The Prevalence of Overanxious Disorder and Separation Anxiety Disorder: Results from the Ontario Child Health Study RUDRADEO C. BOWEN , M.D . , C.M. , F.R.C.P .(C) , DAVID R . OFFORD , M.D . , F .R.C.P. (C), AND MICHAEL H. BOYLE, PH.D.
Abstract. Data from a community epidemi ological study of 1,869 famili es (Ontario Child Health Study) was used to evaluate the effect of different ways of operationalizing DSM-/II-R criteria for over anxiou s disorder (OAD) and separation anxiety disorder (SAD) among adolesc ents aged 12 to 16. The authors determ ined that a high threshold for symptoms to qualify as present, the presence of one or both of the essential symptoms , and the presence of four or more auxiliary sympt oms for OAD and three or more for SAD gave prevalence of OAD of 3.6 % and SAD of 2.4%. There was high overlap between the presence of OAD and SAD and externalizing disorder and depression, but one- half of youth with GAD and SAD had pure anxiety disorder. Youth with GAD and SAD were ju st as impaired as youth with externalizing disorder and depression, except that they admitted to less social isolation and their schoolwork was less affected . J . Am . Acad . ChildAdolesc . Psychiatry. 1990, 29,5 :753-758 . Key Words: prev alence , classificat ion, DSM-II/-R , adolescence , anxiety disorder. Overanxious disorder (OAD) and separation anxiety disorder (SAD) of childhood have rece ived little research attention in clinical studies and even less attention in community studies (Last et aI. , 1987a; Orvaschel and Weissman , 1987). The Ontario Child Health Study (OCHS) data provided an opportunity to evaluate the meaningfulness of different ways of operationalizing DSM-III-R criteria for OAD and SAD in a large community sample (N = 3,294) , and how this affects the prevalence of the disorders (American Psychiatric Association , 1987; Boyle et aI., 1987).
Method Survey Design and Data Coll ection
The OCHS was a cross-sectional interview survey of 1,869 families including 3,294 child ren, ages 4 to 16, conducted in the spring of 1983. The survey methodology and instrumentation have been described by Boyle et aI. (1987) and are briefly summarized here . The target population of the OCHS included all children born between January 1, 1966, through January 1, 1979, whose usual residence was a household dwelling in Ontario . The sampling frame was the 1981 census and the sampling
AcceptedMay 24 , /990 . Dr . Bowen is Professor of Psychiatry. University of Saskatchewan. Saskatoon , Saskatchewan .Dr. Offord isProfessor ofP sychiatry , and Dr . Boyle is Assistant Professor of Psychiatry. McMaster University. Hamilton. Ontario . This work is supported by grams/rom the Ontario Ministry of Community and Social Services and was carried out by the Child Epidemiology Unit. Department of Psychiatry, McMast er University and the Child and Family Center , Chedoke Division, Chedoke-McMaster Hospitals, Hamilton, Ontario. Dr . Bowen was supported by a sabbatical leave research grantfrom the University of Saskatchewan . Ms. Yvonne Racine made valuable contributions to various aspects of this work. especially to data analysis . Reprint requests to Dr . Offord. Chedoke-McMaster Hosp itals , Chedoke Division, Administration Building, Box 2000, Station A , Hamilton. Ontario, Canada. L8N 325. 0890-8567/90/2905-0753$02.00/0© 1990 by the American Academy of Child and Adolescent Psych iatry .
unit was household dwellings . Sampling selection was done by stratified, clustered , random sampling from the census file. Ninety-one percent of all elig ible households agreed to participate , and the refusal rate was low, 3.9%. Interviewing was done by trained interviewers employed by Statistics Canada. The basic pool of items for measuring childhood psychopathology was provided by the Child Behavior Checklist (CBCL) with some modifications and additions (Achenbach and Edelbrock, 1981; Boyle et al., 1987). The OCHS checklist asked about symptoms occurring at present or in the past 6 months. Each item had three response categories: " never or not true " scored as 0, "sometimes or somewhat true" scored as 1, and " often or very true" scored as 2. The first respon se category "never or not true" is different from the CBCL category' 'not true. " Measurement ofDisorder
By comparing the criteria in DSM-III-R with items on the OCHS checklist, the authors concluded that it would be feasible to classify youth as suffering from OAD and/or SAD (American Psychiatric Association , 1987). The DSM-III-R criteria (abbreviated) and corresponding questionnaire items for OAD and SAD are given in Tables 1 and 2. Avoidant disorder and the phobic disorders were represented by single items in the OCHS checklist and obsessive-compulsive disorder by two items. Thu s, the OCHS checklist did not contain enough items (symptoms) on these disorders to operationalize the full DSM-III-R criteria. As the wording of the OCHS checklist items was not identical with that of the DSM-III-R symptoms, the authors decided to define essential symptoms that would be specific for DAD and SAD and would act as gates for qualifying on the less specifi c auxiliary symptoms. The auxiliary symptoms would then count toward meeting the criteria for the disorder (DAD or SAD) only if the essential symptoms were present. Individual DCHS checklist items were then matched with the DSM-III-R symptoms . An attempt was made to find one checklist item per DSM-III-R symptom, based on the match 753
BOWEN ET AL. TABLE
I. Overanxious Disorder: DSM-III-R Symptoms and Items
from the Ontario Child Health Checklist DSM-lll-R Criteria (abbreviated) Essential symptoms Excessive or unrealistic anxiety Excessive or unrealistic worry Auxiliary symptoms I. Worry about future events 2. Concern about the appropriateness of past behavior 3. Worry about competence in one or more areas 4. Somatic complaints
5. Marked self-consciousness 6. Excessive need for reassurance about a variety of concerns 7. Marked feelings of tension or inability to relax
Study Checklist Item I am too fearful or anxious I worry a lot I worry that terrible things might happen.
