Phenomenology of Children and Adolescents With Bipolar Spectrum Disorders

Context  Children and adolescents who present with manic symptoms frequently do not meet the full DSM-IV criteria for bipolar I disorder (BP-I).

Objective  To assess the clinical presentation and family history of children and adolescents with BP-I, bipolar II disorder (BP-II), and bipolar disorder not otherwise specified (BP-NOS).

Design  Subjects and their primary caretaker were assessed by semistructured interview, and family psychiatric history was obtained from interview of the primary caretaker.

Setting  Outpatient and inpatient units at 3 university centers.

Participants  A total of 438 children and adolescents (mean ± SD age, 12.7 ± 3.2 years) with BP-I (n = 255), BP-II (n = 30), or BP-NOS (n = 153).

Main Outcome Measures  Lifetime psychiatric history and family history of psychiatric disorders.

Results  Youth with BP-NOS were not diagnosed as having BP-I primarily because they did not meet the DSM-IV duration criteria for a manic or mixed episode. There were no significant differences among the BP-I and BP-NOS groups in age of onset, duration of illness, lifetime rates of comorbid diagnoses, suicidal ideation and major depression, family history, and the types of manic symptoms that were present during the most serious lifetime episode. Compared with youth with BP-NOS, subjects with BP-I had more severe manic symptoms, greater overall functional impairment, and higher rates of hospitalization, psychosis, and suicide attempts. Elevated mood was present in 81.9% of subjects with BP-NOS and 91.8% of subjects with BP-I. Subjects with BP-II had higher rates of comorbid anxiety disorders compared with the other 2 groups and had less functional impairment and lower rates of psychiatric hospitalization than the subjects with BP-I.

Conclusions  Children and adolescents with BP-II and BP-NOS have a phenotype that is on a continuum with that of youth with BP-I. Elevated mood is a common feature of youth with BP-spectrum illness.

Pediatric bipolar disorder (BP) has become an area of intense clinical and research interest, and the focus of considerable controversy regarding its diagnosis.^1,2 It is becoming more apparent that pediatric BP is a significant public health problem. Studies^3-5 examining the prevalence of pediatric BP in large community samples have found rates varying between 0.1% and 2%. In the study by Lewinsohn and colleagues,³ bipolar adolescents had twice the rate of suicide attempts and significantly lower levels of functioning compared with adolescents with major depressive disorder. Estimates of the lifetime prevalence of BP (including bipolar II disorder [BP-II]) in adults generally range from 1% to 2%, although some recent estimates have been as high as 3.9%.^6-9 In 2 large samples,^10,11 50% to 60% of bipolar adults recalled the initial onset of mania or depression to be 19 years or younger. The onset of BP in childhood or adolescence may be associated with a more severe course compared with onset in adulthood.^12,13 Pediatric BP may become a more serious problem in the future, because the age of onset of BP may be getting younger in more recent birth cohorts.¹¹

Despite the interest, controversy, and public health significance, there are relatively few published studies examining the clinical phenomenology of pediatric BP. Several research groups have published studies^14-16 using semistructured interviews to examine the cross-sectional presentation of bipolar I disorder (BP-I) in child and young adolescent cohorts. Although there were differences in sampling strategies and assessment method, several consistent findings emerged from these studies. Bipolar youth had high rates of irritability, comorbid attention-deficit/hyperactivity disorder, and oppositional defiant disorder. However, there was wide variability in the rates of psychosis, mixed episodes, comorbid conduct disorder, and anxiety disorders and in the presence of elated or elevated mood.

These studies focused almost exclusively on children and adolescents with BP-I. However, as evidenced in the study by Lewinsohn et al,³ most youth with BP in the community do not meet the DSM-IV¹⁷ diagnostic criteria for BP-I. Children and adolescents in clinical and research settings also frequently do not meet the DSM-IV criteria for BP-I for reasons such as inadequate duration of symptoms, insufficient number of symptoms, or lack of a distinct period of abnormal mood in youth who may have chronic irritability.¹⁸ This heterogeneous group of children and adolescents with subthreshold presentations of BP may or may not develop BP-I disorder as they mature into adulthood.

