Research on pediatric bipolar disorder (PBD) has grown over the past decade (Kowatch et al. 2005; Youngstrom et al. 2008). However, the majority of this research has focused on bipolar I and II, leaving the other subtypes – bipolar disorder not otherwise specified (BP NOS) and cyclothymic disorder – less well understood (Youngstrom 2009). Most studies have not differentiated between these subtypes, choosing to combine all presentations that do not meet criteria for bipolar I or II into a broad “NOS” grouping whose definition shifts from paper to paper(Kessler et al. 2001; Youngstrom et al. 2008). Unfortunately, this approach has limited our knowledge of the phenomenology, risk factors, course, and outcomes specific to cyclothymic disorder and BP NOS.

Research suggests that, across the lifespan, cyclothymic disorder and bipolar disorder NOS are often as debilitating as bipolar I (BP I) and bipolar II (BP II) (Axelson et al. 2006; Birmaher et al. 2009; Birmaher et al. 2006; Findling et al. 2005; Judd and Akiskal 2003; Kessler et al. 2001), and that these presentations are more common among both young people and adults than BP I or II (Akiskal et al. 2006; Hantouche 2009; Lewinsohn et al. 1995; Merikangas et al. 2007; Van Meter et al. 2011a). Additionally, presentations of bipolar disorder that have not escalated to mania or major depression may be amenable to preventive interventions that other subtypes are not (Berk et al. 2007; Chang 2008; Klein et al. 1986; Miklowitz and Chang 2008). The term “subthreshold” is often used to refer to bipolar disorders that do not meet criteria for BP I or II. However, this term may be a misnomer, given the symptomatology and level of impairment associated with cyclothymic disorder and BP NOS. In an effort to avoid misrepresenting the severity of these disorders, we prefer to use the acronym “NoMaD,” indicating No Mania or Depression, to refer broadly to bipolar spectrum disorders other than bipolar I and bipolar II.

Cyclothymic disorder is particularly under-investigated. It is rarely included in research studies or diagnosed clinically, particularly among youth Youngstrom et al. (2005). As a result, people who have cyclothymic disorder are likely to be misdiagnosed or not diagnosed at all, which may result in insufficient or inappropriate treatment for up to a third of adults with bipolar spectrum disorder Angst et al. (1997). One factor that undoubtedly contributes to the limited attention paid to cyclothymic disorder is the relative ambiguity of the diagnostic criteria for the disorder; bipolar I and II have clearly delineated A and B criterion, whereas the criteria for cyclothymic disorder are vague, and refer primarily to duration, and to the fact that the patient may not have met criteria for mania or major depression. The criteria for bipolar NOS, similarly, are based on the exclusion of symptoms meeting criteria for mania and depression. Capturing the heterogeneous symptom profiles of bipolar spectrum disorders may require a flexible set of criteria, and no changes have been proposed for the definition of cyclothymic disorder in DSM-5. Learning more about the unique traits of cyclothymic disorder may add precision to this diagnosis in the future.

In an effort to better understand cyclothymic disorder, and to determine whether or not it is diagnostically distinct from BP NOS, we previously conducted a study to examine the validity of cyclothymic disorder in a youth sample Van Meter et al. (2011b). The study focused on five constructs, identified through a literature review, as potential discriminating clinical features of pediatric cyclothymic disorder: Irritability, Comorbidity, Age of Onset, Sleep Disturbance, and Family History (Akiskal et al. 1977; Birmaher et al. 2009; Birmaher et al. 2006; Depue et al. 1981; Findling et al. 2005; Howland and Thase 1993; Klein et al. 1986). This preliminary study found that cyclothymic disorder differed significantly from the non-bipolar spectrum illnesses on these constructs, and it shared many characteristics with the other bipolar subtypes, placing it squarely on the bipolar spectrum.

Youth with cyclothymic disorder had irritability scores much higher than the rest of the sample. Interestingly, depressive irritability was rated as more severe than irritability during elevated mood states in children with cyclothymic disorder, which is perhaps surprising given that other youth studies have focused on the extreme irritability associated with hypomania and manic symptoms (Biederman et al. 2000; Kowatch et al. 2005; Wozniak et al. 1995). For the present study, we hypothesized that youth with cyclothymic disorder would report elevated irritability during both mood phases, and that depressive irritability would be more severe.

