February 2011  << Back  

  Can't open the PDF? Click here for help.



Misdiagnosis of bipolar disorder in children and adolescents: A comparison with ADHD and major depressive disorder

Jagan K. Chilakamarri, MD

Department of Psychiatry, Emory University, Atlanta, GA, USA

Megan M. Filkowski, BA

Department of Psychiatry, Emory University, Atlanta, GA, USA

S. Nassir Ghaemi, MD

Mood Disorders Program, Tufts Medical Center Boston, MA, USA

BACKGROUND: Controversy surrounds the frequency of underdiagnosis vs overdiagnosis of bipolar disorder (BD) in children and adolescents compared with diagnoses of attention-deficit/hyperactivity disorder (ADHD) and major depressive disorder (MDD).

METHODS: Sixty-four children and adolescents (age 7 to 18) treated in a community setting were systematically assessed for diagnostic and treatment histories. Best estimate consensus diagnosis was made using DSM-IV criteria.

RESULTS: ADHD was overdiagnosed (all patients with ADHD had received the diagnosis, as did 38% of patients with MDD and 29% of patients with BD, respectively), while MDD was partially underdiagnosed and partially overdiagnosed (57% of MDD patients received the diagnosis, 43% did not; 33% of patients with BD were incorrectly diagnosed with MDD). BD was underdiagnosed, not overdiagnosed (38% received the diagnosis, 62% did not; BD was not diagnosed in the ADHD sample, and in only 5% of the patients with MDD). The absence of a positive family history predicted misdiagnosis of BD (relative risk = 2.48, 95% confidence interval 1.10 to 5.56). Observational treatment response to stimulants was equally high in all groups (75% to 82%).

CONCLUSIONS: In the first controlled study on this topic, BD was not over-diagnosed in children and adolescents, as it is often claimed, and ADHD was. Stimulant response was nonspecific and diagnostically uninformative. Studies with larger samples are needed to replicate or refute these results.

KEYWORDS: bipolar disorder, ADHD, major depression, misdiagnosis, overdiagnosis, underdiagnosis



There is much debate about whether bipolar disorder (BD) currently is overdiagnosed or underdiagnosed in children and adolescents in the United States.1-3 Recent practice pattern studies demonstrate increased coding for bipolar diagnoses among children.2 Many authors have concluded that this increased diagnosis reflects overdiagnosis,1,3,4 but it is possible that these diagnostic trends reflect correction of earlier underdiagnosis of BD. Some suggest that BD both is overdiagnosed and underdiagnosed in children in different clinical settings.2

Much of this debate may rest on methodologic differences among studies. Studies that claim overdiagnosis based on researchers’ re-assessment of past diagnoses made by clinicians are studies of diagnostic reliability, not validity of misdiagnosis.5 To assess overdiagnosis vs under-diagnosis of BD, one needs to begin with a validated “gold standard” of correct BD diagnoses, made by researchers, not clinicians, then compare rates of BD diagnosis with other conditions, such as major depressive disorder (MDD) and attention-deficit/hyperactivity disorder (ADHD), followed by assessments of prior diagnoses by other clinicians. If other clinicians previously diagnosed all or almost all “true” BD patients as having BD, then underdiagnosis would not be occurring. Given this result, if other clinicians diagnosed some or most of patients from the other 2 diagnostic categories (ADHD or MDD) with BD, then one could conclude that overdiagnosis had occurred. The other underdiagnosis scenario may be the case if the opposite occurred: some or most “true” BD patients previously received other non-bipolar diagnoses, while few or none of patients from the other 2 diagnostic categories (ADHD or MDD) previously were diagnosed with BD by other clinicians. If both overdiagnosis and underdiagnosis of BD are occurring, then some or most “true” BD patients previously would have received other non-bipolar diagnoses, but at the same time, other clinicians would diagnose some or most of patients from the other 2 diagnostic categories (ADHD or MDD) with BD.

No study of misdiagnosis of BD in children and adolescents has been conducted that began with validated bipolar diagnoses compared with other validated diagnostic groups and subsequently assessed prior diagnoses. This methodology, which would represent an assessment of validity rather than reliability, is used in this study.5 Further, no study in children and adolescents specifically has been designed to compare hypotheses of overdiagnosis, underdiagnosis, or both of BD compared with ADHD and MDD, which are common psychiatric diagnoses in children and adolescents.6


The study was performed in a community primary care mental health setting. The treating clinician (J.K.C.) reviewed charts for 64 children and adolescents (age 7 to 18, inpatients and outpatients), all were newly diagnosed patients seen by him from September 2007 to May 2008. The treating clinician, as part of standard clinical care, obtained third-party reports and records for all patients and interviewed family members. Past diagnoses were obtained from these records and from interviews with patients and patients’ families. DSM-IV criteria were applied by the treating clinician, and then confirmed by consensus of 2 experienced researcher/clinicians (J.K.C. and S.N.G.) using de-identified summary data from the charts. Diagnosis was not based on purely unstructured clinical evaluation, however; attempts were made to identify the type of information necessary for all DSM-IV criteria for the diagnoses involved. Family members provided information, a practice that is standard in child and adolescent psychiatry, recommended by mood disorder experts,7 and supported by research studies that find family members more accurately report mania symptoms than patients.8 Diagnoses of psychiatric disorders in family members did not influence research diagnoses because only DSM-IV criteria were used.

