Complex and Rapid Cycling in Bipolar Children and Adolescents:
A Preliminary Study

Barbara Geller, M.D.
Kai Sun, Ph.D.
Betsy Zimerman, M.A.
Joan Luby, M.D.
Jeanne Frazier, B.S.N.
Marlene Williams, R.N.

Washington University School of Medicine
St. Louis, MO, USA

Acknowledgement: This work was supported in part by National Institute on Drug Abuse grant R01 DA04844 to Dr. Geller. The data presented in this communication were the pilot data for National Institute of Mental Health grant R01 MH 53063 "Phenomenology and Course of Pediatric Bipolar Disorders" to Dr. Geller.

The text reproduced here was provided by Dr. Geller. The article is published in the Journal of Affective Disorders, 34 (1995): 259-268. Reprinted with permission from Elsevier Science.


Twenty-six subjects aged 7-18 years old were studied. Diagnoses of bipolar disorders were established using the Kiddie-Schedule for Affective Disorders and Schizophrenia-Present Episode Version-1986 modified for DSM-III-R criteria and for rating the number and duration of manic and hypomanic episodes. Complex cycling patterns were observed. These included numerous, brief episodes suggesting continuous, rapid-cycling in 80.8% of cases. Mean age of onset was early (8.5±4.4 years). Psychotic phenomena, suicidality, hyperactivity, and "mixed mania" were highly prevalent. Data in this report provide support for complex and rapid-cycling patterns in childhood onset BP.

Key Words: rapid-cycling; children; adolescents; bipolar affective disorder; hyperactivity; mixed mania


This paper presents preliminary data on the descriptive characteristics and cycling patterns of bipolar disorders I and II (BP-I and BP-II) occurring in children and adolescents.

By analogy to the evolution of knowledge on major depressive disorders (MDD) in children and adolescents (Puig-Antich, 1980a), it is useful to expect that age-specific differences in BP may emerge (Fristad et al, 1992; Geller et al, 1994; Todd et al, 1993) and that non-data based conceptions may need to be updated.

One of the prevalent conceptions has been that BP disorders are "rare" in the pediatric age group and unlikely to be seen before puberty. However, a number of lines of evidence argue against this idea. First, the only available epidemiological data (Carlson and Kashani, 1988) found a 13.3% rate of Diagnostic and Statistical Manual III (DSM-III) (APA, 1987) hypomania. Secondly, there is increasing evidence for a secular trend for earlier onset of MDD (e.g., Rice et al, 1987) and data that show a high rate of switching from MDD to prepubertal BP (Geller et al, 1994). Based on the above we may expect that BP may become a more frequent diagnosis of childhood.

The long standing question of the differentiation of attention-deficit hyperactivity disorder (ADHD) and BP (Carlson, 1990) has been approached in an important data based paper by Fristad et al (1992). In the latter paper, the BP children often had hyperactivity but were differentiated by mood and cognitive symptoms.

Across age groups, however, there remain diagnostic questions related to the criterion of duration of an episode of BP-I or BP-II (Coryell, 1982; Dunner, 1983; Endicott et al, 1985). Different diagnostic schema have used varying lengths, including the non-specification of episode duration in DSM-III-R (APA, 1987). The latter attests to the difficulty in reaching consensus on the issue of BP cycle length. This has been more problematic for BP-II in which psychotic features and inpatient care cannot be used as markers of illness duration.

During the data collection phase of a previously reported study of BP-I and BP-II occurring during follow-up of prepubertal MDD subjects (Geller et al, 1994), the research nurses noted that there was a wide variation in duration of BP symptomatology and in frequency of episodes. To further characterize these phenomena, a question was added to the assessment instrument (see below) to optimize the ability to obtain data on the frequency and duration of pediatric BP episodes. Data presented below were obtained with use of this question on two new samples of children and adolescents (i.e., other than the prepubertal MDD follow-up study group).

The clinical literature referred to rapid-cycling of juvenile BP (Puig-Antich, 1980a). It is useful to provide data on this prevalent conception because rapid-cycling in adults is associated with a more treatment resistant picture (Wehr et al, 1988). Treatment resistance has also been reported in open trials in bipolar adolescents (Hsu, 1986; Strober et al, 1988). Thus, it is important to establish the relationship of cycling patterns to prognosis and treatment of child and adolescent onset BP.

The definition of rapid cycling used by investigators of the phenomena in adult populations is ³ 4 episodes per year (Wehr et al, 1988). The word "complex" was added to this paper because the number of episodes greatly exceeded four for most subjects (see results section below) and because the duration of individual episodes was sometimes shorter than that given in various diagnostic classifications (e.g., Dunner, 1983).

