Lifetime and 12-Month Prevalence of Bipolar Spectrum Disorder in the National Comorbidity Survey Replication

Context  There is growing recognition that bipolar disorder (BPD) has a spectrum of expression that is substantially more common than the 1% BP-I prevalence traditionally found in population surveys.

Objective  To estimate the prevalence, correlates, and treatment patterns of bipolar spectrum disorder in the US population.

Design  Direct interviews.

Setting  Households in the continental United States.

Participants  A nationally representative sample of 9282 English-speaking adults (aged ≥18 years).

Main Outcome Measures  Version 3.0 of the World Health Organization's Composite International Diagnostic Interview, a fully structured lay-administered diagnostic interview, was used to assess DSM-IV lifetime and 12-month Axis I disorders. Subthreshold BPD was defined as recurrent hypomania without a major depressive episode or with fewer symptoms than required for threshold hypomania. Indicators of clinical severity included age at onset, chronicity, symptom severity, role impairment, comorbidity, and treatment.

Results  Lifetime (and 12-month) prevalence estimates are 1.0% (0.6%) for BP-I, 1.1% (0.8%) for BP-II, and 2.4% (1.4%) for subthreshold BPD. Most respondents with threshold and subthreshold BPD had lifetime comorbidity with other Axis I disorders, particularly anxiety disorders. Clinical severity and role impairment are greater for threshold than for subthreshold BPD and for BP-II than for BP-I episodes of major depression, but subthreshold cases still have moderate to severe clinical severity and role impairment. Although most people with BPD receive lifetime professional treatment for emotional problems, use of antimanic medication is uncommon, especially in general medical settings.

Conclusions  This study presents the first prevalence estimates of the BPD spectrum in a probability sample of the United States. Subthreshold BPD is common, clinically significant, and underdetected in treatment settings. Inappropriate treatment of BPD is a serious problem in the US population. Explicit criteria are needed to define subthreshold BPD for future clinical and research purposes.

The estimated lifetime prevalence of bipolar disorder (BPD) in population surveys using structured diagnostic interviews and standardized criteria is approximately 0.8% for BP-I and 1.1% for BP-II.^1-8 Despite this comparatively low prevalence, BPD is a leading cause of premature mortality due to suicide and associated medical conditions, such as diabetes mellitus and cardiovascular disease.^9,10 Also, BPD causes widespread role impairment.^11,12 The recurrent nature of manic and depressive episodes often leads to high direct and indirect health care costs.^13,14

The large discrepancy between rates of BPD found in large-scale community surveys^1-8 and those derived from prospective longitudinal studies¹⁵ suggests that lifetime prevalence estimates from community surveys may underestimate the true prevalence of BPD. Prospective studies¹⁵ have shown that the symptom inclusion criteria and the diagnostic thresholds for BPD are too restrictive to detect BPD in the general population, particularly in young adults, when the disorder is in evolution. Although the precise definitions are as yet unclear, recent studies^15,16 that have included assessment of subthreshold mood disorders suggest that bipolar spectrum disorder might affect up to 6% of the general population. Therefore, the societal burden of BPD might be far greater than currently estimated because of the lack of inclusion of subthreshold BPD.

The purpose of this study is to present the rates and treatment patterns of bipolar spectrum disorder in the National Comorbidity Survey Replication (NCS-R),¹⁷ a study of a nationally representative sample of the United States.

The NCS-R is a nationally representative survey of mental disorders in English-speaking household residents 18 years and older in the continental United States. Interviews were conducted with 9282 respondents between February 2001 and April 2003. Verbal informed consent was obtained before data collection. Consent was verbal rather than written to maintain consistency with the baseline NCS. The response rate was 70.9%. Respondents were given a $50 incentive for participation. A probability subsample of hard-to-recruit predesignated respondents was administered a brief telephone nonrespondent survey, and results were used to weight the main sample for nonresponse bias. Nonrespondent survey participants were given a $100 incentive. The human subjects committees of Harvard Medical School and the University of Michigan, Ann Arbor, approved these recruitment and consent procedures. The NCS-R interview was administered in 2 parts. Part I included a core diagnostic assessment of all respondents (N = 9282). Part II included questions about correlates and additional disorders administered to all part I respondents who met the lifetime criteria for any core disorder plus a probability (1 in 3) subsample of other respondents (n = 5692). A more detailed discussion of NCS-R sampling and weighting is provided elsewhere.¹⁸