All data were harvested by the treating clinician and recorded in a de-identified manner for review by the coauthor researchers. With this confidential process of retrospective data collection, a waiver of informed consent was obtained from the Emory University Institutional Review Board.

The use of DSM-IV criteria as the "“gold standard” for diagnosis is a standard approach used in misdiagnosis research. The researchers explicitly applied these criteria; clinicians’ diagnoses may or may not have employed those criteria. Stimulant-induced mania was not used to diagnose BD as per DSM-IV criteria that require manic symptoms to occur independent of any direct relationship to an antidepressant.

Misdiagnosis was defined as a patient meeting DSM-IV criteria for a diagnosis different from one that was diagnosed by previous clinicians. We corrected for the possibility of later development of BD if manic episodes occurred late in the course of illness (eg, if depression or ADHD preceded mania, then prior diagnosis of MDD or ADHD was not considered to be a misdiagnosis of later BD). If mood disorder and ADHD criteria were met at the same time, the mood disorder was considered the primary diagnosis based on the consensus view of both mood disorder and ADHD experts on this matter.7 If the primary and secondary diagnoses did not change upon research re-evaluation, then misdiagnosis was not inferred. Change in order of primary vs secondary diagnoses was not counted as misdiagnosis.

Overdiagnosis vs underdiagnosis was defined as follows: if BD is overdiagnosed, then ADHD/MDD commonly would be labeled as BD and most patients with BD would be so diagnosed. If BD is underdiagnosed, BD commonly would be labeled as ADHD/MDD and ADHD/MDD groups would not be labeled as bipolar.

To assess the diagnostic value of treatment response to stimulant medications, retrospective Clinical Global Impressions (CGI) scores were rated by consensus of 2 experienced researcher/clinicians (J.K.C. and S.N.G.) using de-identified summary data from the charts (extracted by J.K.C.).

All data analyses were conducted using descriptive statistics, effect estimation, and confidence intervals with standard statistical software (Statview, Abacus, Cary, NC). Regression modeling explored predictors of misdiagnosis.


Clinical and demographic characteristics can be found in the TABLE. The patient population was preadolescent or early adolescent in age. As would be expected based on the literature on these conditions, there were more males and earlier age of symptom onset in the ADHD group, more substance abuse and hospitalizations in the BD/MDD groups, and segregation by family history for each disorder.

As seen in the FIGURE, BD was underdiagnosed, not overdiagnosed. The most common misdiagnoses made instead of BD were ADHD and MDD, at similar rates. ADHD was overdiagnosed, while MDD was somewhat underdiagnosed, most commonly as ADHD.

The main predictor of misdiagnosis in BD was absence of a BD family history (relative risk [RR] = 2.48, 95% confidence interval [CI] 1.10 to 5.56). Family history was not a predictor of ADHD or MDD. Regression models identified no other predictors of misdiagnosis.

The main secondary diagnoses were ADHD (34.4%) and anxiety disorders (48.4%). ADHD without comorbidity was common (63.6%, 14 of 22 patients), while mood disorders without comorbidity were uncommon (14.3%, 3 of 21 BD patients; 9.5%, 2 of 21 MDD patients). The most common comorbidity with primary ADHD was mood disorder not otherwise specified (36.4%, 8 of 22 patients); for BD it was ADHD (63.6%, 14 of 22 patients); and for MDD it was ADHD (36.4%, 8 of 22 patients).

Stimulants were administered to 76.6% of patients (100% of ADHD patients, 80.9% BD, 76.2% MDD); 64.3% of patients with a primary mood disorder diagnosis who were treated with stimulants had a secondary diagnosis of ADHD (82.3% of BD patients and 70.0% MDD patients). Average duration of stimulant treatment was 3.5 ± 2.9 years (3.2 ± 2.9 years for ADHD, 3.8 ± 3.0 years for BD, 3.5 ± 3.2 years for MDD) and patients tried an average of 2.3 ± 1.5 (range 1 to 6) stimulants. Treatment response to stimulants for distractibility symptoms based on CGI assessments did not appreciably differ among diagnostic groups (81.8% ADHD, 75% BD, and 80% MDD much or very much improved; 9.1% ADHD, 18.8% BD, and 10% MDD worsened; 9.1% ADHD, 6.3% BD, and 10% MDD remained unchanged or minimally improved).