The Course and Outcome of Bipolar Youth (COBY) study, funded by the National Institute of Mental Health, was designed to build on and extend the existing scientific database on the cross-sectional presentation and longitudinal course of pediatric BP. This report is an analysis of the data gathered at the intake assessment about the presentation, psychiatric history, and family history of the COBY study subjects. The primary aims are as follows: (1) to examine similarities and differences between subjects with BP-I, BP-II, and BP not otherwise specified (BP-NOS); and (2) to compare the phenomenological features of the bipolar youth in the COBY study with those in prior published reports.

Subjects were enrolled at 3 academic medical centers: Brown University (n = 144), University of California at Los Angeles (n = 90), and University of Pittsburgh Medical Center (n = 204). Study inclusion criteria were as follows: (1) aged 7 years 0 months to 17 years 11 months and (2) fulfilled the DSM-IV criteria for BP-I or BP-II or the COBY-established criteria for BP-NOS. Because the DSM-IV criteria for BP-NOS are vague, the COBY study investigators set the minimum inclusion threshold for the BP-NOS group as subjects who did not meet the DSM-IV criteria for BP-I or BP-II but had a distinct period of abnormally elevated, expansive, or irritable mood plus the following: (1) 2 DSM-IV manic symptoms (3 if the mood is irritability only) that were clearly associated with the onset of abnormal mood, (2) a clear change in functioning, (3) mood and symptom duration of a minimum of 4 hours within a 24-hour period for a day to be considered meeting the diagnostic threshold, and (4) a minimum of 4 days (not necessarily consecutive) meeting the mood, symptom, duration, and functional change criteria over the subject's lifetime, which could be two 2-day episodes, four 1-day episodes, or another variation. Symptoms and mood changes that occurred during substance use or antidepressant treatment did not count toward a bipolar diagnosis. Study exclusion criteria were as follows: current or lifetime DSM-IV diagnosis of schizophrenia, mental retardation, autism, or severe autistic spectrum disorders or mood disorders due to substance abuse, due to a medical condition, or secondary to use of medications (eg, corticosteroids). Subjects determined to have the onset of BP before comorbid substance use disorder were included. Potential subjects with mild comorbid Asperger disorder or pervasive developmental disorder NOS were included if their mood symptomatology was clearly episodic and best accounted for by the bipolar diagnosis.

Subjects were recruited through clinical referrals from within the medical centers and the community and from advertisements. A telephone or brief face-to-face screening interview and discussion of the study protocol was administered to the parent or caregiver of a potential subject. Those with a history of likely manic symptoms and without an obvious exclusion criterion were scheduled for an in-person assessment at 1 of the 3 sites. Informed consent was obtained before initiation of the assessment from the subject's parent or guardian and from subjects 14 years or older. The study procedures were explained in age-appropriate language to younger subjects, and verbal assent was obtained before the assessment. The institutional review boards at the 3 centers reviewed and approved the study protocol before enrollment of any subject.

Subjects were assessed by semistructured interviews of the child or adolescent and a parent or primary caregiver (about the subject) by a trained research clinician. Nonmood psychiatric disorders were assessed using the Schedule for Affective Disorders and Schizophrenia for School-age Children (KSADS)–Present and Lifetime Version.¹⁹ Mood symptoms were assessed by the mood disorder sections of the KSADS-P (Present Episode, fourth revision)²⁰ plus additional items from the KSADS Mania Rating Scale (MRS).²¹

The KSADS assessments begin with an unstructured interview that reviews lifetime treatment history and course of illness to construct a time line of mood and behavioral problems. Depression and manic symptom severity ratings were recorded on the KSADS-P mood disorder sections and the KSADS MRS for the most severe week in the month before the intake assessment; the 12-item KSADS depression rating was obtained from the KSADS-P.²² The most severe week of depression and manic symptoms in the subjects' lifetime was assessed by the KSADS–Present and Lifetime Version, in the first 84 subjects enrolled in the study, and with the KSADS MRS and KSADS-P mood disorder sections for the remaining subjects. Per the KSADS instructions, mood symptoms that are also in common with other psychiatric disorders (such as motor hyperactivity or distractibility) are not rated as present in the mood sections unless they intensify with the onset of abnormal mood. Comorbid diagnoses were not assigned in the COBY study if they occurred exclusively during a mood episode. The onset of the first and most recent episode of each type of DSM-IV major mood episode (manic, mixed, hypomanic, or major depressive episode [MDE]) was recorded, as was the first time subjects met the criteria for BP-NOS. The age of onset for a subject's BP-spectrum illness was considered to be when the subject first met the DSM-IV criteria for a major mood episode or the COBY study criteria for BP-NOS. Given that the validity of DSM-IV diagnostic criteria for preschool-aged children has not been established, the minimum age of onset for BP-spectrum illness was set at 4 years.