Previously, we found that 98 % of youth with cyclothymic disorder had at least one comorbid diagnosis, and that they had significantly more comorbid diagnoses than the other youth in the sample, on average. Additionally, 83 % had comorbid ADHD, higher than the rate found in most other studies of pediatric bipolar spectrum disorders (Axelson et al. 2006; Findling et al. 2001; Kowatch et al. 2005), and higher than in the other bipolar subtypes in that sample. For the present study, we hypothesized that most youth with cyclothymic disorder would have comorbid diagnoses, and that they would have more comorbid ADHD than the other bipolar subtypes.

Because cyclothymic disorder is, by definition, persistent after onset, and because research has found that NoMaD bipolar disorders may have an earlier onset than BP I and II Lewinsohn et al. (2000), our previous hypothesis was that cyclothymic disorder would have an earlier age of onset than other forms of bipolar disorder. We found that most youth with cyclothymic disorder (73 %) had symptom onset before age 10, and that they had an earlier age of onset than the youth with non-bipolar disorders. All of the bipolar subtypes, in our pediatric sample, tended to have onset before the age of ten, but those with cyclothymic disorder had a significantly younger age of onset than youth with bipolar II. This is consistent with bipolar II having depression as the most pronounced clinical feature, and depression doubling or tripling in incidence during early adolescence Cyranowski et al. (2000).We predict a similar pattern of early onset in the present study.

Sleep disturbance is gaining attention as both a marker of bipolarity, and also a probable mechanism contributing to emotional dysregulation (Harvey 2009). Surprisingly, in our previous study, we found that youth with bipolar II reported more sleep disturbance than youth with other bipolar subtypes. This contradicts the theory that sleep disturbance is related primarily to mania, but is consistent with broader evidence about the importance of sleep (Harvey 2009). In the present study, we hypothesized that youth with cyclothymic disorder would have more severe sleep disturbance than the youth with non-bipolar disorders. As a secondary aim, we also examined whether there were differences between bipolar subtypes on sleep measures.

There is some evidence that, among youth with bipolar disorders, those with cyclothymic disorder are most likely to have a parent with a bipolar spectrum disorder (Findling, et al. 2005). This may be particularly true when comparing cyclothymic disorder to other NoMaD bipolar disorders (Akiskal, et al. 1977; Depue et al. 1981). Previously, we found that youth with cyclothymic disorder had more family history of psychiatric illness and bipolar disorder than the other youth in the sample as a whole. When compared to the other bipolar subtypes, there were no statistically-significant differences in overall psychiatric illness history or bipolar history; however, the analyses had insufficient power to detect small effects that might still be conceptually meaningful Faul et al. (2009). For the present study, we predicted similar findings, but anticipated that youth with cyclothymic disorder may have more family history of psychiatric disorders than the other bipolar subtypes.

Familial history of psychiatric illness, specifically bipolar spectrum disorders, may be an important factor in the phenomenology and course of early onset bipolar disorders (Chang and Steiner 2003; Duffy et al. 2007; Findling et al. 2005). Though research has indicated that familial bipolarity is a risk factor, the effect of having a parent with a bipolar illness – versus not – on a youth suffering from a bipolar illness has not been explored thoroughly. Considering the high rates of familial bipolar disorders found in youth with cyclothymic disorder, it may be that this is an important factor in differentiating youth with NoMaD bipolar disorders. Therefore, we set out to explore whether or not youth with diagnoses of cyclothymic disorder or BP NOS who had a parent with bipolar disorder differed from those with parents who did not have bipolar disorder on the five constructs outlined above. We hypothesized that youth with a family history of bipolar spectrum disorders would exhibit higher levels of irritability, an earlier age of symptom onset, and more sleep disturbance than those youth without a family history of bipolar disorder.

The primary objective of this study was to see if results from our previous study on the validity cyclothymic disorder in youth will remain constant in a new sample. The current sample was recruited from January 1997 through January 2003, and does not overlap with the sample from the previous validation study, which was recruited from July 2003 to March 2008. Based on our prior findings, we assessed differences in irritability, comorbid diagnoses, age of onset, sleep disturbance, and family history across diagnostic categories. Youth with cyclothymic disorder or BP NOS were also compared based on parental history of bipolar spectrum disorders.

Methods

Participants

Participants, aged five to 17, were recruited from a psychiatric research center with a focus on bipolar disorders, and an adult mood disorders clinic for a research study. Participants were excluded if they suffered from a pervasive developmental disorder, or intellectual disability. Telephone screenings were conducted before potential participants were invited to participate (Findling et al. 2001; Findling et al. 2005). All parents provided written consent, and youth participants provided written assent.