The NCS-R diagnoses are based on Version 3.0 of the World Health Organization's Composite International Diagnostic Interview (CIDI),¹⁹ a fully structured, lay-administered diagnostic interview. The DSM-IV criteria were used to define mania (duration ≥1 week), hypomania (duration ≥4 days), and major depressive episode (MDE). The requirement that symptoms do not meet the criteria for a mixed episode (criterion C for mania/hypomania and criterion B for MDE) was not operationalized in making these diagnoses. Respondents were classified as having lifetime BP-I if they ever had a manic episode and as having lifetime BP-II if they ever had a hypomanic but not manic episode and ever had an episode of MDE. The number of lifetime episodes of mania/hypomania and MDE may be overestimated because we did not assess mixed episodes and the rapid cycling subtype of BPD.

Classification of subthreshold BPD included any of the following: (1) recurrent subthreshold hypomania (≥2 criterion B symptoms and all other criteria for hypomania) in the presence of intercurrent MDE, (2) recurrent (≥2 episodes) hypomania in the absence of recurrent MDE with or without subthreshold MDE, and (3) recurrent subthreshold hypomania in the absence of intercurrent MDE with or without subthreshold MDE. The number of required symptoms for a determination of subthreshold hypomania was confined to 2 criterion B symptoms (from the DSM-IV requirement of 3, or 4 if the mood is only irritable) to retain the core features of hypomania in the subthreshold definition. Recurrent hypomania or subthreshold hypomania in the absence of intercurrent MDE was included in the definition because it is part of the DSM-IV definition of BPD not otherwise specified. For the present study, we define the BPD spectrum as a lifetime history of BP-I, BP-II, or subthreshold BPD as defined previously herein. All diagnoses excluded cases with plausible organic causes.

Age at onset of manic/hypomanic episodes and of MDEs was assessed using retrospective self-reports at the syndrome level. Respondents classified as having lifetime BPD were defined as 12-month cases if they had an episode of MDE, mania, hypomania, or subthreshold hypomania at any time in the 12 months before interview. Persistence was assessed by asking respondents to estimate the number of years in their life in which they had a manic or hypomanic episode and, separately, the number of years in which they had an MDE. Clinical reappraisal interviews for BPD using the lifetime nonpatient version of the Structured Clinical Interview for DSM-IV (SCID)²⁰ were administered to a probability subsample of 50 NCS-R respondents. This was a special clinical reappraisal sample that focused exclusively on BPD and was distinct from the larger clinical reappraisal sample used to assess the validity of more common DSM-IV disorders.²¹

The CIDI cases were oversampled in the BPD clinical reappraisal sample, and the data were weighted for this oversampling. As described in more detail elsewhere,²² CIDI-SCID concordance was excellent for any BPD (ie, BP-I, BP-II, or subthreshold BPD), with a κ of 0.94, a positive predictive value (percentage of CIDI cases confirmed using the SCID) of 0.88, a negative predictive value (percentage of CIDI noncases confirmed as not being cases using the SCID) of 1.0, and nonsignificant McNemar test findings (χ²₁ = 0.6; P = .45). The McNemar test evaluates whether the CIDI prevalence estimate differs significantly from the SCID prevalence estimate. Concordances (κ) for individual diagnoses were lower but still acceptable: 0.88 for BP-I, 0.50 for BP-II, and 0.51 for subthreshold BPD, whereas positive predictive values were 0.79, 0.41, and 0.58, respectively. Negative predictive values were consistently greater than 0.99, and the McNemar test value was consistently nonsignificant (χ²₁ = 0.1-0.3; P = .56 to .75).

Symptom severity was assessed in 12-month cases using a self-report version of the Young Mania Rating Scale (YMRS)²³ for mania/hypomania and the Quick Inventory of Depressive Symptomatology Self-Report (QIDS-SR)²⁴ for MDE. The YMRS was based on a fully structured respondent report version developed for parent reports.²⁵ Severity was assessed for the month in the past year when symptoms of either mania or depression were most severe. Standard YMRS and QIDS-SR cutoff points were used to define episodes as severe (including original YMRS and QIDS-SR ratings of very severe, with ratings in the range ≥25 on the YMRS and ≥16 on the QIDS-SR), moderate (15-24 on the YMRS and 11-15 on the QIDS-SR), mild (9-14 on the YMRS and 6-10 on the QIDS-SR), or not clinically significant (0-8 on the YMRS and 0-5 on the QIDS-SR). We did not collect data on differences in the accuracy of recall for the 12 months used in the present study compared with those of the QIDS-SR (30 days) and the YMRS (7 days).