Socioeconomic status was measured using the Hollingshead 4-factor scale.²³ Pubertal status was reported by subjects 10 years and older using the Pubertal Developmental Scale.²⁴ Parents assisted younger children with completing the Pubertal Developmental Scale. The Pubertal Developmental Scale ratings can be converted into associated Tanner stages of sexual development.^25,26 Subjects whose ratings were equivalent to Tanner stage 1 or who were younger than 8 years were considered prepubertal.

The subject’s primary caretaker was interviewed at intake about his or her personal psychiatric history using the Structured Clinical Interview for DSM-IV.²⁷ The primary caretaker was interviewed about the psychiatric status of the subject's first- and second-degree relatives using the Family History Screen.²⁸

Research interviewers were trained to high reliability in administration of the KSADS, the Structured Clinical Interview for DSM-IV, and the Family History Screen before interviewing any subjects or parents. The results of each interview were reviewed by a child psychiatrist or psychologist. Diagnostic reliability was measured by having research interviewers from all sites rate 13 audiotapes of actual COBY study interviews. There was high reliability for differentiating BP from non-BP subjects (κ = 0.90) and from the BP diagnostic subtypes (κ = 0.79). For the nonmood disorders, κ values were 0.80 or higher. The intraclass correlation coefficient was 0.96 for the KSADS MRS and 0.98 for the KSADS Depression Scale.

A commercially available software program (SPSS Release 13.0 for Windows; SPSS Inc, Chicago, Ill) was used for data analysis. The data analyzed for this report were gathered at the intake assessment and represent the cross-sectional view of the subject at intake and the retrospective recall of the subject's history before intake. Only 18 subjects with BP-II had complete KSADS MRS ratings for the most symptomatic week in the past and in the month before intake. Therefore, analyses of individual manic symptoms present during the most severe week of the subject's lifetime were performed for the 220 subjects with BP-I and the 116 subjects with BP-NOS who had ratings for both time frames. Differences in demographic factors among the 3 BP subtypes were analyzed initially using standard parametric and nonparametric univariate tests. For the demographic factors for which potential significant differences existed among the groups (P<.20), models were constructed using the logistic regression or general linear model procedures of a commercially available software program (SPSS Release 13.0 for Windows), controlling for the demographic factors. Because age and pubertal status were highly correlated (r_s = 0.84), only age was entered into the models. The number of relatives for each subject was added as a covariate for the family history analyses. If the overall model was significant, then pairwise comparisons were performed. All values are reported as means ± standard deviations unless otherwise specified. All P values are based on 2-tailed tests with α = .05. To our knowledge, this is the first report comparing pediatric subjects with BP-NOS, BP-II, and BP-I, and identifying potential similarities and differences among the diagnostic groups was a study goal. To avoid inflating the risk of a type II error by applying corrections for multiple comparisons and inaccurately concluding that the diagnostic groups were similar across all domains, uncorrected P values are reported. Estimated effect sizes of the differences between groups adjusted for covariates were calculated as described by Cohen²⁹; effect sizes were converted to d to facilitate comparisons across multiple analyses. Power estimates for the detection of effect sizes posited by Cohen were obtained using a commercially available software program (Sample Power Release 1.2; SPSS Inc).