Clinical, demographic, and treatment characteristics of the sample (n = 64)

Variable ADHD (n = 22) BD (n = 21) MDD (n = 21)
Age, mean ± SD (years) 12.4 ± 3.9 14.2 ± 2.8 15.0 ± 1.7
Sex: female (%); male (%) 27.3; 72.7 52.4; 47.6 66.7; 33.3
Race (%)
  African American



Current substance abuse (%) 4.8 10.5 23.5
Suicide attempts (%) 0 28.6 9.5
Hospitalizations (%) 0 42.9 47.6
Family history (%)
  Substance abuse
Age at onset of depression - 10.1 ± 3.2 10.4 ± 4.6
Age at onset of mania - 11.2 ± 3.0 -
Age at onset of ADHD 6.1 ± 2.1 8.3 ± 4.0 9.5 ± 4.6
Misdiagnosis (%) 0 61.9 28.6
Years to correct diagnosis - 3.1 ± 4.0 1.3 ± .92
ADHD: attention-deficit/hyperactivity disorder; BD: bipolar disorder; MDD: major depressive disorder; SD: standard deviation.

FIGURE Past diagnoses in children with ADHD, MDD, and BD

ADHD: attention-deficit/hyperactivity disorder; BD: bipolar disorder; MDD: major depressive disorder.


When assessed with initially valid diagnoses, BD in this community sample of children and adolescents was underdiagnosed, not overdiagnosed. In contrast, ADHD was overdiagnosed, and MDD was partially both over and underdiagnosed.

Secondary findings include the observation that pure mood disorders—without ADHD or other comorbidities—are uncommon in children, perhaps promoting under-diagnosis of those mood disorders. Benefit with amphetamine stimulants for cognitive symptoms of distractibility also was nonspecific and not diagnostically useful.

A number of studies specifically concerned with misdiagnosis of BD in children and adolescents suggest underdiagnosis. The methodology in these studies involves beginning with children diagnosed with BD by researchers and/or experts, and then assessing prior diagnoses. Results suggest prior underdiagnosis of BD using standard DSM-IV criteria in one-half or more of children who could be diagnosed with BD.9-12 Some studies that reported over-diagnosis of BD in children use a methodology beginning with patients diagnosed by clinicians—not researchers or experts—with BD, and then reassessing their diagnoses with clinicians or experts often with standardized diagnostic instruments. These studies then report that 40% to 60% of such children previously diagnosed with BD do not meet BD criteria when systematically assessed.3,4

These types of studies have several limitations. In studies showing underdiagnosis, one would need to have comparison groups of other validated diagnoses (such as ADHD) in which BD could be shown to be infrequently diagnosed by clinicians. In the case of studies purportedly showing overdiagnosis, the main problem is that these studies begin with an un-validated sample (clinician-diagnosed BD) and then find disagreement with researcher diagnosis. Such disagreement between clinicians and researchers is seen in all psychiatric diagnoses, even depression and schizophrenia,13,14 and in neurologic15,16 diagnoses. This diagnostic disagreement cannot be inferred to represent overdiagnosis of BD specifically; it merely represents poor diagnostic reliability because it generalizes to other diagnoses besides BD. To demonstrate overdiagnosis of BD, one would have to show that BD is diagnosed preferentially relative to other diagnostic categories assessed in a similar manner. Further, one should begin with already validated diagnoses made by researchers (systematically applying DSM-IV criteria, and preferably with structured diagnostic instruments), and then assess prior clinician diagnoses, not vice versa, to avoid confusing reliability with validity.5

Limitations of the current study

There is no single “gold standard” for psychiatric diagnosis, such as pathology, and therefore most nosological studies utilize standardized diagnostic interviews, such as the Structured Clinical Interview for DSM Disorders (SCID). The next best method is to achieve a consensus diagnosis using DSM-IV criteria, as the SCID does; this was the method used in this study. Further, we applied standard adult DSM-IV criteria for diagnosing mania, not broadened definitions. Therefore, these misdiagnosis rates reflect straightforward mania, not subtle or complex forms. Data regarding the impact of family history may be conservative because the lack of a family history was recorded as negative. Treatment response data presented here are suggestive at best and are not the primary focus of the study. The data are limited not only by the observational design, which introduces confounding factors, but also by being partly assessed by the treating clinician as well. The results nonetheless do generalize to the common claim that clinicians’ assessment of treatment response to stimulants may be diagnostically suggestive of ADHD, which we could not confirm in this study. Another limitation is that the first author treated these patients and extracted diagnostic and treatment data from their records for this study; the possibility of implicit bias in that process cannot be excluded, although all attempts were made to capture all the data regarding any diagnosis both in the interviews and in chart review. Larger samples are needed to replicate or refute these results.