Role impairment in 12-month cases was assessed using the Sheehan Disability Scale.²⁶ As with the YMRS and the QIDS-SR, in the Sheehan Disability Scale, the respondents were asked to focus on the 1 month in the past year when their mania/hypomania or MDE symptoms were most severe. Respondents rated the degree to which the condition interfered with their home management, work, social life, and personal relationships using a visual analog scale from 0 to 10 (none = 0, mild = 1-3, moderate = 4 -6, sever e =7-9, and very severe = 10).

Other core DSM-IV disorders assessed using the CIDI included other anxiety disorders, mood disorders, impulse-control disorders, and substance use disorders. Organic exclusion rules and diagnostic hierarchy rules were used in making all diagnoses. As detailed elsewhere,^21,27 blinded clinical repeated interviews using the nonpatient version of the SCID²⁰ with a probability subsample of NCS-R respondents found generally good concordance between CIDI/DSM-IV diagnoses of anxiety, mood, and substance use disorders and independent clinical assessments. Impulse-control disorder diagnoses were not validated because the SCID clinical reappraisal interviews did not include an assessment of these disorders.

Lifetime and 12-month treatment were also assessed by the specialty of the health care professional, including psychiatrists, other mental health professionals, general medical providers, human services professionals, and complementary and alternative medicine providers. Specific medications were also collected for those who endorsed treatment during the past 12 months. Mood stabilizers, anticonvulsants, and antipsychotics were classified as appropriate medications for BPD. Antidepressants and other psychotropic medications in the absence of antimanic agents were classified as inappropriate. Twelve-month treatment was assessed separately for respondents with 12-month BPD and for those with lifetime but not 12-month BPD (ie, maintenance treatment). Appropriateness of medication was examined separately for those being treated by a psychiatrist and a general medical provider.

All the analyses included controls for sex, age (18-29, 30-44, 45-59, and ≥60 years), race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanic, and other), educational level (less than high school, completed high school, some college, and completed college), and occupation (professional, technical, service-clerical, and labor) for the average expected hours of work per week (20-34, 35-44, ≥45).

Subgroup comparisons were used to study lifetime prevalence and persistence of BP-I, BP-II, and subthreshold BPD. Age-at-onset distributions were estimated using the 2-part actuarial method implemented using a software program (SAS 8.2; SAS Institute Inc, Cary, NC).²⁸ The actuarial method differs from the more familiar Kaplan-Meier²⁹ method in using a more accurate way of estimating the timing of onsets in a given year.³⁰ The actuarial method, similar to the Kaplan-Meier method, assumes constant conditional risk of onset at a given year of life across cohorts. Persistence was examined by calculating means and interquartile ranges of reported years in episode, number of lifetime episodes, and the proportion of lifetime episodes that were manic/hypomanic vs MDE. Comorbidity was assessed by calculating odds ratios between BPD and other CIDI/DSM-IV disorders. Clinical severity, severity of role impairment, and treatment were examined by calculating distributions within the BPD subgroups. Because the NCS-R sample design used weighting and clustering, all statistical analyses were performed using the Taylor series linearization method,³¹ a design-based method implemented using a software system (SUDAAN version 8.01; RTI International, Research Triangle Park, NC). Significance tests of sets of coefficients were performed using Wald χ² tests based on design-corrected coefficient variance-covariance matrices. Statistical significance was evaluated using 2-sided design-based .05-level tests.

Lifetime prevalence estimates are 1.0% for BP-I, 1.1% for BP-II, 2.4% for subthreshold BPD, and 4.4% overall (Table 1). Sex-specific (male and female) prevalence estimates are 0.8% and 1.1% for BP-I, 0.9% and 1.3% for BP-II, and 2.6% and 2.1% for subthreshold BPD. Approximately one third (36.7%) of subthreshold cases have a history of recurrent subthreshold hypomania in the presence of MDEs, whereas 41.9% have a history of recurrent hypomania in the absence of recurrent MDEs. The remaining 21.4% have a history of recurrent subthreshold hypomania in the absence of MDEs. Twelve-month prevalence estimates are 0.6% for BP-I, 0.8% for BP-II, and 1.4% for subthreshold BPD.