Measures

Schedule for Affective Disorders and Schizophrenia for School-Age Children (K-SADS)

The K-SADS was used to measure Diagnosis, Comorbidity, and Age of Onset. All participants were interviewed using one of two versions of the K-SADS, the Schedule for Affective Disorders and Schizophrenia for School-Age Children-Epidemiological version (K-SADS-E) (Orvaschel 1994), or the Present and Lifetime version (K-SADS-PL) (Kaufman et al. 1997). The protocol switched to the KSADS-PL when it became available because it captured information about past episodes of all conditions. The PL uses three point anchors instead of four point anchors for most items, collapsing moderate and severe into “threshold.” Because we used other measures to quantify the severity of depressive (i.e., CDRS-R) and manic (i.e., YMRS) symptoms, the change in scaling on the KSADS items did not affect any measures used in analyses. Interviewers were highly-trained research assistants, who maintained high rates of inter-rater reliability throughout the study (kappa ≥ 0.85) Findling et al. (2001). Youth with mood symptoms not meeting criteria for mania or depression were considered for a diagnosis of cyclothymic disorder – before BP NOS – as dictated by DSM-IV-TR. In order for a participant to receive a diagnosis of cyclothymic disorder, s/he would have to meet report both hypomanic symptoms and depressive symptoms for at least 10 of the preceding 12 months. The symptoms would have to co-occur, to assure the presence of a mood episode, rather than a more chronic profile of irritability with brief periods of depressed mood, which might be more consistent with a diagnosis of ADHD with comorbid depression, for example. Finally, these symptoms would have to occur in the context of a change in functioning because, though cyclothymic disorder is associated with symptoms present over an extended period of time, all bipolar spectrum disorders require a mood episode. If the youth had symptoms of hypomania and depression present for less than 10 months of the preceding year, a diagnosis of BP NOS was considered. A diagnosis of BP NOS was made in cases where the participant failed to meet duration criteria, but met all other criteria for another bipolar spectrum disorder, or in cases where the participant was one symptom short of meeting criteria for mania or hypomania (e.g. elated mood, plus two other symptoms; or irritable mood plus three other B criterion symptoms).

The other diagnostic categories used in this study were bipolar I, bipolar II, bipolar NOS, depression (comprised of major depression, dysthymia, and depression NOS), depression with comorbid ADHD, ADHD, and disruptive behavior disorders (DBD; comprised of oppositional defiant disorder (ODD), conduct disorder (CD), and disruptive behavior disorder not otherwise specified (DBD NOS). All diagnoses were reviewed by a board-eligible or board-certified child/adolescent psychiatrist.

Age of onset often refers to the age at which a participant first met diagnostic criteria for a depressive, manic, hypomanic, or mixed episode (Goldstein et al. 2005). However, given the significant delay that often occurs before an accurate bipolar diagnosis is made Marchand et al. (2006), in this study, age of symptom onset was determined by the youngest age of mood symptomatology, rather than diagnosis, as reported in the K-SADS interview.

Parent General Behavior Inventory (P-GBI)

The P-GBI was used to measure Irritability and Sleep Disturbance. Irritability was assessed during periods of both depressive and elevated mood using select items from the P-GBI Youngstrom et al. (2001), a modified version of the General Behavior Inventory (GBI) Depue et al. (1981) that parents completed about their child. The questions comprise two scales, the depressive and the hypomanic/biphasic, both scales with high construct validity and very high internal consistency (alphas of 0.97 for depression and 0.94 for hypomanic/biphasic) (Youngstrom et al. 2001). Due to changes in funding, the P-GBI was interrupted for a period of more than a year; analyses including the P-GBI were performed on a subset of the whole sample (n = 640).

Items representing depressive or hypomanic irritability were selected in the following way. Three independent raters, a clinical psychologist, a psychiatrist, and a doctoral student in clinical psychology identified items from these subscales that were believed to evaluate/describe irritability. For example, an item representing depressive irritability is, “Have you had periods of sadness and depression when almost everything gets on your nerves and makes you irritable or angry (other than related to the menstrual cycle)?” A sample item from the elevated irritability scale is, “Have there been times of several days or more when, although you were feeling unusually happy and intensely energetic (clearly more than your usual self), you also had to struggle very hard to control inner feelings of rage or an urge to smash or destroy things?” Reliability was high for both depressive irritability (Cronbach’s α = 0.87, seven items) and for elevated irritability (Cronbach’s α 0.71, four items)Van Meter et al. (2011b). Correlation between the depressive and hypomanic irritability scales was r = 0.68, p < 0.05. Sleep disturbance was assessed using the seven-item sleep subscale from the P-GBI, (Cronbach’s α = 0.83) (Meyers and Youngstrom 2008).

