Bipolar Disorders In a Community Sample of Older Adolescents: Prevalence, Phenomenology, Comorbidity, and Course PETER M. LEWINSOHN, PH.D., DANIEL N. KLEIN, PH.D.,
AND
JOHN R. SEELEY, M.S.
ABSTRACT Objective: The purpose of this study was to examine the prevalence, clinical characteristics, and mental health treatment
services utilization of adolescents with bipolar disorders and manic symptoms. Method: Structured diagnostic interviews were administered to a representative community sample of 1,709 older adolescents (aged 14 through 18 years). Results: The lifetime prevalence of bipolar disorders (primarily bipolar II disorder and cyclothymia) was approximately
1%. An additional 5.7% of the sample reported having experienced a distinct period of abnormally and persistently elevated, expansive, or irritable mood even though they never met criteria for bipolar disorder ("core positive" subjects). The rate of manic symptoms in these subjects was similar to that reported in clinical samples, and the course of bipolar disorder was relatively chronic. Compared with adolescents with a history of major depression (n = 316) and a "never mentally ill" group (n
= 845), the bipolar and core positive subjects both exhibited significant functional impairment and
high rates of comorbidity (particularly with anxiety and disruptive behavior disorders), suicide attempts, and mental health services utilization. Conclusions: These data highlight the clinical and public health significance of even the milder and subthreshold cases of bipolar disorder in adolescence. J. Am. Acad. Child Ado/esc. Psychiatry, 1995, 34, 4:454-463. Key Words: bipolar, affective, elation.
In 1960, Anthony and Scott published a seminal paper demonstrating that there are some children who meet carefully specified diagnostic criteria for bipolar disorder. Since then, many investigators have described children and adolescents with full-blown bipolar illness (Ballenger et al., 1982; Carlson and Strober, 1979; Gammon et al., 1983; Hsu and Starzynski, 1986; Krasa and Tolbert, 1994; Strober and Carlson, 1982; Weinberg and Brumback, 1976; White and O'Shanick, 1977). This literature is concordant with both classic (e.g., Kraepelin, 1921) and more contemporary (Loranger and Levine, 1978; Weissman et al., 1988) reports indicating that substantial numbers of adults with
bipolar disorder date the onset of their disorder in childhood or adolescence. Hence, it is currently accepted that bipolar disorder can be manifested in children and adolescents, although it appears to be rare, particularly before adolescence (Bowring and Kovacs, 1992; Carlson, 1990; Strober et al., 1993). Almost all available data on bipolar disorder in childhood and adolescence have been derived from patient samples. There are almost no data available from epidemiological studies of community samples of children and adolescents. Epidemiological studies are important for several reasons. First, they are necessary to establish the prevalence of the disorder. In treatment settings, it is likely that children and adolescents with full-blown bipolar (bipolar I) disorder and those whose bipolar disorders are comorbid with other mental disorders are overrepresented and that the milder forms such as bipolar II and cyclothymia are underrepresented. These milder conditions may be the most common forms of bipolar disorder (Akiskal and Mallya, 1987; Egeland and Hostetter, 1983). Basing generalizations on clinic samples may also be problematic because it is known that some cases of childhood bipolar
Accepted September 9, 1994. Dr. Lewinsohn and Mr. Seeley are with the Oregon Research Institute, Eugene, and Dr. Klein is with the Department ofPsychology, State University of New York at Stony Brook. This research was supported in part by NIMH awards MH40501 and MH50522. Reprint requests to Dr. Lewinsohn, Oregon Research Institute, 1115 Franklin Blvd., Eugene, OR 91403-1983. 0890-8567/95/3404-0454$03.00/0© 1995 by the Ametican Academy of Child and Adolescent Psychiatry.
454
J.
AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:4, APRIL 1995
BIPOLAR DISORDERS IN ADOLESCENTS
disorder may be misdiagnosed as attention-deficit hyperactivity disorder or conduct disorder (Carlson, 1990; Garvey and Tuason, 1980; Joyce, 1984; Mukherjee et aI., 1983). Finally, there is evidence suggesting that treatment itself may influence the prevalence of the disorder. Thus, a number of investigators have argued that antidepressants can precipitate manic and hypomanic episodes in children and adolescents (Strober and Carlson, 1982) as well as adults (Wehr and Goodwin, 1987). Hence, epidemiological studies are necessary to provide an accurate estimate of the prevalence and clinical characteristics of bipolar disorder in children and adolescents. None of the three major epidemiological studies done in the recent past using DSM-lII or DSM-III-R criteria (Hall et al., 1980; McGee et al., 1990; Whitaker et al., 1990) have reported the prevalence of bipolar disorder and, to our knowledge, only one epidemiological study of psychiatric disorders in community samples of children or adolescents has reported data on bipolar disorder or symptomatology. In a sample of 150 adolescents, Carlson and Kashani (1988) reported that 13.3% reported periods of at least 2 days in which they experienced four or more manic symptoms. While none of these adolescents exhibited sufficient impairment to meet criteria for a manic episode, 3 (1.5%) appeared to qualify for a diagnosis of bipolar II disorder or cyclothymia. The adolescents with manic symptoms exhibited high rates of comorbidity, and 70% were judged by the interviewers to need treatment. In a previous publication describing results from the Oregon Adolescent Depression Project (OADP), we reported the prevalence (point and lifetime) and incidence of bipolar disorders as per DSM-lII-R criteria. The OADP is a community-based, longitudinal investigation of the epidemiology of psychiatric disorders in a cohort of 1,709 high school students (Lewinsohn et al., 1993). In the previous report (Lewinsohn et al., 1993), we presented data on the point and lifetime prevalence and 12-month incidence of DSM-III-R bipolar disorder (including cyclothymia). For the purposes of this report, we have broadened the definition of bipolar disorder to include the DSM-IV category of bipolar II (hypomanic episode and at least one major depressive episode). Hence, additional information on the prevalence and incidence of bipolar disorder is presented in this report. Primarily, however, in this