Mean retrospectively reported age at onset of the first manic/hypomanic or major depressive episode is somewhat earlier for BP-I (18.2 years) and BP-II (20.3 years) than for subthreshold BPD (22.2 years). The interquartile range (25th-75th percentiles of the cumulative age-at-onset distribution) is between the late teens and the early 40s for all 3 disorders, with the increase in cumulative lifetime prevalence being fairly linear in that age range (Figure). Persistence, indirectly indicated by the ratio of 12-month prevalence to lifetime prevalence, is higher for BP-I (63.3%) and BP-II (73.2%) than for subthreshold BPD (59.5%) (results available on request). The same pattern is found for retrospectively reported mean number of years in episode (10.3 for BP-I, 11.6 for BP-II, and 6.8 for subthreshold BPD) and mean number of lifetime episodes (77.6 for BP-I, 63.6 for BP-II, and 31.8 for subthreshold BPD). More detailed analysis (results available on request) showed that low persistence of subthreshold BPD is limited to participants without a history of MDE. The ratio of lifetime manic/hypomanic episodes to total episodes is in the range of 0.5 to 0.6 for respondents with lifetime BP-I or BP-II and considerably higher (0.8) for subthreshold BPD due to the inclusion of hypomania in the absence of MDE. Twelve-month ratios are similar to lifetime ratios.

The sociodemographic correlates of BPD in the NCS-R are modest in magnitude but fairly consistent across the BPD spectrum. Bipolar disorder is inversely related to age and educational level and is elevated in previously married individuals compared with the currently married (only for subthreshold BPD) and in the unemployed-disabled compared with the employed. Bipolar disorder is unrelated to sex, race/ethnicity, and family income (results not presented but available on request).

Lifetime comorbidity with other DSM-IV/CIDI disorders was reported by most respondents with a history of threshold (95.8%-97.7%) and subthreshold (88.4%) BPD (Table 2). Odds ratios of BPD with individual disorders are uniformly significant and generally higher for BP-I (5.2-13.7) and BP-II (2.6-16.7) than for subthreshold BPD (2.2-5.0). An exception is odds ratios with substance use disorders being much higher for BP-I than for either BP-II or subthreshold BPD. Comorbidity with 3 or more other disorders is dramatically higher than comorbidity with only 1 disorder across the BPD spectrum.

The YMRS 12-month episode severity was rated as severe in a higher proportion of BP-I (70.2%) than BP-II (55.4%) or subthreshold BPD (31.5%) (Table 3). Most manic/hypomanic episodes were classified as either severe or moderate across the BPD spectrum (87.3%-94.6%). The QIDS-SR 12-month severity was rated as more severe for BP-II (84.0%) than for BP-I (70.5%) or subthreshold BPD (46.4%). As with mania/hypomania, most MDEs were classified as either severe or moderate across the spectrum (92.3%-100.0%).

Severe role impairment due to 12-month mania/hypomania was reported by 73.1% of those with 12-month BP-I, 64.6% with BP-II, and 45.9% with subthreshold BPD (Table 4). Severe role impairment due to 12-month MDE in people with BPD was reported by even higher proportions of 12-month cases. As with clinical severity, severe role impairment due to MDE was more common in participants with BP-II (91.4%) vs BP-I (89.3%) or subthreshold BPD (78.8%). Reports of moderate or severe impairment were also more common for MDE (98.6%-100.0%) than for mania/hypomania (87.3%-100%). Impairment was common in all the domains assessed using the Sheehan Disability Scale.

Lifetime treatment of emotional problems by the time of the interview was reported by 80.1% of respondents with lifetime BPD (Table 5). A history of treatment is more common for BP-I and BP-II (89.2%-95.0%) than for subthreshold BPD (69.3%), although even the latter is high compared with other DSM-IV/CIDI disorders.³² Psychiatrists were the most common providers for BP-I (64.9%) and BP-II (62.2%) and general medical professionals for subthreshold BPD (37.5%).

Summing treatment proportions across sectors shows multiple sectors to be the norm, with a 2.2-sector average among patients (2.7 for BP-I, 2.4 for BP-II, and 1.9 for subthreshold BPD). Treatment of 12-month BPD was more frequent than treatment of other DSM-IV/CIDI disorders (67.3% for BP-I, 65.8% for BP-II, and 36.7% for subthreshold BPD).³³ Unlike lifetime treatment, nonpsychiatrist mental health professionals were the most common providers (35.4% for BP-I, 33.8% for BP-II, and 20.6% for subthreshold BPD).

