, which is much more frequent
when one or both parents are depressed (Rutter & Quinton, 1984). Thus, it might be suggested that the comorbidity
between depression and conduct disorder could arise, at least in part, because parental depression is associated
with a genetic risk for depression in the offspring and an environmentally mediated risk (when there is associated
discord) for conduct disorders. In order to test this hypothesis, it would be necessary to examine comorbidity after
taking account of these two routes to disorder in the child; thus there should be no comorbidity when there is
family harmony and cohesion if the postulated mechanisms are operating in the way hypothesized.
Chiles, Miller and Cox (1980) provided data that are consistent with the risk factor overlap concept. Delinquent
adolescents with and without comorbid depression were compared. The two groups could not be differentiated
on the basis of risk factors for delinquency (a family history of divorce, parental death, child abuse and incest) but
the depressed delinquents did differ in being more likely to have factors thought to constitute risk variables for
depression (depression or alcoholism in a first degree relative). Thus, it could be argued that depressed delinquents
had the risk factors for both disorders whereas the nondepressed delinquents had them for only one. It is of
importance to notice here that assortative mating may also be another mechanism by which two disorders are
transmitted to the offspring.

The comorbid pattern constitutes a meaningful syndrome

The basic assumption that appears to underly the DSM-III-R policy of making diagnoses on the presence
of particular constellations of symptoms irrespective of the presence or absence of other constellations is that
comorbidity does not alter the meaning of any of the diagnoses involved in the comorbidity pattern. That assumption
may be tested by comparing comorbid and “pure” diagnoses on features such as course or family history or
responses to treatment that might reflect diagnostic meaning. This strategy has been used relatively infrequently
but there are two examples of syndromes where the few available data seem to negate the assumption. There is a
12

large body of data demonstrating that stimulant drugs usually bring about major, sometimes dramatic, short-term
benefits in children with attention deficit hyperactivity disorders (ADHD) (Klein, 1987). Taylor et al. (1987) showed
that the benefits were not evident when the system pattern included marked anxiety; and Pliszka (1989) found
that children with ADHD comorbid with anxiety had no significant benefit from methylphenidate in a double blind
control trial (whereas those without anxiety showed the usual good drug response). The findings from both these
studies are striking for two separate reasons. First, as they showed, comorbidity with conduct disorder had no effect
on drug response, so that this is not a nonspecific effect of comorbidity of any type. Second, the co-occurrence of
anxiety disorder removed one of the most characteristic features of one of the two disorders in the comorbid
pattern. The implication seems to be either that the comorbid pattern constitutes a variety of anxiety disorder
(systematic comparisons with “pure” anxiety disorders are needed to test the hypothesis) or that it constitutes a
meaningfully distinctive syndrome in its own right. Either way, the implication is that the comorbidity has altered
the meaning to be attached to ADHD and, therefore, that it warrants separate coding in the classification system.
A second example is provided by the comorbid pattern of depression and conduct disorder in childhood. The
follow-up into adult life undertaken by Harrington, Fudge, Rutter, Pickles & Hill (in press) showed that the risk
of adult criminality associated with conduct disorder was unaffected by the presence or absence of comorbid
depression. However, the risk of major depressive disorder in adult life associated with childhood depression was
affected by comorbid conduct disorder; the low risk in the comorbid group did not differ from that in the group
without depression. Puig-Antich et al. (1989) also found that the comorbid group differed from children with a’
‘pure’’ depressive disorder with respect to family history of major depressive disorder. This difference in familiality
applied only to the comorbidity with conduct disorder and not to that with separation anxiety. The findings to
date are too sparse to warrant firm conclusions and in any case do not differentiate between the alternative
hypotheses that comorbid depression and conduct disorder constitute a meaningfully different syndrome and that
the depression in the comorbid group is secondary to or part of the conduct disorder. Again, however, it seems that
the comorbid pattern may change the meaning of the depressive disorder; it needs to be identified as a separate
group and studied further.

