Stability and Natural History of DSM-III Childhood Diagnoses

10
Stability and Natural History of DSM-III Childhood Diagnoses DENNIS P. CANTWELL, M.D., AND LORIAN BAKER, PH.D. Abstract. Follow-up or natural history outcome data for various DSM-lIl child and adolescent psychiatric diagnoses are presented.The data are relevant not only to our understandingof the specific disorders but also to the validityof the DSM-llI diagnostic categories. "Semi-blind" psychiatric evaluationsof 151 children were made as they presented to a community speech/language clinic and again approximately 4 years later. The follow-up data revealed highstabilityfor only three diagnoses: infantileautism, attention deficitdisorderwith hyperactivity, and oppositionaldisorder. The data revealed that severalof the DSM-Ill subcategories lacked predictivevalidity. This wastrue for the distinctionsbetweenattention deficitdisorderwith versuswithout hyperactivity; and between avoidant, separationanxiety,and overanxiousdisorders. Surprisingly lowstabilitywasfound for conduct disorder diagnosesas were surprisingly poor prognoses for parent-childproblems and adjustment disorders. J. Am. Acad. Child Adolesc. Psychiatry, 1989, 28, 5:691-700. Key Words: DSM-lIl, prognosis, validity, attention deficit disorder, conduct disorder. Little is known about the natural history (or outcome) of many of the childhood and adolescent psychiatric disorders. Such data are not only necessary to our understanding of the various psychiatric disorders, but also relevant to the estab- lishment of a psychiatric diagnostic classification system. In a recent publication in the Journal, Cantwell and Baker (I988a) discussed the criteria (including reliability, validity, and feasibility) for a successful psychiatric diagnostic classifi- cation system. One of the more important criteria is that the diagnostic categories have external validity, that is, that they can be verified as unique categories by external data. As outlined elsewhere (Cantwell, 1975), various types of data are relevant to the external validity of psychiatric diagnostic cat- egories. These include epidemiological data, family aggrega- tion data, biological and laboratory data, psychosocial data, natural history and outcome data, and response to treatment studies. The present paper presents data of one such type, namely, natural history and outcome data, for various DSM- III childhood and adolescent diagnostic categories. In particular, this paper presents data from a prospective 4- to 5-year follow-up study of a cohort of children drawn from a large community speech and hearing clinic. Six hundred children were seen initially when they presented for assess- ment of speech and/or language disorders. All were initially under the age of 16 and without significant hearing impair- ment. The results of the initial psychiatric assessment of these children, presented in the Journal and elsewhere (Baker and Cantwell, 1982a, b; Cantwell and Baker, 1980; Cantwell et aI., 1979, 1981) revealed a psychiatric illness rate of approxi- mately 50%. The preliminary results of follow-up evaluations of 300 of these children were presented recently in the Journal (Baker and Cantwell, 1987). It was found that the overall prevalence of psychiatric disorder and the overall prevalence of certain Accepted April 4. 1989. Dr. Cantwell is Joseph Campbell Professor ofChild Psychiatry and Director of Residency Training in Child Psychiatry at the UCLA Neuropsychiatric Institute. Dr. Baker is a Research Psycholinguist also at the UCLA Neuropsychiatric Institute (760 Westwood Plaza, Los Angeles, CA 90024), where reprints may be requested. 0890-8567/89/2805-0691 $02.00/0© 1989 by the American Acad- emy of Child and Adolescent Psychiatry. specific diagnoses had increased significantly between the initial and follow-up evaluations. However, the specific out- comes for the various initial psychiatric diagnoses were not examined. The present paper examines 151 of those children who initially had various DSM-III childhood psychiatric diagnoses and traces the stability of the various diagnoses found. These data are relevant to the external validity of these DSM-III diagnostic categories. Method The methodology of the studies has been described in previous publications (Baker and CantwelI, 1982a, b, 1987; Cantwell and Baker, 1980, 1985; Cantwell et aI., 1979, 1981). Both initial and follow-up evaluations consisted of speech/ language assessment, academic achievement testing, and in- tellectual testing (administered by L.B.) and psychiatric as- sessment by a board-certified child psychiatrist (D.P.C. or others in initial study; D.P.C. alone in the follow-up study). Psychiatric diagnoses were made initially and at follow-up using DSM-III diagnostic criteria and based on data collected with four types of instruments: parent and child interviews (the DICA, Orvaschel, 1985) and parent and teacher behavior rating scales (two forms of each: Conners, 1973; Rutter et aI., 1970). The reliability of the psychiatric diagnosis procedure was tested at the time of the initial study; it was 96% for the presence or absence of psychiatric disorder, and 94% for the specific diagnoses (Cantwell and Baker, 1985). The same procedures were used for the folIow-up evaluation, but they were not retested for reliability at that time. The follow-up psychiatric diagnoses were made "semi- blind" to the initial psychiatric diagnosis. (That is, the clini- cian did not have the initial evaluation materials and diagnosis available to him. Because of the large number of children in the initial study, most were not remembered by the time of follow-up; however, in a few outstanding cases the child's initial visit and diagnosis were remembered). Ninety percent of the entire initial sample was seen for follow-up. However, this present paper reports only on 150 cases who had certain specific psychiatric diagnoses initially. The initial diagnoses, and age and sex distributions for these cases are summarized in Table I below. 691

Transcript of Stability and Natural History of DSM-III Childhood Diagnoses

Stability and Natural History of DSM-III Childhood Diagnoses

DENNIS P. CANTWELL, M.D., AND LORIAN BAKER, PH.D.

Abstract. Follow-upor natural history outcome data for various DSM-lIl child and adolescent psychiatricdiagnoses are presented. The data are relevant not only to our understanding of the specific disordersbut also tothe validityof the DSM-llI diagnostic categories. "Semi-blind" psychiatric evaluationsof 151 children weremadeas they presented to a community speech/language clinic and again approximately 4 years later. The follow-updata revealed high stabilityfor only three diagnoses: infantileautism, attention deficitdisorder with hyperactivity,and oppositionaldisorder. The data revealed that severalof the DSM-Ill subcategories lacked predictivevalidity.This wastrue for the distinctionsbetweenattention deficitdisorderwith versuswithout hyperactivity; and betweenavoidant, separation anxiety,and overanxiousdisorders. Surprisingly lowstabilitywas found for conduct disorderdiagnosesas were surprisingly poor prognoses for parent-child problems and adjustment disorders. J. Am. Acad.Child Adolesc. Psychiatry, 1989, 28, 5:691-700. Key Words: DSM-lIl, prognosis, validity, attention deficitdisorder, conduct disorder.

Little is known about the natural history (or outcome) ofmany of the childhood and adolescent psychiatric disorders.Such data are not only necessary to our understanding of thevarious psychiatric disorders, but also relevant to the estab­lishment of a psychiatric diagnostic classification system.

In a recent publication in the Journal, Cantwell and Baker(I988a) discussed the criteria (including reliability, validity,and feasibility) for a successful psychiatric diagnostic classifi­cation system. One of the more important criteria is that thediagnostic categories have external validity, that is, that theycan be verified as unique categories by external data. Asoutlined elsewhere (Cantwell, 1975), various types of data arerelevant to the external validity of psychiatric diagnostic cat­egories. These include epidemiological data, family aggrega­tion data, biological and laboratory data, psychosocial data,natural history and outcome data, and response to treatmentstudies. The present paper presents data of one such type,namely, natural history and outcome data, for various DSM­III childhood and adolescent diagnostic categories.