J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:4, APRIL 1995
report we examine (1) the prevalence and dimensionality of manic symptoms in the high school population; (2) the age of onset and duration of episodes of bipolar disorder; and (3) the demographic and clinical correlates of adolescent bipolar disorder, including comorbidity, impairment, and mental health services utilization. METHOD Subjects and Procedure Participants were randomly selected from nine senior high schools representative of urban and rural districts in western Oregon. A total of 1,709 adolescents completed the initial (T 1) assessments (interview and questionnaires) between 1987 and 1989, with an overall participation rate of 61 %. At the second assessment (T 2 ), 1,507 participants (88.2%) returned for a readministration of the interview and questionnaire (mean T 1-T2 interval = 13.8 months, SD = 2.3). With minor exceptions, the adolescents in the T 1 sample were representative of high school students in western Oregon. Small but statistically significant differences emerged due to attrition in the T 1-T2 panel sample; attrition was associated with lower parental socioeconomic status, smaller household number, male gender, past diagnoses of disruptive behavior disorders, and past substance use disorders (males only). Additional details regarding the sample are provided in Lewinsohn et al. (1993). Approximately half of the T 1-T 2 panel sample was female (53.7%), with an average T 1 age of 16.6 (SD = 1.2). A total of 8.9% were nonwhite; 71.3% were living with two parents and 53% were living with two biological parents; 12.3% had repeated a grade in school. Parental education level (maximum value for mother or father) was as follows: 1.9% did not complete high school, 16.1 % completed high school, 35.1 % had a partial college education, and 46.9% had an academic or professional degree.
Diagnostic Interview Adolescents were interviewed at T 1 with a version of the Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS) that combined features of the Epidemiologic version (K-SADS-E) (Orvaschel et al., 1982) and the Present Episode version (K-SADS-P) and included additional items to derive diagnoses of most disorders as per DSM-III-R criteria (American Psychiatric Association, 1987). Parents were not interviewed in light of evidence that the reliability of children's reports of psychopathology increase with age, while the reliability of parents' reports decreases (Edelbrock et al., 1985). At T 2 , subjects were interviewed using the Longitudinal Interval Follow-up Evaluation (Keller et al., 1987), which provides detailed information about the course of psychiatric symptoms and disorders since the initial K-SADS interview, with rigorous criteria for recovery from a disorder (i.e., symptom-free, or one or two symptoms to a mild degree, for 8 or more weeks). Interviewers also elicited information on, and rated the presence and severity of, depressive symptoms and other psychiatric disorders since T 1 using the K-SADS format. Therefore, lifetime diagnostic information was available regarding the occurrence and duration of all disorders before and at T 2•
455
LEWINSOHN ET AL.
Interrater Reliability Diagnostic interviewers were carefully trained and supervised. For reliability purposes, all interviews were either audio- or videotaped and a second interviewer reviewed the recordings of 233 randomly selected interviews. Interrater reliability was evaluated by using the K statistic (Cohen, 1960) for diagnosis of bipolar disotder and for DSM-III-R manic symptoms. The K value for the agreement on a lifetime diagnosis of bipolar disorder was .49; the K value for the agreement on a lifetime occurrence of a distinct period of elevated, expansive, or irritable mood was .57. Among the 23 subjects rated as having experienced an elevated, expansive, or irritable mood by the original interviewer, K values for the agreement on the lifetime occurrence of the seven DSMIll-R-associated symptoms were as follows: inflated self-esteem (.44), decreased need for sleep (.62), more talkative than usual (.66), flight of ideas (.41), distractibility (.51), inctease in goaldirected activity or psychomotor agitation (.34), and excessive involvement in pleasurable activities with a high potential for painful consequences (.44).
Diagnostic Groups A total of 18 cases of bipolar disorder were identified at either the time 1 or time 2 assessment. These included 2 individuals who met full DSM-III-R criteria for both a lifetime manic and major depressive episode; 11 who met all criteria for a manic episode except for marked impairment in social, family, or school functioning or hospitalization, and had a histoty of major depression, and received a diagnosis of bipolar not otherwise specified (bipolar II in DSM-IV); and 5 who received a diagnosis of cyclothymia. For comparative purposes, three additional mutually exclusive diagnostic groups were constructed: subjects who reported having experienced a distinct period of abnormally and persistently elevated, expansive, or irritable mood, but never met criteria for bipolar disorder (core positive, n = 97); subjects with no lifetime history of a distinct period of abnormally and persistently elevated, expansive, or irritable mood (core negative, n = 1,594); subjects who reported never experiencing a period of elevated, expansive, or irritable mood but had a histoty of major depressive disorder (MDD, n = 316); and subjects who were never mentally ill and did not qualifY for the core positive group (NMI, n = 845). It should be noted that a subset of the core positive subjects met symptom criteria for hypomania but did not receive the diagnosis of bipolar II disorder. Because the period of mood disturbance was less than 4 days, there was no histoty of major depression, or one or more DSM-IVexciusion criteria were met.
