Disentangling chronological age from age of onset in children and adolescents with obsessive compulsive disorder

Transcription

1 International Journal of Neuropsychopharmacology (2001), 4, Copyright 2001 CINP Disentangling chronological age from age of onset in children and adolescents with obsessive compulsive disorder SPECIAL SECTION Daniel A. Geller 1,2,3, Joseph Biederman 2,3,4, Stephen V. Faraone 3, Christine A. Bellordre 2, Grace S. Kim 2, Lisa Hagermoser 2, Kathleen Cradock 2, Jean Frazier 2,3 and Barbara J. Coffey 2,3 Obsessive Compulsive Disorder Clinic and Pediatric Psychopharmacology Clinic, McLean Hospital, Belmont, MA, USA Harvard Medical School, Boston, MA, USA Massachusetts General Hospital, Fruit Street, Boston, MA, USA Abstract Although paediatric obsessive compulsive disorder (OCD) is increasingly recognized as a putative developmental subtype of the disorder, it remains uncertain as to whether additional subtyping by age at onset in childhood or adolescence is warranted. Subjects included children and adolescents meeting DSM-III-R and DSM-IV criteria for OCD referred to a specialized OCD clinic. All youth were systematically evaluated with structured diagnostic interviews and clinical assessment by an OCD expert. Irrespective of current age, an earlier age at onset predicted increased risk for attention deficit hyperactivity disorder, simple phobia, agoraphobia and multiple anxiety disorders. In contrast, mood and psychotic disorders were associated with chronological age and were more prevalent in older subjects. Tourette s disorder showed associations with both chronological age and age at onset. Chronological age and age at onset predicted different patterns of comorbidity and dysfunction in children and adolescents with OCD. Considering the heterogeneity of OCD, age at onset may help identify meaningful developmental subtypes of the disorder beyond chronological age. Received 11 September 2000; Reviewed 13 December 2000; Revised 6 February 2001; Accepted 7 February 2001 Key words: Obsessive compulsive disorder, paediatric, developmental, age at onset, chronological age. Introduction As we recently reviewed, obsessive compulsive disorder (OCD) affecting children and adolescents (termed paediatric ) differs from its adult counterpart in terms of male preponderance, familial aggregation with OCD and Tourette s syndrome, and psychiatric comorbidity with disruptive behaviour and developmental disorders (Geller et al., 1998; Nestadt et al., 2000). These differences suggested that paediatric OCD may be developmentally discontinuous with the adult-onset condition despite shared phenotypic features. Support for this notion also derives from the observation of distinct peaks of onset of OCD. Studies of paediatric OCD have reported a mean age at onset of around 10 yr (Geller et al., 1996; Hanna, 1995; Last and Strauss, 1989; Riddle et al., 1990; Swedo Address for correspondence: Dr D. A. Geller, Pediatric OCD Clinic, McLean Hospital, 115 Mill Street, Belmont, MA 02478, USA. Tel.: (617) Fax: (617) geller mclean.harvard.edu et al., 1989; Thomsen, 1993; Toro et al., 1992). In contrast, studies of adults with OCD report a mean age at onset of 21 yr (Karno et al., 1988; Minichiello et al., 1990; Pauls et al., 1995; Rasmussen and Eisen, 1992; Thyer et al., 1985). These differing age-at-onset distributions suggest that adult- and paediatric-onset OCD, although sharing phenotypic features, may represent different developmental subtypes of the disorder. Despite this compelling evidence for syndromatic discontinuity between paediatric-onset and adult-onset OCD, uncertainties remain as to where the line of developmental discontinuity should be drawn. Adolescent subjects are an intermediate group that could resemble either children or adults with OCD. Yet, current studies of OCD combine all child and adolescent subjects into one developmental category. This approach assumes homogeneity within the child adolescent age range; an assumption that has not yet been validated. Another barrier to understanding the concomitants of age at onset for OCD is the confounding effect of