Most subjects (65.5%) were referred from outpatient programs; 14.8% of the subjects were recruited from inpatient units, 14.2% from advertisements, and 5.5% from other sources. Most subjects (255 [58.2%]) met the criteria for BP-I at intake, with a few (30 [6.8%]) diagnosed as having BP-II and a substantial proportion (153 [34.9%]) diagnosed as having BP-NOS. The subjects with BP-II were older than the subjects with BP-I, and both groups were older than the subjects with BP-NOS; this pattern was replicated in the ratings of pubertal status (Table 1). Comparisons of sex and household composition among the 3 groups showed potential differences at P<.20. These 3 demographic factors were controlled for in all of the analyses. Although the BP-II group had a later age of onset compared with the BP-I and BP-NOS groups, the 3 groups did not differ when controlling for demographic variables.

There were no significant differences in the mean KSADS severity ratings of manic or depressive symptoms during the most symptomatic week in the month before intake (Table 1). Subjects with BP-I had a small but significantly higher level of functional impairment during the month before intake than the other groups, as measured on the Child Global Assessment Scale, and more impairment during the most severe period in their lifetime.

The only significant difference in the rates of comorbid disorders was that subjects with BP-II had higher rates of anxiety disorders compared with subjects with BP-I and BP-NOS (Table 1).

The rate of lifetime history of suicidal ideation did not differ among the 3 BP subtypes; however, subjects with BP-I were significantly more likely to attempt suicide than subjects with BP-NOS (Table 1). Compared with subjects with BP-NOS, subjects with BP-I had significantly higher rates of a lifetime history of psychosis, psychiatric hospitalization, and psychotropic medication treatment. There was no significant difference in the lifetime rates of an MDE between the BP-I and BP-NOS groups. The subjects with BP-II did not have any significant differences compared with the other groups, except for a lower rate of hospitalization than the subjects with BP-I.

As shown in Table 2, the only significant difference among the 3 BP subtypes in the psychiatric family history was a higher proportion of subjects with BP-II with a history of suicide attempts in any first-degree relatives compared with the subjects with BP-NOS.

The subjects with BP-NOS had significantly lower KSADS MRS ratings during their most severe period of manic symptoms during their lifetime compared with the subjects with BP-I (rating, 27.8 ± 6.6 vs 37.3 ± 7.3; F_1,329 = 123, P<.001, d = 1.30). Table 3 shows the presence of individual manic symptoms during the most severe lifetime episode. There were small differences between the subjects with BP-I and those with BP-NOS when mild (KSADS score, ≥3) was set as the threshold for the presence of a symptom. The odds ratios were most increased in the subjects with BP-I relative to the subjects with BP-NOS for decreased need for sleep and hypersexuality; these were the only 2 symptoms in which d > 0.40. Subjects with BP-I had significantly higher rates of all symptoms rated at an intensity of moderate (KSADS score, ≥4) or higher compared with the subjects with BP-NOS.

During the most severe lifetime episode, subjects with BP-I had significantly more of the 7 DSM-IV symptom criteria for a manic or hypomanic episode rated at a mild or higher severity, compared with the subjects with BP-NOS (6.1 ± 1.1 vs 5.2 ± 1.4; F_1,331 = 39.0, P<.001, d = 0.74). The difference between the BP-I and BP-NOS groups was more substantial when the DSM-IV criteria threshold was set at moderate or higher (5.3 ± 1.4 vs 3.6 ± 1.6; F_1,331 = 95.2, P<.001, d = 1.14).

There were 3 ways in which subjects could meet the COBY study criteria for BP-NOS but would not meet the DSM-IV criteria for BP-I or BP-II: (1) the subject met the DSM-IV criteria for a hypomanic episode but did not have a history of an MDE, (2) the subject had periods of manic symptoms that were of sufficient duration (≥4 days for hypomania or ≥7 days for mania) but had 1 symptom criterion less than the DSM-IV threshold, or (3) the subject had a sufficient number and types of manic symptoms, but the duration of episode(s) was insufficient. Of the 153 subjects with BP-NOS, 18 (11.8%) had a hypomanic episode but no MDE. For the remaining subjects with BP-NOS who had KSADS MRS ratings for the most severe lifetime episode, 96.6% met the DSM-IV symptom criteria (elated or elevated mood plus 3 symptom criteria or irritable mood plus 4 symptom criteria) for a manic or hypomanic episode when the threshold for presence of a DSM-IV criterion was set at mild or higher. Even with the DSM-IV symptom threshold set at moderate or higher, 61.2% of the subjects with BP-NOS met the DSM-IV symptom criteria for a manic or hypomanic episode.