One disorder creates an increased risk for the other

A further possible explanation for comorbid patterns is that one disorder creates an increased risk for the
other. For example, Cadoret et al. (1990), using an adoptee design, showed that an adult diagnosis of antisocial
personality or substance abuse was associated with a four-fold increase in risk for depressive symptomatology
even though the genetic origins of the two disorders were distinct. Robins and McEvoy (1990), using retrospective
data from the Epidemiological Catchment Area Study, showed that conduct disorder in childhood was a powerful
predictor of later substance abuse. The data were not such as to allow determination of the extent to which these
were truly separate disorders. However, the evidence suggested that the mechanisms involved both duration and
timing of exposure to illicit drugs, plus, probably, the generation of psychosocial stressors and adversity that serve
as risk factors for heavy use of psychoactive substances.
There is an extensive literature on the effects of acute and life stressors as risk factors for psychiatric disorder
in both childhood and adult life (Brown & Harris, 1989; Goodyer, 1990). However, less attention has been paid
to the origins of these stressors (Champion, 1990; Rutter, 1986). There is increasing evidence that people shape
and select their own environments and that various psychiatric disorders play a part in generating stress and
adversity. For example, follow-up studies of children with conduct disorder have shown markedly increased rates
of unemployment and marital breakdown in adult life (Robins, 1966, 1986), both well established risk factors for
depressive disorders. Neuroticism has also been found to be associated with an increased likelihood of marital
breakdown (Kelly & Conley, 1987).
It will be appreciated that an adequate testing of this causal chain hypothesis requires several steps. Clearly, a
start is provided by longitudinal time sequence data showing that condition A both precedes and is associated with
an increased risk for condition B; whereas the reverse does not apply. However, as clearly noted, this could simply
mean that A and B constitute different stages in the progression of a single disorder. Data showing that comorbid A
+ B has correlates that approximate those of ‘’pure’’ A, but not those of “pure” B,indicate that B cannot have caused
A (a strategy used to examine the comorbidity of conduct disorder and reading retardation-Rutter et al., 1970) but,
although consistent with the hypothesis that A caused B the findings are open to other interpretations.
In particular, this pattern of results indicates that the meaning of “pure” B and comorbid A + B must be
somewhat different. What are needed are data showing that the origins of “pure” B and comorbid A + B involve
the same risk factors but that condition A generates those risk factors. In other words, it is necessary to test the
several links in the postulated causal chain. This full research strategy has yet to be pursued in the investigation of
comorbidity in child psychiatric disorders.
13

CONCLUSIONS

Many studies document the pervasiveness of patterns of comorbidity in child, as well as adult, psychiatry.
However, it is clear that the extent of comorbidity can be assessed only through the use of fully representive
epidemiological data. That is primarily because calculation of the chance expectation of comorbidity requires data
on the general population base rates of each of the conditions involved in the comorbidity. In addition, the Berkson
effect and a variety of referral biases make the use of clinic data hazardous and potentially misleading.
Until relatively recently, comorbidity has received little attention in the research literature and considerations
of psychiatric classification issues have generally ignored the extreme frequency with which psychiatric disorders
seem to co-occur. This neglect has had two unfortunate consequences. First, the findings of many studies of one
specified psychiatric condition are likely to be misleading because the correlates of the disorder being investigated
may represent the correlates of some unspecified comorbid condition. Second, there is the unsafe assumption that
the meaning of any given disorder is exactly the same regardless of the presence or absence of other · disorders.
It is sometimes claimed that the DSM-III-R classification system avoids the difficulties associated with diagnostic
hierarchies jut because it allows any level of comorbidity. However, it is evident that it includes more hierarchies
than generally appreciated and that the encouragement of comorbid diagnoses introduces further problems.
In this paper we have outlined some of the detection artifacts that may produce a false picture of comorbidity
and have gone on to discuss some of the nosological considerations that apply to comorbidity. These include:
the concepts of disorders as categories or dimensions; overlapping diagnostic criteria; artificial subdivisions of
syndromes; one disorder representing an early manifestation of the other; and one disorder being part of the
other. Finally, we have considered some of the possible explanations of true comorbidity in terms of: shared risk
factors; overlapping risk factors; comorbid patterns constituting a meaningful syndrome; and one disorder creating
an increased risk for the other. In each case, we have put forward some general guidelines for testing alternative
hypotheses. It is evident that it may be quite difficult to differentiate between competing explanations but equally it
is clear that an improved understanding of the varied mechanisms underlying comorbidity should shed important
light on the processes involved in the genesis and contribution of psychiatric disorders.