2 170 D. A. Geller et al. chronological age. Because some psychiatric disorders show a greater prevalence with increasing age, differences attributed to age at onset could be artifacts of age at assessment. So, a systematic examination of developmental variability in OCD requires additional efforts aimed at disentangling chronological age from age of onset. A further methodological issue in age-at-onset studies regards definition of onset given that many children will have a gradual subclinical onset. Clarifying these issues could have important implications. For example, it is possible that stratifying paediatric OCD on the basis of age or age at onset could reduce the heterogeneity of study samples and thereby yield insights into the aetiology, correlates, and appropriate treatments for patients. These issues have not been adequately examined in the OCD literature. To help fill this gap in the literature, we examined the association between age at onset of OCD and gender, psychiatric comorbidity, and measures of functioning. Our analyses take into account the confounding effects of chronological age upon age at onset. Based on the prior literature, we hypothesized that an earlier age at onset would be associated with male gender and increased rates of disruptive behaviour, tic and developmental disorders. Methods The sample consisted of consecutive referrals to a specialized paediatric OCD clinic at a major academic medical centre that underwent systematic evaluation. Youth with mental retardation, autism or severe neurological disorder were excluded from this study. All subjects were clinically evaluated by the lead author (D.G.), the director of this programme, a board-certified child and adolescent psychiatrist and paediatrician knowledgeable in paediatric OCD. In addition, all subjects were also evaluated using a standardized assessment battery that included a structured diagnostic interview (Kiddie- Schedule for Affective Disorders and Schizophrenia- Epidemiological Version (K-SADS-E), fifth revision) (Orvaschel and Puig-Antich, 1994) using the parents as informants. The K-SADS-E is a widely used, structured assessment interview. IRB-approved informed parental consent and child assent was obtained for subjects. For every DSM-IIIR IV disorder, information was collected regarding present (last month) and lifetime diagnosis, age at onset and offset of symptoms as well as associated impairment (mild, moderate, severe) and treatment history. Interviews were performed by highly trained and supervised Bachelor s degree-level interviewers who underwent a rigorous training programme under the supervision of the lead author (D.G.) and programme chief ( J.B.). Training included mastery of the instruments, learning about diagnostic criteria, watching training tapes, observing interviews performed by experienced raters, and rating several subjects under the direct supervision of the lead author. All structured clinical interviews were audiotaped with parental consent and all were reviewed at weekly clinical review meetings attended by at least two board-certified child and adolescent psychiatrists, who were blind to the identity of the subject, before final diagnoses were assigned. Agreement between raters and experienced clinicians has been excellent (n 175 interviews, mean overall kappa coefficient of correlation 0 9). We obtained a kappa of 0 99 for attention deficit hyperactivity disorder (ADHD), 0 97 for oppositional defiant disorder and panic disorder, and 0 95 for overanxious disorder. A kappa of 1 0 was obtained for tics. A sample of 40 structured diagnostic interviews (K-SADS- E), were blindly rated by the lead author who listened to taped interviews of the DSM-IV OCD modules and independently assigned diagnoses, achieving a kappa correlation of 0 87 with the interviewers. In addition, all structured diagnostic interview information was subsequently re-reviewed by the lead author to ensure that it was clinically meaningful for each individual subject. To be included in this review, subjects had to meet all DSM-IV criteria for the full clinical syndrome of OCD both at structured interview with the parent as well as at clinical evaluation with both parent and youth. This required that typical obsessional thoughts and compulsive rituals be present every day, be time consuming, and cause significant subjective distress as well as functional impairment in at least one important domain of living. Prior OCD symptoms that were sub-threshold for full clinical diagnosis were not included in the calculation of age at onset. Age at onset was determined by a review of all data including both structured and clinical interview information from multiple informants, using a best estimate method (Leckman et al., 1982), by the lead author. In the few instances where there were discrepancies between parental and youth reports in reported age at onset, this was resolved in favour of the informant that the lead author deemed most reliable. In addition, OCD symptoms were quantitatively recorded using the Children s Yale Brown Obsessive Compulsive Scale (CY-BOCS) which includes a symptom checklist and a 10-item ordinal scale to rate symptom severity that has been validated in paediatric subjects (Scahill et al., 1997) We used regression analysis to examine the effects of age at onset (a continuous variable) on our outcome variables. For binary outcome data we used logistic