To our knowledge, this is the first report on the systematic assessment and comparison of children and adolescents with BP-I, BP-II, and BP-NOS. Children and adolescents with bipolar spectrum disorders present with a history of substantial impairment and high rates of prior hospitalization, medication treatment, major depression, suicidal ideation, and comorbid disruptive and anxiety disorders. In addition, they frequently have histories of psychotic symptoms and suicide attempts.

Youth with BP-NOS were similar to those with BP-I on many factors, including age of onset, duration of illness, rates of comorbid diagnoses and prior MDEs, family history, and the types of manic symptoms that were present during the most serious lifetime episode. Of the specific manic symptoms, decreased need for sleep and hypersexuality showed the largest difference in the subjects with BP-I. During the most serious lifetime episode of manic symptoms, subjects with BP-I on average met 1 more DSM-IV manic symptom criterion compared with subjects with BP-NOS; the symptoms were more intense and functional impairment was more severe in the BP-I group. Rates of psychosis and psychiatric hospitalization were higher in the subjects with BP-I, although this is not unexpected given that these features are part of the DSM-IV criteria for a manic episode. While rates of suicidal ideation were similar for both groups, subjects with BP-I had higher rates of suicide attempts.

It is of particular interest how the subjects with BP-NOS did not reach the full threshold for a BP-I or BP-II diagnosis. Although the COBY study criteria for BP-NOS allowed subjects to be included if they had 1 DSM-IV symptom criterion less than the full criteria for a manic or hypomanic episode, nearly every subject with BP-NOS met the full DSM-IV symptom criteria during the most serious lifetime episode. A small proportion had a lifetime history of a hypomanic episode but no history of an MDE. This indicates that the subjects with BP-NOS did not have the full criteria for BP-I or BP-II primarily because they did not meet the DSM-IV duration criteria of 7 consecutive days of manic symptoms for a manic or mixed episode or 4 consecutive days for a hypomanic episode.

The proportion of subjects meeting the full DSM-IV criteria for BP-II at intake was small compared with the BP-I and BP-NOS groups. The subjects with BP-II were older, generally late or postpubertal, and were more likely to be female. Functional impairment was mildly less in the subjects with BP-II compared with the subjects with BP-I, but overall there were few detectable differences between the subjects with BP-II and the subjects with either BP-I or BP-NOS. Whether the diagnosis of BP-II in children and adolescents has distinct implications from that of BP-I will require a larger sample of subjects with BP-II.

The results of the study must be considered in the context of certain limitations. The data were obtained at the initial intake assessment and represent history and phenomenology obtained via the retrospective recall of the subjects and their caretakers. Although the size of the BP-I and BP-NOS samples was relatively large, we did not have robust power to detect differences of small effect size. The fact that few statistically significant differences were found between the 2 groups does not prove that they are equivalent. The BP-II sample was substantially smaller, which only allowed for detecting differences of large effect size. The study sample was recruited by a variety of methods, including referral from clinical programs. Therefore, the subjects may represent a more severely affected cohort of bipolar youth and these results may not generalize to an epidemiological sample of bipolar youth. Although nearly all subjects had significant manic and/or depressive symptoms during the month before intake, subjects were not required to meet the criteria for a manic, mixed, or hypomanic episode at the time of study intake. This may have identified a cohort of subjects whose illness was less chronic or severe than bipolar subjects obtained from clinical settings during a full criteria episode. Because the DSM-IV does not have clearly operationalized criteria for cyclothymic disorder, we did not systematically separate subjects who might meet the criteria for cyclothymic disorder from those who fulfilled the COBY study BP-NOS criteria. Detailed characterization of manic symptoms was performed for the worst week in the past month and the most serious week in the subject's lifetime. We chose this method because we believed that subjects and their parents would have relatively accurate recall of symptoms that occurred in the recent past and when they were most severe. The circumscribed period of 1 week would help ensure that the symptoms occurred concomitantly. However, it is possible that different manic symptoms were present at other times. Finally, the family history assessment method relied on the report of one caregiver for the other relatives.