This is the first report of systematic data on cycling patterns in children and adolescents.


Study Populations

Two populations were studied. One was specifically recruited to obtain pilot data for a larger study of BP children and adolescents (NIMH 1 R01 MH53063, "Phenomenology and Course of Pediatric Bipolar Disorders", PI: Barbara Geller, M.D.). These data were also used for a parallel pilot study of MRI (Botteron et al, 1995). This population will be referred to as the BP Pilot subjects. A second population were consecutive BP subjects entered into a double-blind placebo controlled study of lithium for bipolar adolescents with substance dependency disorders (Geller et al, 1992a). This group will be referred to as the Adol BP. Thus, for the pilot data purposes in this manuscript, these two populations were a convenience sample.

Inclusion and Exclusion Criteria

Inclusion criteria for the BP Pilot group were males and females; 6-16 years old; current DSM-III-R mania or BP-I 0R current DSM-III-R BP-II with a past history of BP-I (i.e., lifetime diagnosis of BP-I); Children's Global Assessment Scale Score (CGAS) £ 60 (definitely a case) (Bird et al, 1987; Shaffer et al, 1983); and duration ³ two weeks. A diagnosis of BP-II or hypomania without a history of lifetime BP-I or mania was not included in the BP Pilot to be sure of a BP-I or manic diagnosis for the parallel MRI study (Botteron et al, 1995). Thus, the diagnosis of a current BP-II disorder for this group was only given if the hypomania was current. The latter assured that only the most conservative approach was used for this preliminary study. If a past history of BP-I was accurate, all of the BP-II subjects in this group had a lifetime diagnosis of BP-I and a current diagnosis of BP-II. Thus, diagnoses of current BP-I or II were assigned on the basis of psychopathology observed by the research team. Exclusion criteria for the BP Pilot group were IQ < 75 (to assure ability to understand the assessment questions), autism, pervasive developmental disorder, schizophrenia, major medical disorder, or pregnancy.

Inclusion criteria for the Adol BP group were males and females; 12-18 years old; current DSM-III-R BP-I, BP-II, manic or hypomanic disorders; DSM-III-R substance dependency disorder; mood disorder preceded substance disorder or was present for at least two weeks without substance use; CGAS £ 60; and duration of mood disorder ³ two months. Because this was a drug treatment study, a longer duration of mood disorder was used as an inclusion criterion than for the BP Pilot group. Exclusion criteria for the Adol BP group were IQ < 75, autism, pervasive developmental disorder, schizophrenia, major medical disorder, or pregnancy.

Assessment Methods

The Kiddie-Schedule for Affective Disorders and Schizophrenia-Present State Version-1986 (K-SADS-P-1986)(Puig-Antich and Ryan, 1986) modified for DSM-III-R criteria was administered separately to mothers and subjects by research nurses with established interrater reliability. The K-SADS-P utilized a time frame of the entire episode of mood disorders. Because these subjects had chronic, unremitting illness (albeit with multiple, rapid cycles), the time frame on the KSADS-P was equivalent to lifetime diagnoses. The CGAS was completed by the rater who performed the KSADS-P-1986 interviews. Disparities between mother and subject were handled by using all reliable information based on the data of Bird et al (1992). Similar to our prior publications on MDD in children (e.g., Geller et al, 1992b), the research nurses had virtually 100% agreement on all BP diagnoses and on all BP criteria items. Details of the training process were provided in Geller et al (1992b). Training of raters to consistent inter-rater reliability in our unit was a three month, full-time process. As is customary with semi-structured interview techniques, any uncertainty about the rating of an item that occurred after training was completed was discussed with another trained rater until consensus was reached. Annual recalibration of raters was also performed. Test-retest reliability of the BP category was not performed. Reliability of data from the substance dependent subjects was enhanced (compared to similar ratings of adults with substance dependent disorders) by the following: Ratings included interviews with parents, school reports and court reports. Additionally, unlike adults patients, these adolescent subjects had always lived in the parental household. Thus, these teenage substance dependent subjects were continuously observed by their parents from birth until the time of entry into the study.

To meet diagnostic criteria for BP-I, KSADS-P-1986 items needed to be rated ³ 4 (moderate or severe). To meet BP-II, KSADS-P-1986 items needed to be rated =3 (mild).