3 Age and age at onset in paediatric OCD 171 Table 1a. Association of chronological age and age at onset of OCD with demographic features in paediatric OCD subjects (n 101) Children (n 45) childhood onset (n 33) adolescent onset (n 23) n % n % n % Male Intact family SES [mean (S.D.)] 2 0 (1 0) 1 6 (0 8) 1 5 (0 8) Table 1b. Association of chronological age and age at onset with demographic features in paediatric OCD subjects (n 101) Chronological age Age at onset z score t score p value z score t score p value Gender a a Intact family a a SES (1 5) b b a p values derived from logistic regression. b p values derived from ordinal logistic regression. regression, for ordinal outcome data, ordinal logistic regression and for continuous outcome data, linear regression. All tests were two-tailed. One statistical problem with age-at-onset data is that this variable is, by definition, confounded by age at interview (e.g. it is impossible to have an age at onset of 13 if the age at interview is less than 13). This will confound results if the age of the patient is positively correlated with outcome variables. To deal with this problem, we followed the guidelines of Weinburg (1993) by determining if current age showed a significant positive association with any outcomes. For those variables showing significance, current age was included as a covariate in the analysis for the regression models assessing the significance of age at onset. This allows us to assess the effect of age at onset not confounded by age at accession. This analytic method has the additional benefit of improving power by using age at onset as a continuous variable rather than analysing the dichotomous variable child vs. adolescent onset. Stratification is, however, useful for describing the results in a clinically meaningful manner. Thus, to describe the effects of age and age at onset on the outcome variables, we report the prevalence of outcomes in the three groups: children (age 12 yr) with OCD, adolescents (12 18 yr) with childhood-onset OCD and adolescents with adolescent-onset OCD. Results The sample comprised 101 paediatric OCD patients consecutively referred to a specialized OCD programme. Of these, 66% were male, 45% were younger (children) and 55% were 12 yr. Of the 56 adolescent OCD subjects, 33 had onset of their OCD in childhood and the other 23 in adolescence. No associations were identified in gender distribution, social class, rate of separation or divorce in the family by age at onset or chronological age (Tables 1a and 1b). As expected, age at onset was positively associated with age at interview. Also as expected, significant associations were identified between duration of symptoms and chronological age (positive) and age at onset (negative). Although OCD severity scores showed no age associations, Global Assessment of Function (GAF) scores were more impaired and rates of hospitalization increased with advancing chronological age (Table 2b). Table 2a

4 172 D. A. Geller et al. Table 2a. Association of chronological age and age at onset of OCD with clinical features in paediatric OCD subjects (n 101) Children (n 45) childhood onset (n 33) adolescent onset (n 23) Mean S.D. Mean S.D. Mean S.D. Age of onset of OCD Chronological age Duration of OCD OCD severity Lifetime (1 minimal, 3 severe) Current (1 minimal, 3 severe) Global assessment of function (GAF) Lifetime Current n % n % n % Psychiatric hospitalization Table 2b. Association of chronological age and age at onset with clinical features in paediatric OCD subjects (n 101) Chronological age Age at onset z score t score p value d.f. z score t score p value d.f. Age of onset of OCD c 98 na na na Chronological age na na na c 98 Duration of OCD c cd 98 OCD severity Lifetime b na b na (1 minimal, 3 severe) Current b na b na (1 minimal, 3 severe) Global assessment of function (GAF) Lifetime c cd 99 Current c cd 99 Psychiatric hospitalization a na ad na p values are derived from: a logistic regression; b ordinal regression; c linear regression. d Controlled for covariate chronological age. Degrees of freedom (d.f.) apply to t scores. describes the mean age at onset, chronological age, duration of symptoms, OCD severity and global functioning scores and treatment history in three groups. Treatment history found that 12% of subjects had received medication only, 41% had received both medication and some form of therapy (not necessarily cognitive behavioural) and only 20% had received no treatment at all.