In particular, this paper presents data from a prospective 4­to 5-year follow-up study of a cohort of children drawn froma large community speech and hearing clinic. Six hundredchildren were seen initially when they presented for assess­ment of speech and/or language disorders. All were initiallyunder the age of 16 and without significant hearing impair­ment. The results of the initial psychiatric assessment of thesechildren, presented in the Journal and elsewhere (Baker andCantwell, 1982a, b; Cantwell and Baker, 1980; Cantwell etaI., 1979, 1981) revealed a psychiatric illness rate of approxi­mately 50%.

The preliminary results of follow-up evaluations of 300 ofthese children were presented recently in the Journal (Bakerand Cantwell, 1987). It was found that the overall prevalenceof psychiatric disorder and the overall prevalence of certain

Accepted April 4. 1989.Dr. Cantwell is Joseph Campbell ProfessorofChild Psychiatry and

Director of Residency Training in Child Psychiatry at the UCLANeuropsychiatric Institute. Dr. Baker is a Research Psycholinguistalso at the UCLA Neuropsychiatric Institute (760 Westwood Plaza,Los Angeles, CA 90024), where reprints may be requested.

0890-8567/89/2805-0691 $02.00/0© 1989 by the American Acad­emy of Child and Adolescent Psychiatry.

specific diagnoses had increased significantly between theinitial and follow-up evaluations. However, the specific out­comes for the various initial psychiatric diagnoses were notexamined.

The present paper examines 151 of those children whoinitially had various DSM-III childhood psychiatric diagnosesand traces the stability of the various diagnoses found. Thesedata are relevant to the external validity of these DSM-IIIdiagnostic categories.

Method

The methodology of the studies has been described inprevious publications (Baker and CantwelI, 1982a, b, 1987;Cantwell and Baker, 1980, 1985; Cantwell et aI., 1979, 1981).Both initial and follow-up evaluations consisted of speech/language assessment, academic achievement testing, and in­tellectual testing (administered by L.B.) and psychiatric as­sessment by a board-certified child psychiatrist (D.P.C. orothers in initial study; D.P.C. alone in the follow-up study).

Psychiatric diagnoses were made initially and at follow-upusing DSM-III diagnostic criteria and based on data collectedwith four types of instruments: parent and child interviews(the DICA, Orvaschel, 1985) and parent and teacher behaviorrating scales (two forms of each: Conners, 1973; Rutter et aI.,1970). The reliability of the psychiatric diagnosis procedurewas tested at the time of the initial study; it was 96% for thepresence or absence of psychiatric disorder, and 94% for thespecific diagnoses (Cantwell and Baker, 1985). The sameprocedures were used for the folIow-up evaluation, but theywere not retested for reliability at that time.

The follow-up psychiatric diagnoses were made "semi­blind" to the initial psychiatric diagnosis. (That is, the clini­cian did not have the initial evaluation materials and diagnosisavailable to him. Because of the large number of children inthe initial study, most were not remembered by the time offollow-up; however, in a few outstanding cases the child'sinitial visit and diagnosis were remembered).

Ninety percent of the entire initial sample was seen forfollow-up. However, this present paper reports only on 150cases who had certain specific psychiatric diagnoses initially.The initial diagnoses, and age and sex distributions for thesecases are summarized in Table I below.

691

692 CANTWELL AND BAKER

The group studied consisted of 151 children, 105 (70%) of the group ranged in age from 5.2 to 20.6 years (mean age 9.7;whom were males, and 46 (30%) of whom were females. standard deviation 2.9). Initially, all of the children had someInitially the children ranged in age from 2.3 to 15.9 years impairment in either speech and/or language functioning.(mean age 5.9 years; standard deviation 2.9). At follow-up, Twenty-three percent of the group had only disorders of

speech production; 65% had disorders involving both speechproduction and language, and 12% of the children had dis-

TABLE 1. Initial Data for Cases Followed orders involving only language development. The intellectual

Age (in years)functioning of the group was within normal limits, and the

Sex socioeconomic distribution was approximately one third up-Diagnosis N Range Mean SD (% Male) per or upper-middle social class level, one third middle-class

ADDH 35 2.3 to 9.6 5.5 1.8 91 level, and one third lower- or lower-middle social class levelADDNO 5 5.7 to 8.3 6.7 1.2 40 (using Hollingshead two-factor scoring of social class.)Conduct disorders 9 3.2 to 9.4 5.7 2.0 89 The majority of children in the sample (approximatelyOppositional disorder 15 3.4 to 13.1 5.4 3.3 80 85%) received at least 4 months of speech/language therapy,Total behavioral 64 2.3 to 13.1 5.6 2.3 84 either within the school setting, in a private speech clinicSeparation anxiety 9 2.4 to 6.6 3.6 1.3 56 setting, or both. Conversely, the majority of children did notAvoidant disorder 14 3.1 to 9.7 5.0 1.9 43 receive psychiatric treatments, although 10%of the group hadOveranxious disorder 8 4.0 to 11.2 7.3 2.3 63 received medication trials (usually under the supervision of aTotal anxiety 31 2.4 to 11.2 5.2 2.3 48 pediatrician) and 15% had had at least one visit with aAffective disorders 7 6.5 to 15.8 10.4 3.9 43 psychologist, psychiatrist, or family therapist.Total emotional 38 2.4 to 15.8 6.1 3.3 58

ResultsAdjustment disorders 19 3.0 to 15.8 6.4 4.2 53Pervasive devel disorder 3 5.4 to 11.8 8.3 3.2 100 Outcome ofthe Behavioral Disorder DiagnosesParent-child problem 15 2.8 to 7.7 4.1 1.3 80 The outcome data for all of the cases (N = 139) withMixed psychiatric uncomplicated or "pure" initial diagnoses are summarized in

diagnoses 12 2.8 to 15.9 8.1 3.8 67 Table 2 below. The behavioral psychiatric disorders (whichTotal cases followed 151 2.3 to 15.9 5.9 2.9 70 include the DSM-lII diagnostic categories of attention deficit

TABLE 2. Outcomes for Cases with Pure Initial Disorders

Follow-up Status"

Other Diagnoses (N)

EmotionalInitial

Diagnosis N

Well

N %

SameDisorder

N % Behavioral Anxiety Affective PDD Other

BehavioralADDH 35 3 9 28 80 1 3 4 2 IADDNO 5 1 20 0 0 5 0 I 0 0Conduct 9 3 33 I II 3 0 0 0 2Oppositional 15 1 7 6 40 10 I 2 0 ITotal behavioral 64 8 13 35 55 18 3 7 2 5

EmotionalAvoidant 14 5 36 4 29 3 4 2 0 0Separation anxiety 9 4 44 I 11 2 3 0 0 0Overanxious 8 2 25 2 25 3 2 I 0 0Affective 7 2 29 2 29 I 2 2 I 0Total emotional 38 13 34 9 24 9 II 5 I 0

Parent/child 15 3 20 0 0 14 I 0 0 3Pervasive developmental disorder 3 0 0 3 100 0 0 0 0 0Adjustment disorders with

Depressed mood 2 2 100 0 0 0 0 0 0 0Emotional problems 8 2 25 0 0 5 3 0 0 0Withdrawl I 0 0 0 0 I I 0 0 0Conduct and emotional problems 5 I 20 0 0 4 2 0 0 0Conduct disorder 2 0 0 0 0 2 I 0 0 0Atypical disorder I 0 0 0 0 I 0 0 0 0Total adjustment disorders 19 5 26 0 0 13 7 0 0 0

• Numbers under "Follow-up status" refer to the number of times this diagnosis occurred; hence, due to comorbid diagnoses, these numberswill not tally with the number of children in the N column.