For the Adol BP group, a diagnosis of ADHD was established using the corresponding items from the KSADS-E, the lifetime version of the KSADS (Puig-Antich et al, 1980b). Due to the preliminary nature of the BP Pilot group, the K-SADS-E was not administered to this population. Rather, hyperactivity in the BP Pilot group was ascertained by use of the ten item Conners' Parent Questionnaire (Conners, 1973). A cut-off of ³ 15 was used to rate hyperactivity as present. Although these two methods of assessing hyperactivity were different in format, both are widely accepted in child psychiatry as research measures of hyperactivity (see e.g., Biederman et al, 1991).

Number and Duration of Manic and Hypomanic Episodes Item

The following item was included in the K-SADS-P-1986 rating of mothers and subjects:

Rater should record the number of episodes lasting the specified period of time during the past year. A duration of ³ 4 hours per day was needed to count as a daily episode of < 1 day.

1. < 1 day ____
2. 1 to < 2 days ____
3. 2 to < 3 days ____
4. 3 to 6 days ____
5. 7 to 13 days ____
6. ³ 14 days ____

As noted in the introduction, the above item was developed because families gave high numbers of episodes only for the BP diagnostic categories. Thus, the above item was not added to any other diagnostic categories because a high multiple number of episodes was not spontaneously given or elicited for any other diagnostic category. For the BP time frame and for all time frames on the K-SADS-P-1986, the following time markings were used. For parents and older children, time frames were broken into days, weeks, months and years. For younger children standard frames of reference were Easter, Thanksgiving, Christmas and summer vacations.

Statistical Analyses

Categorical variables were analyzed using the continuity-adjusted chi-square statistic. Fisher's exact test was used for small cell sizes (< 5). Continuous variables were analyzed using Student's t-test. Statistical tests were two-tailed at a significance level of p < 0.05.


Twenty-six consecutive subjects were entered. Thirteen were from the BP Pilot group and 13 were from the Adol BP group. Seventeen subjects were ³ 13 years old (13 subjects from the Adol BP group and four from the BP Pilot group). Nine subjects, all from the BP Pilot group, were £ 2 years old.

Table 1 presents the descriptive and diagnostic characteristics by age at entry to the study protocol. Chronicity of mood disorders was evidenced by the early age of onset compared to the age at entrance to the study protocol. Severity was demonstrated in both groups by CGAS scores in the range of severe impairment (0 is worst, 100 is best). "Mixed mania", which refers to simultaneous mania and depression (Himmelhoch and Garfinkel, 1986), was highly prevalent. There were no significant differences between the £ 12 and ³ 13 years old groups on any of the diagnostic items in Table 1. In addition, there were no significant differences between the BP Pilot and Adol BP groups on any of these diagnostic items.

In the ³ 13 year old group one subject had mania (i.e., without MDD) and in the £ 12 year old group one subject had hypomania. For purposes of clarity, these subjects were analyzed with the BP-I and BP-II groups, respectively. Thus, BP-I and BP-II cases had similar global severity, suicidality, and comorbidities. They differed in that BP-II cases had scores of ³ 3 rather than ³ 4 on one or more criteria items. As noted above under the methods section, BP-II refers to the current diagnosis and not to the lifetime diagnosis.

Five subjects (all in the BP Pilot group) were on psychoactive medications prescribed by their referring physicians at the time of assessment. Four subjects were on lithium. All four were concurrently receiving other medications (one was also on Navane and Cogentin; one was also on Tegretol; one was also on Dexedrine and one was also on clonidine). One subject was receiving only desipramine 25 mg. at bedtime. All five subjects had BP psychopathology prior to medications and had continued current BP diagnoses even on the medication.

Table 2 shows the psychotic phenomena by age. The age distribution permits comparison to the Chambers et al (1982) data on psychotic phenomena in prepubertal MDD children (see Discussion below). Only pathological delusions and hallucinations are presented, i.e. hypnogogic and hypnopompic phenomena were not included and simply hearing one's name called was not included. For the Adol BP group, only non-substance induced phenomena were included. There were no significant differences in the prevalence of psychotic phenomena between the two age groups.

Tables 3 and 4 show data obtained from the KSADS-P-1986 item "Number and Duration of Manic and Hypomanic Episodes" described in the Method section. Cases were divided by BP-I and BP-II to demonstrate the presence of complex cycling patterns in both BP-I and BP-II categories. These phenomena were reported by both mothers and subjects.

To further elucidate the contents of Tables 3 and 4, Case 6 of the BP-II series can be used. This subject reported 104 episodes that lasted ³ 4 hours on 104 separate days; 52 episodes that lasted < 2 days; 52 episodes that lasted < 3 days; 40 episodes of 3-6 days duration; and 5 episodes of 7-13 days in length. Thus, there were 253 episodes occurring on 491 days (number of episodes times number of days/episode).