When comparing the COBY study results with those of the 3 large cross-sectional phenomenology studies of pediatric BP that used KSADS-based interviews, the methodological differences could affect the comparison. Biederman and colleagues,¹⁴ from Massachusetts General Hospital (Mass General), Boston, described a sample of 186 children and adolescents meeting the DSM-III-R criteria for mania; these subjects were identified from consecutive referrals to a pediatric psychopharmacology clinic. Additional details from the initial 110 subjects were the focus of an earlier report.³⁰ Geller and colleagues,¹⁵ from Washington University, St Louis, Mo, described the cross-sectional presentation of the 93 children and early adolescents with BP-I who were recruited by consecutive case ascertainment from psychiatric and pediatric sites for a longitudinal follow-up study.³¹ Subjects were required to meet DSM-IV criteria for mania for at least 2 weeks or for hypomania for at least 2 months and to have either elated mood or grandiosity. Findling and colleagues,¹⁶ from Case Western Reserve University, Cleveland, Ohio, described a referred cohort of 90 children and adolescents meeting the DSM-IV criteria for BP-I. There was some variability among the studies as to how the presence or absence of a particular manic symptom was coded. The Mass General study used the KSADS–Epidemiologic Version,³² which has a dichotomous presence or absence choice. The Washington University study used the WASH-U KSADS,³³ which has similar symptom severity coding as the KSADS MRS. Symptoms met the threshold for hypomania if the severity rating was mild and met the threshold for mania if the rating was moderate or higher. The Case Western Reserve University study used the KSADS–Present and Lifetime Version, which has thresholds for manic symptoms that are roughly equivalent to either mild or moderate on the KSADS-P. The samples for all 3 studies were younger than the COBY study BP-I sample. To reduce the effect of age, we selected a random subsample of the COBY study BP-I adolescents (35% of 13-14 year olds and 25% of 15-16 year olds) in addition to the children with BP-I to include in the comparison. Of the 150 subjects in the age-matched sample, manic symptom data on the most serious lifetime episode were available for 133. Table 4 summarizes the findings of the 3 studies and of the younger COBY study BP-I sample.

The COBY study sample was similar to the others as far as the high rates of comorbid attention-deficit/hyperactivity disorder and oppositional defiant disorder, and the high frequency of irritability. Age of onset of mania was older in the COBY study subjects, despite the selection of a younger subsample. The current or most recent DSM-IV major mood episode for the COBY study subjects was more likely to be a manic episode (46%) than a mixed episode (34%). Rates of subjects meeting DSM-IV criteria for mixed episodes were significantly lower than what was found in the Mass General sample. Although the presence of elated or elevated mood and/or grandiosity was not required for inclusion, the COBY study sample had high rates of these symptoms. This was true even if the conservative threshold of moderate or more severe is used (elated or elevated mood, 84%; and grandiosity, 53%). Rates of comorbid conduct disorder were lower than in the Mass General sample. However, the COBY and the Mass General samples had high rates of comorbid anxiety disorders, in contrast to the other 2 groups.

Methodological differences among the studies may account for some of the variation in results (Kowatch et al³⁴ provide additional details). For instance, subjects in the COBY study were explicitly required to have a distinct episode of abnormal mood during which the manic symptoms had to occur. In addition, manic symptoms that were also in common with other psychiatric disorders were counted as satisfying the criteria for mania in the COBY study only if they clearly intensified with the onset of abnormal mood. This is in contrast to the Washington University study, in which symptoms that satisfied the criteria for more than 1 diagnosis were counted toward each diagnosis.³⁵ Because none of the studies used epidemiological ascertainment, sampling biases may also have influenced the results.