Assessment of Clinical Characteristics Mental Health Treatment Utilization. Information about treatment utilization, which was defined as teceiving outpatient psychotherapy or counseling, medications, or hospitalization for a mental disorder, was obtained as part of the diagnostic interview at both T) and T 2• Treatment urilization was assessed whenever a diagnosis was obtained. A total of 272 subjects (15.9%) reported receiving some form of treatment for one or more mental disorders. Global Assessment ofFunctioning (GAF). GAF scores were determined as per DSM-llI-R Axis V criteria for level of functioning both at the T 2 interview and during the previous year. Scores could range from 1 to 90, with ratings less than or equal to 70 indicating mild symptomatology or some degree of role impairment in social, school, or occupational functioning.
456
Impairment. As part of the K-SADS interview, it was determined whether subjects exhibited impairment in social, family, and school functioning as part of an affective episode. History of Suicide Attempt. As part of both the T 1 and T 2 interview assessments of depression, adolescents were asked, "Have you ever tried to kill yourself or done anything that could have killed you?" A total of 133 subjects (7.8%) reported a history of having made one or more suicide attempts at either assessment.
RESULTS Prevalence (Lifetime and Point) and Incidence of Bipolar Disorders in Adolescents
Of the 1,709 adolescents evaluated in the initial phase of the study, we identified 16 cases of bipolar disorders, yielding a lifetime prevalence of 0.94%. Among the 1,507 adolescents reevaluated in the second wave approximately 12 months later, there were 15 cases of bipolar disorders, yielding a lifetime prevalence of 0.99%. The point prevalence of bipolar disorder was 0.64% (11 cases) at T 1 and 0.53% (8 cases) at T z• Two of the adolescents with bipolar disorders identified in the second wave were new cases, yielding a I-year incidence rate of 0.13% (2/1,507 - 13 preexisting cases = 2/1,494 = 0.13%). Overall, between the two evaluations, we identified a total of 18 cases of bipolar disorders. Of the 18 bipolar cases, 2 met full DSM-III-R criteria for a lifetime manic episode. Eleven adolescents met full criteria for a major depressive episode, and all criteria for a manic episode except for marked impairment in social or occupational functioning or hospitalization. In these individuals, bipolar disorder not otherwise specified (bipolar II in DSM-IV) was diagnosed. Of these 11 bipolar II cases, three had met criteria for cyclothymia before their major depressive episode. Five adolescents received diagnoses ofcyclothymia with no lifetime history of major depression. Finally, there were 97 cases (5.7% of the sample) who reported having experienced a distinct period of elevated, expansive, or irritable mood but who did not meet criteria for bipolar or bipolar II disorder or cyclothymia. Their mean number of associated manic symptoms was 2.9 (SO = 2.1; range = 0 to 7). These subjects will be referred to as "core positive" throughout the rest of the article. Only one of the bipolar subjects and none of the core positive subjects were treated with lithium; none from either group had been hospitalized; and one of
]. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:4, APRIL 1995
BIPOLAR DISORDERS IN ADOLESCENTS
the bipolar subjects was rated as having experienced psychotic symptoms. To wit, this participant reported that the characters from the television show "Star Trek" were talking to her and that certain fellow students were threatening to harm her. Demographic and Clinical Characteristics
The demographic and clinical characteristics of the adolescents with bipolar disorders and the core positive subjects are presented in Table 1. For comparative purposes, data are also presented for two groups from the larger OADP project with no history of distinct periods of elevated, expansive, or irritable mood: 316 adolescents with a history of major depression (MDD) and 845 subjects with no history of mental disorder (NMI). Significant differences between groups were followed by comparisons of bipolar versus core positive, MDD versus bipolar and core positive groups combined, and NMI versus bipolar and core positive groups combined. There were significantly fewer females in the combined bipolar and core positive group than in the major depressive group, but a significantly greater proportion of females than in the NMI group (Table 1).
The core positive subjects were also significantly older than the bipolar subjects. Finally, the combined bipolar and core positive group was significantly less likely to live with both parents than the NMI subjects. With regard to clinical characteristics, the bipolar subjects exhibited significantly poorer functioning on the GAF scale, both currently and during the past year, than the core positive group. Furthermore, the combined bipolar and core positive groups had poorer functioning on the GAF than the MDD and NMI groups. Compared with the NMI group, the combined bipolar and core positive groups were significantly more likely to have received mental health treatment. Finally, a significantly greater proportion of bipolar cases than core positive subjects had a history of suicide attempts; in the combined bipolar and core positive groups, the proportion with a history of suicide attempts was also significantly greater than in the NMI group. The participants were also rated on impairment in social, family, and school functioning as part of their episode. A substantial proportion of bipolar and core positive subjects exhibited impairment in all three spheres, with the rates of impairment being comparable with that of the MDD group (Table 2). The proportion
TABLE 1 Diagnostic Group Differences on Demographic Variables, Global Assessment of Functioning, Treatment Utilization, and Suicide Attempt Contrasts"
Diagnostic Group Variable n % Female Mean age (yr) % White % Live with 2 biological parents Parental education (%) < High school graduate High school graduate Partial college College degree (B.S./B.A.) Professional degree Global assessment of functioning Current During past year % Received treatment % Attempted suicide
NMI (I)
MDD (2)
Core Positive (3)
845 47.9 16.4 92.2
316 69.9 16.7 88.3
97 56.7 16.9 92.8
18 66.7 16.0 94.4
NA X2(3) = 46.29*** F(3,1272) = 10.03*** X2(3) = 4.97
59.5
42.4
46.4
38.9
X2(3) = 31.25*** X2 (l2) = 19.53
0.34
0.28
1.4 14.0 33.9 26.6 24.0
3.2 20.4 37.7 23.6 15. I
2.3 17.0 38.6 23.9 18.2
0.0 17.6 35.3 29.4 17.6
87.5 87.6 3.4 1.2
83.1 83.6 33.2 22.2
82.0 83.0 39.2 14.4
76.3 74.9 55.6 44.4
F(3,1237) = 55.40*** F(3,1233) = 64.63*** X2(3) = 248.32**~ X2(3) = 177.40***
3.06** 4.96*** 1.67 8.84**
3.90*** 8.73*** 5.29*** 10.33*** 2.67 201.34*** 0.46 101. 18***
BD (4)
Test Statistic
3 vs. 4
2 vs. 3,4 I vs. 3,4
0.62 2.79**
5.18* 1.74
4.32* -0.27
8.44**
Note: NMI = never mentally ill; MDD = major depressive disorder; BD = bipolar disorder; NA = not applicable. " t Tests are reported for the continuous-type variables; X2 tests are reported for the categorical variables. * p < .05; ** P < .01; *** P < .001.
