Readers can revisit our narrative analysis of the first 12 epidemiological studies (Parry et al. 2018) and these are represented along with the cited prevalence rates from Van Meter et al. (2011) and Van Meter et al. (2019a) in Table 1. In this article we have performed the same qualitative narrative analysis on the newer eight studies.

Table 1 The original 12 epidemiological studies reported by Van Meter et al. (2011)Reanalysis of the eight new studies used by Van Meter et al. (2019a) meta-analysis  sults

The 19 surveys were largely unsuitable for meta-analysis because the methodologies varied in instrumentation, ages of subjects, concordance between informants, prevalence period, and diagnostic criteria. There was wide variation in the reported prevalence rates. BD prevalence rates were zero or close to zero in surveys of prepubertal children.

The eight new surveys covered by Van Meter et al. (2019a) are illustrated in Table 2 and narrative comments follow.

Table 2 The eight extra epidemiological studies reported on by Van Meter et al. (2019a)Päären et al. (2014) Conducted 1991–1993, follow-up 2006–2008, Sweden

This Swedish study screened 2,300 16–17-year-olds for lifetime depression. Three hundred and fourteen who screened positive then participated in the Diagnostic Interview for Children and Adolescents (revised for DSM-III-R, adolescent version, DICA-R-A) for hypomanic symptoms (“hypomania spectrum”). Additionally, 317 controls who had not screened positive for depression were also assessed with the DICA-R-A. The combined and enriched (not strictly community) cohort of 631 adolescents’ findings were: “90 participants with hypomania spectrum (40 full-syndromal, 18 with brief episode, and 32 subsyndromal), 197 participants with major depressive disorder (MDD) and 229 controls” (p. 4).

Fifteen years later, at ages 31–33 years-old, the cohort was diagnostically reviewed with the Mini International Neuropsychiatric Interview Plus (MINI Plus) as well as screening for personality disorders. The authors found: “continued mood disorder in adulthood (MDD, bipolar disorder, or dysthymia) was reported by 60.9% of the hypomania spectrum group and 70.0% of the MDD group” (p. 9). Although many had been prescribed antidepressants, just 7% of hypomania spectrum (n = 4) and 2% of MDD and no controls had been prescribed antipsychotics. No hypomania spectrum and just 2% of MDD had been prescribed anticonvulsants and nobody prescribed lithium.

In a separate article Päären et al. (2013) reported that of the 64/90 adolescents at follow-up who had reported lifetime hypomanic syndromes at baseline, 38 adults (59%) reported an adult MDD, but only four (6%) reported recurrence of hypomania and only two (3%) reported a manic episode. The authors stated it was unlikely that hypomania spectrum adolescents lost to follow-up had converted to BD as the register data showed “only a few received inpatient or outpatient care for mental disorders” (p. 196) and concluded: “The results indicate that only a small proportion of adolescents with hypomania spectrum episodes continue to have (hypo)mania in adulthood” (p. 190). In comparison, for the 130/197 adolescents with MDD, three (2%) reported manic episodes and thirteen (10%) reported hypomanic episodes at follow-up at ages 31–33 years-old (p. 194). Both groups had similar rates of adult MDD (56% and 55% respectively) and of no adult mood disorder (33% for both). Thus, the rates of conversion to BD-I or BD-II disorder in adulthood from adolescent hypomania spectrum compared with MDD episodes was similar.

Tijssen et al. (2010) Conducted 1994–2002 (follow up over 8 years) Germany

This German study interviewed 1,395 14–17-year-old adolescents on four occasions over eight years with the World Health Organisation’s German version Composite International Diagnostic Interview (DIA-X/M-CIDI) to derive lifetime DSM-IV diagnoses and subsyndromal mental symptoms. Minimum duration criteria was 2 weeks for depressive symptoms and 4 days for hypomanic/manic symptoms. At first follow-up, a parent was interviewed for family psychiatric history and whether ADHD diagnosed, otherwise interviews were with the adolescents.

The study focussed on risk factors (presence of family psychiatric history, negative life events, substance use, ADHD diagnosis and personality factors) related to the new appearance of “subclinical expression of bipolar psychopathology” (p. 255). Therefore, all 37 (2.65%) of cases with hypomanic/manic episodes were excluded after the baseline interview. Additionally, any participants (n = 653) with less than full data were excluded.