I feel that I have to be perfect Nausea, feel sick, stomach aches or cramps, vomiting, throwing up,headaches I am self-conscious or easily embarrassed I worry about doing the wrong thing I am afraid that I might think or do something bad I am nervous or tense
between the content of the item and the symptom. Where two or more checklist items appeared to represent a symptom, correlation coefficients were determined between the items. If the correlation was low «0.3), indicating unrelatedness or high (>0.7) indicating redundancy, the item with the best face resemblance to the DSM-/lI-R criterion was used, otherwise multiple items were retained. For OAD, the DSM-I/l-R symptom "excessive or unrealistic concern about the appropriateness of past behavior," was not represented by an item in the OCHS checklist (Table I). The somatic complaints symptom is represented by four individual checklist items. This seems consistent with DSMI/l-R, where anyone of several somatic complaints could establish the presence of the symptom. "Excessive need for reassurance about a variety of concerns" was represented by two OCHS checklist items with a Pearson's correlation coefficient of 0.34, both of which were retained. For SAD, using DSM-/lI-R, two essential symptoms were defined (Table 2). These essential symptoms were repeated in the DSM-III-R eighth and ninth auxiliary symptoms, "complaints of excessive distress in anticipation of separation" and" complaints of excessive distress when separated, " and consequently the authors did not include them in the list of auxiliary symptoms. "Unrealistic and persistent worry that an untoward calamitous event will separate the child from a major attachment figure" was not represented by a checklist item. The somatic complaints auxiliary symptoms are shared by OAD and SAD with the provision, of course, that the essential symptoms are present. Measures ofImpairment and Utilization
Seven measures of impairment and utilization were constructed. An impairment indicated inferior functioning in a 754
TABLE
2. Separation Anxiety Disorder: DSM-III-R Symptoms and Items from the Ontario Child Health Study Checklist
DSM-lll-R Criteria (abbreviated) Essential symptoms Excessi ve anxiety concerning separation from those to whom the child is attached Auxiliary symptoms 1. Worry about harm to attachment figures or fear that they will leave and not return 2. Worry that a calamity will separate the child from a major attachment figure 3. Reluctance or refusal to go to school 4. Reluctance or refusal to go to sleep or sleep away from home 5. Avoidance of being alone 6. Nightmares on the theme of separation 7. Complaints of physical symptoms
Study Checklist Item I become overly upset when leaving someone I am close to I become overly upset while away from someone I am close to I worry that something bad will happen to people I am close to
I am afraid of going to school I have trouble sleeping
I don't like to be alone I have nightmares Nausea, feel sick, stomachaches or cramps, vomiting, throwing up, headaches
8. Signs or complaints or distress in anticipation of separation 9. Signs or complaints of distress when separated
particular area of a youth's life and was measured independently of disorder. Each measure referred to the time period now or in the past 6 months. Based on replies by the youths themselves, these were: (1) "problems getting along," defined as youths with frequent problems getting along with any one of other children, teachers or his/her family and derived from three items; (2) "isolated youth," defined as youth with limited number and contact with friends derived from two items; (3) "poor competence," defined as how well the youth performed in sports and other activities such as music and hobbies, compared to other youth of the same age and measured by two items; (4) "needs professional help," identified youth perceived by themselves as having had emotional or behavioral problems for which professional help is or was needed and is derived from three items; (5) "poor participation, " described youth who do not take part in sports, lessons, or clubs with adult coaching or leadership and is derived from three items. The parent source was used for: (6) "poor school performance," which described a perception by the parent of the youth not doing well at school and derived from one item; and (7) "use of mental health or social services," identified the youth being the focus of a consultation at a mental health, social, or judicial facility and was derived from six items. Criteria for the diagnoses of conduct disorder and hyperactivity (combined as externalizing disorders) are described by Boyle et al. (1987). Briefly, scale scores were derived from the relevant checklist items and thresholds for disorders were developed from a sample of the children who were actually J. Am. Acad. Child Adolesc. Psychiatry, 29:5, September I 990
PREYALENCE OF ADOLESCENT ANXIETY DISORDER TABLE
3. Prevalence ofOveranxious Disorder (OAD) and Separation Anxiety Disorder (SAD) by Duration/Intensity Threshold Disorder
Source
Threshold A U N (%)
Threshold If N (%)
Threshold CC N (%)
Threshold A N (%)
Threshold B N (%)
Threshold C N(%)
Parent Youth
0 46 (3.6)
19 (1.5) 199 (15.3)
100 (7.3) 421 (32.3)
0 31 (2.4)
35 (2.7) 339 (26.0)
159 (12.2) 549 (42.2)
Note: The diagnoses of OAD and SAD required the presence of one or both of the essential symptoms and four and three auxiliary symptoms for OADandSAD, respectively. "Threshold A: At least "often or very true" for essential and auxiliary items. bThreshold B: At least "often or very true" for essential and at least "sometimes or somewhat true" for auxiliary items. 'Threshold C: At least "sometimes or somewhat true" for essential and auxiliary items.
seen by child psychiatrists. The criteria for the diagnosis of depression are described by Fleming et al. (1989). This was derived from the relevant checklist items. A core symptom of depression was required, and, in addition, four DSM-III symptoms had to be present. Steps in the Analysis
The first objective was to evaluate the effect on prevalence of different ways of operationalizing DSM-III-R criteria for OAD and SAD. The authors first changed the duration/intensity threshold for symptoms to qualify as present. The parent and youth sources were analyzed separately. (The thresholds are specified in the legend to Table 3). Second, the authors examined the requirement for the presence of one or both essential symptoms. Third, the number of auxiliary symptoms required was changed. Fourth, the responses of the youth classified as suffering from OAD and SAD were examined for the presence of other psychiatric disorders and the presence of the measures of impairment. Finally, the DSM-III-R diagnostic criteria were treated as scales and the internal consistencies were determined. Results Questionnaires were completed by parents and teachers of the children aged 4 to 11 years. Parents and teachers identified three children and two children, respectively, as suffering from OAD, and four and one children as suffering from SAD, when the criteria required were high duration/intensity for each symptom, the presence of one or both of the essential features of the disorders, and the presence of four and three auxiliary symptoms, respectively, for OAD and SAD. These are the criteria described later in this paper that gave the most plausible prevalences for OAD and SAD in youth aged 12 to 16 years. Because of the extremely small prevalences in children aged 4 to 11 years, the authors assume that either the criteria or method or both were not appropriate for young children. The rest of the paper focusses on youth aged 12 to 16 years. Among the 1,299 youth, there were 557 (42.9%) in the 12 to 13-year age range and 742 (57.1 %) in the 14 to 16-yearage range. There were almost equal numbers of males and females in both age ranges. Questionnaires were completed by parents, youth, and teachers on the youth aged 12 to 16 years. Parents and youth completed the questionnaires at home, in private, under the supervision of an interviewer. The parent l.Am.Acad. ChildAdolesc. Psychiatry, 29:5, September1990
and youth sources were used in the analysis of duration/intensity thresholds; subsequently, the youth source only was used. The teacher source was not used because of the large amount of missing data. Varying the thresholds for duration/intensity for the essential and auxiliary symptoms. The prevalences of OAD and SAD, using the published DSM-III-R criteria, for the three thresholds that the authors investigated, are presented in Table 3. The duration/intensity threshold A for the youth source that yielded a prevalence of 3.6% for OAD, and 2.4% for SAD was consistent with published prevalences from the literature. If threshold B was applied to the parent source alone, the authors would also have obtained plausible prevalences of 1.5% forOAD and 2.7% for SAD. The authors compared the overlap of cases between threshold A for the youth source and threshold B for the parent source. Only two youths were classified as suffering from OAD by both the parents and youth. No youth was classified as suffering from SAD by both parents and youth. The authors did compare the youth identified duration/intensity threshold A cases and the parent identified threshold B cases (Table 3) on the measures of impairment discussed below and in Table 6 (data not presented). The authors found that parents and youth were consistent in reporting impairments among cases identified by that particular source, but that there was little agreement across sources. Mindful of the fact that parents might not possess useful information on the internal symptoms of anxiety disorders, and that the information was being obtained from youths aged 12 to 16 and that they might therefore place more reliance on self-report, the authors decided to use only the youth source in additional analysis (Kashani and Orvaschel, 1988). Requiring one or both ofthe essential symptoms. ForOAD, using the at least" often or very true" threshold for responses, the youth source only, and considering the essential symptoms alone, 17.4% of the youth qualified when either or both essential symptoms were required, and 2.5% qualified when both essential symptoms were required. Since the application of the auxiliary symptoms would have reduced the prevalence even further, the authors decided to require either or both of the essential symptoms. The Pearson's r correlation between the two essential symptoms was 0.36. For SAD, using the same strategy, the prevalences were 33.5% when either one or both essential symptoms were required and 17.2% when both essential symptoms were required. The correlation be-
755
BOWEN ET AL.