5 Age and age at onset in paediatric OCD 173 Table 3a. Association of chronological age and age at onset of OCD with other psychiatric disorders in paediatric OCD subjects (n 101) Children (n 45) childhood onset (n 33) adolescent onset (n 23) Other psychiatric disorders n % n % n % Disruptive behaviour disorders ADHD Conduct disorder Oppositional defiant disorder Mood disorders Major depression Bipolar disorder Psychosis Tourette s disorder Anxiety disorders Multiple anxiety disorders Panic disorder Agoraphobia Social phobia Simple phobia Overanxious disorder Separation anxiety Avoidant disorder Alcohol and substance disorders Alcohol abuse dependence Substance abuse dependence Smoking Analysis of chronological age revealed significant associations between chronological age with mood disorders (major depression, bipolar disorder), psychosis, Tourette s disorder and smoking, irrespective of age at onset of OCD (Table 3b). Regression analyses of age at onset (controlling for chronological age for those disorders above that showed a positive association with this) showed that ADHD, agoraphobia, simple phobia and multiple anxiety disorders were positively correlated with an earlier age at onset irrespective of chronological age (Table 3b). Tourette s disorder showed an almost significant association with age at onset (p 0 06). Table 3a shows the frequency of lifetime comorbid psychiatric disorders in three groups. The pattern of results after stratification is consistent with the regression analyses. Although the correlation between Tourette s disorder and age at onset did not reach significance, inspection of Table 3a shows a much higher prevalence among children (27%) compared with adolescent cases with childhood-onset (9%) and adolescent-onset cases (0%). Neither non- Tourette s tic disorders (including chronic and simple tics) nor all tic disorders when combined showed significant associations with age or age at onset. The relative proportion of subjects with age at onset of OCD before and after comorbid disorders is depicted in Figure 1. In a subset of subjects (n 55) neither age at onset nor chronological age predicted severity as measured by the CY-BOCS (Table 4). Discussion In a systematic evaluation of the effects of chronological age and age at onset on OCD features and correlates, we found that younger age at onset of OCD was associated with higher rates of ADHD and non-ocd anxiety

6 174 D. A. Geller et al. Table 3b. Association of chronological age and age at onset with other psychiatric disorders in paediatric OCD subjects (n 101) Chronological age Age at onset Other psychiatric disorders z score p value z score p value Disruptive behaviour disorders ADHD Conduct disorder Oppositional defiant Mood disorders Major depression a Bipolar disorder a Psychosis a Tourette s disorder a Anxiety disorders Multiple anxiety disorders Panic disorder Agoraphobia Social phobia Simple phobia Generalized anxiety Separation anxiety Avoidant disorder Alcohol substance abuse disorders Alcohol abuse dependence Substance abuse dependence Smoking a p values and z scores are derived from logistic regression. a Controlled for covariate chronological age. Comorbid disorders ADHD Separation anxiety Simple phobia Social phobia Conduct disorder Tourette s disorder Oppositional disorder GAD Bipolar disorder Major depression Agoraphobia Psychosis Panic disorder Avoidant disorder Smoking Substance abuse/dependence Alcohol abuse/dependence 0% 10% 20% 30% 40% 50% 60% Percentage 70% 80% 90% 100% n = 46 n = 39 n = 25 n = 12 n = 8 n = 15 n = 44 n = 6 n = 15 n = 58 n = 26 n = 13 n = 21 n = 4 n = 6 n = 1 n = 1 OCD onset first Concurrent onset Comorbid Dx. onset first Figure 1. Relative age at onset of comorbid disorders in children and adolescents with obsessive compulsive disorder.

7 Age and age at onset in paediatric OCD 175 Table 4. Association of chronological age and age at onset with CY-BOCS scores in paediatric OCD subjects (n 55) Children (n 21) childhood onset (n 18) adolescent onset (n 16) n Mean n Mean n Mean Mean CY-BOCS score Chronological age Age at onset t score p value t score p value CY-BOCS total score Obsessions subtotal Compulsions subtotal p values and t scores are derived from linear regression. disorders, independent of chronological age, while chronological age was significantly associated with increased rates of mood disorders, psychotic symptoms and hospitalization, independent of age at onset. The prevalence of Tourette s disorder showed significant association with decreasing age as well as an almost significant association with earlier ages at onset of OCD. These findings suggest that both chronological age and age at onset are meaningful developmental variables in OCD within the child to adolescent age range. This extends prior work showing the relevance of age at onset within the child to