Acknowledgement - Dr Caron was supported by Fellowship No. 891069 from Fonds de la Recherche en Santé du
Québec.
14

REFERENCES

American Psychiatric Association (1987). Diagnostic and statistical manual of mental disorders (DSM-III) (3rd edn).
Washington, DC: Author.
Anderson,]. C., Williams, S., McGee, R. & Silva, P.A. (1987). DSM-III disorders in preadolescent children. Prevalence
in a large sample from the general population. Archives of General Psychiatry, 44, 69-78.
Barlow, D. H. (1988). Anxiery and its disorders. The nature and treatment of anxiery and panic. New York: Guilford.
Berkson, J. (1946). Limitations of the application of fourfold table analysis to hospital data. Biometrics,
2, 47-53.
Bird, H., Canino, G., Rubio-Stipec, M., Gould, M. S., Ribera, J., Sesman, M., Woodbury, M., Huertas-Goldman, S.,
Pagan, A., Sanchez-Lacay, A. & Moscoso, M. (1988). Estimates of the prevalence of childhood maladjustment in
a community survey in Puerto Rico. Archives of General Psychiatry, 45, 1120-1126.
Boyd, J. H., Burke, J. D., Gruenberg, E., Holzer, C. E., Rae, D. S., George, L. K., Karno, M., Stoltzman, R., McEvoy, L. &
Nestadt, G. (1984). Exclusion criteria of DSM-III. Archives of General Psychiatry, 41, 983-989.
Brown, G. & Harris, T. (Eds) (1989). Life events and illness. New York: Guilford.
Bukstein, 0. G., Brent, D. A. & Kaminer, Y. (1989). Comorbidity of substance abuse and other psychiatric disorders
in adolescents. American Journal of Psychiatry, 146, 1131-1141.
Cadoret, R., Troughton, E., Moreno, L. & Whitters, A. (1990). Early life psychosocial events and adult affective
symptoms. In L. Robins & M. Rutter (Eds), Straight and devious pathways.from childhood to adulthood (pp. 300-
313). New York: Cambridge University Press.
Champion, L. (1990). The relationship between social vulnerability and the occurrence of severely threatening life
events. Psychological Medicine, 20, 157-161.
Chiles, J. A., Miller, M. L. & Cox, C. B. (1980). Depression in an adolescent delinquent population. Archives of General
Psychiatry,37, 1179-1184.
Costello, E.J., Costello, A.J., Edelbrock, C., Burns, B.J., Dulcan, M. K., Brent, D. &Janiszewski, S. (1988). Psychiatric
disorders in pediatric primary care. Archives of General Psychiatry, 45, 1107-1116.
Flament, M. F., Whitaker, A., Rapaport,]. L., Davies, M., Zaremba Berg, C., Kalikow, K., Sceery, W. & Shaffer, D. (1988).
Obsessive compulsive disorder in adolescence: an epidemiological study. Journal of the American Academy of
Child Psychiatry, 27, 764-771.
Goodyer, I. M. (1990). Life experiences, development and childhood psychopathology. Chichester: Wiley.
Grayson, D., Bridges, K., Cook, D. & Goldberg, D. (1990). The validity of diagnostic systems for common mental
disorders: a comparison between the ID-CATEGO and the DSM-III systems. Psychological Medicine, 20, 209-218.
Harrington, R., Fudge, H., Rutter, M., Pickles, A. & Hill,J. (in press). Adult outcome of childhood and adolescent
depression: II. Links with antisocial disorders. Journal of the American Academy of Child and Adolescent Psychiatry.
Holland, A.J., Hall, A.,Murray, R.,Russell, G.F.M. & Crisp, A.H.(1984). Anorexianervosa: a study of 34 twin pairs and
one set of triplets. British Journal of Psychiatry, 145, 414-419.
Jaspers, K. (1963) General psychopathology. Manchester: Manchester University Press.
Kashani, J. H., Beck, N. C., Hoeper, E. W., Fallahi, C., Corcoran, C. M., McAllister, J. A., Rosenberg, T. K. & Reid, J. C.
(1987). Psychiatric disorders in a community sample of adolescents. American Journal of Psychiatry, 144, 584-
589.
Kelly, E. L. & Conley, J. J. (1987). Personality and compatability: a prospective analysis of marital stability and marital
satisfaction. Journal of Personality and Social Psychology, 52, 27-40.
Kendell, R. E. (1975). The role of diagnosis in psychiatry. Oxford: Blackwell Scientific.
Klein, R. G. (1987). Pharmacotherapy of childhood hyperactivity: an update. In H. Y. Meltzer, W. Bunney, J. Coyle,
K. Davies, I. Kopin, C. R. Schuster, R. I. Shader & G. Simpson (Eds), Psychopharmacology: the third generation of
progress (pp. 1215-1224). New York: Raven Press.
Kohnstamm, G. A., Bates,J. E. &Rothbart, M. K. (1990). Temperament in childhood. Chichester: Wiley.
Last, C. G., Hersen, M., Kazdin, A. E., Finkelstein, R. & Strauss, C. C. (1987). Comparison of DSM-III separation anxiety
and overanxious disorders: demographic characteristics and patterns of comorbidity. Journal of the American
Academy of Child and Adolescent Psychiatry, 26, 528-531.
Loeber, R. (1990). Development and risk factors of juvenile antisocial behavior and delinquency. Child Psychology
Review, 10, 1-141.
15