J.
AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:4, APRIL 1995
457
LEWINSOHN ET AL.
TABLE 2 Percent of Participants Rated as Showing Impairment for the Diagnosric Groups Impairment Social Family School
BD (n
=
18)
66.7 55.6 83.3
MDD (n
Group = 316)
Core Positive (n
54.7 64.9 66.1
97)
=
X2 (df = 2) 2.43, NS 4.32, NS 9.01*
48.5 (30.W 53.6 (34.0) 52.6 (30.0)
Note: BD
= bipolar disorder; MDD = major depressive disorder; NS = not significant. Numbers in parentheses are percentages for core positive group without MDD (n = 50). * P < .01. a
of bipolar subjects showing impairment in school functioning was especially high. It is important to note, however, that the degree of impairment exhibited by those core positive subjects who did not have a lifetime history of MOD was substantially less (n = 50; 30%, 34%, and 30% for social, family, and school impairment, respectively).
irritable but not elevated mood. The most common manic symptom in both groups was an increase in goaldirected activity. Other frequent symptoms included increased speech, inflated self-esteem, decreased need for sleep, and distractibility. Flight of ideas and excessive involvement in pleasurable activities with a high potential for painful consequences were the least common symptoms. It is not surprising that almost all symptoms were more frequent among the bipolar than core positive subjects, with significant differences emerging for inflated self-esteem, decreased need for sleep, and increase in goal-directed activity. However, the relative frequency of symptoms was similar across the two groups (Spearman p = .74, P = .03, one-tailed). Overall, manic symptoms were relatively highly intercorrelated.
Symptomatology The prevalence of specific DSM-III-R manic symptoms during the most recent episode in the 18 bipolar and 97 core positive subjects appears in Table 3. Subjects in both groups were much more likely to report elevated or expansive than irritable mood, and relatively few subjects (11 % of the bipolar subjects; 21 % of the core positive subjects) reported periods of
TABLE 3 Prevalence of DSM-Ill-R Manic SymptOms by Group Group BD (n Symptom Core Elevated/expansive mood Irritable mood Inflated self-esteem Decreased need for sleep More talkative than usual Flight of ideas Distracribility Increase in goal-directed activity or psychomotor agitation Excess pleasurable acrivities with painful consequences
=
18)
Core Positive (n
=
97) 2
No.
%
No.
%
X
16 4 11 13 8 11
88.9 22.2 61.1 61.1 72.2 44.4 61.1
71 41 31 30 51 41 36
73.2 42.3 32.0 30.9 52.6 42.3 37.1
2.03 2.56 5.57* 6.03*a 2.37 0.03 3.62
17
94.4
64
66.0
5.91*
6
33.3
29
29.9
0.08
11
= bipolar disorder. Significant group X sex interaction X2(l, N = 115) = 4.40, P < .05. Prevalence for female cases was 75.0% compared with 25.5% for female controls, X2 (l, N = 67) = 10.73, P < .01. Prevalence for male cases was 33.3% compared with 38.1 % for male controls, x2(l, N = 48) = 0.05, P < .05. * P < .05. Note: BD
a
458
J.
AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:4, APRIL 1995
BIPOLAR DISORDERS IN ADOLESCENTS
Treating the seven symptoms as a scale, and combining the 115 bipolar and core positive subjects, Cronbach's a was .73.' It is of interest to compare the relative prevalence of manic symptoms in our sample with those reported by Krasa and Tolbert (1994) in a sample 008 bipolar adolescent inpatients. The rank order of the frequency of manic symptoms reported by Krasa and Tolbert (1994) correlated .63 and .75 with the frequencies of symptoms in the bipolar and core positive groups, respectively. Males and females did not differ on the prevalence of any manic symptoms or on the total number of manic symptoms. However, there was a significant group by sex interaction on one symptom, decreased need for sleep. Seventy-five percent of the bipolar, but only 25.5% of the core positive, women reported a decreased need for sleep, X2 (1, N = 67) = 10.73, P < .01. In contrast, the frequency of this symptom was 33.3% and 38.1 % in male bipolar and core positive subjects, respectively, X2 (1, N = 48) = 0.05, P > .10. Dimensionality of Manic Symptoms
The dimensional structure of the manic symptoms during the most recent episode for the bipolar and core positive subjects (total n = 115) was examined via principal-components analysis. Both a one- and a two-component solution were evaluated. The onecomponent solution accounted for 38.2% of the variance with all seven items loading greater than .40 (range = .48 to .7I). Because two components had eigenvalues greater than 1.0, a two-component solution was also examined. The second component accounted for an additional 14.6% of the variance. The component loadings from an oblique rotation are shown in Table 4. The first component had relatively high and unique loadings on the following symptoms: decreased need for sleep, flight of ideas, distractibility, and poor judgment. The second component had high and unique loadings on inflated self-esteem and increase in goaldirected activity. The two components were significantly correlated, r = .36, P < .001. However, neither component was significantly correlated with age or sex. The first component appears to reflect behavioral disorganization and impaired functioning. Consistent with this interpretation, it is correlated with the impairment ratings in Table 2: total score (range 0 to 3), r = .27, P = .003; social impairment, r = .17, P = .08;
]. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:4, APRIL 1995
TABLE 4 Principal-Components Analysis of DSM-III-R Manic Symproms Component Loadings Symptom
Component 1
Component 2
Inflated self-esteem Decreased need for sleep More talkative than usual Flight of ideas Distractibility Increase in goal-directed activity or psychomoror agitation Excess pleasurable activities with painful consequences
-.11 .63 .44 .78 .81
.84 .09 .40 -.02 -.14
.11
.74
.53
.03
Eigenvalue % of variance
2.68 38.2
1.02 14.6
family impairment, r = .22, P = .017; and school impairment, r = .20, P = .03. It is interesting to speculate that individuals with elevated scores on this component may be particularly likely to exhibit psychotic symptoms during more severe manic episodes. Component 2 appears to represent the grandiosehyperactive part of the manic syndrome. This dimension resembles the behavioral facilitation (or behavioral engagement) system that has been hypothesized to represent the core neurobehavioral disturbance in bipolar disorder (Depue and Iacono, 1988). None of the correlations of the component 2 scores with the abovementioned impairment ratings attained statistical significance. Comorbidity
The rates of other lifetime disorders in the bipolar and core positive subjects, and in subjects with no lifetime history of distinct periods of elevated, expansive, or irritable mood (core negatives), are presented in Table 5. Both the bipolar and core positive groups exhibited significantly higher rates of comorbid anxiety disorders (especially separation anxiety and panic disorder) and disruptive behavior disorders (especiallyattention-deficit hyperactivity disorder) than the core negative group. In addition, the core positive subjects exhibited significantly higher rates of comorbid overanxious, obsessive-compulsive, conduct, and oppositional defiant disorder, substance use disorders (including both alcohol and drug abuse/dependence), and eating disorders than the core negative subjects. All of the bipolar subjects (except the cyclothymic subjects), by
459
LEWINSOHN ET AL.
TABLE 5 Lifetime Prevalence Rates for DSM-III-R Disorders by Group Group Core Negative (n = 1,594)
Core Positive (n = 97)
(n
Disorder
No.
%
No.
%
No.
Anxiety Phobias Separation anxiety Overanxious Panic Obsessive-compulsive Disruptive behavior ADHD Conduct Oppositional defiant Substance use Alcohol Drug Eating Depression MDD Dysthymia
122 60 50 16 14 6 110 43 48 40 165 94 121 16 357 336 43
7.r 3.8 3.1 " 1.0" 0.9" 0.4" 6.9" 2.7" 3.0" 2.5" 10.4" 5.9" 7.6" 1.0" 22.4 2U 2.7
31 4 18 8 3 3 18 8 8 7 23 12 18 5 51 47 12
32.0 b 4.1 18.6 b 8.2 b 3.1 b 3.1 b 18.6 b 8.2 b 8.2 b 7.2 b 23.7 b 12.4 b 18.6 b 5.2 b 52.6 48.5 12.4
6 2 4 0 2 0 4 2 1 1 4 2 3 0
%
33.31 IU 22.2 b 0.0,1 lUI O.Oal 22.2 1 lUI 5.6,1 5.6'1 22.2,1 11.1 al 16.7,1 0.0,1 NA NA NA
Note: Percents with different subscripts differ significantly. BP = bipolar disorder; ADHD disorder; MDD = major depressive disorder; NA = not applicable. * p < .05; ** P < .01; *** P < .001.