adult age range while emphasizing the value of considering age when evaluating the effects of age at onset. Our results showing that the association between OCD and ADHD is accounted for by those subjects with early-onset OCD suggest that the presence of ADHD in OCD subjects may help identify a developmentally unique subgroup of OCD with early-onset disorder. These results extend previous findings documenting an important association in juvenile OCD with ADHD (Geller et al., 1996). For example, in our sample 57% of OCD children younger than 12 yr had ADHD. Although this rate is higher than in previous reports (Hanna, 1995; Last and Strauss, 1989; Riddle et al., 1990; Swedo et al., 1989; Toro et al., 1992), these previous studies often used many exclusion criteria, including comorbid Tourette s disorder, to select homogenous samples of OCD youth. Since Tourette s disorder is frequently associated with ADHD, this exclusion would be expected to result in a marked under-representation of ADHD. In addition, since all epidemiological studies of paediatric OCD have drawn upon community samples of adolescents only (Apter et al., 1996; Douglass et al., 1995; Flament et al., 1988; Thomsen, 1993; Valleni-Basile et al., 1994) this ascertainment scheme could also have resulted in a lower rate of ADHD. Furthermore, in the present study, in virtually all cases of youth with OCD and ADHD, the ADHD predated the onset of OCD by many years so that the symptomatic picture of ADHD in these comorbid cases cannot be assumed to be simply an artifact of intrusive obsessions. Considering that ADHD and OCD respond to different treatments, these findings have important clinical applications. Although anxiety disorders in general, and agoraphobia and simple phobia in particular, aggregated in younger onset cases, the prevalence of multiple anxiety disorders among adolescents with adolescent-onset OCD was still substantial (35%), indicating that non-ocd anxiety disorders are important correlates of both younger- and older-onset OCD cases. These results suggest that age at onset may be associated with a gradient of anxiety severity, leading to more anxiety disorders being expressed in children having earlier onset ages. In our sample, Tourette s disorder showed a negative association with increasing chronological age supporting recent reports in the literature (Coffey et al., unpublished observations, 2000; Spencer et al., 1999) of declining tic severity over time. Although the effect of age at onset of OCD on Tourette s disorder did not reach statistical significance, we found a strong trend for Tourette s disorder to be associated with younger ages of OCD onset (p 0 06). The effect of both chronological age

8 176 D. A. Geller et al. and age at onset can also be seen in our presentation of frequency data (Table 3a). There are very high rates of Tourette s disorder among children with OCD (27%) but much lower rates among adolescents with childhood onset (9%) suggesting that tics appear to improve with increasing age. These results illustrate the importance of examining both age and age at onset in the evaluation of clinical correlates of OCD subjects (Table 3b). Irrespective of age at onset, OCD youth showed increasing rates of comorbid mood disorders with advancing age. Both major depression and bipolar disorder aggregated significantly in adolescent OCD subjects independently of age of onset of OCD. These findings correspond well to our results showing that adolescent OCD subjects also had significantly lower GAF scores and higher rates of psychiatric hospitalization, which underscore the severe morbidity associated with concurrent mood disorders in OCD youth. The rates of mood disorders in our adolescent OCD subjects were similar to those described in adult samples (Pigott et al., 1994), indicating that by adolescence, rates of major depression reach adult levels. Although novel, our finding of a high rate of mania in our sample of adolescents with OCD (22%) compared with the general population fits well with a recent report documenting a similar overlap between OCD and mania in adults (Kruger et al., 1995). Despite controversy, an emerging literature documents that mania is not rare in youth, but it may be difficult to diagnose (Wozniak et al., 1995). Moreover, the relatively high rate of mania in our sample is not entirely surprising considering not only the high rate of major depression but also that major depression in children and adolescents frequently has a bipolar outcome (Geller et al., 1993; Strober and Carlson, 1982; Strober et al., 1988). Since many previous paediatric OCD studies excluded subjects with both major depression and bipolar disorder, rates of the latter may have been under-reported. Considering that serotonin uptake inhibitors (SRIs) are the mainstay of pharmacological treatment for OCD and that