This is the first controlled study to assess the validity of overdiagnosis of BD vs ADHD vs MDD in children and adolescents. Contrary to common claims, BD was not overdiagnosed in patients in this study. In contrast, ADHD was overdiagnosed in children. MDD was partially underdiagnosed and partially overdiagnosed. The absence of a positive family history appears to predict underdiagnosis of BD, even in the context of manic episodes. Treatment response to stimulants was not informative diagnostically. Larger samples are needed to replicate or refute these results.

DISCLOSURES: Dr. Chilakamarri receives grant/research support from Pfizer, Inc. Dr. Ghaemi receives grant/research support from Pfizer, Inc. and has been a consultant for Sepracor. Ms. Filkowski reports no financial relationship with any company whose products are mentioned in this article.


  1. Harris J. The increased diagnosis of “juvenile bipolar disorder”: what are we treating?. Psychiatr Serv. 2005;56:529–531.
  2. Moreno C, Laje G, Blanco C, et al. National trends in the outpatient diagnosis and treatment of bipolar disorder in youth. Arch Gen Psychiatry. 2007;64:1032–1039.
  3. Zimmerman M, Ruggero CJ, Chelminski I, et al. Is bipolar disorder overdiagnosed?. J Clin Psychiatry. 2008;69:935–940.
  4. Pogge DL, Wayland-Smith D, Zaccario M, et al. Diagnosis of manic episodes in adolescent inpatients: structured diagnostic procedures compared to clinical chart diagnoses. Psychiatry Res. 2001;101:47–54.
  5. Dohrenwend B. The problem of validity in field studies of psychological disorders revisited. In: Tsuang M, Tohen M, Zahner G, eds. Textbook in psychiatric epidemiology. New York, NY: Wiley-Liss; 1995;3–22.
  6. Kessler RC, Avenevoli S, Ries Merikangas  K. Mood disorders in children and adolescents: an epidemiologic perspective. Biol Psychiatry. 2001;49:1002–1014.
  7. Goodwin F, Jamison K. Manic depressive illness, 2nd ed. New York, NY: Oxford University Press; 2007.
  8. Keitner GI, Solomon DA, Ryan CE, et al. Prodromal and residual symptoms in bipolar I disorder. Compr Psychiatry. 1996;37:362–367.
  9. Faedda GL, Baldessarini RJ, Glovinsky IP, et al. Treatment-emergent mania in pediatric bipolar disorder: a retrospective case review. J Affect Disord. 2004;82:149–158.
  10. Wozniak J, Biederman J, Kiely K, et al. Mania-like symptoms suggestive of childhood-onset bipolar disorder in clinically referred children. J Am Acad Child Adolesc Psychiatry. 1995;34:867–876.
  11. Marchand WR, Wirth L, Simon C. Delayed diagnosis of pediatric bipolar disorder in a community mental health setting. J Psychiatr Pract. 2006;12:128–133.
  12. Post RM, Chang KD, Findling RL, et al. Prepubertal bipolar I disorder and bipolar disorder NOS are separable from ADHD. J Clin Psychiatry. 2004;65:898–902.
  13. Anthony JC, Folstein M, Romanoski AJ, et al. Comparison of lay Diagnostic Interview Schedule and a standardized psychiatric diagnosis. Experience in eastern Baltimore. Arch Gen Psychiatry. 1985;42:667–675.
  14. Kendler KS, Gallagher TJ, Abelson JM, et al. Lifetime prevalence, demographic risk factors, and diagnostic validity of nonaffective psychosis as assessed in a US community sample. The National Comorbidity Survey. Arch Gen Psychiatry. 1996;53:1022–1031.
  15. Flossmann E, Redgrave JN, Briley D, et al. Reliability of clinical diagnosis of the symptomatic vascular territory in patients with recent transient ischemic attack or minor stroke. Stroke. 2008;39:2457–2460.
  16. Lopez OL, Litvan I, Catt KE, et al. Accuracy of four clinical diagnostic criteria for the diagnosis of neurodegenerative dementias. Neurolog. 1999;53: 1292–1299.

CORRESPONDENCE: S.N. Ghaemi, MD, Professor of Psychiatry, Mood Disorders Program, Tufts Medical Center, 800 Washington Street, Box 1007, Boston, MA 02111 USA E-MAIL: nghaemi@tuftsmedicalcenter.org