definition, had major depression. However, the core positive subjects also exhibited high rates of major depression (49%) and dysthymia (12%). For the nine bipolar subjects with a history of nonaffective disorders, the nonaffective condition preceded the onset of the first affective episode in seven (78%) cases. The level of comorbidity in the core positive and bipolar subjects was very comparable, with no significant differences. However, due to the larger N, a greater number of comparisons involving the core positive group were statistically significant. We also examined whether there were any interactions with sex. There was only one significant interaction: disruptive behavior disorders X group X sex, X2 (2, N = 1,709) = 12.00, P = .0025. Four of the 12 female bipolar subjects had comorbid disruptive behavior disorders, as opposed to none of the male bipolar subjects. The opposite pattern was found for the core positive subjects. Of 42 male core positive subjects, 15 had disruptive behavior disorders, compared with only 3 of 55 female core positive subjects. The high rate of comorbidity raises the question of whether the high levels of functional impairment, suicide attempts, and treatment utilization in the bipolar
460
BD = 18)
=
X2 74.46*** 2.61 68.50*** 34.91 *** 20.62*** 12.98** 22.79*** 13.26** 8.05* 7.92* 18.57*** 7.18* 16.10*** 13.21 ** 45.50*** 39.11 *** 27.19***
attention-deficit hyperactivity
subjects described above are due to bipolar disorder per se, or to coexisting nonaffective disorders. To address this issue, we compared the bipolar subjects with (n = 9) and without (n = 9) lifetime nonaffective disorders on treatment, suicide attempts, and current and past year GAF. None of these differences were significant, although the comorbid bipolar subjects exhibited greater impairment on each of these variables (66.7% of the comorbid versus 44.4% of the noncomorbid cases had been treated; 44.4% of the comorbid versus 11.1 % of the non-comorbid cases had attempted suicide; the current GAF for the comorbid subjects was 73.6 versus 78.8 for the non-comorbid cases; and the past year GAF for the comorbid subjects was 72.9 versus 76.6 for the non-comorbid cases). We also divided the core positive subjects into those with (n = 74) and without (n = 23) a lifetime history of any affective or nonaffective disorder and compared them on the four indices of impairment. A significantly greater proportion of core positive subjects with (50%) than without (4.3%) diagnosable psychopathology had received treatment, 2 (1, N = 97) = 15.35, P < .OOL However, there were no differences between the two subgroups on attempted suicide (16.2% of those with
x
]. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:4, APRIL 1995
BIPOLAR DISORDERS IN ADOLESCENTS
versus 8.7% of those without diagnoses) or on the current and past year GAF (current GAF: mean = 81.2, SD = 10.7 with versus mean = 85.3, SD = 8.1 without diagnoses; past year GAF: mean = 82.0, SD = 9.3 with versus mean = 86.6, SD = 6.5 without diagnoses). Contrasts were also made between the 23 core positive subjects without a history of mental illness and the NMI group. Those core positive subjects without a history of mental illness had a significantly greater proportion who attempted suicide than the NMI group (8.7% versus 1.2%, X2 [1, N = 868J = 9.27, P < .01). However, no significant differences were found on the GAF measures between these two groups. Course The mean age of onset of the first affective episode for the 18 bipolar cases was 11.75 (SD = 2.96). This was significantly earlier than the mean age of onset for the 316 adolescents with a history of major depression but no periods of elated or irritable mood in the OADP sample (mean = 14.95; SD = 2.75), t(332) = 4.79, P < .001. Mean onset age for females (11.74; SD = 3.5) did not differ from that of males (11.75; SD = 1.62). Of the 18 adolescents with bipolar disorder, the condition first started with a manic or hypomanic episode for 1 (5.5%), a major or minor depressive episode for 11 (61.1 %), and could not be determined for 6 (33.3%). The polarity of the onset episode was particularly difficult to determine for the cyclothymic subjects; it was indeterminate for 4 of the 5 subjects. The duration of illness, calculated as age at recovery (or current age if still in the midst of an episode) minus age at onset, for the 18 bipolar adolescents ranged from 0.7 to 125.0 months, with a median of 48.3 months and a mean (SD) of 50.4 (38.0) months. Total time in episode, calculated as the sum of the duration of each episode, ranged from 0.7 to 96.0 months, with a median of 28.0 months and a mean (SD) of 36.3 (28.9) months. The duration of the most recent episode ranged from 0.2 to 96.0 months, with a median of 10.8 months, and a mean (SD) of 20.0 (24.5) months. The cyclothymic subjects did not differ from the bipolar I and II subjects on these variables. The mean illness duration was not significantly different for females versus males. At the time of the index assessment, 7 subjects had recovered and 11 were still in the midst of an episode.
]. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:4, APRIL 1995
It is not surprising that the currently ill group exhibited a more chronic course than the recovered subjects. For example, the median duration ofillness for the currently ill group was 69.0 months, compared with 13.2 months for the recovered group. Similarly, the median total duration of an episode was 39.7 months for the currently ill group, compared with 13.1 months for the recovered subjects. Compared with the 316 core negative major depressive subjects in the OADP sample, the bipolar subjects exhibited a poorer course. Using the KaplanMeier product-limit method, the mean estimated duration of illness for the bipolar subjects was 80.2 months (SD = 13.8), compared with a mean duration of 15.7 months (SD = 1.6) for the major depressive subjects, Mantel-Cox statistic = 27.23, df= 1, P < .001. Similarly, the mean estimated total episode duration was 61.6 months (SD = 10.4) for the bipolar subjects, compared with 7.5 months (SD = 1.0) for the major depressive subjects, Mantel-Cox statistic = 36.03, df= 1, P < .001.
DISCUSSION
To our knowledge, this is the first report of a large-scale epidemiological study of bipolar disorder in adolescents. The lifetime prevalence of bipolar disorders in our sample was approximately 1%, which was similar to that reported in the Epidemiological Catchment Area study (Weissman et al., 1988) and other modern epidemiological studies of adult samples (Smith and Weissman, 1992). In general, the severity of the 18 bipolar cases detected in our sample was relatively mild. Only two subjects met full criteria for bipolar I disorder, only one had experienced psychotic symptoms, and none had been hospitalized. While more than half of the bipolar subjects had received some form of mental health treatment, only one of these subjects had been treated with lithium. This suggests that many of them were not recognized as having a bipolar disorder by the mental health professionals with whom they had come into contact. In addition, the fact that only 11 % of our bipolar cases had experienced a full-blown manic episode raises the possibility that many of them will have more severe episodes in the future. Hence, it will be important to follow these subjects into adulthood to observe the evolution and course of the disorder.
461
LEWINSOHN ET AL.