The Schedule for Affective Disorders and Schizophrenia (SADS-LB)

The SADS-LB (Endicott and Spitzer 1978) was used to determine parental history of psychiatric disorder when parents were available to be interviewed directly. Reliability for diagnoses made using the SADS is very high (Cronbach’s α 0.85-0.99, depending on diagnosis). If a parent was not available to be interviewed, their diagnoses were based on the Family History Research Diagnostic Criteria interview (FH-RDC) Andreasen et al. (1977) or on prior assessment by a psychiatrist at University Hospitals of Cleveland. The FHRDC was also used to assess for psychiatric disorder in other family members including grandparents and siblings.

Child’s Global Assessment Scale (C-GAS)

The C-GAS (Shaffer et al. 1983) was used to gauge youth’s current level of overall functioning. C-GAS scores were assigned by the interviewer after completing the KSADS interview with both the parent and the youth, integrating all findings and observations from both portions of the interview. Participants’ C-GAS scores were used to control for degree of impairment, which may account for greater variability in symptom severity than diagnosis.

Analytic Plan

Youth with cyclothymic disorder were compared to all other youth in the sample, and to test more specifically whether cyclothymic disorder belongs on the bipolar spectrum or if it shares more in common with the non-bipolar disorders, we undertook two additional sets of analyses for each of these comparisons. First, we contrasted youth with cyclothymic disorder and youth with other bipolar disorders (bipolar I, bipolar II, bipolar NOS). Next, youth with cyclothymic disorder were compared to youth with non-bipolar spectrum disorders, including depression, ADHD, disruptive behavior disorders (ODD, CD), and given the phenomenological overlap, youth with comorbid depression and ADHD. Age, gender, substance use and C-GAS score were included as covariates. Though it is accepted that bipolar occurs in children, prevalence is thought to increase through adolescence (Shankman et al. 2009; Van Meter et al. 2011b). Previous research on adults has not found sex differences in the prevalence of bipolar disorder (Merikangas et al. 2007), but there is some evidence that, in children, boys are more likely than girls to receive a diagnosis of bipolar (Moreno et al. 2007). Substance use is often associated with psychiatric disorders, including pediatric bipolar disorder (Brook et al. 1998; Kandel et al. 1997; Wilens et al. 1999) and may affect symptomatology.

Chi-squared analyses were used to compare rates and categorical variables across diagnostic groups. Scores from the P-GBI were converted into a percent of the maximum possible score (POMP). Converting summed scores into percents provides a standard scale to interpret differences (Cohen et al. 1999. ANOVA was used to compare means on rating scales, linear regression models were used to compare mean differences, when controlling for gender, age, and C-GAS. Repeated-measures ANOVA investigated the pattern of irritability across elevated and depressive periods. Post-hoc comparisons used Tukey’s HSD procedure, which allows for between-group comparisons following an ANOVA, correcting for the increased likelihood of Type I error due to multiple comparisons.

Results

Preliminary Analyses

The sample was comprised of 894 participants, 62.3 % (565) of whom were male. The sample ranged in age from just under five to 17, with an average age of 11.3 years (SD = 3.4). Seventy-seven percent (n = 702) of the sample reported their race as White, 13.8 % Black (n = 125), 2.4 % Hispanic (n = 22), 0.3 % Asian (n = 3), 0.2 % Native American (n = 2), and 3.9 % (n = 35) reported race as Other. Five people declined to report race information. Diagnostic subgroups did not differ by race; see Table 1.