Loeber, R. & Lahey, B. B. (1989). Recommendations for research on disruptive behavior disorders of childhood and
adolescence. In B. B. Lahey & A. E. Kazdin (Eds), Advances in clinical child psychology (Vol. 12, pp. 221-257). New
York: Plenum Press.
Mannuzza, S., Fyer, A. J., Liebowitz, M. R. & Klein, D. F. (1990). Delineating the boundaries of social phobia: its
relationship to panic disorder and agoraphobia. Journal of Anxiery Disorders, 4, 41-59.
Merikangas, K. R., Leckman, J. F., Prusoff, B. A., Pauls, D. L. & Weissman, M. M. (1985). Familial transmission of
depression and alcoholism. Archives of General Psychiatry, 42, 367-372.
Merikangas, K. R., Prusoff, B. A. & Weissman, M. M. (1988). Parental concordance for affective disorders:
psychopathology in offspring. journal of Affective Disorders, 15, 279-290.
Newman, S. C., Shrout, P. E. & Bland, R. C. (1990). The efficiency of two-phase designs in prevalence surveys of
mental disorders. Psychological Medicine, 20, 183-193.
Offord, D. R., Poushinsky, M. F. & Sullivan, K. (1978). School performance, IQ and delinquency. British Journal of
Criminology, 18, 110-127.
Pfeffer, C.R. & Plutchik, R. (1989). Co-occurrence of psychiatric disorders in child psychiatric patients and nonpatients:
a circumplex model. Comprehensive Psychiatry, 30, 275-282.
Pliszka, S. (1989). Effect of anxiety on cognition, behavior, and stimulant response in ADHD. Journal of the American
Academy of Child and Adolescent Psychiatry, 28, 882-887.
Puig-Antich, J. (1982). Major depression and conduct disorder in prepuberty. Journal of the American Academy of
Child Psychiatry, 21, 118-128.
Puig-Antich, J., Goetz, D., Davies, M., Kaplan, T., Davies, S., Ostrow, L., Asnis, L., Twomey, J., Lyengar, S. & Ryan,
N. D. (1989). A controlled family history study of prepubertal major depressive disorder. Archives of General
Psychiatry, 46, 406-418.
Puig-Antich, J., Lukens, E., Davies, M., Goetz, D. & Todak, G. (1985). Psychosocial functioning in prepubertal major
depressive disorders: II. Interpersonal relationships after sustained recovery from the affective episode.
Archives of General Psychiatry, 42, 511-517.
Puig-Antich, J., Perels, J.M., Chambers, W. J., Tabrizi, M.A., King, J., Goetz, R., Davies, M. & Stiller, R. L. (1987). Imipramine
in prepubertal major depressive disorders. Archives of General Psychiatry, 44, 81-89.
Richman, N., Stevenson, J. & Graham, P. J. (1982). Pre-school to school: a behavioural study. London: Academic Press.