4. Frequency and Percent of Youth Meeting the Criteria for Overanxious Disorder (OAD) and Separation Anxiety Disorder (SAD), According to the Number ofAuxiliary Symptoms Present. One ofBoth Essential Symptoms Having Been Present TABLE
No. of Auxiliary Symptoms
Frequency
I
2 3 4
5 6 or more
37 55 48 34 31 13 2
16.9 14.1 9.9 6.2 3.6 a 1.2 0.2
147 65 20 9 2
BothOAD and SAD (N= 8) N (%)
1 (3) 8 (21)
5 (22) 2 (9)
0 4 (50)
7 (18)
1 (4)
1 (13)
22 (59)
15 (65)
3 (38)
18.7 7.4
2.4' 0.9 0.2
"Threshold of auxiliary symptoms for DSM-III-R.
tween the essential symptoms was 0.69, so, as with OAD, it was decided to require either one or both essential symptoms. Requiring different numbers ofauxiliary symptoms. Using the at least" often or very true" response threshold for youth who had either one or both of the essential symptoms present (Threshold A, Table 3), the authors examined the prevalences when different numbers of auxiliary symptoms scored" often or very true" were required for the diagnosis of OAD and SAD. The results are presented in Table 4. For OAD, setting the threshold at four or more auxiliary symptoms yields a population prevalence of 3.6% which is consistent with the existing literature. For SAD, setting the threshold at three or more auxiliary symptoms yields a population prevalence of2.4%, also consistent with the literature. Among the 46 youths classified with OAD, there were 37 females and nine males and among 31 youths with SAD, 25 females and six males. Overlap with Other Disorders and Impairment
The authors investigated the overlap of the classified cases of OAD and SAD with conduct disorder and hyperactivity (combined as externalizing disorders), and depression. The data are presented in Table 5. There is frequent overlap with the externalizing disorders and depression, but 59% of the cases ofOAD and 65% of the cases of SAD have no depression or externalizing disorder. The authors then created four groups of youth with pure anxiety disorders, externalizing disorder, depression, and no disorder for comparison on seven impairments. The cases with' 'pure" OAD and SAD were combined in a single category to increase the numbers. The seven impairments were conceived as external validators of the categories of disorder. 756
One or both externalizing disorders Depression Depression and one or both externalizing disorders No depression or externalizing disorder (pure anxiety disorder)
SAD only, NoOAD W=23) N (%)
Note: Percentages do not always add to 100% because of rounding of digits.
Separation anxiety disorder (SAD) 1 2 3 4 5 or more
OADonly, No SAD (N= 38) N (%)
Cumulative % in Population Prevalence
Overanxious disorder (OAD)
o
5. Overlap ofDiagnoses in Youth with Overanxious Disorder (OAD) and Separation Anxiety Disorder (SAD) with Externalizing Disorders (Conduct Disorder and Hyperactivity) and Depression
TABLE
The results are presented in Table 6. In general, the youth with anxiety disorders are almost as impaired as youth with externalizing disorders and more impaired than youth with no disorder. All of the overall chi-squares for the 4 by 2 tables (four diagnostic groups by impairment, present or not) were significant at the p < 0.01 level with 3 degrees offreedom, except for poor participation. It appears that anxious youth do not report being socially isolated or friendless (isolated youth), while depressed youth score relatively high on this impairment. The anxiety disorders do not appear to interfere with school performance as perceived by parents, while 11.1 % of the parents of depressed youth report poor school performance. Internal Consistency ofthe Scales
The diagnostic instruments for OAD and SAD were treated as scales and examined for internal consistency by multiple correlations among symptoms and calculation ofCronbach's alpha coefficient. Items were scored as a scale score of 2 = 1 (positive) and a scale score of 0 or 1 = 0 (negative). When there was more than one checklist item per DSM-III-R symptom, a positive score on any item counted as positive for the symptom. For OAD, the mean interitem correlation for all of items for the whole sample of adolescents with complete data (N = 1221)wasO.21 and the standardized alpha was 0.68. The item that correlated least with the sum of other items was "somatic complaints" which had a squared multiple correlation of 0.07, and alpha of 0.68 if the item was deleted. "Somatic complaints" correlated better with items on the SAD scale (squared multiple correlation 0.13, alpha 0.46, if item deleted). The mean interitem correlation of the SAD items was 0.12 and the standardized alpha 0.51. For SAD, the two items that correlated least with the sum of other items were "I don't like to be alone" (squared multiple correlation 0.01, alpha 0.61, ifitem deleted), and "I am afraid to go to school" (squared multiple correlation 0.02, alpha 0.51, if item deleted). J. Am. Acad. Child Adolesc . Psychiatry. 29:5, September 1990
PREVALENCE OF ADOLESCENT ANXIETY DISORDER TABLE
6. Youth with Anxiety Disorders, Externalizing Disorders, and Depression Only, Compared by Measures of Impairment Use of Mental Health or Social Services %
Isolated Youth %
Poor Competence %
Needs Professional Help %
Poor Participation %
20.5
0
10.3
10.3
10.8
2.5
15.8
26.3
1.6
13.1
7.1
8.3
16.1
6.8
30.6
8.0
18.0
6.3
6.1
11.1
12.8
6.2
1.0
4.9
1.6
4.4
6.1
5.0
Problems Getting Along %
Poor School Performance %
Anxiety disorder only (N= 40)
Externalizing disorder only (N= 61)
Depression only (N= 50)
No disorder or normals (N = 1,048)
Discussion Prevalence
The authors determined the prevalence of OAD and SAD to be 3.6% and 2.4%, respectively, among youth aged 12 to 16 years. Comparison of rates between different studies is difficult because of different age ranges studied and different methodologies. A New Zealand study obtained rates ofOAD and SAD of2.9% and 3.5% among l l-year-olds (Anderson et al., 1987). An American study of 14 to 16-year-olds found that 17.3 % fulfilled DSM-1II-R criteria for one of the anxiety diagnoses, but 8.7% were defined as likely cases (Kashani and Orvaschel, 1988). A Puerto Rican study estimated that 6.8% of 4 to 16-year-olds suffered from SAD, but4.7% were defined as cases (Bird et al., 1988). The rate of anxiety disorder among 10- and ll-year-olds was 1.4% in the Isle of Wight, but the criteria and instruments were different, and parent and teacher sources were used (Rutter et al., 1970). The female to male ratio of 4: 1 for OAD and 6: 1 for SAD is higher than the ratio found in most other studies, but the trend is consistent with most other community studies (Orvaschel and Weissman, 1987). The higher female to male ratio becomes established in late adolescence for depression and presumably for anxiety also (Rutter, 1988). The authors are less confident about the classification of SAD compared with OAD because our ability to operationalize the DSM-III-R criteria for SAD was not as close as for OAD, e.g., the OCHS checklist items for symptoms four and six for SAD (Table 2) appear overinclusive compared with the DSM-III-R criteria. In addition, the symptom cohesiveness is not as high for SAD as for OAD. Two of the ambiguities on the boundaries of SAD are whether this condition is an early manifestation of adult panic and agoraphobia and the extent of the overlap between SAD and school refusal (Gittleman, 1987; Last et al., 1987b). DSM-III-R Diagnostic Criteria It is generally accepted that prevalence rates are sensitive to differences in sources and thresholds used to define the disorder (Fleming et al., 1989). First, the authors determined that requiring a high level of certainty for the presence of individual essential and auxiliary symptoms yields the most plausible prevalence. l.Am.Acad. Child Adolesc. Psychiatry, 29:5, September 1990
Second, defining essential and auxiliary symptoms was necessary in this study because our questionnaire items did not match the DSM-lII-R criteria exactly. This practical requirement seems to reflect the spirit of the DSM-III -R criteria where a general descriptive statement is made about the nature of the disorder as "evidenced" or "indicated" by more specific criteria (American Psychiatric Association, 1987, p. 21). The authors determined that meeting a minimum of only one of the essential symptoms was necessary. This seems to be the intent of the description for OAD and SAD in DSM-lII-R, although the manual is not explicit on this point (American Psychiatric Association, 1987, pp. 60 and 64). Third, when the number of auxiliary symptoms was varied, it was found that the number of required symptoms that produced the most plausible prevalence was the number specified by DSM-lII-R. For SAD, the authors were less certain that requiring four auxiliary symptoms rather than three, which would have reduced the prevalence to 0.9% from 2.4% was not appropriate. The authors determined the internal consistencies of the checklist items to understand better the relationship between the items used to operationalize the DSM-III -R criteria. These were low in the whole sample and even lower in the subgroup with one of the core symptoms present, probably because of the reduced variability of scores in the latter group. The data suggest that somatic symptoms are more closely tied to SAD than to OAD. A similar finding has been reported in a study of 95 hospitalized children with psychiatric disorders. The number of somatic symptoms did not differ between those with OAD and the rest, while those with SAD had more somatic symptoms than the rest of the sample (Livingston et al., 1987). In the present study, for SAD, "school refusal" and "avoids being alone" had the lowest correlations with the rest of the items. The nature of the overlap between school refusal and SAD is not clear (Last et al., 1987b). On conceptual grounds, it would seem that not wanting to separate from attachment persons is not the same as not wanting to be alone. Overlap with Other Psychiatric Disorders
The overlap of anxiety with depression and the externalizing disorders observed here is consistent with other research findings and has been extensively discussed in the child and adult psychiatric literature (Anderson et al., 1987; Last et al. ,
757
BOWEN ET AL.