STABILITY OF osu-m DIAGNOSES 693

disorders, oppositional disorder , and conduct disorders) arediscussed first.

Overall, the behavioral psychiatric disorders were amongthe most stable categories of diagnoses. Children with earlybehavioral disorders were highly likely to have some psychi­atric disorder 4 to 5 years later. In fact, 87% of those childrenwith behavioral disorders initially were still psychiatrically illat follow-up. Within the general group of behavioral disorders,however, the various specific diagnostic categories showed aconsiderable range in stability.

Attention Deficit Disorder with Hyperactivity

Attention deficit disorder with hyperactivity (ADDH) wasthe most stable subtype of the behavioral disorders. Of the 35children with this pure diagnosis initially, a total of 28 (or80%) still had ADDH at follow-up, 23 in pure form, and fivewith some additional diagnosis. Furthermore, the recoveryrate for ADDH was very low; only three of the 35 children(or 9%) were psychiatrically well at follow-up.

The other psychiatric outcomes for the ADDH childrenwere diverse. The most commonly occurring other outcomefor the ADDH children was some type of emotional disorder.The emotional disorders included both anxiety disorders (twocases with overanxious disorder , and one case with obsessivecompulsive disorder) and affective disorders (one case withdysthymic disorder, three with major depression). Of the sixADDH children who had emotional disorders at follow-up,five had comorbid ADDH diagnoses and only one child hada pure emotional disorder. Among the other psychiatric ill­nesses found in the ADDH children at follow-up were: otherbehavioral disorders (one case with undersocialized aggressiveconduct disorder); pervasive developmental disorders (twocases with infantile autism); and transient tic disorder (onecase).

Comorbid diagnoses at follow-up were found in a total offive of the ADDH children; three children had dual diagnosesand two children had triple diagnoses. For all of these children ,the comorbid diagnoses consisted of ADDH plus some otherassociated psychiatric illness. The other associated illnesseswere usually emotional disorders (three cases of major depres­sion, and one case each of dysthymic disorder, overanxiousdisorder and obsessive compulsive disorder). However therewas one ADDH child who had ADDH and transient ticdisorder at follow-up.

Attention Deficit Disorder without Hyperactivity

As can be seen in Table 2, the psychiatric outcome forattention deficit disorder without hyperactivity (ADDNO)showed very different recovery and stability rates from thoseof ADDH. First, the recovery rate of ADDNO appeared tobe higher than for ADDH . One child (constituting 20%oftheADDNO group) was psychiatrically well at follow-up, asopposed to only 9% of the ADDNO group. Unfortunatelythe ADDNO group was so small (only five cases) that statis­tical analysis of this comparison was not possible.

ADDNO was the least stable of the behavioral diagnoses.At follow-up, none of the children from the initial ADDNOgroup had maintained the same diagnoses. However, all fourof the ADDNO children who were still psychiatrically ill at

follow-up had a diagnosis of ADDH. In addition to theADDH follow-up diagnosis, one child had two other associ­ated diagnoses at follow-up. These were undersocialized un­aggressive conduct disorder, and dysthymic disorder.

Conduct Disorders

Nine children had pure initial diagnoses of conduct disor­ders. Of these, one had socialized unaggressive conduct dis­order, one had undersocialized unaggressiveconduct disorder ,and eight had undersocialized aggressive conduct disorder.The conduct disorders, as a group, showed a better rate ofrecovery than either ADDH or ADDNO. Three (33%) of thenine conduct disordered children (two with undersocializedaggressive conduct disorder and one with socialized unaggres­sive conduct disorder) were well at follow-up.

The conduct disorders were not stable, either as individualdiagnoses or as a general group. At follow-up, only one of thechildren (with initial undersocialized aggressive conduct dis­order) still had any conduct disorder diagnosis (in fact, thesame diagnosis). Of those conduct disordered children whowere still psychiatrically ill at follow-up, three had otherbehavioral disorders (ADDH), and two had other types ofdisorders (organic personality syndrome and unspecified psy­chiatric disorder).

Oppositional Disorder

Oppositional disorder showed the poorest recovery rate ofall the behavioral psychiatric disorders. At follow-up, onlyone of the fifteen oppositional children (7%) had becomepsychiatrically well.

The oppositional disorder diagnosis showed an intermedi­ate degree of stability over the 4- to 5-year period. Six of theoppositional children (or 40%) retained the same diagnosis atfollow-up. However, other behavioral disorders were alsocommon outcomes for those children with initial diagnosesof oppositional disorder. At follow-up, six of the oppositionalchildren had ADDH and four had some type of conductdisorder. Three oppositional children had emotional disordersat follow-up (two had dysthymic disorder and one had over­anxious disorder), and one child had organic personalitysyndrome.

Three of the 15 oppositional children had comorbid diag­noses at follow-up. These were ADDH, oppositional disorderand dysthymic disorder in one case; ADDH and undersocial­ized conduct disorder in the second case, and ADDH, unag­gressiveconduct disorder, and dysthymic disorder in the thirdcase.

Emotional Disorder Diagnoses

In contrast to the behavioral psychiatric disorders (whichinvolve overt disturbances in behaviors), the emotional psy­chiatric disorders involve disturbances of mood or feelings.The emotional psychiatric disorders fall into two major sub­classes: anxiety disorders (including separation anxietydisorder, avoidant disorder, and overanxious disorder) andaffective disorders (including major depression, dysthymicdisorder, cyclothymic disorder, and bipolar disorder). Thirty­eight of the children in this study initially had pure emotionaldisorders. They were slightly older (aged 2.4 to 15.8 years, X

694 CANTWELL AND BAKER

= 6.1) than the children with pure behavioral disorders (aged2.3 to 13.1, X = 5.6), and they were more likely to be girls(58% were males in the emotional group versus 84% in thebehavioral group).

In general, the emotional disorders were characterized byhigher rates of recovery, and , for those cases who remainedpsychiatrically ill, lower rates of stability than the behavioraldisorders. Within the general category of emotional disorders,the various specific diagnostic categories were less dissimilarwith regard to outcome than the various categories of behav­ioral disorders. The specific categories of emotional disordersare each discussed below.

Avoidant Disorder

The most common of the pure emotional disorders occur­ring initially was avoidant disorder with 14 cases. Thesechildren ranged in age from 3.1 years to 9.7 years (X = 5.0,SO = 1.9) and consisted of eight girls and six boys. At follow­up, these children were most commonly still psychiatricallyill (with only five of the group or 36% being well).

Avoidant disorder was the most stable of the childhoodanxiety disorders. At follow-up, four of the avoidant childrenstill had the same diagnosis. This represents 29% of the groupof avoidant children, and 44% of the subgroup of avoidantchildren who had not recovered from psychiatric illness atfollow-up.

The most common other follow-up diagnoses for thesechildren were other types of emotional disorders. Four of theinitially avoidant children had overanxious disorder at follow­up, and two had dysthymic disorder . Three behavioral disor­der diagnoses were found: ADDH , ADDNO, and oppositionaldisorder.

Of the six avoidant children who were ill at follow-up withother disorders, three had more than one diagnosis. Thecomorbid psychiatric diagnoses were (I) overanxious disorder,dysthymic disorder, and ADONO; (2) avoidant disorder andoveranxious disorder; and (3) oppositional disorder, dys­thymic disorder, and overanxious disorder.