Table 1 Demographic information for diagnostic subgroups
Full size table

The rate of bipolar spectrum disorders, including BP I, BP II, cyclothymic disorder, and BP NOS, did not differ by sex (X 2(1) = 0.72, p = 0.40). The average age of participants was different between those who were diagnosed with any bipolar disorder (M = 10.91) and those who were not (M = 11.67), (F(1,837) = 10.62, p < 0.005). Four percent of the sample was diagnosed with a substance use disorder, there was no difference in the frequency of comorbid substance use between youth with bipolar and those without (X 2(1) = 2.23, p = 0.14). The level of current functioning differed significantly between youth with bipolar disorder (M = 50.15) and those with non-bipolar spectrum disorders (M = 54.29), (F(1,359) = 10.46 p < 0.005). Covariates were accounted for using logistic regression for analyses reported as chi squared statistics; covariates did not affect the results found in the chi square analyses.

See Table 2 for effect sizes comparing cyclothymic disorder to the other bipolar diagnoses, and to the non-bipolar diagnoses, on irritability scores, comorbid diagnoses, sleep disturbance, and family history.

Table 2 Effect sizes comparing cyclothymic disorder to other bipolar disorders and to non-bipolar disorders
Full size table

Irritability

The mean POMP hypomanic/mixed irritability P-GBI score for youth with cyclothymic disorder was 59.38 (SD = 18.82), significantly higher than the mean score of 45.70 (SD = 20.30) for the rest of the sample (all youth without a diagnosis of cyclothymic disorder) (t(635) = -4.05, p < 0.001). The mean depressive irritability P-GBI score for youth with cyclothymic disorder was 62.22 (SD = 17.02), also significantly higher than the mean score of 53.43 (SD = 20.54) for the rest of the sample (t(635) = -2.58, p < 0.01). There were also significant differences in total irritability scores across groups, (p < 0.001); see Table 3.

Table 3 Mean Parent GBI irritability scores (Percent Of Maximum Possible) by diagnostic group
Full size table

Repeated-measures ANOVA, investigating the pattern of irritability across elevated and depressive periods, was not significant (F(3,288) = 2.11, p = 0.09), the level of irritability across elevated and depressed episodes was equivalent across bipolar subtypes. Related, the hypothesis that youth with cyclothymic disorder would have similar levels of irritability in both elevated and depressive episodes, was supported, (t(37) = -1.21, p = 0.24).

Comorbidity

One hundred percent of the youth with cyclothymic disorder in the sample met criteria for at least one comorbid diagnosis, with an average number of 2.42 (SD = 1.18) comorbid diagnoses. This is significantly higher than the rest of the sample, F(1,892) = 4.73, p < 0.05). Six (11 %) youth with cyclothymic disorder had a comorbid anxiety disorder. This is not significantly different from the rest of the sample (X 2(1) = 0.07 p = 0.79). Thirty-one of the youth with cyclothymic disorder (58 %) had comorbid ADHD; this was not significantly different from the rest of the sample (X 2(1) = 0.006, p = 0.94).

As a group, the bipolar subtypes had significantly more comorbid diagnoses than the other youth in the sample, F(1,886) = 253.98, p < 0.001. Within the bipolar spectrum disorders, those youth with BP I had the most comorbid diagnoses (M = 2.81), which was significantly different from youth with BP II (M = 2.09, p < 0.05) and youth with BP NOS (M = 2.41, p < 0.05), but not different from youth with cyclothymic disorder (p = 0.10) according to post hoc analyses. Related, youth with BP I were the most likely to have comorbid ADHD (X 2(3) = 16.60 p < 0.001). See Table 4.

Table 4 Comparison of number and type of comorbid diagnoses
Full size table

Age of Onset

Of the youth with cyclothymic disorder for whom age of onset data were available, 83 % had symptom onset prior to the age of 10. The average age of onset for youth with cyclothymic disorder was 6.1 (SD = 3.3). Though this was not significantly different from most of the sample, it was significantly younger than the age of onset for youth with depression (M = 10.1, p < 0.001). This result supports the theory that, if depression presents very young, it may be more likely to progress to bipolar disorder.

We used Cox regression to determine whether or not youth with cyclothymic disorder had an earlier age of mood symptom onset than youth with other bipolar subtypes. Youth with cyclothymic disorder had a significantly earlier age of onset than youth with bipolar II (Wald = 3.78, p = 0.05); the difference from the other subtypes was not significant.

Sleep Disturbance

Youth with cyclothymic disorder had a mean POMP score of 54.32 (SD = 15.77) on the P-GBI sleep subscale, which was significantly higher than the rest of the study sample (M = 45.53, SD = 17.55) (t(635) = -3.01, p < 0.005).