Despite the fact that few of our subjects met criteria for bipolar I disorder, these adolescents exhibited significant impairment. Compared to the NMI group, a significantly greater proportion of bipolar subjects had attempted suicide and received mental health treatment, and they exhibited significantly poorer global functioning both currently and during the past year. During their most recent episode, the majority of bipolar subjects reported impaired functioning in social situations, with family, and at school. In addition, adolescent bipolar disorder was associated with a high degree of comorbidity. In particular, the bipolar subjects exhibited significantly elevated rates of comorbid anxiety (especially separation anxiety and panic) disorders and disruptive behavior (especially attention-deficit hyperactivity) disorders. Moreover, the bipolar subjects were at least as impaired as the major depressive comparison group on every variable examined. Indeed, a greater proportion of bipolar than major depressive subjects had attempted suicide, and the bipolar subjects exhibited significantly greater impairment than the major depressive subjects on the GAP both at the second assessment and during the previous year. Finally, the bipolar subjects exhibited a relatively chronic course. It was particularly noteworthy that the median duration of illness in this group was more than 4 years and that these adolescents had already spent a median total of 28 months in an affective episode. Thus, these data are consistent with previous studies (Akiskal et aI., 1985; Akiskal and Mallya, 1987; Depue et aI., 1981; Klein et aI., 1986) indicating that even relatively mild forms of bipolar disorder in adolescents are serious conditions that are associated with substantial impairment and comorbidity. The most surprising aspect of this study was that the core positive group also exhibited high levels of impairment and comorbidity that were often comparable with that of the bipolar and major depressive subjects. While in many of these cases, the impairment may have been due to coexisting disorders, even the core positive subjects without any form of diagnosable psychopathology had a significantly elevated rate of suicide attempts compared with the NMI group. Thus, it appears that even if they lack a sufficient number of symptoms for a bipolar diagnosis, adolescents who report experiencing distinct periods ofelated, expansive, or irritable mood are a clinically significant group. These data are consisted with the findings of Carlson
462
and Kashani (1988), who also reported high rates of comorbidity and impairment (as indexed by interviewers' judgments of need for treatment) in adolescents with manic symptoms, most of whom did not qualify for a bipolar diagnosis. Taken together, these data highlight the importance of probing for these symptoms even when the initial presentation does not suggest bipolar disorder and of taking such "subthreshold" cases seriously. It will be particularly important to follow these subjects into adulthood to ascertain their risk for developing diagnosable bipolar conditions and other forms of psychopathology. Comparisons between the bipolar and major depressive subjects were generally consistent with the unipolar-bipolar differences reported in the adult literature (Perris, 1992). Thus, there was a significantly greater proportion of females among the major depressive than bipolar subjects. In addition, the bipolar subjects reported a significantly earlier age of onset, a greater proportion had attempted suicide, and they manifested a much more severe course of illness. The fact that the relative prevalence rates of manic symptoms shown in our community sample are consistent with clinical populations of adolescents is important. It suggests that we were detecting milder forms of the more serious cases seen in clinical practice. In addition, the factor analysis yielded interesting, albeit preliminary, findings. These data suggested that bipolar disorder, at least in an adolescent community sample, can be described in terms of two dimensions, one reflecting behavioral disorganization and impairment and the other reflecting behavioral facilitation or engagement (Andrews et aI., 1993). Finally, there were few gender differences in our sample. Although our N was small, the prevalence, age of onset, phenomenology, and course of bipolar disorder in adolescents appeared to be similar for both males and females. In conclusion, the prevalence of bipolar disorder in a large community sample of adolescents was very similar to that in recent epidemiological studies of adult samples. Although most of our bipolar cases only met criteria for bipolar II disorder and cyclothymia, they exhibited considerable impairment; high rates of attempted suicide, comorbidity, and mental health care utilization; and a relatively chronic course. In addition, a larger group of subjects was identified who reported distinct periods of elevated, expansive, or irritable mood
J.
AM. ACAD. CHILD ADOLESC. PSYCHIATRY. 34:4. APRIL 1995
BIPOLAR DISORDERS IN ADOLESCENTS
but did not meet criteria for any form of bipolar disorder. These subjects also exhibited considerable impairment. These data highlight the clinical significance of even the milder and subthreshold forms of bipolar disorder in adolescence.