Separation Anxiety Disorder

The group of children with an initial pure diagnosis ofseparation anxiety disorder consisted of four girls and fiveboys, aged 2.4 to 6.6 years (X = 3.6, SD = 1.3). At follow-up,four of these children (44% of the group) were psychiatricallywell, and one child (II %) still had separation anxiety disorder.These figures represent the highest rate of recovery and thelowest rate of stability of any of the emotional disorders.

Other psychiatric diagnoses at follow-up were approxi­mately equally distributed between anxiety and behavioraldisorders. Two children had behavioral disorders (AODH andADDNO) and three children had overanxious disorder . Onlyone of the children with separation anxiety disorder initiallyhad comorbid diagnoses at follow-up. These were ADDNOand overanxious disorder.

Overanxious Disorder

Eight children (three girls and fiveboys) received pure initialdiagnoses of overanxious disorder. These children tended tobe older initially (4.0 to 11.2 years, mean age 7.3, SD 2.3)

than the other anxiety disordered children . Of all thesubgroups of emotionally disordered children, the overanx­ious group had the lowest recovery rate. Only two overanxiouschildren (or 25%) were psychiatrically well at follow-up.

The stability of the overanxious disorder diagnosis was alsoquite low. Only two children (25%) continued to have over­anxious disorder at follow-up, whereas an equal number hadanother type of anxiety disorder (avoidant disorder). Otherfollow-up diagnoses found in the overanxious group wereADDH (one case), ADDNO (two cases),and major depression(one case). Two of the overanxious children received multiplediagnoses at follow-up: avoidant disorder and overanxiousdisorder in one case, and avoidant disorder and ADONO inthe other.

Affective Disorders

Seven older children (ages 6.5 to 15.8 years, mean age 10.4years, SO 3.9), consisting of four girls and three boys, hadpure initial diagnoses of some type of affective disorder . Thespecificaffectivedisorder diagnoses found were: major depres­sion (two cases), dysthymic disorder (two cases), cyclothymicdisorder (one case), and bipolar disorder (one case).

At follow-up, two of the affective disorder cases (one ofmajor depression and one of dysthymic disorder) had becomepsychiatrically well and two (one of dysthymic disorder andone of bipolar disorder) were stable. In addition, two casesstill had affective disorders, but a different subtype thaninitially (the initial diagnoses of major depression and cyclo­thymic disorder both became dysthymic at follow-up).

Other follow-up diagnoses in the affective group included:pervasive developmental disorder, ADDH , overanxious dis­order, and avoidant disorder . Although the numbers are toosmall for statistical analysis, it appears that children withaffective disorders may be less likely to develop behavioraldisorders than children with anxiety disorders. Conversely, itappears that children with affective disorders may be morelikely to have multiple diagnoses at follow-up. Three of theaffective disordered children (43%) had comorbid psychiatricdisorders at follow-up. Two were cases of initial major depres­sion (one case with ADDH and pervasive developmentaldisorder at follow-up and the other with dysthymic disorderand overanxious disorder) and one was an initial case ofbipolar disorder (with two follow-up diagnoses, bipolar dis­order and avoidant disorder).

Parent-Child Problem

The DSM-lII "V-code" diagnosis of parent-child problemocurred as the single initial diagnosis in 15 young (2.8 to 7.7years, X 4.1, SD 1.3) children. Three of these children weregirls and the remainder were boys. Although parent-childproblem is not considered a "true" psychiatric disorder , it isof interest here because the prognosis for these children wasas poor or poorer than that for the children with true psychi­atric disorders. As can be seen in Table 2, only three of thechildren with early parent children problems (or 20%) werefound to be psychiatrically well at follow-up.

Of the 12 children with parent-child problem initially whoremained psychiatrically ill at follow-up, all but one receiveda behavioral disorder diagnosis. Of the II children with

STABILITY OF DSM-llI DIAGNOSES 695

parent-child problem initially who had behavioral disordersat follow-up, ten had ADDH, one had ADONa, and threehad secondary diagnoses of oppositional disorder. Other psy­chiatric disorders found at follow-up included pica (two cases),avoidant disorder (one case) and adjustment disorder withdisturbance of conduct and emotions (one case).

Multiple psychiatric diagnoses were very common amongthe children with initial parent-child problems. One-third ofthe group had comorbid psychiatric disorders at follow-up:two cases had ADDH and pica, one had ADDH and opposi­tional disorder, one had ADONa and oppositional disorder,and one had ADDH, oppositional disorder and avoidantdisorder.

Pervasive Developmental Disorders

Three boys received initial diagnoses of infantile autism.Not surprisingly (insofar as DSM-Ill considers the pervasivedevelopmental disorders to be chronic) this group of disordershad the lowest recovery rate (0% were well at follow-up) andthe highest stability rate (100%). All three of the autisticchildren had the same diagnosis at follow-up as initially.

Adjustment Disorders

Nineteen children in the sample had a pure diagnosis ofsome type of adjustment disorder. These included adjustmentdisorder with disturbance of conduct (two cases), adjustmentdisorder with depressed mood (two cases), adjustment disor­der with mixed emotional features (eight cases), adjustmentdisorder with mixed distusrbance of emotions and conduct(five cases), adjustment disorder with withdrawal (one case),and adjustment disorder with atypical features (one case).

The adjustment disorders had the lowest stability rate (0%)of any of the psychiatric disorders. The recovery rate for theadjustment disorders was also relatively low (26%).

The outcome data for the children with adjustment disor­ders were also examined in order to determine if there wasany stability in the general types of symptoms, i.e., if childrenwith initial adjustment disorders with emotional symptomstended to have follow-up emotional diagnoses and if childrenwith initial adjustment disorders with conduct symptomstended to have follow-up behavioral disorders. In fact, as thedata in Table 2 show, there was little correspondence betweeninitial symptomatology and follow-up diagnosis. The adjust­ment disordered children were most likely to have follow-up

diagnoses of behavioral disorders or anxiety disorders, regard­less of their specific initial type of adjustment disorder. Thefollow-up diagnoses of the adjustment disordered childrenwere ADDH (11 cases), oppositional disorder (two cases),overanxious disorder (six cases), and avoidant disorder (onecase).

Six of the 19 children (32%) with initial adjustment disor­ders had comorbid diagnoses at follow-up. These includedthree cases with ADDH and overanxious disorder, two caseswith ADDH and oppositional disorder, and one case withavoidant disorder and overanxious disorder.

Outcome ofCases with Initial Comorbidity

Twelve children had comorbid psychiatric diagnoses ini­tially. The initial and follow-up diagnoses of these casesalong with their (initial) age and sex data are summarized inTable 3.

ADDH, the most common of the pure initial diagnoses,was also the most common comorbid diagnosis. Seven chil­dren had an initial diagnosis of ADDH along with some othercoexisting diagnosis. The outcomes for these seven childrenlargely mirrored those of the children with the pure ADDHdiagnoses: low rate of recovery and high rate of stability. Ofthe seven comorbid ADDH children, none werewellat follow­up, and five (71 %) had some type of ADD diganosis at follow­up.

Affective disorders were also common among the initialcomorbid group of children. Six children had some type ofinitial comorbid affectivedisorder (four with major depressionand one each with dysthymic disorder and cyclothymic dis­order). As with the children with initial pure affective disor­ders, these children had follow-up diagnoses most commonlyin the behavioral or anxiety disorders groups. None of thechildren with initial comorbid affectivedisorders had the sameaffective diagnosis at follow-up, although there was one childwhose initial major depression became dysthymic disorder atfollow-up.