Although the full implications of the COBY study await the extended longitudinal follow-up phase, several clinically relevant observations can be made from the cross-sectional assessment. Children and adolescents with a bipolar spectrum disorder have a severe illness marked by high rates of comorbid psychiatric disorders, psychosis, suicidality, and functional impairment. Children and adolescents with significant manic symptoms who do not meet the full criteria for BP-I or BP-II usually met the DSM-IV symptom criteria for mania or hypomania but not the duration criteria. Whether the duration criteria for manic or hypomanic episodes are appropriate for pediatric populations remains an empirical question for which the longitudinal component of the COBY study may provide some answers. Last, most children and adolescents with BP-NOS have similar, albeit less severe, presentations as youth with BP-I disorder and have similar comorbidity and family histories. The effect size of the differences between the subjects with BP-I and BP-NOS (when they existed) was generally of small to medium magnitude (d = 0.2-0.5). The fact that the subjects with BP-NOS differed from those with BP-I primarily on duration and severity of manic symptoms, but not on the fundamental phenomenology of manic symptoms, comorbid disorders, or family history, provides preliminary evidence that many of these subthreshold pediatric cases have a phenotype that is on the same continuum as that of youth with BP-I. Samples of adults with bipolar spectrum illness indicate that many bipolar adults report brief (1-3 day) periods of hypomania.^36,37 Therefore, it is not clear whether most youth with BP-NOS will continue into adulthood with short periods of manic symptoms or progress into full-threshold BP-I disorder.

The COBY study inclusion criteria for BP-NOS defined minimum criteria that were intentionally set at a low threshold for research purposes to obtain a broad range of subjects with subthreshold manic symptoms; they do not establish a recommended boundary for clinical diagnosis. Furthermore, many subjects with BP-NOS substantially exceeded this minimum threshold but did not meet the DSM-IV criteria for BP-I or BP-II. The COBY study is assessing which of the subjects within the broad BP-NOS group will go on to have definite BP. Preliminary results from the longitudinal phase of the first 92 subjects with BP-NOS recruited into the COBY study indicate that a significant number go on to meet the DSM-IV criteria for BP-I during follow-up.³⁸ The proportion of subjects with BP-NOS who eventually convert to DSM-IV BP-I, and the factors that moderate and/or mediate conversion, will require extended follow-up.

Correspondence: David Axelson, MD, Western Psychiatric Institute and Clinic, 3811 O’Hara St, Pittsburgh, PA 15213 (axelsonda@upmc.edu).

Submitted for Publication: August 23, 2005; final revision received January 27, 2006; accepted January 30, 2006.

Financial Disclosures: Dr Keller has been a consultant to or received honoraria from Bristol-Myers Squibb, Collegium, Cyberonics, Inc, Cypress Bioscience, Inc, Eli Lilly and Company, Forest Laboratories, Inc, Janssen, LP, Merck & Co, Inc, Organon USA, Otsuka Pharmaceutical Co, Ltd, Pfizer Inc, PharmaStar, Sepracor Inc, Vela Pharmaceuticals Inc, and Wyeth Pharmaceuticals; has received grants from or performed research for Eli Lilly and Company, Forest Laboratories, Inc, Merck & Co, Inc, Organon USA, Pfizer Inc, and Wyeth Pharmaceuticals; and has served on the advisory boards of Abbott Laboratories, Bristol-Myers Squibb, Cephalon, Inc, Cyberonics, Inc, Cypress Bioscience, Inc, Eli Lilly and Company, Forest Laboratories, Inc, GlaxoSmithKline, Janssen, LP, Merck & Co, Inc, Mitsubishi Pharma Corporation, Novartis, Organon USA, Pfizer Inc, Sanofi-Synthelabo, Scirex Corporation, Sepracor Inc, Somerset Pharmaceuticals Inc, Vela Pharmaceuticals Inc, and Wyeth Pharmaceuticals.

Funding/Support: This study was supported by grants MH59929 (Dr Birmaher), MH59977 (Dr Strober), and MH59691 (Dr Keller) from the National Institute of Mental Health.

Acknowledgment: We thank Carol Kostek for her assistance with manuscript preparation; COBY faculty, including Kristin Bruning, MD, and Jennifer Dyl, PhD; raters, including Mathew Arruda, BA, Mark Celio, BA, Jennifer Fretwell, BA, Michael Henry, BS, Risha Henry, PhD, Norman Kim, PhD, Marguerite Lee, BA, Marilyn Matzko, EdD, Heather Schwickrath, MA, Anna Van Meter, BA, and Matthew Young, BA; data personnel, including Amy Broz, AS, Colleen Grimm, BA, and Nicole Ryan, BA; and Editha Nottelmann, PhD, and Regina James, MD, for their continued support.