Of the remaining 705 participants, 162 (23.0%) had a lifetime history of subsyndromal hypomanic/manic symptoms at baseline and 125 (17.7%) subsyndromal depressive symptoms. The three follow-up interviews examined for onset of new hypomanic/manic symptoms or depressive symptoms and then persistence of such symptoms and relationship to the risk factors. It appears from Tijssen et al.’s article that although many of the 705 subjects had manic symptoms, Tijssen et al. do not report if any developed a full hypomanic/manic episode over the 8-year follow-up.

As there was no follow-up data on the 37 (2.65%) of the original 1,395 at baseline who recorded a lifetime hypomanic/manic episode of > 4 days duration, it is unclear how many of this group might have had recurrent episodes in the 8-year follow-up period, which could have been a means of validating true BD cases.

The authors postulated that the transitory nature of subsyndromal hypomanic/manic symptoms and a link with novelty seeking temperament in adolescence that faded by young adulthood represented physiological dopaminergic mechanisms of adolescent development. Tijssen et al. (2010) concluded: “This hypothesis would fit well with the observed high prevalence of manic symptoms reported in adolescents that are transitory for the great majority of individuals” (p. 263).

Roberts et al. (2007) Conducted 2000, USA

This study utilised data from a representative sample of 4,175 youths in Texas aged 11–17-years-old and one of their caregivers. The DISC-IV was used to examine for 12-month prevalence of DSM-IV diagnoses. Impairment was measured using the DISC-IV impairment scale and the Child Global Assessment Scale (CGAS).

The authors only used youth reports for making DSM-IV diagnoses, adding caregiver reports to ascertain level of impairment. They state: “we did not interview parents about the DSM-IV disorders assessed by youth interview”, noting that “many studies have demonstrated considerable discordance in parent–child reports of psychopathology” and “there is little consensus on how parent and youth reports should be combined in epidemiologic studies” (p. 7).

The findings were divided according to DISC-IV without impairment, with DISC-IV impairment criteria, or with CGAS impairment criteria. They found a 12-month prevalence of all DSM-IV disorders of 17.1% (no impairment), 11.1% (DISC-IV impairment) and 5.3% (CGAS impairment). The respective findings for mania were 0.39% (95% C.I. 0.18–0.61), 0.31% (0.12–0.51), 0.22% (0.05–0.39) and for hypomania were 0.81% (0.50–1.12), 0%, 0.09% (0–0.20). The youngest (11–12-year-olds) had roughly half the odds ratios of having a mood disorder than both the 13–15-year-old and 16–17-year-old cohorts, but hypomania/mania were not distinguished from MDD/Dysthymia.

Kozloff et al. (2010) Conducted 2002, Canada

This study extracted data from the representative Canadian Community Health Survey: Mental Health and Well-being (CCHS 1.2) survey of 36,984 people aged 15-years and older. The CCHS 1.2 used the World Mental Health-Composite International Diagnostic Interview (WMH-CIDI) based on DSM-IV criteria. Caregivers were not interviewed. Kozloff et al. looked at data from the 5,673 participants aged 15–18-years-old and 19–24-years-old.

The CCHS 1.2 used DSM-IV criteria for diagnosis of a lifetime manic episode but rather than full 7-day duration criteria, defined a manic episode of lasting “several days or longer” (p. 351). Kozloff et al. presumed that:

[T]he sample likely included subjects with [BD-I] as well as BD-II and ‘not otherwise specified’ but the CCHS 1.2 interview did not include criteria required to accurately differentiate these subgroups of BD (p. 351).

Kozloff et al. found the “overall weighted lifetime prevalence rate of BD was: 2.1% (1.4–2.7) among adolescents 15–18-years and 3.8% (3.0–4.6) among young adults aged 19–24” (p. 352). There were high lifetime comorbid anxiety disorders (46.6%) and 12-month problematic substance use prevalence (50.4%).