Discussion

Attention Deficit Disorder with Hyperactivity

ADDH was one of the most stable disorders seen, with 80%of the ADDH disorder still having the same diagnosis four tofive years later. This finding is not surprising in view of thelarge number of reports of childhood ADDH continuing into

TABLE 3. Datafor Cases with Initial Comorbid Diagnoses

Initial Diagnoses Follow-up Diagnoses Age Sex

ADDH/separation anxietyADDH/major depressionADDH/e1ective mutismADDH/organic personality syndromeADDH/major depressionADDH/major depressionADDH/major depressionADDNO/gender identity disorderAvoidant/cyclothyrnic disorderOppositional/overanxiousOppositional/avoidant disorderDysthymic/parent-child pblm

ADDNOavoidant disorder/ADDNOADDH/overanxious disorderorganic brain syndromeADDH/dysthymic disorderADDH/overanxious disorderADDHADDH/avoidant/oppositionalwellwellelective mutism/oppositionalADDH/generalized anxiety

4.310.72.89.89.0

11.19.68.28.53.24.6

15.9

mmmmmmmffffm

696 CANTWELL AND BAKER

adolescence and even adulthood (Amado and Lustman, 1982;Borland and Heckman, 1976; Cowart, 1982; Thorley, 1984;Weiss and Hechtman, 1986).

The other psychiatric disorders that developed in theADDH children included transient tic disorder, pervasivedevelopmental disortler, anxiety disorders and depressive dis­orders. An association between tics and ADDH has beenobserved in the Comings' research on Tourette syndrome. Asreported in Comings and Comings (1987), in the naturalcourse of the disease, in the majority of Tourette syndromepatients, ADDH develops first, then 2.4 years later motor andvocal tics develop. Although the patient in this study devel­oped a transient (motor) tic rather than chronic motor andvocal tics, it is likely that a similar mechanism may beinvolved. It is relevant to note that this particular patient hadnot received stimulant medication.

The association between ADDH and anxiety and affectivedisorders has also been reported in the literature (Biedermanet al., 1987; Carlson and Cantwell, 1982; Lahey et al., 1987).The appearance of these disorders at follow-up in certain ofthe ADDH children, as well as the high frequency of initialcomorbid diagnoses of ADDH plus emotional disorders maybe relevant to the hypotheses that ADDH children withaffective disorders constitute a distinct subgroup (Biederman,et al., 1987). Unfortunately, the numbers of children in thisstudy are too small for statistical comparisons of other back­ground and family variables that could confirm this hypoth­esis.

The development at follow-up of infantile autism in two ofthe initially ADDH patients is somewhat surprising, althoughless so considering that infantile autism is a major cause ofspeech/language disturbances and that these children weredrawn from a speech clinic sample. Both of these patients hadbeen quite young initially (2.3 years and 3.2 years of age),both were boys, and both had severe disturbances in bothspeech and language. Retrospective examination of thesepatients' initial charts revealed an inability to test either childusing formal procedures due to apathetic noninvolvement inone child and tantrums in the other child. These two casespoint out the difficulty of establishing a differential diagnosisbetween hyperactivity and pervasive developmental disorderin young children. They also confirm the findings of Konnoand Ohno's (1981) developmental study of the symptomsof hyperactivity and autism. These authors reported that,according to parental reports, the earliest symptoms in hyper­active children were hyperactivity, followed by temper tan­trums and perseveration, whereas the earliest symptom ofautistic children was temper tantrums.

Despite a well-documented association between hyperactiv­ity and conduct disorders (Langner et al., 1974; Reeves et aI.,1987; Rutter et al., 1970; Sandbert et al., 1980; Werry et al.,1987a), only one of the ADDH children in this study had aconduct disorder at follow-up. Thus, the data from this groupof children support those authors arguing for the independ­ence of hyperactivity and conduct syndromes (Loney et aI.,1978; Taylor et aI., 1986; Taylor, 1988; Trites and Laprade,1983). Insofar as the children studied were still fairly youngat follow-up, it is still possible, however, that conduct disor­ders may develop in some of these children as they becomeolder and face learning and social difficulties in the school

setting that might predispose to later development of conductdisorder.

Attention Deficit Disorder without Hyperactivity

The ADDNO category is a controversial one, and, in fact,the ADDH versus ADDNO dichotomy (as defined in DSM­III) is no longer maintained in DSM-III-R. The evidence thatADDNO is valid as a psychiatric disorder distinct fromADDH is primarily symptom data having to do with eitheremotional or cognitive functioning (Berry et al., 1985; Carl­son, 1986; Carlson et al., 1987; Edelbrock et al., 1984; Kingand Young, 1982; Sergeant and Scholten, 1985). However,one recent study (Frank and Ben-Nun, 1988) found thatcertain neurological correlates did distinguish the two disor­ders.

The present study is the only one of which the authors areaware that examines the natural history of both ADDH andADDNO. These data revealed very different recovery andstability rates for the two disorders. The recovery rate forADDNO was considerably higher than that for ADDH; andthe stability rate was considerably lower. Twenty percent ofthe ADDNO group became well versus 9% of the ADDHgroup; and 0% of the ADDNO group remained stable versus80% of the ADDH group. Although these data would appearto argue for the distinctness of the two disorders, closerexamination reveals that the outcomes for the two disorderswere actually rather similar. In fact, all of the ADDNOchildren who were still psychiatrically ill at follow-up receiveda diagnosis of ADDH. Thus, the rate for ADDH at follow-upwas exactly the same, (80%), among the children with initialADDNO as it was for the children with initial ADDH. Thissuggests that ADDH and ADDNO may not have validity asdistinct psychiatric syndromes.

Conduct Disorders

Of the nine purely conduct disordered children in thisstudy, three (33%) recovered and only one (II %) remainedconduct disordered at follow-up. These findings contradictthe generally held belief that aggressive behaviors and/orconduct disorders in children are relatively stable over time(Halverson and Waldrop, 1976; Kagan and Moss, 1962;Minde and Minde, 1977; Olweus, 1979; Zeitlin, 1986).

Furthermore, the finding has relevance to the validity ofthe DSM-III conduct disorder definition. Rutter and Tuma(1988) have observed that a diagnostic system for childhoodpsychiatric disorders needs to differentiate between those con­duct disorders that constitute transient problems during child­hood or adolescence and those that represent the precursorsof sociopathic or personality disorders in adult life. Thepresent data would indicate that the definition of conductdisorders provided by DSM-III applies to transient childhoodproblems.

There are two other factors that may be relevant to theobserved lack of continuity of conduct disorders in this sampleof children. First, the children were very young when firstdiagnosed (the mean age was 5.7 years), and did not yetmanifest some of the more "serious" or "delinquency" symp­toms of conduct disorder (such as firesetting, stealing outsidethe home, mugging, vandalism, or rape). Furthermore, be­cause of the young age of the children initially, they did not

STABILITY OF DSM-llI DIAGNOSES 697

display school failure or learning disability, factors which,when present with nondelinquent conduct disorder, are typi­cally associated with persistence (Wolff, 1985).

Finally, it must be remembered that all of the children inthis study had, at least initially, some type of speech and/orlanguage dysfunction, and that this dysfunction may haveplayed a role in the outcome and indeed in the diagnosis ofthese children's psychiatric disorders. With the conductdisorder diagnosis in particular, it is possible the conductsymptomatology manifested by these young children was aresponse to the frustration of being unable to express them­selves due to a language disorder . Although the number ofchildren here are too small for statistical analysis, the raw datado give some support to the hypothesis that frustration/language disorder plays a role in the continuity of the conductdisorder diagnoses. In fact, the three conduct disordered chil­dren who became psychiatrically well had speech/languagedisorders that were milder initially and that had markedlyrecovered at follow-up. Those conduct disordered childrenwho remained ill at follow-up had more severe speech/lan­guage disorders both initially and at follow-up. Thus far,however, there has been only one published study providingany suggestion that improvement in communication is linkedwith improvement in emotional/behavioral symptoms (Carrand Durand, 1985).