Robins, L. (1966). Deviant children grown up. Baltimore: Williams & Wilkins.
Robins, L. N. (1978). Sturdy childhood predictors of adult antisocial behavior: replications from longitudinal studies.
Psychological Medicine, 8, 611-622.
Robins, L. N. (1986). The consequences of conduct disorder in girls. In D. Olweus, J. Block & M. Radke-Yarrow (Eds),
Development of antisocial and prosocial behavior: research, theories and issues (pp. 385-414). New York: Academic
Press.
Robins, L. N. & McEvoy, L. (1990). Conduct problems as predictors of substance abuse. In L. Robins & M. Rutter (Eds),
Straight and devious pathways from childhood to adulthood (pp. 182-204). New York: Cambridge University Press.
Rutter, M. (1986). Meyerian psychobiology, personality development and the role of life experience. American
Journal of Psychiatry, 143, 1077-1087.
Rutter, M. (1988). DSM-III-R: a postscript. In M. Rutter, A. H. Tuma & I. S. Lann (Eds), Assessment and diagnosis in child
psychopathology (pp. 453-464). New York: Guilford.
Rutter, M. (1989a). Isle of Wight revisited: twenty-five years of child psychiatric epidemiology. Journal of the American
Academy of Child and Adolescent Psychiatry, 28, 633-653.
Rutter, M. (1989b). Psychiatric disorder in parents as a risk factor for children. In D. Shaffer, J. Philips & N. B. Enzer
(Eds), with M. M. Silverman & V. Anthony (Associate Eds), Prevention of mental disorders, alcohol and other drug
use in children and adolescents (pp. 157-189). OSAP Prevention Monograph 2. Rockville, MD: Office for Substance
Abuse Prevention, U.S. Department of Health and Human Sciences.
Rutter, M. & Giller, H. (1983). Juvenile delinquency: trends and perspectives. Harmondsworth: Penguin.
Rutter, M. & Gould, M. (1985). Classification. In M. Rutter & L. Hersov (Eds), Child and adolescent psychiatry: modern
approaches (pp. 304-321). Oxford: Blackwell Scientific.
Rutter, M., MacDonald, H., Le Couteur, A., Harrington, R., Bolton, P. & Bailey, A. (1990). Genetic factors in child
psychiatric disorders-II. Empirical findings. Journal of Child Psychology and Psychiatry, 31, 39-83.
Rutter, M. & Quinton, D. (1984). Parental psychiatric disorder: effects on children. Psychological Medicine, 14, 853-
880.
Rutter, M., Tizard, J. & Whitmore, K. (Eds) (1970). Education, health and behavior. London: Longman (reprinted 1981,
Melbourne, FL: Krieger).
16