1987a; Bird et al. , 1988; Kashani and Orvaschel, 1988; Last, 1988). The authors classified more than one-half of subjects with anxiety disorders alone , in agreement with the findings from the New Zealand and Puerto Rican studies (Anderson et al., 1987; Bird et al., 1988). Thefinding of a large proportion of pure anxiety cases lends support to the validity of anxiety disorders as distinct syndromes in adolescence. External Validity
On five out of seven measures of impairment, anxiety disordered youth were as impaired as depressed youth and youth with externalizing disorders, and more impaired than youth with no disorders (Table 6). Two measures of impairment appear to distinguish the youth with anxiety from those with major depression. Anxious youth appear to be less likely than depressed youth to report themselves as isolated or friendless, or to be perceived by their parents as performing poorly in school. These data provide some tentative support from the area of social functioning for a distinction between anxiety and depressive disorders. A study that compared children with anxiety disorders to children with conduct disorders found that the anxious children were less handicapped than the externalizing children, but more handicapped than normal children (Reeves et al., 1987). Children with anxiety are reported to be more socially withdrawn in reviews of the literature but in this study did not admit to being socially isolated (Reeves et al., 1987; Werry et al., 1987). The authors emphasize that their method was a reasonable attempt to operationalize the DSM-III-R criteria and was not identical to the DSM -III-R . In summary, this study found the prevalence of overanxious disorder (OAD) and separation anxiety disorder (SAD) to be 3.6% and 2.9 %, respectively, among adolescents aged 12 to 16 years. Anxiety disorders are more common among girls. Youth perception of limited contact with friends, and parent perception of poor school performance distinguish anxiety disordered youth from depressed youth . It was found that by requiring a high duration! intensity criterion for a symptom to count as positive, either one or both of the DSM-III-R essential features of the disorders and the presence of four and three auxiliary symptoms, respectively, for OAD and SAD yielded the most plausible community prevalences.
758
References Achenbach, T. M. & Edelbrock , C. S . (1981), Behavioral problems and competencies reported by parents of normal and disturbed children aged 4 through 16. Monogr.Soc.Res.ChildDev .,46:1-78. American Psychiatric Association (1987), Diagnostic and Statistical Manual of Mental Disorders (3rd ed, Revised) . Washington, DC: American Psychiatric Association . Anderson, J. C. , Williams , S., McGee , R. & Silva, P. A. (1987) , DSM-III disorders in pre-adolescent children . Arch . Gen . Psychiatry, 44:69-76. Bird, H. R., Canino, G., Rubio-Stipec, M. et al. (1988), Estimates of the prevalence of childhood maladjustment in a community survey in Puerto Rico. Arch. Gen. Psychiatry, 45:1120-1126. Boyle, M. H., Offord, D. R., Hoffman, H. G. et al. (1987) , Ontario Child Health Study: 1. methodology. Arch. Gen. Psychiatry, 44:826831. Fleming, J. E. , Offord, D. & Boyle, M. H. (1989) , Prevalence of childhood and adolescent depression in the community. Br .J . Psychiatry , 155:647-654. Gittleman, R. (1987), Childhood anxiety disorders. In: Anxiety Disorders of Childhood, ed. R. Gittleman . New York: Guilford, pp. 101125. Kashani, J. & Orvaschel,H . (1988), Anxiety disorders in mid-adolescence. Am.J. Psvchiatry , 145:960-964. Last, C. G. (1988), Anxiety disorders in childhood and adolescence. In: Handbook ofAnxiety Disorders, eds . C. G. Last & M. Hersen, New York: Pergamon , pp. 531-540. - - Hersen , M. , Kazdin, A. E.,-·Finkelstein, R. & Strauss , C . C. (l987a), Comparison of DSM-III separation anxiety and overanxious disorders . J . Am . Acad. Child Adolesc . Psychiatry, 26:527531. -. -Francis , G., Hersen, M., Kazdin, A. E. & Strauss , C. C . ( 1987b), Separation anxiety and school phobia : a comparison using DSM-III criteria. Am .J. Psychiatry , 144:653-667. Livingston, R. Taylor, J. L., & Crawford, S. L. (1987), A study of somatic complaints and psychiatric diagnosis in children . J . Am . Acad. ChildAdolesc . Psychiatry , 27:185-187 . Orvaschel, H. & Weissman, M.M. (1987), Epidemiology of anxiety disorders in children . In: Anxiety Disorders of Childhood, cd. R. Gittleman. New York: Guilford, pp. 58-72. Reeves , J. C ., Werry, J. S., Elkind , G. S. & Zametkin, A. (1987), Attention deficit, conduct, oppositional, and anxiety disorders in children. II . Clinical characteristics . J . Am. Acad. Child Adolesc .
Psychiatry , 26:144-155 . Rutter, M. (1988), Epidemiological approaches to developmental psychopathology . Arch .Gen .Psy chiatry,45 :483-495 . --Tizard ,1. & Whitmore , K. (1970) , Education, Health and Behavior . London : Longmans, pp. 183-187. Werry, J. S., Reeves , 1. C. & Elkind, G. S. (1987), Attention deficit , conduct, oppositional, and anxiety disorders in children: I. a review of research on differentiating characteristics. J . Am. Acad. Child Adolesc. Psychiatry. 26: 133-143.