The natural history data for conduct disorder presentedhere are of particular interest when compared to the naturalhistory data for ADDH and ADDNO. It will be recalled thatof those conduct disordered children who were still psychiat­rically ill at follow-up, three had other behavioral disorders(ADDH), and two had other types of disorders (organicpersonality syndrome and unspecified psychiatric disorder).

Thus, although the conduct disordered children did notshow stability of the conduct disorder over the four yearfollow-up period, they did show differential outcomes fromthe ADDH and ADDNO children. Unlike the children withinitial ADDH or ADDNO, none of the initial conduct disor­dered children had follow-up disorders in the emotional orpervasive developmental categories. Also, unlike a number ofthe ADDH and ADDNO children , none of the children withinitial conduct disorders showed comorbid psychiatric disor­ders at follow-up.

Oppositional Disorder

Oppositional disorder is one of the less well researched ofthe DSM-III childhood categories. Little empirical data areavailable regarding the prevalence and incidence of the dis­order, or the genetic, neurological. biochemical, or demo­graphic factors that may be associated with it (Paez andHirsch, 1988). In addition to the lack of established validity,this diagnosis has reliability problems having to do with itsdistinction from normal developmental behaviors on the onehand and from conduct disorder on the other hand (Rey etal.• 1988; Rutter, 1988).

In the present study, oppositional disorder showed a naturalhistory that was quite distinct from that of conduct disorderand from normal development. Oppositional disorder hadthe poorest recovery rate (7%) of all the behavioral psychiatricdisorders. Unlike conduct disorder, oppositional disorder hada relatively high degree of stability (45%) over time. Those

oppositional disordered children who developed other psy­chiatric illnesses at follow-up tended to have other behavioraldisorders. either ADDH alone or ADDH plus conduct disor­der.

Emotional Disorder Diagnoses (Anxiety Disorders andAffective Disorders)

The natural history data presented here lend additionalvalidity to the already emp irically validated (Achenbach andEdelbrock, 1981) "broad band" distinction between emo­tional and behavioral disorders . The emotional disorders as agroup were characterized by higher rates of recovery. andlower rates of stability than the behavioral disorders. Althoughthere was little stability for specific subtypes of emotionaldisorders, there was some stability within the general categoryof emotional disorders. This finding supports those authors(Rutter and Gould . 1985; Werry et al., 1987b) who suggestthat subdivisions between types of emotional disorders arepremature.

Children with initial anxiety disorders showed a mild trendtowards anxiety disorder at follow-up, but with very limitedstability for the particular subtypes of anxiety disorder. Sepa­ration anxiety disorder was particularly unstable , with 44%recovering and only II % remaining stable. The very youngage of patients with this diagnosis (this was the youngestdiagnosis group with a mean age of 3.6 years) may haveplayed a role in the lack of stability. Even overanxious disor­der, which has been thought to be a forerunner of generalizedanxiety disorder in adults, was quite unstable (25%). Theanxiety disorder data is discussed further elsewhere (Cantwelland Baker, 1988b).

As with anxiety disordered children, the affectively disor­dered children were most likely to have emotional (ratherthan behavioral) disorders at follow-up. As with the anxietydisorders, the affective disorders showed both recovery andstability rates of approximately one in three. Almost half ofthe affectively disordered children had comorbid diagnoses atfollow-up. This is in keeping with other studies (Cantwell andCarlson, 1983; Kovacs et al., 1984) that have found thatdepressive disorder in children frequently occurs with otherpsychiatric disorders including anxiety disorders and behav­ioral disorders.

Adjustment Disorders

Adjustment disorders are, by definition. unstable. TheDSM-III definition of adjustment disorders specifies thatthese are disturbances that are associated with stressors andwill remit either when the precipitating stressor ceases orwhen a new level of adaptation occurs. Thus, it was notexpected that these disorders would show much stability. Asexpected, none of the adjustment disordered children hadretained this diagnosis at follow-up.

However, recovery from adjustment disorders was lesscom­mon than had been hoped, with only 26% of the adjustmentdisordered children being well at follow-up. The poor prog­nosis for the adjustment disordered children in this studyreplicates the findings of Andreasen and Hoenk (1982) thatadjustment disorders are unstable and carry a poor prognosisfor children and adolescents, In Andreasen and Hoenk'sstudy, adjustment disordered adolescents and adults were

698 CANTWELL AND BAKER

followed five years after initial evaluation. While 71 % of theadults were psychiatrically well at follow-up, only 44% of theadolescents were psychiatrically well and 50% had a differentdiagnosis.

Adjustment disorders appear to be a rather common diag­nosis in children and adolescents (Jacobsen et al., 1980;Weiner and DeIGaduio, 1976). The findings here indicatethat it is often not a self-limiting condition as the DSM-IIIdefinition implies but can be a serious disorder carrying apoor prognosis. Furthermore, the diversity of outcomes sug­gests that rather than a single disease entity, adjustmentdisorders may in fact be precursors or preliminary stages ofvarious other diagnoses. Furthermore, there is an apparentlack of predictive value for the various subclassifications ofadjustment disorders. These findings, coupled with the doc­umented lack of reliability for the adjustment disorder diag­noses (Werry et al., 1983) suggest the need for serious recon­sideration of this diagnostic category.

Parent-Child Problem

A final comment is necessary regarding the DSM-III "V­code" diagnosis parent-child problem. This diagnosis was usedin cases where there were problems in parent-child interac­tions but where the child did not manifest definitive psychi­atric illness. Thus, all of these children were clearly withoutpsychiatric illness initially.

Nonetheless, even though parent-child problem is not con­sidered a true psychiatric disorder, its natural history was aspoor or poorer than that of many of the true psychiatricdisorders. Parent-child problem had a recovery rate of only20% and a stability rate of 0%. Of 15 children with initialparent-child problem, three were well at follow-up, II had abehavioral disorder and one had an adjustment disorder.Among the II children with follow-up behavioral disorders,there were 10 cases of ADDH, three cases of oppositionaldisorder, and one case ofADDNO. Thus these follow-up datasuggest that the parent-child problem may be a precursor tolater full-blown behavioral psychiatric disorders.

It has been suggested that these children may have hadbehavioral disorders initially, i.e., that the diagnostic thresholdfor ADDH used initially may have been too high. However,retrospective chart review provided no evidence ofany behav­ioral psychiatric illness initially.

A search for common family or background factors revealedno consistent trends in the parent-child group. There was nospecific common pattern of family structure, environmentalstressors, or family pathology in the group, although there didseem to be somewhat elevated rates of maternal depression(present in five of the 15 mothers) and paternal alcoholism(present in three of the 15 fathers). Nonetheless, fully one halfof the parents were psychiatrically well and the majority offamilies were without psychosocial stressors (other than par­ent-child conflict). The parental psychiatric diagnoses werethe same at follow-up as initially.