Appropriate medication use could be analyzed only for 12-month treatment (Table 6). A significantly higher proportion of patients receiving psychiatric (45.0%) vs general medical (9.0%) treatment received appropriate medication. Conversely, the proportion of patients who received inappropriate treatment during the past 12 months was significantly greater in those who received treatment from general medical physicians (73.1%) vs psychiatric specialists (43.4%). A significant gradient was found in the proportion of all 12-month cases (ignoring whether they received treatment) that received appropriate medication: 25.0% for BP-I, 15.4% for BP-II, and 8.1% for subthreshold BPD. The proportion receiving inappropriate medication was also higher for BP-I (38.7%) and BP-II (38.9%) than for subthreshold BPD (23.8%). The opposite pattern was found for the proportion receiving no medication (36.3% for BP-I, 45.7% for BP-II, and 68.1% for subthreshold BPD). The numbers were too small to distinguish 12-month cases with mania/hypomania only, MDE only, and both mania/hypomania and MDE with adequate statistical power. A significant gradient was also found in the proportion of lifetime cases without a 12-month episode (ignoring whether they received 12-month treatment) that received appropriate maintenance medication: 17.9% for BP-I, 15.6% for BP-II, and 3.2% for subthreshold BPD. The proportion of all lifetime cases that received inappropriate medication was also higher for BP-I (35.3%) than for BP-II (24.5%) or subthreshold BPD (21.5%). The opposite pattern was found for the proportion receiving no medication (46.8% for BP-I, 59.9% for BP-II, and 75.3% for subthreshold BPD).

There are several limitations of this study that should be considered in interpreting these findings. First, use of the fully structured, lay-administered CIDI precluded the collection of information on the full spectrum of expression of BPD proposed in recent studies.^16,34-36 Although we could not modify the thresholds for some of the diagnostic criteria for mania and depression, our definition of subthreshold BPD is still more restrictive than the definitions proposed by clinical researchers. Therefore, our prevalence estimate of subthreshold BPD is likely to underestimate bipolar spectrum disorder in the population. A related limitation is the absence of information on mixed episodes, rapid cycling, and brief episodes that could only be assessed in more flexible semistructured clinical interviews. Although the clinical reappraisal study found good concordance of CIDI diagnoses with blinded clinical diagnoses based on the SCID, concordance was lower for BP-II and subthreshold BPD than for BP-I.²² The less flexible nature of the CIDI than that of clinical interviews could have also led to overestimation of comorbidity and bias in clinical severity and persistence. Finally, differences between the 1-year recall period of the symptom scales used in the present study and those for which the scales were standardized diminish the comparability of the present findings with those of previous clinical samples.

In the context of these limitations, the results provide the first nationally representative US general population prevalence estimates of subthreshold BPD. Prevalence estimates of BP-I and BP-II (1.0%-1.1%) are consistent with estimates from earlier population-based studies,^1,3-8 except for a much higher lifetime prevalence estimate of BP-I (3.3%) in a large recent national survey of the United States.² The lower NCS-R BP-I prevalence estimate was confirmed by the clinical reappraisal study, whereas no clinical validation of the 3-fold higher estimate in the former study was reported.

The inclusion of symptom severity measures also demonstrates the validity of the spectrum concept of bipolarity. The direct association between increasingly restrictive definitions of BPD and the indicators of clinical validity, including number of episodes, chronicity, symptom severity, impairment, comorbidity, and treatment,^1,34-36 provides evidence of the underlying dimensional nature of bipolar illness. For example, the severity of work impairment increased from 19.8% for subthreshold BPD to 47.5% for BP-II to 62.3% for BP-I, and the severity of manic/hypomanic episodes rose from 31.5% for subthreshold BPD to 55.4% for BP-II to 70.2% for BP-I. Likewise, the mean age at onset ranged from 18.2 years for BP-I to 20.3 years for BP-II to 22.2 years for the subthreshold BPD subgroup, and the estimated average numbers of lifetime episodes were 77.6 for BP-I, 63.6 for BP-II, and 32.0 for subthreshold BPD. Although these retrospective estimates may reflect recall bias to some extent, they are still within the range of estimates reported in prospective studies and family studies.^37,38 Since the mean age at onset of BPD occurs at one of the most critical periods of educational, occupational, and social development, its consequences often lead to lifelong disability. Not only does the disorder begin at an early age but affected individuals spend an average of a decade of their lives in episodes of illness.