In their discussion the authors noted these rates of BD meant that “BD in youth is a fairly common, highly comorbid disorder” (p. 353) and higher than in older adults. They postulated that attrition from suicide and misadventure may reduce rates of BD in older adults, but also listed study limitations that may have over-estimated the BD rate in this youth and young adult sample: 1) the CCHS 1.2 survey used a more liberal duration criteria than strict DSM-IV; 2) the WMH-CIDI was not calibrated for adolescents and “may not accurately distinguish BD from other disorders [such as] ADHD and conduct disorder [which are] more common in adolescents”; 3) episodes may be difficult to distinguish from “extreme and frequent mood swings” that “adolescents tend to experience more than adults”; 4) “the CCHS 1.2 survey relied on self-report without the support of collateral information” (p. 353).

Anselmi et al. (2009) Conducted 2005–2006, Brazil

This study involved children from a Brazilian 1993 Birth Cohort Study. Anselmi et al. screened 4,452 preadolescents (mean age 11.3-years-old) and their mothers with the Strengths and Difficulties Questionnaire (SDQ) in 2004/2005 and later conducted a diagnostic phase of interviews with the children (mean age 12.4-years-old) and mothers using the Development and Well-Being Assessment of Children and Adolescents (DAWBA) in 2005/2006. Prevalence period was not stated but based on citation (Goodman et al. 2000) was 1 month for ‘emotional disorders’.

All who screened positive above an a-priori severity of SDQ symptoms (n = 122), and a random selection of those below the cut-off (n = 158) for a total of 280 children and their mothers, were interviewed using the DAWBA. Diagnoses were based on both parent and child report with adjudication by a child psychiatrist author where parent interviews were discordant.

Extrapolating, Anselmi et al. (2009) calculated: “479 preadolescents out of the 4,448 participants of the 1993 birth cohort would present at least one psychiatric disorder according to either the ICD-10 or DSM-IV” (p. 138). The main findings were: any DSM-IV/ICD-10 diagnosis 10.8% (95% C.I. 7.1–14.5); any anxiety disorder 6.0% DSM-IV/6.2% ICD-10; any DSM-IV/ICD-10 depressive disorder 1.6% (0.4–3.6); ADHD/hyperkinetic disorder 4.1% (1.6–6.4) DSM-IV/2.7% (0.9–5.0) ICD-10; DSM-IV/ICD-10 oppositional-conduct disorder 4.4% (1.6–6.4); eating disorders 0.1% (0.3–0.5); and tic disorders 1.3% (0.2–2.2).

There were no cases of hypomania/mania reported.

Pan et al. (2014) Conducted 2009–2011, Brazil

This Brazilian study screened 9,937 parents of 6–12-year-old children using the Family History Survey (FHS) and then interviewed 2,512 subjects comprised of “a high-risk subgroup (n = 1,554) and a random-selection group (n = 958)” (p. 626) with the parent-version DAWBA. Impairment was measured using the SDQ impact score.

Pan et al. found 479 (19.1%) of the 2,503 subjects screened positive for lifetime manic symptoms and five children (0.2%), mean age 9.4 ± 1.34 years, met criteria for lifetime BD-I/BD-II, while 41 subjects (1.6%) met criteria for BD-NOS.

Other psychiatric disorders lifetime prevalence rates for the whole sample of 2,503 were: any disorder 25.7%; any anxiety disorder 9.9%; any depressive disorder 2.9% ; ADHD 10.9%; any CD/ODD 6.8%. For the 479 screened positive for lifetime manic symptoms the comorbidity was higher: any disorder 43.6%; any anxiety disorder 18.4%; any depressive disorder 6.3%; ADHD 19.8%; any CD/ODD 11.9%.

Pan et al. also examined two dimensions of manic symptomatology: 1) an ‘under-control’ subscale including irritable, distractible, risk taking, less self-control, poor concentration, invades other people’s personal space, bossy, less concerned about getting into trouble, overly sexed, constant changes of plans and activities, flight of ideas, talking to strangers, overconfident, and restless; 2) an ‘exuberant’ subscale including cheerful, joking and laughing more than usual, outgoing, active, fast talk, noisier, gets more done, full of energy, excitable, and restless. Only the ‘under-control’ subscale was associated with psychiatric morbidity and psychosocial impairment. The authors commented that the ‘exuberant’ episodic manic symptoms being mild and not associated with impairment may need to be severe and frequent to be of significance. They noted that “several under-control symptoms of mania are also symptoms of ADHD and ODD/CD. Therefore, we may have ascertained symptoms of externalizing disorders rather than manic symptoms” (p. 631).