Youth with cyclothymic disorder differed from the youth with non-bipolar spectrum disorders (post hoc p < 0.05), with the exception of those with depression (M = 48.20, p = 0.25) and those with disruptive behavior disorders (M = 45.76, p = 0.64).When compared to other youth with bipolar spectrum disorders, youth with cyclothymic disorder differed from those with BP NOS (M = 42.43, p < 0.005).

Family History

Ninety-four percent of the youth with cyclothymic disorder in the sample, for whom data were available, reported a family history of psychiatric disorder. See Table 5. However, youth with cyclothymic disorder were not more likely than other youth in the sample to have a family history of psychiatric disorder, with the exception of those youth with DBD (p < 0.05). Parental bipolar disorder was prevalent among youth with cyclothymic disorder; 84.40 % had at least one parent with a bipolar spectrum disorder. Youth with cyclothymic disorder were more likely than the youth without bipolar disorder in the sample to have a parent with bipolar disorder (X 2(4) = 90.06, p < 0.001).

Table 5 Percent with family risk factors for psychiatric disorder by diagnosis
Full size table

There was a difference in the likelihood of familial bipolar risk among the bipolar subtypes, (X 2(3) = 19.36, p < 0.001); youth with cyclothymic disorder were significantly more likely than youth with BP I to have a parent with bipolar disorder. There was not a significant difference between bipolar subtypes in familial psychiatric illness, though cyclothymic disorder had the highest percentage.

Finally, a series of analyses was conducted to determine whether or not youth with NoMaD bipolar disorders (a diagnosis of cyclothymic disorder or BP NOS) and a family history of bipolar spectrum disorders differ from youth with NoMaD diagnoses and no family history of bipolar spectrum disorders. See Table 6. Unfortunately, though the data suggest that there may be some important differences, particularly in the level of irritability (higher in youth with a family history of bipolar disorder) and age of onset (earlier in youth with a family history of bipolar disorder), there were only 20 youth with NoMaD bipolar disorders and no family history of bipolar disorder, therefore these analyses are underpowered. There were no statistically significant differences between the two groups on the following: irritability (depressive, hypomanic, total), total number of comorbid diagnoses, comorbid anxiety, comorbid ADHD, age of onset, sleep disturbance.

Table 6 Comparing presence or absence of family history of bipolar disorder in youth with cyclothymic disorder or bipolar NOS
Full size table

Discussion

The primary aim of this study was to cross-validate prior work using research diagnoses of cyclothymic disorder in youth seeking outpatient services. This is an important addition to the literature, doubling the number of studies investigating the presentation of cyclothymic disorder in youth. Previous studies of youth have not included cyclothymic disorder as a diagnostic category, choosing instead to lump cyclothymic disorder and BP NOS into one category.

Both this study and the other validation study by this group used the diagnostic validation system proposed by Robins and Guze (1970) and explored the characteristics of cyclothymic disorder across the three following categories: Clinical Description, Delimitation from Other Disorders, and Family Studies. The current study expanded on prior work by exploring differences in youth with diagnoses of cyclothymic disorder or BP NOS who have a family history of bipolar spectrum disorders, versus those who do not. Table 7 summarizes the findings from both studies.

Table 7 Summary of key findings from present data and prior investigation of pediatric cyclothymic disorder
Full size table

The first study found that cyclothymic disorder was firmly situated on the bipolar spectrum, sharing many characteristics with the other bipolar spectrum disorders. The current findings largely support the same conclusion. Across multiple characteristics associated with bipolar disorder, the youth with cyclothymic disorder were not statistically different from those youth diagnosed with other bipolar spectrum disorders. The previous validation study found that cyclothymic disorder differed from non-bipolar disorders on the same set of criteria. The exception was youth with both depression and ADHD, with whom those with cyclothymic disorder shared many characteristics. In the current study, the delineation between cyclothymic disorder and the non-bipolar disorders was less definitive: The pattern of results was similar, but the effect sizes were smaller.

Irritability

Parents of youth with cyclothymic disorder retrospectively reported high levels of irritability during both depressive and hypomanic mood states. Cyclothymic disorder was the only bipolar diagnosis that significantly predicted total irritability score. Results were consistent with the hypothesis that cyclothymic disorder is associated with irritability during elevated and depressed mood states, contributing to an apparent persistence of irritability despite changes in mood or energy.