J . Am. Acad. Child Ado/esc. Ps ychiatry, 29 :5, September 1990
Abstract. Data from a community epidemi ological study of 1,869 famili es (Ontario Child Health Study) was used to evaluate the effect of different ways of operationalizing DSM-/II-R criteria for over anxiou s disorder (OAD) and separation anxiety disorder (SAD) among adolesc ents aged 12 to 16. The authors determ ined that a high threshold for symptoms to qualify as present, the presence of one or both of the essential symptoms , and the presence of four or more auxiliary sympt oms for OAD and three or more for SAD gave prevalence of OAD of 3.6 % and SAD of 2.4%. There was high overlap between the presence of OAD and SAD and externalizing disorder and depression, but one- half of youth with GAD and SAD had pure anxiety disorder. Youth with GAD and SAD were ju st as impaired as youth with externalizing disorder and depression, except that they admitted to less social isolation and their schoolwork was less affected . J . Am . Acad . ChildAdolesc . Psychiatry. 1990, 29,5 :753-758 . Key Words: prev alence , classificat ion, DSM-II/-R , adolescence , anxiety disorder. Overanxious disorder (OAD) and separation anxiety disorder (SAD) of childhood have rece ived little research attention in clinical studies and even less attention in community studies (Last et aI. , 1987a; Orvaschel and Weissman , 1987). The Ontario Child Health Study (OCHS) data provided an opportunity to evaluate the meaningfulness of different ways of operationalizing DSM-III-R criteria for OAD and SAD in a large community sample (N = 3,294) , and how this affects the prevalence of the disorders (American Psychiatric Association , 1987; Boyle et aI., 1987).
Method Survey Design and Data Coll ection
The OCHS was a cross-sectional interview survey of 1,869 families including 3,294 child ren, ages 4 to 16, conducted in the spring of 1983. The survey methodology and instrumentation have been described by Boyle et aI. (1987) and are briefly summarized here . The target population of the OCHS included all children born between January 1, 1966, through January 1, 1979, whose usual residence was a household dwelling in Ontario . The sampling frame was the 1981 census and the sampling
AcceptedMay 24 , /990 . Dr . Bowen is Professor of Psychiatry. University of Saskatchewan. Saskatoon , Saskatchewan .Dr. Offord isProfessor ofP sychiatry , and Dr . Boyle is Assistant Professor of Psychiatry. McMaster University. Hamilton. Ontario . This work is supported by grams/rom the Ontario Ministry of Community and Social Services and was carried out by the Child Epidemiology Unit. Department of Psychiatry, McMast er University and the Child and Family Center , Chedoke Division, Chedoke-McMaster Hospitals, Hamilton, Ontario. Dr . Bowen was supported by a sabbatical leave research grantfrom the University of Saskatchewan . Ms. Yvonne Racine made valuable contributions to various aspects of this work. especially to data analysis . Reprint requests to Dr . Offord. Chedoke-McMaster Hosp itals , Chedoke Division, Administration Building, Box 2000, Station A , Hamilton. Ontario, Canada. L8N 325. 0890-8567/90/2905-0753$02.00/0© 1990 by the American Academy of Child and Adolescent Psych iatry .
unit was household dwellings . Sampling selection was done by stratified, clustered , random sampling from the census file. Ninety-one percent of all elig ible households agreed to participate , and the refusal rate was low, 3.9%. Interviewing was done by trained interviewers employed by Statistics Canada. The basic pool of items for measuring childhood psychopathology was provided by the Child Behavior Checklist (CBCL) with some modifications and additions (Achenbach and Edelbrock, 1981; Boyle et al., 1987). The OCHS checklist asked about symptoms occurring at present or in the past 6 months. Each item had three response categories: " never or not true " scored as 0, "sometimes or somewhat true" scored as 1, and " often or very true" scored as 2. The first respon se category "never or not true" is different from the CBCL category' 'not true. " Measurement ofDisorder
By comparing the criteria in DSM-III-R with items on the OCHS checklist, the authors concluded that it would be feasible to classify youth as suffering from OAD and/or SAD (American Psychiatric Association , 1987). The DSM-III-R criteria (abbreviated) and corresponding questionnaire items for OAD and SAD are given in Tables 1 and 2. Avoidant disorder and the phobic disorders were represented by single items in the OCHS checklist and obsessive-compulsive disorder by two items. Thu s, the OCHS checklist did not contain enough items (symptoms) on these disorders to operationalize the full DSM-III-R criteria. As the wording of the OCHS checklist items was not identical with that of the DSM-III-R symptoms, the authors decided to define essential symptoms that would be specific for DAD and SAD and would act as gates for qualifying on the less specifi c auxiliary symptoms. The auxiliary symptoms would then count toward meeting the criteria for the disorder (DAD or SAD) only if the essential symptoms were present. Individual DCHS checklist items were then matched with the DSM-III-R symptoms . An attempt was made to find one checklist item per DSM-III-R symptom, based on the match 753
BOWEN ET AL. TABLE
I. Overanxious Disorder: DSM-III-R Symptoms and Items
from the Ontario Child Health Checklist DSM-lll-R Criteria (abbreviated) Essential symptoms Excessive or unrealistic anxiety Excessive or unrealistic worry Auxiliary symptoms I. Worry about future events 2. Concern about the appropriateness of past behavior 3. Worry about competence in one or more areas 4. Somatic complaints
5. Marked self-consciousness 6. Excessive need for reassurance about a variety of concerns 7. Marked feelings of tension or inability to relax
Study Checklist Item I am too fearful or anxious I worry a lot I worry that terrible things might happen.