They also commented on the similar methodology and findings to Stringaris et al. (2010) (one of the original 12 epidemiological studies), who found very low rates of BD-I/BD-II and a larger group of BD-NOS children/youth who may or may not progress to BD proper. In their contribution to the CAMH debate on PBD, Pan et al. (2019) state:

BD-NOS prevalence was 1.6% in the Brazilian sample, compared to 1.1% by parent report and 1.5% by youth report in the British B-CAMHS (Pan et al. 2014; Stringaris et al. 2010). We have also found that overall BD prevalence was 1.8% in the [Brazilian] study, exactly the same prevalence rate reported in Van Meter et al. (2011) metanalysis. This result adds to the pool of non-US studies in which youth BD could be identified using both narrow (0.2%) and broad (1.6%) criteria. However, this finding does not necessarily mean that all these subjects are ‘true’ bipolar cases. (p. 104).

In the same article they add:

[U]ntil we do fully understand the pathophysiology of BD, impairment may help guide our judgement when we face the hard task of distinguishing manic symptoms from normative (and perhaps developmentally essential) exuberant and under-controlled behaviour in youth. (p. 104)

Vizard et al. (2018) conducted 2017, England

This UK study surveyed 9,117 2–19-year-old children and youth using the DAWBA. Informants (Vizard et al. 2018, p. 6) were parent for 2–10-year-olds; parent, teacher and child/youth for 11–16-year-olds; and parent and youth for 17–19-year-olds. Trained lay interviewers conducted the interviews and clinician raters made the ICD-10 diagnoses.

The study did not distinguish between types of BD due to the small numbers and reported zero cases (to first decimal place) amongst all 5–17-year-olds. However, by gender and age group the prevalence rate was 0.1% amongst 11–16-year-old boys (no girls) and 0.3% amongst 17–19-year-old girls (no boys) (NHS Digital 2018).

Prevalence rates for 7,654 children and youth in the 5–19-year-old range then were (with 95% CI): All: rate of 0.0% (0.0 – 0.1); boys: 0.0% (0.0 – 0.1); girls: 0.1% (0.0 – 0.2). In comparison the rates for any anxiety disorder was: All: 7.2% (6.6 – 7.9); boys: 5.4% (4.7 – 6.2); girls: 9.1% (8.1 – 10.1). For any depressive disorder: All: 2.1% (1.7 – 2.5); boys: 1.4% (1.0 – 1.8); girls: 2.8% (2.2 – 3.5). However, disorders increased with age, for instance 20.3% and 6.5% of older adolescent girls were recorded as having an anxiety or depressive disorder respectively. These rates could be considered low in comparison to some other nations. The authors note a limitation was the response rate of only 52% to the stratified random selection across England.

Karacetin et al. (2018) conducted 2014–2015, Turkey

This study, representing 44 Turkish child and adolescent psychiatry departments, drew upon The Epidemiology of Childhood Psychopathology in Turkey (EPICPAT-T) Study to focus on the prevalence of affective disorder in preadolescent children. The subjects were 5,842 primary school students (mean age 8.7 ± 1.2-years). Parents were interviewed with the Kiddie Schedule for Affective Disorders and Schizophrenia for School Age Children-Present and Lifetime Version (K-SADS-PL). The authors noted that the K-SADS-PL covered the full gamut of bipolar spectrum disorders. They did not discover a single case.

They attributed the lack of bipolar spectrum disorders to the younger age of the cohort, comparing it to the absence of mania in the Great Smoky Mountains Study in the US (Costello et al. 1996) and stated: “Consistent with these findings, there were no cases of BP disorder in our study sample, in which school-age children with a mean age of 8.7 ± 1.2 years were included” (p.519).

Karacetin et al. noted that “irritability can be a part of many psychiatric disorders” and that “the prevalence of ADHD in the EPICPAT-T Study was found to be 12.4% … some of the cases with BP might have been misdiagnosed as ADHD” (p. 519). The prevalence of depressive disorders in this study was (with/without impairment): All depressive disorders 2.5%/1.6%; major depressive disorder 1.06%/1.7%; dysthymia 0.2%/0.2%; adjustment disorder with depressive features 0.4%/0.17%; depressive disorder NOS 0.15%/0.14%.