When compared to those youth with non-bipolar disorders, youth with cyclothymic disorder had similar levels of irritability to those youth with disruptive behavior disorders. Though irritability is a common symptom of DBD, our previous study found significantly lower levels of irritability in the youth with DBD. More in line with our hypotheses, the youth with cyclothymic disorder were more irritable overall than the youth with depression and the youth with ADHD. As found in the previous validation study, those youth with depression and ADHD experienced similar levels of irritability to the youth with cyclothymic disorder, though their levels of elevated mood irritability were lower.

Irritability associated with mania and hypomania, not depression, is discussed most often in pediatric bipolar disorder studies (Biederman et al. 2000; Kowatch et al. 2005; Wozniak et al. 1995). Given the result in our previous validation study that indicated that depressive irritability was more severe, and the present result that the two were equivalent, it seems that further research on the role of depressive irritability in bipolar spectrum disorders is warranted(Birmaher et al. 2006; Judd et al. 2002).

Comorbidity

Youth diagnosed with cyclothymic disorder almost always had at least one comorbid diagnosis, consistent with previous findings. However, the rates of comorbid anxiety (11 %) and ADHD (59 %) were lower than in the first validation study, though not inconsistent with other studies of pediatric bipolar disorder (Birmaher et al. 2006; Findling et al. 2005). With the exception of cyclothymic disorder, which was always comorbid with another Axis I disorder, the rate of comorbidity in this study was fairly consistent across diagnoses, both bipolar spectrum and not, which is somewhat different from other studies that have found that youth with bipolar spectrum disorders are likely to have greater comorbidity than other youth (Lewinsohn et al. 1995; Spencer et al. 2001; Wozniak et al. 1995).

Age of Onset

Youth with cyclothymic disorder had an average age of symptom onset of 6 years, one month. This was consistent with the earlier validation study, and supports the theory that bipolar spectrum disorders, particularly more chronic variations, may often begin in childhood (Lewinsohn et al. 2000; Perlis et al. 2004). The other bipolar subtypes had similarly early ages of onset; only bipolar II was significantly different from cyclothymic disorder. Similar to the previous study, youth with bipolar II had an onset of more than 3 years later than the youth with cyclothymic disorder. The age of onset for BP II was similar for that of MDD, and both were later than the rest of the bipolar subtypes. This result could have diagnostic implications; affective symptoms presenting before the age of ten seem to be associated more with cyclothymic disorder, BP I and BP NOS, whereas later onset is associated with a more depressive presentation. Prospective studies of pre-symptomatic youth are needed to further explore this finding.

Sleep Disturbance

Youth with cyclothymic disorder had high levels of parent-reported sleep disruption, significantly higher than the averages for youth with ADHD and youth with BP NOS. Findings support the theory sleep disturbance discriminates between cyclothymic disorder and ADHD (Findling et al. 2005; Harvey et al. 2006; Kowatch et al. 2005). It is possible that circadian rhythm dysregulation is a factor in determining course of the disorder. Further research on differences between subtypes of bipolar disorder is necessary to better understand this phenomenon.

Family History

As predicted, youth with cyclothymic disorder had greater family history of bipolar disorder than the non-bipolar youth, with the exception of those youth with DBD. The high rate of bipolar spectrum disorders in the parents of the youth with DBD is unexpected. If the youth with DBD were very young, they may be exhibiting prodromal PBD. However, their average age is older than that of the rest of the sample. Psychiatric illness heritability is often nonspecific (DelBello and Geller 2001). Having a parent with a bipolar spectrum disorder increases risk for psychiatric illness in general, not just for BP. Youth with cyclothymic disorder had similar rates of family history of general psychiatric disorder to the other bipolar subtypes. Youth with cyclothymic disorder had significantly higher rates of familial BP than did youth with BP I, consistent with the possibility that earlier onset bipolar spectrum disorders may lead to a more chronic presentation and a worse prognosis.

NoMaD Bipolar Disorders Comparisons

We compared youth with a diagnosis of cyclothymic disorder or BP NOS who had at least one parent with a bipolar spectrum disorder to those who did not have a parent with bipolar disorder. Comparing familial versus non-familial BP status found plausible trends, though the size of the subsamples compared limited power. First, youth with a family history of bipolar spectrum disorders had irritability scores six and a half points higher than the other group, suggesting a greater tendency towards mood dysregulation. Second, the group with a family history of bipolar had onset of symptoms more than a year and a half earlier than the other group.