Methodological Considerations

There are several possible explanations for the low stabilityof many of the DSM-III childhood diagnoses. First, the DSM­III system was developed without detailed consideration ofdevelopmental factors in diagnosis. Nonetheless, it is likely

that the symptomatology of the various psychiatric disorderswill vary for different age levels. The initial diagnoses reportedhere were made when the majority of children were ratheryoung (approximately half being in the preschool age range).Furthermore, given the young age of the sample, schoolreports were not available for some of the children, makingdiagnosis at that time still more difficult. If indeed the lowstability of the diagnoses were due to an age factor, the presentstudy may contribute to improved validity for DSM-IV bypointing out the need for attention to the developmentalfactor.

A second explanation of the low stability is that some ofthe cases may have been misdiagnosed. It has been observed,for example, that the three cases who "developed" pervasivedevelopmental disorder (PDD) are particularly suspicious,given that PDD typically has early onset. However, retrospec­tive examination of these children's charts confirmed thatinitially none met the diagnostic criteria for PDD. Two wereyoung boys (2.3 and 3.2 years ofage) who had marked ADDHand some type of language disorder (the younger child wasnonverbal and the older had significant but not gross deficits).The third was an older boy (6.6 years) with clear mooddisorder. All three of these children had initial behavioralabnormalities including severe tantrums. None of the childreninitially had ritualistic behaviors, stereotypic movements, orpervasive lack of responsiveness to people, although thesebecame prominent later on.

Finally, it must be remembered that the sample in thisstudy is an unusual one, neither a clinically-referred samplenor a community sample. By definition, all ofthe children inthe sample had a coexisting speech/language disorder at somepoint in time. This not only means that reaching an accuratediagnosis is more difficult in these children, but it may alsomean that these children are different in other ways. However,given that several researchers have reported a high prevalenceof undiagnosed speech/language disorders among both inpa­tient and outpatient psychiatrically-referred children (Grin­nell et al., 1983, Gualtieri et al., 1983; Love and Thompson,1988), this difference may be more apparent than real.

Conclusions

Although there are certain methodological issues to beconsidered with regard to this study, the study provides im­portant information about the outcomes of various of theDSM-III childhood diagnoses. The follow-up data revealedhigh stability for only three diagnoses: infantile autism, atten­tion deficit disorder with hyperactivity, and, more surpris­ingly, oppositional disorder. The study revealed that a numberof psychiatric diagnoses lacked predictive validity; that is,children with these disorders were highly likely to have otherdifferent disorders at follow-up. This was true for attentiondeficit disorder without hyperactivity, avoidant disorder, sep­aration anxiety disorder, overanxious disorder, and conductdisorder. The study also revealed surprisingly poor prognosesfor parent-child problems and adjustment disorders.

ReferencesAchenbach, T. M. & Edelbrock, C. S. (1981), Behavioral problems

and competencies reported by parents of normal and disturbedchildren aged 4 through 16. Monogr. Soc. Res. Child Dev., 46,

STABILITY OF DSM-JIl DIAGNOSES 699

Serial No. 188.Amado, H. & Lustman, P. J. (1982), Attention deficit disorders

persisting in adulthood. Compr. Psychiatry. 23:200-214.Andreasen, N. C. & Hoenk, P. R. (1982), The predictive value of

adjustment disorders. Am. J . Psychiatry, 139:584-590.American Psychiatric Association (1980), Diagnostic and Statistical

Manual ofMental Disorders. Ird Ed. (DSM-Ill) . Washington, DC:APA.

Baker, L. & Cantwell, D. P. (I 982a), Developmental, social. andbehavioral characteristics of speech and language disordered chil­dren. Child Psychiatry Hum. Dev., 12:195-207.

-- -- (l982b), Language acquisition, cognitive development,and emotional disorder in childhood. In: Children's Language.Volume 3, ed. K. E. Nelson . Hillsdale, NJ: Lawrence ErlbaumAssociates, pp. 286-321.

-- -- (1987), A prospective psychiatric follow-up of childrenwith speech/language disorders. J . Am. Acad. Child Adolesc. Psy­chiatry, 26:546-553.

Berry, C. A., Shaywitz, S. E. & Shaywitz, B. A. (1985), Girls withattention deficit disorder: a silent minority? A report on behavioraland cognitive characteristics. Pediatrics, 76:801-809.

Biederman, J., Munir, K., Knee, D. et al. (1987), High rate ofalTectivedisorders with probands with attention deficit disorder and in theirrelatives: a controlled family study . Am. J . Psychiatry, 144:330­333.

Borland, B. L. & Heckman, H. K. (1976), Hyperactive boys and theirbrothers: a 25 year follow-up study . Arch. Gen. Psychiatry, 33:669­675.

Cantwell , D. P. (1975), A model for the investigation of psychiatricdisorders of childhood: its application in genetic studies of thehyperkinetic syndrome. In: Explorations in Child Psychiatry. ed.E. J. Anthony. New York: Plenum, pp, 57-79.

-- and Baker, L. (1980), Psychiatric and behavioral characteristicsof children with communication disorders . 1. Pediatar. Psychol.,5:161-178.

-- -- (1985), Psychiatric and learning disorders in children withspeech and language disorders : a descriptive analysis. Advances inlearning and behavioral disabilities, 4:29-47.

-- -- (1988a), Anxiety disorders in children with communica­tion disorders. Journal ofAnxiety Disorders, 2:135-146.

-- -- (1988b), Issues in the classification of child and adolescentpsychopathology. J. Am. Acad. Child Adolesc. Psychiatry, 26:546­553.

-- --Mattison, R. (1979), The prevalence of psychiatric disorderin children with speech and language disorder. J. Am. Acad. ChildAdolesc. Psychiatry, 18:450-461.

-- -- -- ( 1981), Prevalence, type and correlates of psychiatricdisorder in 200 children with communication disorder. J. Dev.Behav. Pediatr. 2:131-136.

-- Carlson , G . A. (1983), Affective Disorders in Childhood andAdolescence-an Update. New York : Spectrum Publications.

Carlson, C. L. (1986), Attention deficit disorder without hyperactivity.In: Advances in Clinical Child Psych ology, Vol. 9, ed. B. Lahey &A. Kazdin. New York : Plenum, pp. 153-175.

- Lahey, B. B., Frame, C. L.. Walker, J. Hynd , G. W. (1987),Sociometric status of clinic-referred children with attention deficitdisorders with and without hyperactivity. J. Abnorm. Child Psy­chol., 15:537-547.

Carlson, G. A. Cantwell, D. P. (1982) , Suicidal behavior and depres­sion in children and adolescents. J . A m . Acad. Child Psychiatry,21:36 1- 368.

Carr , E. & Durand, V. (1985), The social-communicative basis ofbehavior problems in children . In: Theoretical Issues in BehavioralTherapy, ed. S. Reiss & R. Bootzin . New York: Academic Press.

Comings, D. E. & Comings, B. G. (1987), A controlled study ofTourette's syndrome. I. Attention-deficit disorder, learning disor­ders, and school problems. Am. J . /fum. Genet., 41:701-741.

Conners, C. K. (1973), Rating scales for use in drug studies withchildren. Psychopharmacol Bull. (Special Issue), 24-34.

Cowart, V. S. (1982), ADD: Not limited to children. JAMA, 16:248­286.

Edelbrock, c., Costello, A. J. & Kessler, M. D. (1984), Empiricalcorroboration of the attention deficit disorder. J. Am. Acad. Child

Psychiatry, 23:285-290.Frank, Y. & Ben-Nun , Y. (1988), Toward a clinical subgrouping of

hyperactive and non hyperactive attention deficit disorder. Resultsof a comprehensive neurological and neuropsychological assess­ment. Am. 1. Dis. Child., 142:153-155.