Only five of the 17 subjects ³ 13 years old and none of the 9 subjects £ 12 years old had a single episode that lasted ³ two weeks as their only episodes. Eleven of the 15 manic subjects had episodes that lasted at least one week in addition to episodes that were more frequent and of shorter duration. Eight of the 11 hypomanic subjects had episodes that lasted at least one week in addition to episodes of shorter duration and high frequency. Therefore, 80.8% of the total 26 subjects had rapid cycling patterns (100% of the younger and 70.6% of the older age groups).

Case Vignettes of a 24 Hour Period

Case 1: A nine year old boy was sent to the principal's office 14 times during one school day for disobeying his teacher. The patient informed the principal that the school authorities could not harm him because he was "Bill Clinton's son". The child's high energy level at school continued at home, where he rearranged the furniture in his room until midnight. At 2:00 a.m., the patient began calling multiple 900 numbers, which caused a very high telephone bill.

Case 2: A 16 year old female was noted to be irritable during the school day. At 4:00 p.m. she arrived home. At home, the patient talked "non-stop" about going to California to become a model. Of note, this patient was 50 lbs. overweight. It was noted that she would "jump from subject to subject" and that "no one could get a word in edgewise". At 11:30 p.m., she informed her family that a male teacher was sexually attracted to her and wanted to have an affair with her.

The above examples were typical of the cases. They were selected to be representative and not extreme.


Descriptive and Diagnostic Characteristics

Clearly, this work is preliminary. The investigators, nevertheless, believed that it was important to share with the field because of the clinical urgency regarding recognition of child and adolescent bipolar patients.

The chronicity and severity of these subjects argues against transient, developmental "highs" or "adolescent turmoil" (Rutter et al, 1976) as the reason for the clinical BP picture.

The low percent of females in the £ 12 year old group compared to the equal gender representation in the ³ 13 years old group and in adult mania (Tohen et al, 1990a,b), may be the result of the small sample size, ascertainment bias (i.e. maybe mothers bring sons to treatment for BP symptoms earlier than daughters), or developmental (i.e., early onset BP may be more common in males).

Subjects on medication at the time of assessment still had major, severe psychopathology and were often on polypharmacy. The latter is consistent with the relative lithium resistance of adolescent BP subjects with prepubertal psychopathology (Hsu, 1986; Strober et al, 1988).

"Mixed mania" (coexistent mania and depression) was reported to have a 30% prevalence in adults (Himmelhoch and Garfinkel, 1986) compared to the much higher 57.7% rate in this report. Because "mixed mania" has been associated with worse outcome in adults with mania (Tohen et al, 1990a,b), the high prevalence in children and adolescents is consistent with the above cited treatment resistance.

High prevalence of suicidality (46.2%) is consistent with the 35 times greater risk of suicide among adult BP patients compared to adult depressed patients (Bipolar Disorder, NIH Guide, 1993).

There is a question of whether the import of a BP-II diagnosis is different at younger ages (Geller et al, 1994). It is possible that BP-II in children is the earliest manifestation of a future BP-I course. The latter would differ from the stability of the BP-II diagnosis reported for adults (Coryell et al, 1989).

The high rate of dysthymia may reflect the ease of meeting DSM-III-R criteria for this disorder.

For the Adol BP group, the substance dependency is a potential confound to the data. However, even in this preliminary study there were no significant differences between the descriptive and psychotic characteristics of the BP Pilot and Adol BP groups. All of the Adol BP group had mood disorders independent of the substance dependency by time course, i.e. either the mood disorder came first or the mood symptomatology existed when no substance abuse or use was occurring.

Psychotic Phenomena (Hallucinations and Delusions)

Although the prevalence of hallucinations in the £ 12 year olds shown in Table 2 was similar to the 23.1% reported for prepubertal subjects with MDD (Chambers et al, 1982), the MDD group had benign types, e.g. hearing their name called. By contrast, the BP subjects sustained pathological hallucinations including command, conversing, religious, and thoughts heard aloud.

Delusions for the £ 12 year old BP group presented in Table 2 were substantially more prevalent than the < 5% noted by Chambers et al (1982) for prepubertal children with MDD. Of note, for the BP groups £ 12 and ³ 13 years old, delusions were grandiose, persecutory, somatic and others in addition to ideas of reference. The latter is important because ideas of reference have been described in a community sample of adolescents (Rutter et al, 1976). Thus, the BP subjects had multiple types of delusions that have not been reported in community based samples.

The marked perceptual distortions seen in these BP subjects are consistent with the multiple reports in the literature that describe misdiagnosis of adolescents with BP as schizophrenic (Horowitz 1975, 1977).