I feel that I have to be perfect Nausea, feel sick, stomach aches or cramps, vomiting, throwing up,headaches I am self-conscious or easily embarrassed I worry about doing the wrong thing I am afraid that I might think or do something bad I am nervous or tense
between the content of the item and the symptom. Where two or more checklist items appeared to represent a symptom, correlation coefficients were determined between the items. If the correlation was low «0.3), indicating unrelatedness or high (>0.7) indicating redundancy, the item with the best face resemblance to the DSM-/lI-R criterion was used, otherwise multiple items were retained. For OAD, the DSM-I/l-R symptom "excessive or unrealistic concern about the appropriateness of past behavior," was not represented by an item in the OCHS checklist (Table I). The somatic complaints symptom is represented by four individual checklist items. This seems consistent with DSMI/l-R, where anyone of several somatic complaints could establish the presence of the symptom. "Excessive need for reassurance about a variety of concerns" was represented by two OCHS checklist items with a Pearson's correlation coefficient of 0.34, both of which were retained. For SAD, using DSM-/lI-R, two essential symptoms were defined (Table 2). These essential symptoms were repeated in the DSM-III-R eighth and ninth auxiliary symptoms, "complaints of excessive distress in anticipation of separation" and" complaints of excessive distress when separated, " and consequently the authors did not include them in the list of auxiliary symptoms. "Unrealistic and persistent worry that an untoward calamitous event will separate the child from a major attachment figure" was not represented by a checklist item. The somatic complaints auxiliary symptoms are shared by OAD and SAD with the provision, of course, that the essential symptoms are present. Measures ofImpairment and Utilization
Seven measures of impairment and utilization were constructed. An impairment indicated inferior functioning in a 754
TABLE
2. Separation Anxiety Disorder: DSM-III-R Symptoms and Items from the Ontario Child Health Study Checklist
DSM-lll-R Criteria (abbreviated) Essential symptoms Excessi ve anxiety concerning separation from those to whom the child is attached Auxiliary symptoms 1. Worry about harm to attachment figures or fear that they will leave and not return 2. Worry that a calamity will separate the child from a major attachment figure 3. Reluctance or refusal to go to school 4. Reluctance or refusal to go to sleep or sleep away from home 5. Avoidance of being alone 6. Nightmares on the theme of separation 7. Complaints of physical symptoms
Study Checklist Item I become overly upset when leaving someone I am close to I become overly upset while away from someone I am close to I worry that something bad will happen to people I am close to
I am afraid of going to school I have trouble sleeping
I don't like to be alone I have nightmares Nausea, feel sick, stomachaches or cramps, vomiting, throwing up, headaches
8. Signs or complaints or distress in anticipation of separation 9. Signs or complaints of distress when separated
particular area of a youth's life and was measured independently of disorder. Each measure referred to the time period now or in the past 6 months. Based on replies by the youths themselves, these were: (1) "problems getting along," defined as youths with frequent problems getting along with any one of other children, teachers or his/her family and derived from three items; (2) "isolated youth," defined as youth with limited number and contact with friends derived from two items; (3) "poor competence," defined as how well the youth performed in sports and other activities such as music and hobbies, compared to other youth of the same age and measured by two items; (4) "needs professional help," identified youth perceived by themselves as having had emotional or behavioral problems for which professional help is or was needed and is derived from three items; (5) "poor participation, " described youth who do not take part in sports, lessons, or clubs with adult coaching or leadership and is derived from three items. The parent source was used for: (6) "poor school performance," which described a perception by the parent of the youth not doing well at school and derived from one item; and (7) "use of mental health or social services," identified the youth being the focus of a consultation at a mental health, social, or judicial facility and was derived from six items. Criteria for the diagnoses of conduct disorder and hyperactivity (combined as externalizing disorders) are described by Boyle et al. (1987). Briefly, scale scores were derived from the relevant checklist items and thresholds for disorders were developed from a sample of the children who were actually J. Am. Acad. Child Adolesc. Psychiatry, 29:5, September I 990
PREYALENCE OF ADOLESCENT ANXIETY DISORDER TABLE
3. Prevalence ofOveranxious Disorder (OAD) and Separation Anxiety Disorder (SAD) by Duration/Intensity Threshold Disorder
Source
Threshold A U N (%)
Threshold If N (%)
Threshold CC N (%)
Threshold A N (%)
Threshold B N (%)
Threshold C N(%)
Parent Youth
0 46 (3.6)
19 (1.5) 199 (15.3)
100 (7.3) 421 (32.3)
0 31 (2.4)
35 (2.7) 339 (26.0)
159 (12.2) 549 (42.2)
Note: The diagnoses of OAD and SAD required the presence of one or both of the essential symptoms and four and three auxiliary symptoms for OADandSAD, respectively. "Threshold A: At least "often or very true" for essential and auxiliary items. bThreshold B: At least "often or very true" for essential and at least "sometimes or somewhat true" for auxiliary items. 'Threshold C: At least "sometimes or somewhat true" for essential and auxiliary items.
seen by child psychiatrists. The criteria for the diagnosis of depression are described by Fleming et al. (1989). This was derived from the relevant checklist items. A core symptom of depression was required, and, in addition, four DSM-III symptoms had to be present. Steps in the Analysis
The first objective was to evaluate the effect on prevalence of different ways of operationalizing DSM-III-R criteria for OAD and SAD. The authors first changed the duration/intensity threshold for symptoms to qualify as present. The parent and youth sources were analyzed separately. (The thresholds are specified in the legend to Table 3). Second, the authors examined the requirement for the presence of one or both essential symptoms. Third, the number of auxiliary symptoms required was changed. Fourth, the responses of the youth classified as suffering from OAD and SAD were examined for the presence of other psychiatric disorders and the presence of the measures of impairment. Finally, the DSM-III-R diagnostic criteria were treated as scales and the internal consistencies were determined. Results Questionnaires were completed by parents and teachers of the children aged 4 to 11 years. Parents and teachers identified three children and two children, respectively, as suffering from OAD, and four and one children as suffering from SAD, when the criteria required were high duration/intensity for each symptom, the presence of one or both of the essential features of the disorders, and the presence of four and three auxiliary symptoms, respectively, for OAD and SAD. These are the criteria described later in this paper that gave the most plausible prevalences for OAD and SAD in youth aged 12 to 16 years. Because of the extremely small prevalences in children aged 4 to 11 years, the authors assume that either the criteria or method or both were not appropriate for young children. The rest of the paper focusses on youth aged 12 to 16 years. Among the 1,299 youth, there were 557 (42.9%) in the 12 to 13-year age range and 742 (57.1 %) in the 14 to 16-yearage range. There were almost equal numbers of males and females in both age ranges. Questionnaires were completed by parents, youth, and teachers on the youth aged 12 to 16 years. Parents and youth completed the questionnaires at home, in private, under the supervision of an interviewer. The parent l.Am.Acad. ChildAdolesc. Psychiatry, 29:5, September1990
and youth sources were used in the analysis of duration/intensity thresholds; subsequently, the youth source only was used. The teacher source was not used because of the large amount of missing data. Varying the thresholds for duration/intensity for the essential and auxiliary symptoms. The prevalences of OAD and SAD, using the published DSM-III-R criteria, for the three thresholds that the authors investigated, are presented in Table 3. The duration/intensity threshold A for the youth source that yielded a prevalence of 3.6% for OAD, and 2.4% for SAD was consistent with published prevalences from the literature. If threshold B was applied to the parent source alone, the authors would also have obtained plausible prevalences of 1.5% forOAD and 2.7% for SAD. The authors compared the overlap of cases between threshold A for the youth source and threshold B for the parent source. Only two youths were classified as suffering from OAD by both the parents and youth. No youth was classified as suffering from SAD by both parents and youth. The authors did compare the youth identified duration/intensity threshold A cases and the parent identified threshold B cases (Table 3) on the measures of impairment discussed below and in Table 6 (data not presented). The authors found that parents and youth were consistent in reporting impairments among cases identified by that particular source, but that there was little agreement across sources. Mindful of the fact that parents might not possess useful information on the internal symptoms of anxiety disorders, and that the information was being obtained from youths aged 12 to 16 and that they might therefore place more reliance on self-report, the authors decided to use only the youth source in additional analysis (Kashani and Orvaschel, 1988). Requiring one or both ofthe essential symptoms. ForOAD, using the at least" often or very true" threshold for responses, the youth source only, and considering the essential symptoms alone, 17.4% of the youth qualified when either or both essential symptoms were required, and 2.5% qualified when both essential symptoms were required. Since the application of the auxiliary symptoms would have reduced the prevalence even further, the authors decided to require either or both of the essential symptoms. The Pearson's r correlation between the two essential symptoms was 0.36. For SAD, using the same strategy, the prevalences were 33.5% when either one or both essential symptoms were required and 17.2% when both essential symptoms were required. The correlation be-
755
BOWEN ET AL.