REFERENCES Akiskal HS, Downs ], Jordan P, Watson S, Daugherty 0, Pruitt DB (1985), Affective disorders in referred children and younger siblings of manic-depressives: mode of onset and prospective course. Arch Gen
Psychiatry 42:996-1003 Akiskal HS, Mallya G (1987), Criteria for the "soft" bipolar spectrum: treatment implications. Psychopharmacol Bull 23:68-73 American Psychiatric Association (1987), Diagnostic and Statistical Manual ofMental Disorders, 3rd edition (DSM-III). Washington, DC: American Psychiatric Association Andrews ]A, Hops H, Aty 0, Tildesley E, Harris] (1993), Parental influence on early adolescent substance use: specific and nonspecific effects. J Early Adolesc 13:285-310 Anthony], Scott P (1960), Manic depressive psychosis in childhood. J Child Psychol Psychiatry 1:52-72 Ballenger ]e, Reus VI, Post RM (1982), The atypical clinical picture of adolescent mania. Am J Psychiatry 139:602-606 Bowring MA, Kovacs M (1992), Difficulties in diagnosing manic disorders among children and adolescents. J Am Acad Child Adolesc Psychiatry 31:611-614 Carlson GA (1990), Annotation: child and adolescent mania-diagnostic considerations. J Child Psychol Psychiatry 31 :331-341 Carlson GA, Kashani ]H (1988), Manic symptoms in a non-referred adolescent population. J Affect Disord 15:219-226 Carlson GA, Strober M (1979), Affective disorders in adolescence. Psychiatr
Clin North Am 2:511-526 Cohen ]A (1960), A coefficient of agreement for nominal scales. Educ
Psychol Meas 20:37-46 Depue RA, Iacono WG (1988), Neurobehavioral aspects of affective disorders. Annu Rev PsychoI40:457-492 Depue RA, Slater ]F, Wolfsetter-Kausch H, Klein 0, Goplerud E, Farr o (1981), A behavioral paradigm for identifYing persons at risk for bipolar depressive disorder: a conceptual framework and five validation studies. J Abnorm PsychoI90:381-437 Edelbrock C, Costello A], Dulcan MK, Kalas R, Conover NC (1985), Age differences in the reliability of the psychiatric interview of the child. Child Dev 56:265-275 Egeland ]A, Hostetter AM (1983), Amish study: affective disorder among the Amish, 1976-1980. Am J Psychiatry 140:56-61 Gammon GO, John K, Rothblum ED, Mullen K, Tischler GL, Weissman MM (1983), Use of a structured diagnostic interview to identifY bipolar disorder in adolescent inpatients: frequency and manifestations of the disorder. Am J Psychiatry 140:543-547 Garvey MJ, Tuason VB (1980), Mania misdiagnosed as schizophrenia. J
Clin Psychiatry 41:75-78
J.
AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 34:4, APRIL 1995
Hall RCW, Gardner ER, Stickney SK (1980), Physical illness manifesting as psychiatric disease: II. Analysis of a state hospital inpatient population.
Arch Gen Psychiatry 37:989-995 Hsu LKG, Starzynski JM (1986), Mania in adolescence. J Clin Psychia-
try 47:596-599 Joyce BR (1984), Age of onset in bipolar affective disorder and misdiagnosis as schizophrenia. Psychol Med 14:145-149 Keller MB, Lavori PW, Friedman B, Nielsen E, Endicott J, McDonaldScott PA (1987), The Longitudinal Interval Follow-up Evaluation: a comprehensive method for assessing outcome in prospective longitudinal studies. Arch Gen Psychiatry 44:540-548 Klein ON, Depue RA, Krauss SP (1986), Social adjustment in the offspring of parents with bipolar affective disorder. J Psychopathol Behav Assess 8:355-366 Kraepelin E (1921), Manic-Depressive Insanity and Paranoia. Edinburgh: E & S Livingstone. Reprinted New York: Arno Press, 1976 Krasa NR, Tolbert HA (1994), Adolescent bipolar disorder: a nine-year experience. J Affect Disord 30:175-184 Lewinsohn PM, Hops H, Roberts RE, Seeley JR, Andrews JA (1993), Adolescent psychopathology: I. Prevalence and incidence of depression and other DSM-IIJ-R disorders in high school students. J Abnorm
Psychol102:133-144 Loranger A, Levine P (1978), Age at onset of bipolar affective illness. Arch
Gen Psychiatry 35:1345-1348 McGee R, Feehan M, Williams S, Partridge F, Silva PA, Kelly J (1990), DSM-IJI disorders in a large sample of adolescents. JAm Acad Child
Adolesc Psychiatry 29:611-619 Mukherjee S, Shukla S, Woodle J, Rosen AM, Olarte S (1983), Misdiagnosis of schizophrenia in bipolar patients: a multiethnic comparison. Am J
Psychiatry 140:1571-1574 Orvaschel H, Puig-AntichJ, Chambers WJ, Tabrizi MA, Johnson R (1982), Retrospective assessment of prepubertal major depression with the Kiddie-SADS-E. JAm Acad Child Psychiatry 21:392-397 Perris C (1992), Bipolar-unipolar distinction. In: Handbook of Affective Disorders, Paykel ES, ed. New York: Guilford Press, pp 57-75 Smith AL, Weissman MM (1992), Epidemiology. In: Handbook ofAffective Disorders, Paykel ES, ed. New York: Guilford Press, pp 111-129 Strober M, Carlson GA (1982), Bipolar illness in adolescents with major depression: clinical, genetic, and psychopharmacologic predictors in a three- to four-year prospective follow-up investigation. Arch Gen
Psychiatry 39:549-555 Strober M, Lampert C, Schmidt S, Morrell W (1993), The course of major depressive disorder in adolescents: I. Recovery and risk of manic switching in a follow-up of psychotic and nonpsychotic subtypes. J
Am Acad Child Adolesc Psychiatry 32:34-42 Wehr TA, Goodwin FK (1987), Can antidepressants cause mania and worsen the course of affective illness? Am J Psychiatry 144: 1403-1418 Weinberg WA, Brumback RA (1976), Mania in childhood: case studies and literature review. Am J Dis Child 130:380-385 Weissman MM, Leaf Pi, Tischler GL et al. (1988), Affective disorders in five US communities. Psychol Med 18:141-153 Whitaker A, Johnson J, Shaffer 0 et al. (1990), Uncommon troubles in young people: prevalence estimates of selected psychiatric disorders in a nOnteferred adolescent population. Arch Gen Psychiatry 47:487-496 White JH, O'Shanick G (1977), Juvenile manic-depressive illness. Am J
Psychiatry 134:1035-1036
463