Conclusion

This replication and expansion of our previous study examining the validity of cyclothymic disorder in a youth sample largely confirmed the position of cyclothymic disorder on the bipolar spectrum. Findings showed more similarities than differences between cyclothymic disorder and the other bipolar subtypes; see Table 7. Based on the results of this, and the previous validation study, closer examination of comorbid ADHD, family history and age of onset, in particular, is indicated.

Most surprising were the number of similarities found between the youth with DBD and those with cyclothymic disorder. These results further support the conclusion made in the previous study, that more precise collection of information related to episode polarity and duration is needed. The episodic nature of cyclothymic disorder is important; episodicity may be the best way to distinguish cyclothymic disorder, and other bipolar subtypes, from other childhood disorders.

Finally, comparisons of youth with cyclothymic disorder and bipolar NOS, who had a family history of bipolar disorder and those who did not, found few significant differences, likely due to small sample sizes, but trends including younger age of onset and greater irritability in those with a family history were found. Previous research has suggested that genetic risk may be an important determinant of symptomatology in bipolar spectrum disorders (Findling et al. 2005), and that bipolar risk may be transmitted primarily through cyclothymic temperament (Evans et al. 2005). Both early onset of symptoms and persistent or trait-like irritability would be consistent with cyclothymic temperament (Akiskal 1996; Akiskal and Akiskal 1992; Geller et al. 2008; McElroy et al. 1997; Perlis et al. 2004). Further investigation, including temperament measures and a prospective design, would help to better elucidate the etiological and prognostic differences between those with and those without parental bipolar disorder.

Prognostic and Treatment Implications

Failure to recognize cyclothymic disorder in clinical practice can have significant negative impact on those affected. Prognostic errors resulting in conceptualization of the child as suffering from something other than a major mood disorder will limit the diligence with which risk for later developing bipolar I or II is managed, and will likely minimize the fact that the child is at higher risk of suicidality than s/he would be if suffering from ADHD or another non-mood disorder.

The lack of attention paid to cyclothymic disorder clinically is also likely to result in treatment errors. These are harder to state with confidence, given the lack of a developed treatment literature for cyclothymic disorder, but it is safe to say that there may be a missed opportunity for targeted prevention (secondary intervention), along with the use of psychological interventions, like teaching emotion regulation strategies and psychoeducation about mood disorder and warning signs indicating need for more acute treatment. Additionally, though it may be premature to talk about specific pharmacotherapy, there is evidence to suggest that treatment with a medication, other than those indicated for the true disorder, will – at best – limit treatment effectiveness, and – at worst – may exacerbate symptoms or worsen prognosis.

Limitations and Future Directions

The present study added to the sparse literature on cyclothymic disorder in youth. Though care was taken to ensure the reliability and validity of the diagnoses, we acknowledge the lack of group reliability data (as opposed to individual level reliability) as a limitation and a future direction for study. In addition, a reality of the field is that the diagnostic criteria for cyclothymic disorder and bipolar NOS represent a heterogeneous group of symptoms presentations, and as a result, may be less clear than those for BP I and II. Though we closely followed DSM-IV-TR guidelines, fallibility is unavoidable when relying on retrospective report and human interpretation.

In the present study, as in the previous validation study, the measurement of family history was based primarily on interview of one family member who provided history about all relatives. Such “family history” methods are reasonably accurate, but definitely not a gold standard compared to direct interview of all relatives as is done in some genetic studies Baker et al. (1987).

More study of longitudinal episode patterns and duration associated with cyclothymic disorder may offer diagnostic utility for clinicians. Another important area of future research is the investigation of differences between bipolar spectrum disorders. Some youth with cyclothymic disorder and BP NOS may remit Cicero et al. (2009), while others will progress to BP I or II (Birmaher et al. 2009; Birmaher et al. 2006), and others will continue with a chronic course (Lewinsohn et al. 1995). Determinants of these different courses would have important prevention and intervention implications.

Unfortunately, most studies of pediatric bipolar disorder fail to differentiate between cyclothymic disorder and NOS subtypes, in spite of the fact that these subtypes may represent the greatest opportunity for preventative intervention, and they are likely the most prevalent form of bipolar spectrum disorder. Furthermore, though research is limited, data clearly indicate that those with cyclothymic disorder are affected to a similar degree as those with BP I and II. Prospective studies, including family history, genetic, and biologic data, are crucial to the goal of better understanding the etiology and course of cyclothymic disorder in order to inform treatment and improve outcomes for those youth affected.