Grinnell, S. W., Scott-Hartnett, D. & Glasier, J. L. (1983), Languagedisorders (Letter to the Editor) . J. Am. Acad. Child Psychiatry ,22:580-581.

Gualtieri, C. T., Koriath, U., Van Bourgondien , M. & Saleeby, N.(1983), Language disorders in children referred for psychiatricservices. J. Am. Acad. Child Psychiatry, 22:165-171.

Halverson, C. F. & Waldrop, M. F. (1976), Relations between pre­school activity and aspects of intellectual and social behavior at age7Jh, Developmental Psychology, 12:107-112.

Jacobsen, A. M., Goldberg. I. D., Burns. B. J., Hoeper, E. W., Hankin,J. R. & Hewitt, K. (1980), Diagnosed mental disorders in childrenand use of health services in four organized health care settings.Am. J . Psychiatry, 137:559-565.

Kagan, J. & Moss, H. A. (1962), Birth 10 Maturity. New York: Wiley.King, C. & Young. R. D. (1982), Attentional deficits with and without

hyperactivity : teacher and peer perceptions. J. Ahnorm. ChildPsychol., 10:483-495 .

Konno, Y. & Ohno, K. (1981), A comparative and developmentalstudy of hyperactivity and its related symptoms in autistic andhyperactive children . Japanese Journal of Special Education,19:37-47 .

Kovacs, M., Feinberg, T. L. , Crouse-Novak, M. A., Paulauskas, S. L.& Finkelstein, R. (1984). Depressive disorders in childhood. I. Alongitudinal prospective study of characteristics and recovery. Arch.Gen. Psychiatry, 41 :229-237.

Lahey. B. B.. Schaughency, E. A.. Hynd, G. W.• Carlsonk, C. L. &Nieves. N. (1987), Attention deficit disorder with and withouthyperactivity: Comparison of behavioral characteristics of clinic­referred children . J . Am. Acad. Child Adolesc. Psychiatry, 26 :718­723 .

Langner, T. S., Gersten, J. C. & Eisenberg, J. G. (1974), Approachesto measurement and definition in the epidemiology of behaviordisorders : ethnic background and child behavior. Int . J. HealthServ ., 4:483-50 I.

Loney. J.• Langhorne, J. E. & Paternite, C. E. (1978). Empirical basisfor subgrouping the hyperkinetic/minimal brain dysfunction syn­drome. J. Abnorm. Psychol., 87:431-441.

Love, A. J. & Thompson. M. G. (1988), Language disorders andattention deficit disorders in young children referred for psychiatricservices: analysis of prevalence and a conceptual synthesis. Am. J.Orthopsychiatry. 58:52-64.

Minde, R. & Minde, K. (1977). Behavioral screening of pre-schoolchildren: A new approach to mental health? In: EpidemiologicalApproaches to Child Psychiatry, ed. P. Graham. London: AcademicPress, pp. 139-164.

Olweus, D. ( 1979), Stability of aggressive reaction patterns in males:A review. Psychol. Bul/., 86:852-875.

Orvaschel , H. (1985), Psychiatric interviews suitable for use in re­search with children and adolescents . Psychopharmacol. Bull. ,21:737-746.

Paez, P. & Hirsch, M. (1988), Oppositional disorder and electivemutism. In: Handbook of Clinical Assessment of Children andAdolescents, ed. C. J. Kestenbaum & D. T. Williams. Vol. 2. NewYork: New York University Press, pp. 800-811.

Rey, J. M., Bashir, M. R., Schawrz, Moo Richards, I. N.. Plapp, J. M.& Stewart, G. W. (1988), Oppositional disorder: fact or fiction? J.Am. Acad. Child Adolesc. Psychiatry, 27 :157-162.

Reeves, J . c.. Werry, J. 5., Elkind, G. A. & Zarnetkin, A. (1987),Attention deficit, conduct, opositional, and anxiety disorders inchildren : II. Clinical characteristics. J. Am. Acad. Child Adolesc.Psychiatry, 26:144-155.

Rutter, M. (1988), DSM-III-R : A postcript. In: Assessment andDiagnosis in Child Psychopathology, ed. M. Rutter, A. H. Tuma& I. S. Lann. New York: Guilford Press, pp. 453-464.

-- Gould, M. (1985), Classification. In: Child and AdolescentPsychiatry: Modern Approaches, ed. M. Rutter & L. Hersov , Bos­ton: Blackwell Scientific. pp, 304-324.

-- Graham, P. & Yule. W. (1970). A Neuropsychiatric Study in

700 CANTWELL AND BAKER

Childhood. Lavenham, Suffolk: The Lavenham Press.-- Tizard, J. & Whitmore, K. (1970), Education, Health, and

Behavior. London: Longman.-- Turna, A. H. (1988), Diagnosis and classification: Some out­

standing issues. In: (Eds.), Assessment and Diagnosis in ChildPsychopathology, ed. M. Rutter, A. H. Tuma & I. S. Lann. NewYork: Guilford Press, pp, 437-452.

Sandberg, S. T., Wieselberg, M. & Shaffer, D. (1980), Hyperkineticand conduct problem children in a primary school population. 1.Child Psychol. Psychiatry, 21:293-311.

Sergeant, J. A. & Scholten, C. A. (1985), On resource strategy limi­tations in hyperactivity: cognitive impulsivity reconsidered. J. ChildPsychol. Psychiatry, 26:97-109.

Taylor, E. (1988), Attention deficit disorder and conduct disordersyndromes. In: Assessment and Diagnosis in Child Psychopathol­ogy, ed. M. Rutter, A. H. Tuma & I. S.Lann. New York: GuilfordPress, pp., 377-407.

-- Schachar, R., Thorley, G. & Wieselberg, M. (1986), Conductdisorder and hyperactivity: I. Separation of hyperactivity and an­tisocial conduct in British child psychiatric patients. Br. J. Psychia­tar}',149:760-767.

Thorley, G. (1984), Review of follow-up and follow-back studies ofchildhood hyperactivity. Psychol. Bull., 96:116-132.

Trites, R. & Laprade, K. (1983), Evidence for an independent syn­drome of hyperactivity. J. Child Psychol. Psychiatry, 24:573-586.

Weiner, I. B. & DelGaudio, A. C. (1976), Psychopathology in adoles­cence. Arch. Gen. Psychiatry, 33:187-193.

Weiss, G. & Hechtman, L. T. (1986), Hyperactive Children GrownUp: Empirical Findings and Theoretical Considerations. NewYork: Guilford Press.

Werry, J. S., Elkind, G. S. & Reeves, J. C. (I 987a), Attention deficit,conduct, oppositional, and anxiety disorders in children: III. Lab­oratory differences. J. Abnorm. Child Psychol., 15:409-428.

--Methven, R. J., Fitzpatrick, J. & Dixon, H. (1983), The interraterreliability of DSM-III in children. J. Abnorm. Child Psychol.,11:341-354.

-- Reeves, J. C. & Elkind, G. S. (1987b), Attention deficit, conduct,oppositional, and anxiety disorders in children: I. A review ofresearch on differentiating characteristics. J. Am. Acad. Child Ado­lese. Psychiatry, 26:133-143.

Wolff, S. (1985), Non-delinquent disturbances of conduct. In: Childand Adolescent Psychiatry: Modern Approaches (2nd ed.), ed. M.Rutter & L. Hersov. Oxford: Blackwell Scientific, pp. 400-413.

Zeitlin, H. (1986), The Natural History of Psychiatric Disorder inChildhood. New York: Oxford University Press.