4. Frequency and Percent of Youth Meeting the Criteria for Overanxious Disorder (OAD) and Separation Anxiety Disorder (SAD), According to the Number ofAuxiliary Symptoms Present. One ofBoth Essential Symptoms Having Been Present TABLE
No. of Auxiliary Symptoms
Frequency
I
2 3 4
5 6 or more
37 55 48 34 31 13 2
16.9 14.1 9.9 6.2 3.6 a 1.2 0.2
147 65 20 9 2
BothOAD and SAD (N= 8) N (%)
1 (3) 8 (21)
5 (22) 2 (9)
0 4 (50)
7 (18)
1 (4)
1 (13)
22 (59)
15 (65)
3 (38)
18.7 7.4
2.4' 0.9 0.2
"Threshold of auxiliary symptoms for DSM-III-R.
tween the essential symptoms was 0.69, so, as with OAD, it was decided to require either one or both essential symptoms. Requiring different numbers ofauxiliary symptoms. Using the at least" often or very true" response threshold for youth who had either one or both of the essential symptoms present (Threshold A, Table 3), the authors examined the prevalences when different numbers of auxiliary symptoms scored" often or very true" were required for the diagnosis of OAD and SAD. The results are presented in Table 4. For OAD, setting the threshold at four or more auxiliary symptoms yields a population prevalence of 3.6% which is consistent with the existing literature. For SAD, setting the threshold at three or more auxiliary symptoms yields a population prevalence of2.4%, also consistent with the literature. Among the 46 youths classified with OAD, there were 37 females and nine males and among 31 youths with SAD, 25 females and six males. Overlap with Other Disorders and Impairment
The authors investigated the overlap of the classified cases of OAD and SAD with conduct disorder and hyperactivity (combined as externalizing disorders), and depression. The data are presented in Table 5. There is frequent overlap with the externalizing disorders and depression, but 59% of the cases ofOAD and 65% of the cases of SAD have no depression or externalizing disorder. The authors then created four groups of youth with pure anxiety disorders, externalizing disorder, depression, and no disorder for comparison on seven impairments. The cases with' 'pure" OAD and SAD were combined in a single category to increase the numbers. The seven impairments were conceived as external validators of the categories of disorder. 756
One or both externalizing disorders Depression Depression and one or both externalizing disorders No depression or externalizing disorder (pure anxiety disorder)
SAD only, NoOAD W=23) N (%)
Note: Percentages do not always add to 100% because of rounding of digits.
Separation anxiety disorder (SAD) 1 2 3 4 5 or more
OADonly, No SAD (N= 38) N (%)
Cumulative % in Population Prevalence
Overanxious disorder (OAD)
o
5. Overlap ofDiagnoses in Youth with Overanxious Disorder (OAD) and Separation Anxiety Disorder (SAD) with Externalizing Disorders (Conduct Disorder and Hyperactivity) and Depression
TABLE
The results are presented in Table 6. In general, the youth with anxiety disorders are almost as impaired as youth with externalizing disorders and more impaired than youth with no disorder. All of the overall chi-squares for the 4 by 2 tables (four diagnostic groups by impairment, present or not) were significant at the p < 0.01 level with 3 degrees offreedom, except for poor participation. It appears that anxious youth do not report being socially isolated or friendless (isolated youth), while depressed youth score relatively high on this impairment. The anxiety disorders do not appear to interfere with school performance as perceived by parents, while 11.1 % of the parents of depressed youth report poor school performance. Internal Consistency ofthe Scales
The diagnostic instruments for OAD and SAD were treated as scales and examined for internal consistency by multiple correlations among symptoms and calculation ofCronbach's alpha coefficient. Items were scored as a scale score of 2 = 1 (positive) and a scale score of 0 or 1 = 0 (negative). When there was more than one checklist item per DSM-III-R symptom, a positive score on any item counted as positive for the symptom. For OAD, the mean interitem correlation for all of items for the whole sample of adolescents with complete data (N = 1221)wasO.21 and the standardized alpha was 0.68. The item that correlated least with the sum of other items was "somatic complaints" which had a squared multiple correlation of 0.07, and alpha of 0.68 if the item was deleted. "Somatic complaints" correlated better with items on the SAD scale (squared multiple correlation 0.13, alpha 0.46, if item deleted). The mean interitem correlation of the SAD items was 0.12 and the standardized alpha 0.51. For SAD, the two items that correlated least with the sum of other items were "I don't like to be alone" (squared multiple correlation 0.01, alpha 0.61, ifitem deleted), and "I am afraid to go to school" (squared multiple correlation 0.02, alpha 0.51, if item deleted). J. Am. Acad. Child Adolesc . Psychiatry. 29:5, September 1990
PREVALENCE OF ADOLESCENT ANXIETY DISORDER TABLE
6. Youth with Anxiety Disorders, Externalizing Disorders, and Depression Only, Compared by Measures of Impairment Use of Mental Health or Social Services %
Isolated Youth %
Poor Competence %
Needs Professional Help %
Poor Participation %
20.5
0
10.3
10.3
10.8
2.5
15.8
26.3
1.6
13.1
7.1
8.3
16.1
6.8
30.6
8.0
18.0
6.3
6.1
11.1
12.8
6.2
1.0
4.9
1.6
4.4
6.1
5.0
Problems Getting Along %
Poor School Performance %
Anxiety disorder only (N= 40)
Externalizing disorder only (N= 61)
Depression only (N= 50)
No disorder or normals (N = 1,048)
Discussion Prevalence
The authors determined the prevalence of OAD and SAD to be 3.6% and 2.4%, respectively, among youth aged 12 to 16 years. Comparison of rates between different studies is difficult because of different age ranges studied and different methodologies. A New Zealand study obtained rates ofOAD and SAD of2.9% and 3.5% among l l-year-olds (Anderson et al., 1987). An American study of 14 to 16-year-olds found that 17.3 % fulfilled DSM-1II-R criteria for one of the anxiety diagnoses, but 8.7% were defined as likely cases (Kashani and Orvaschel, 1988). A Puerto Rican study estimated that 6.8% of 4 to 16-year-olds suffered from SAD, but4.7% were defined as cases (Bird et al., 1988). The rate of anxiety disorder among 10- and ll-year-olds was 1.4% in the Isle of Wight, but the criteria and instruments were different, and parent and teacher sources were used (Rutter et al., 1970). The female to male ratio of 4: 1 for OAD and 6: 1 for SAD is higher than the ratio found in most other studies, but the trend is consistent with most other community studies (Orvaschel and Weissman, 1987). The higher female to male ratio becomes established in late adolescence for depression and presumably for anxiety also (Rutter, 1988). The authors are less confident about the classification of SAD compared with OAD because our ability to operationalize the DSM-III-R criteria for SAD was not as close as for OAD, e.g., the OCHS checklist items for symptoms four and six for SAD (Table 2) appear overinclusive compared with the DSM-III-R criteria. In addition, the symptom cohesiveness is not as high for SAD as for OAD. Two of the ambiguities on the boundaries of SAD are whether this condition is an early manifestation of adult panic and agoraphobia and the extent of the overlap between SAD and school refusal (Gittleman, 1987; Last et al., 1987b). DSM-III-R Diagnostic Criteria It is generally accepted that prevalence rates are sensitive to differences in sources and thresholds used to define the disorder (Fleming et al., 1989). First, the authors determined that requiring a high level of certainty for the presence of individual essential and auxiliary symptoms yields the most plausible prevalence. l.Am.Acad. Child Adolesc. Psychiatry, 29:5, September 1990
Second, defining essential and auxiliary symptoms was necessary in this study because our questionnaire items did not match the DSM-lII-R criteria exactly. This practical requirement seems to reflect the spirit of the DSM-III -R criteria where a general descriptive statement is made about the nature of the disorder as "evidenced" or "indicated" by more specific criteria (American Psychiatric Association, 1987, p. 21). The authors determined that meeting a minimum of only one of the essential symptoms was necessary. This seems to be the intent of the description for OAD and SAD in DSM-lII-R, although the manual is not explicit on this point (American Psychiatric Association, 1987, pp. 60 and 64). Third, when the number of auxiliary symptoms was varied, it was found that the number of required symptoms that produced the most plausible prevalence was the number specified by DSM-lII-R. For SAD, the authors were less certain that requiring four auxiliary symptoms rather than three, which would have reduced the prevalence to 0.9% from 2.4% was not appropriate. The authors determined the internal consistencies of the checklist items to understand better the relationship between the items used to operationalize the DSM-III -R criteria. These were low in the whole sample and even lower in the subgroup with one of the core symptoms present, probably because of the reduced variability of scores in the latter group. The data suggest that somatic symptoms are more closely tied to SAD than to OAD. A similar finding has been reported in a study of 95 hospitalized children with psychiatric disorders. The number of somatic symptoms did not differ between those with OAD and the rest, while those with SAD had more somatic symptoms than the rest of the sample (Livingston et al., 1987). In the present study, for SAD, "school refusal" and "avoids being alone" had the lowest correlations with the rest of the items. The nature of the overlap between school refusal and SAD is not clear (Last et al., 1987b). On conceptual grounds, it would seem that not wanting to separate from attachment persons is not the same as not wanting to be alone. Overlap with Other Psychiatric Disorders
The overlap of anxiety with depression and the externalizing disorders observed here is consistent with other research findings and has been extensively discussed in the child and adult psychiatric literature (Anderson et al., 1987; Last et al. ,
757
BOWEN ET AL.