The prognostic import of psychosis in young BP needs consideration because first episode adult patients with mania had a worse outcome if they were psychotic (Tohen et al, 1990a,b).

Cycling Patterns in Child and Adolescent Bipolarity

Although the large number of cycles reported seems to lack "face validity", these histories were given independently (i.e., families did not know what other families had answered). Thus, it would seem equally implausible to ignore these parental and subject reports. Why else would so many independently assessed families answer the item of "how many episodes" with large numbers of episodes for the BP category when only one or a few episodes were given for any other disorder assessed by the K-SADS-P-1986?

These data suggest that the subjects were continuously cycling from mania (or hypomania) to euthymia or depression.

If the cycling patterns represent rapid cycling phenomena, this would be consistent with clinical reports in the literature of rapid cycling among children and adolescents (Puig-Antich, 1980a). Because of the poor prognosis and treatment resistance of adults with rapid-cycling (Wehr et al, 1988), this is an important feature to fully characterize.

Independent corroboration of the cycling pattern data in this report was provided by findings recently presented by Kutcher (1993) from an adolescent inpatient unit in which subjects were monitored on a daily basis for changes in mood states. Kutcher (1993) reported cycling pattern data that is similar to our findings, i.e. multiple brief episodes occurring continuously in many subjects.

Clearly, the almost continuous nature of the episodic phenomena described by the mothers and subjects presents a further challenge to the distinction between ADHD and BP. The latter is especially true for the £ 12 year old group. Various authors (e.g. Carlson, 1990) have suggested that discrete episodes could be an aid in differential diagnosis between these two disorders. This suggestion will most likely continue to be true for some BP pediatric aged subjects, but not for others.

Future Validation Studies

Validation of pediatric BP by symptomatology (e.g. psychotic phenomena); family history; and longitudinal course (i.e., stability of BP phenomena over time and continuity with adult BP) is underway in our five year study (Phenomenology and Course of Pediatric Bipolar Disorders, NIMH R01 MH53063, P.I. Barbara Geller, M.D.).

Answers to questions about precise timing of episodes and interepisode states (e.g., depressed, euthymic) will require prospective, controlled studies. To accomplish the latter objective, in our NIMH funded study daily logs will be obtained from parents and subjects by interview. Thereafter, these daily logs will be compared to recall during every six month follow-up assessments. The latter methodology will permit quantification and detailed characterization of cycling patterns in children and adolescents. Further, our study will allow investigation of the prognostic significance of rapid cycling and of the relationship between naturalistically prescribed medications and cycling phenomena. The rate of hyperactivity is consistent with the report of Fristad et al (1992) on the high prevalence of ADHD among bipolar patients. Because the prevalence of ADHD was lower in the adolescents than in the younger age group (see Table 1), it is possible that ADHD is an age-specific manifestation of BP. In our ongoing NIMH study of pediatric bipolar disorders, the control groups are ADHD and community controls. Follow-up of the BP, ADHD and community control groups will provide data on age-specificity of comorbid BP and ADHD disorders.

Heterotypy, i.e. the same illness appearing differently at specific ages, is also possible. Thus, these subjects later in life may present with the spectrum of mood labile personality disorders described in adults (e.g. Akiskal et al, 1995).

Clinical Implications

It will be important for practitioners to give serious attention to improbable sounding histories of complex cycling patterns in children and adolescents with symptoms of BP-I and BP-II. How many days per year of BP phenomena will qualify as a BP diagnosis will have to be decided at the present time by individual clinical judgement on a case by case basis. Thus, a strong family history of BP, psychotic phenomena and 180 episodes per year lasting ³  4 hours per episode would likely be considered as BP by most clinicians. But suppose the cycling pattern prevails for only 90 non-contiguous days per year and the clinical picture does not include psychosis. Decisions about the latter clinical situation will be better informed when further systematic investigations of child and adolescent BP become available.

Until data become available, practitioners may not be able to differentiate ADHD from comorbid BP with ADHD on a cross-sectional basis. Families can be educated about diagnostic possibilities in order to better aid the clinician with longitudinal evaluation. Education of families about the possibility of increasing BP manifestations and consideration of the potential effects of tricyclic antidepressant (TCA) treatment (Geller et al, 1993) on BP course are warranted. The latter is useful because TCAs are widely prescribed for ADHD.

The reference list and four tables exist as separate pages in order to reduce the amount of time this document takes to load into many web browsers. The following are links to these pages:

Return to Top

For suggestions and corrections, please contact WebMaster