these antidepressants can precipitate mood changes and mania in predisposed individuals, the appropriate identification of bipolar disorder in OCD patients is of serious clinical relevance in the management of these complicated OCD youth. In our sample, we found that a psychotic symptomatology was associated with increasing chronological age but not age at onset. Reports in adults from Janet (1903) to Rasmussen and Eisen (1992) have found an overlap between psychosis and OCD that ranges from 8 to 14%. It is important to stress that the absence of insight regarding OCD symptomatology was not part of our definition of psychosis, which relied on the presence of delusions or hallucinations. In nearly all cases a major mood disorder was also present. Considering the morbidity associated with psychotic phenomenology and their unique therapeutic implications, these results highlight the importance of identifying these OCD youth and treating them appropriately. Although gender was not associated with either chronological age or age at onset in this study, the proportion of males in our sample (66%) was significantly greater than the 50% (p by an exact binomial test) reported in a number of adult studies (Karno et al., 1988; Minichiello et al., 1990; Pauls et al., 1995; Rasmussen and Eisen, 1992; Thyer et al., 1985). This confirms previous findings of male predominance in paediatric OCD (Geller et al., 1998). The findings reported in this study should be interpreted in light of several methodological limitations. Since the sample consisted of children and adolescents referred to a specialized paediatric OCD programme, they may not generalize to community samples or to other mental health clinics; however they should generalize to other speciality clinic samples. Although our findings derived from structured diagnostic interview data collected from interviews with the mothers, all subjects included in this study had direct clinical assessment and diagnostic confirmation by the lead author, who also reviewed the structured interview information for clinical relevance. Information on age at onset was retrospective and thus subject to recall bias (Angold and Costello, 1993). It is also possible that age-at-onset reporting is less accurate for adults than children given the longer period of elapsed time from onset to interview. Although such recall biases cannot account for the pattern of results reported in this study, more work with longitudinally ascertained samples may help clarify some of these issues. Also, age-at-onset data is, by definition, confounded by age at interview. Although this could have confounded our results, our method of analysis addressed this potential confound. Despite these limitations, our findings indicate that both age at onset and chronological age are meaningful developmental variables affecting the expression of OCD. We found that early-onset OCD is associated with higher rates of ADHD and anxiety disorders while rates of mood and psychotic disorders increased with chronological age. Tourette s disorder showed associations with both chronological age and age at onset. Clinical and research studies of paediatric OCD should take into account both age at onset of the disorder as well as chronological age in examining correlates of this disorder. Acknowledgements This work has been funded in part by an Eli Lilly Pilot Research Award from the American Academy of Child

9 Age and age at onset in paediatric OCD 177 and Adolescent Psychiatry, a Tourette Syndrome Association Foundation award, an Obsessive Compulsive Foundation award and grant no. NIMH K08 MH References Angold A, Costello E (1993). Depressive comorbidity in children and adolescents: empirical, theoretical, and methodological issues. Journal of American Psychiatry 150, Apter A, Fallon Jr. TJ, King RA, Ratzoni G, Zohar AH, Binder M, Weizman A, Leckman JF, Pauls DL, Kron S, Cohen DJ (1996). Obsessive compulsive characteristics: from symptoms to syndrome. Journal of the American Academy of Child and Adolescent Psychiatry 35, Coffey B, Biederman J, Geller D, Spencer T, Park K, Shapiro S, Garfield S (2000). The course of Tourette s Disorder: a literature review. Harvard Review of Psychiatry 8, Douglass HM, Moffitt TE, Dar R, McGee R, Silva P (1995). Obsessive compulsive disorder in a birth cohort of 18- year-olds: prevalence and predictors. Journal of the American Academy of Child and Adolescent Psychiatry 34, Flament M, Whitaker A, Rapoport J, Davies M, Berg C, Kalikow K, Sceery W, Shaffer D (1988). Obsessive compulsive disorder in adolescence: an epidemiological study. Journal of the American Academy of Child and Adolescent Psychiatry 27, Geller B, Fox L, Fletcher M (1993). Effect of tricyclic antidepressants on switching to mania and on the onset of bipolarity in depressed 6 to 12-year-olds. Journal of the American Academy