1987a; Bird et al. , 1988; Kashani and Orvaschel, 1988; Last, 1988). The authors classified more than one-half of subjects with anxiety disorders alone , in agreement with the findings from the New Zealand and Puerto Rican studies (Anderson et al., 1987; Bird et al., 1988). Thefinding of a large proportion of pure anxiety cases lends support to the validity of anxiety disorders as distinct syndromes in adolescence. External Validity
On five out of seven measures of impairment, anxiety disordered youth were as impaired as depressed youth and youth with externalizing disorders, and more impaired than youth with no disorders (Table 6). Two measures of impairment appear to distinguish the youth with anxiety from those with major depression. Anxious youth appear to be less likely than depressed youth to report themselves as isolated or friendless, or to be perceived by their parents as performing poorly in school. These data provide some tentative support from the area of social functioning for a distinction between anxiety and depressive disorders. A study that compared children with anxiety disorders to children with conduct disorders found that the anxious children were less handicapped than the externalizing children, but more handicapped than normal children (Reeves et al., 1987). Children with anxiety are reported to be more socially withdrawn in reviews of the literature but in this study did not admit to being socially isolated (Reeves et al., 1987; Werry et al., 1987). The authors emphasize that their method was a reasonable attempt to operationalize the DSM-III-R criteria and was not identical to the DSM -III-R . In summary, this study found the prevalence of overanxious disorder (OAD) and separation anxiety disorder (SAD) to be 3.6% and 2.9 %, respectively, among adolescents aged 12 to 16 years. Anxiety disorders are more common among girls. Youth perception of limited contact with friends, and parent perception of poor school performance distinguish anxiety disordered youth from depressed youth . It was found that by requiring a high duration! intensity criterion for a symptom to count as positive, either one or both of the DSM-III-R essential features of the disorders and the presence of four and three auxiliary symptoms, respectively, for OAD and SAD yielded the most plausible community prevalences.
758
References Achenbach, T. M. & Edelbrock , C. S . (1981), Behavioral problems and competencies reported by parents of normal and disturbed children aged 4 through 16. Monogr.Soc.Res.ChildDev .,46:1-78. American Psychiatric Association (1987), Diagnostic and Statistical Manual of Mental Disorders (3rd ed, Revised) . Washington, DC: American Psychiatric Association . Anderson, J. C. , Williams , S., McGee , R. & Silva, P. A. (1987) , DSM-III disorders in pre-adolescent children . Arch . Gen . Psychiatry, 44:69-76. Bird, H. R., Canino, G., Rubio-Stipec, M. et al. (1988), Estimates of the prevalence of childhood maladjustment in a community survey in Puerto Rico. Arch. Gen. Psychiatry, 45:1120-1126. Boyle, M. H., Offord, D. R., Hoffman, H. G. et al. (1987) , Ontario Child Health Study: 1. methodology. Arch. Gen. Psychiatry, 44:826831. Fleming, J. E. , Offord, D. & Boyle, M. H. (1989) , Prevalence of childhood and adolescent depression in the community. Br .J . Psychiatry , 155:647-654. Gittleman, R. (1987), Childhood anxiety disorders. In: Anxiety Disorders of Childhood, ed. R. Gittleman . New York: Guilford, pp. 101125. Kashani, J. & Orvaschel,H . (1988), Anxiety disorders in mid-adolescence. Am.J. Psvchiatry , 145:960-964. Last, C. G. (1988), Anxiety disorders in childhood and adolescence. In: Handbook ofAnxiety Disorders, eds . C. G. Last & M. Hersen, New York: Pergamon , pp. 531-540. - - Hersen , M. , Kazdin, A. E.,-·Finkelstein, R. & Strauss , C . C. (l987a), Comparison of DSM-III separation anxiety and overanxious disorders . J . Am . Acad. Child Adolesc . Psychiatry, 26:527531. -. -Francis , G., Hersen, M., Kazdin, A. E. & Strauss , C. C . ( 1987b), Separation anxiety and school phobia : a comparison using DSM-III criteria. Am .J. Psychiatry , 144:653-667. Livingston, R. Taylor, J. L., & Crawford, S. L. (1987), A study of somatic complaints and psychiatric diagnosis in children . J . Am . Acad. ChildAdolesc . Psychiatry , 27:185-187 . Orvaschel, H. & Weissman, M.M. (1987), Epidemiology of anxiety disorders in children . In: Anxiety Disorders of Childhood, cd. R. Gittleman. New York: Guilford, pp. 58-72. Reeves , J. C ., Werry, J. S., Elkind , G. S. & Zametkin, A. (1987), Attention deficit, conduct, oppositional, and anxiety disorders in children. II . Clinical characteristics . J . Am. Acad. Child Adolesc .
Psychiatry , 26:144-155 . Rutter, M. (1988), Epidemiological approaches to developmental psychopathology . Arch .Gen .Psy chiatry,45 :483-495 . --Tizard ,1. & Whitmore , K. (1970) , Education, Health and Behavior . London : Longmans, pp. 183-187. Werry, J. S., Reeves , 1. C. & Elkind, G. S. (1987), Attention deficit , conduct, oppositional, and anxiety disorders in children: I. a review of research on differentiating characteristics. J . Am. Acad. Child Adolesc. Psychiatry. 26: 133-143.
J . Am. Acad. Child Ado/esc. Ps ychiatry, 29 :5, September 1990