Ryan, N. D., Puig-Antich, J., Cooper, T., Rabinovich, H., Ambrosini, P., Davies, M., Torres, D. & Fried, J. (1986) Imipramine
in adolescent major depression: plasma level and clinical response. Acta Paediatrica Scandinavica, 73, 275-288.
Shaffer, D. (1973). The association between enuresis and emotional disorder: a review of the literature. In I. Kolvin,
R. C. MacKeith & S. R. Meadow (Eds), Bladder control and enuresis (pp. 118-136). London: Heinemann.
Shaffer, D. (1985). Enuresis. In M. Rutter & L. Hersov (Eds), Child and adolescent psychiatry: modern approaches (2dn
edn) (pp. 465-481). Oxford: Blackwell Scientific.
Shaffer, D., Campbell, M., Cantwell, D., Bradley, S., Carlson, G., Cohen, D., Denckla, M., Frances, A., Garfinkel, B.,
Klein, R., Pincus, H., Spitzer, R. L., Volkmar, F. & Widiger, T. (1989). Child and adolescent psychiatric disorders
in DSM-IV: issues facing the work group. Journal of the American Academy of Child and Adolescent Psychiatry, 28,
830-835.
Shepherd, M., Oppenheim, B. & Mitchell, S. (1971). Childhood behavior and mental health. London: University of
London Press.
Strober, M., Lampert, C., Morrell, W., Burroughs, J. & Jacobs, C. (in press). A controlled family study of anorexia
nervosa: evidence of familial aggregation and lack of shared transmission with affective disorders. International
Journal of Eating Disorders.
Sturge, C. (1982). Reading retardation and antisocial behaviour. Journal of Child Psychology and Psychiatry, 22, 21-31.
Szatmari, P., Boyle, M. & Offord, D. R. (1989). ADDH and conduct disorder: degree of diagnostic overlap and
differences among correlates. Journal of the American Academy of Child and Adolescent Psychiatry, 28, 865-872.
Taylor, E. (1988). Attention deficit and conduct disorder syndromes. In M. Rutter, A. H. Tuma & I. S. Lann (Eds),
Assessment and diagnosis in child psychopathology (pp. 377-407). New York: Guilford.
Taylor, E., Schachar, R., Thorley, G., Wieselberg, H. M., Everitt, B. & Rutter, M. (1987). Which boys respond to
stimulant medication? A controlled trial of methylphenidate in boys with disruptive behaviour. Psychological
Medicine, 17, 121-143.
Weinstein, S. R., Stone, K., Noam, G. G., Grimes, K. & Schwab-Stone, M. (1989). Comparison of DISC with clinicians’
DSM-III diagnoses in psychiatric inpatients. Journal of the American Academy of Child and Adolescent Psychiatry,
28, 53-60.
Weissman, M. M. (1988). Psychopathology in the children of depressed parents: direct interview studies. In D. L.
Dunner, E. S. Gehrson &J. E. Barret (Eds), Relatives at risk for mental disorder (pp. 143-159). New York: Raven
Press.
Weissman, M. M., Gammon, G.D., John, K., Merikangas, K. R., Warner, V., Prusoff, B. A. & Sholomskas, D. (1987).
Children of depressed parents. Increased psychopathology and early onset of major depression. Archives of
General Psychiatry, 44, 847-853.
Weissman, M. M., Warner, V. & Fendrich, M. (1990). Applying impairment criteria to children’s psychiatric diagnosis.
Journal of the American Academy of Child and Adolescent Psychiatry, 29, 789-795.
World Health Organization (1978). International classification of diseases (9th revision). Geneva: World Health
Organization.
World Health Organization (1990). ICD-10 1990 Draft of Chapter V, Categories F00-F99, Mental and Behavioural
Disorders (including disorders of Psychological Development). Clinical Descriptions and Diagnostic Guidelines.
(WHO/MNH/MEP/87.1, Rev.4). Geneva: World Health Organization, Division of Mental Health.
Yule, W. & Rutter, M. (1985). Reading and other learning difficulties. In M. Rutter & L. Hersov (Eds), Child and
adolescent psychiatry: modem approaches (2nd edn) (pp. 444-464). Oxford: Blackwell Scientific.
17

This document is a scanned copy of a printed document. No warranty

is given about the accuracy of the copy. Users should refer to the

original published version of the material.

1.1