of Child and Adolescent Psychiatry 32, Geller D, Biederman J, Griffin S, Jones J, Lefkowitz TR (1996). Comorbidity of juvenile obsessive compulsive disorder with disruptive behavior disorders. Journal of the American Academy of Child and Adolescent Psychiatry 35, Geller D, Biederman J, Jones J, Park K, Schwartz S, Shapiro S, Coffey B (1998). Is juvenile obsessive compulsive disorder a developmental subtype of the disorder?: a review of the pediatric literature. Journal of the American Academy of Child and Adolescent Psychiatry 37, Hanna GL (1995). Demographic and clinical features of obsessive compulsive disorder in children and adolescents. Journal of the American Academy of Child and Adolescent Psychiatry 34, Janet P (1903). Les Obsessions et La Psychasthenie, vol. I. Paris: Felix Alcan. Karno M, Golding J, Sorenson S, Burnam A (1988). The epidemiology of obsessive compulsive disorder in five U.S. communities. Archives of General Psychiatry 45, Kruger S, Cooke RG, Hasey GM, Jorna T, Persad E (1995). Comorbidity of obsessive compulsive disorder in bipolar disorder. Journal of Affective Disorders 34, Last CG, Strauss CC (1989). Obsessive compulsive disorder in childhood. Journal of Anxiety Disorders 3, Leckman J, Sholomskas D, Thompson W, Belanger A, Weissman M (1982). Best estimate of lifetime psychiatric diagnosis: a methodological study. Archives of General Psychiatry 39, Minichiello WE, Baer L, Jenike MA, Holland A (1990). Age of onset of major subtypes of obsessive compulsive disorder. Journal of Anxiety Disorders 4, Nestadt G, Samuels J, Bienvenu JO, Grados M, Hoehn-Saric R, Liang K, LaBuda M, Riddle M, Walkup J (2000). A family study of obsessive compulsive disorder. Archives of General Psychiatry 57, Orvaschel H, Puig-Antich J (1994). Schedule for Affective Disorder and Schizophrenia for School-age Children, Epidemiologic Version. Ft. Lauderdale, FL: Nova University, Center for Psychological Study. Pauls D, Alsobrook II J, Goodman W, Rasmussen S, Leckman J (1995). A family study of obsessive compulsive disorder. American Journal of Psychiatry 152, Pigott TA, L Heureux F, Dubbert B, Bernstein S, Murphy DL (1994). Obsessive compulsive disorder: comorbid conditions. Journal of Clinical Psychiatry 55 (Suppl. 10), Rasmussen S, Eisen J (1992). The epidemiology and clinical features of obsessive compulsive disorder. Psychiatric Clinics of North America 15, Riddle M, Scahill L, King R, Hardin M, Towbin K, Ort S, Leckman J, Cohen D (1990). Obsessive compulsive disorder in children and adolescents: phenomenology and family history. Journal of the American Academy of Child and Adolescent Psychiatry 29, Scahill L, Riddle M, McSwiggin-Hardin M, Ort SI, King RA, Goodman WK, Cicchetti D, Leckman JF (1997). Children s Yale Brown Obsessive Compulsive Scale: reliability and validity. Journal of the American Academy of Child and Adolescent Psychiatry 36, Spencer T, Biederman J, Coffey B, Geller D, Wilens T, Faraone S (1999). The four-year course of tic disorders in boys with ADHD. Archives of General Psychiatry 56, Strober M, Carlson G (1982). Bipolar illness in adolescents with major depression: Clinical, genetic, and psychopharmacological predictors in a three- to four-year prospective follow-up investigation. Archives of General Psychiatry 39, Strober M, Morrell W, Burroughs J, Lampert C, Danforth H, Freeman R (1988). A family study of bipolar I disorder in adolescence: Early onset of symptoms linked to increased familial loading and lithium resistance. Journal of Affective Disorders 15, Swedo S, Rapoport J, Leonard H, Lenane M, Cheslow D (1989). Obsessive compulsive disorder in children and adolescents: clinical phenomenology of 70 consecutive cases. Archives of General Psychiatry 46, Thomsen P (1993). Obsessive compulsive disorder in children and adolescents: self-reported obsessive compulsive behaviour in pupils in Denmark. Acta Psychiatrica Scandinavica 88, Thyer BA, Parrish RT, Curtis GC, Nesse RM, Cameron OG

10 178 D. A. Geller et al. (1985). Ages of onset of DSM-III anxiety disorders. Comprehensive Psychiatry 26, Toro J, Cervera M, Osejo E, Salamero M (1992). Obsessive compulsive disorder in childhood and adolescence: a clinical study. Journal of Child Psychology and Psychiatry and Allied Disciplines 33, Valleni-Basile L, Garrison C, Jackson K, Waller J, McKeown R, Addy C, Cuffe S (1994). Frequency of obsessive compulsive disorder in a community sample of young adolescents. Journal of the American Academy of Child and Adolescent Psychiatry 33, Weinburg CR (1993). Toward a clearer definition of confounding. American Journal of Epidemiology 137, 1 8. Wozniak J, Biederman J, Kiely K, Ablon S, Faraone S, Mundy E, Mennin D (1995). Mania-like symptoms suggestive of childhood onset bipolar disorder in clinically referred children. Journal of the American Academy of Child and Adolescent Psychiatry 34,