The prevalence estimate of subthreshold BPD in the NCS-R is broadly consistent with the findings of 2 large community epidemiologic surveys in Europe.^39,40 However, it is likely to be a lower bound because we did not include reduced duration of hypomanic episodes. The NCS-R results clearly document the clinical significance of subthreshold BPD, as most subthreshold cases had moderate-severe symptom profiles and role impairment based on standard rating scales. As might be expected, there was lower episode persistence and a lower severe-moderate ratio in subthreshold vs threshold cases. Consistent with previous research,¹⁵ the proportions of those with depressive episodes that were rated as severe were higher for BP-II than for BP-I and lower for subthreshold BPD. The more striking results from the perspective of subthreshold BPD are that the clinical severity, role impairment, and comorbidity of subthreshold BPD are all high and, indeed, comparable with those of nonbipolar major depression reported in previous NCS-R analyses.⁴¹ The findings on the clinical significance of subthreshold BPD strongly argue for inclusion of the BPD spectrum concept in the diagnostic classification system.

After controlling for time at risk,⁴² the high magnitude of comorbidity of threshold BPD with other mental disorders is consistent with previous clinical^43,44 and population-based^2-4,45-47 studies. However, although the association between BPD and substance abuse has been well-established, the high magnitude of comorbidity between BPD and anxiety disorders, particularly panic disorder, generalized anxiety disorder, and obsessive-compulsive disorder, has not been widely recognized.^48,49 Despite the higher disorder-specific comorbidity of threshold than subthreshold BPD, comorbidity with at least 1 other disorder was nearly as common in subthreshold (88.4%) as in threshold (95.8%-97.7%) cases. This means that the generally lower bivariate comorbidity of subthreshold than threshold BPD is due to lower multimorbidity (ie, comorbidity with multiple conditions).⁵⁰ This pervasive comorbidity across the BPD spectrum is suggestive of disturbances in multiple regulatory systems and should be a topic for future research.

Most NCS-R respondents with threshold (87.1%-91.5%) or subthreshold (67.8%) BPD reported lifetime treatment for emotional problems. However, treatment in the year before the interview was lower than lifetime treatment for both threshold (65.8%-67.3%) and subthreshold (36.7%) cases, and only a few received appropriate medication (25.0% for BP-I, 15.4% for BP-II, and 8.1% for subthreshold BPD). Appropriate maintenance medication of currently asymptomatic lifetime cases was even lower than inappropriate maintenance medication (17.9% for BP-I, 15.6% for BP-II, and 3.2% for subthreshold BPD). The proportions receiving inappropriate medication (primarily antidepressants in the absence of antimania agents) were considerably higher than those receiving appropriate treatment (31.4% for 12-month cases and 25.1% for asymptomatic lifetime cases), especially in patients undergoing general medical treatment (73.1% for 12-month cases and 65.6% for asymptomatic lifetime cases). Appropriate medication use was much higher and inappropriate medication use lower in patients receiving psychiatric treatment than those receiving general medical treatment, although fewer than half of the psychiatric patients took appropriate medications (45.0% of 12-month cases and 41.1% of asymptomatic lifetime cases). The high use of inappropriate medications is a concern given the dangers associated with the use of antidepressants in the absence of mood stabilizers to treat BPD.⁵¹ Note that the treatment percentages represent patients who took medications. The number of patients who were prescribed but did not take medications was not recorded in the survey. It is possible that higher proportions were prescribed antimanic agents but did not take them because subjective distress is far lower in mania/hypomania than in depression.¹¹

The present results reinforce the argument of other researchers^16,34-36 that clinically significant subthreshold BPD is at least as common as threshold BPD. Although most individuals with BPD receive treatment owing to comorbid disorders, the lack of recognition of their underlying bipolarity leads to only a few receiving appropriate treatment. More comprehensive screening of bipolar symptoms is needed in people seen for treatment of other Axis I disorders. The failure to recognize subthreshold BPD and pervasive comorbidity of BPD can also reduce the precision of estimates and lead to bias in genetic and other etiologic studies of mood disorders.^52-54 Additional research is needed to resolve uncertainty regarding the most appropriate threshold and boundary distinctions for BPD. This uncertainty remains a major impediment to advancing the understanding of the BPD spectrum in the population.

Correspondence: Kathleen R. Merikangas, PhD, Section on Developmental Genetic Epidemiology, National Institute of Mental Health, 35 Convent Dr, MSC#3720, Bethesda, MD 20892 (kathleen.merikangas@nih.gov).

Submitted for Publication: May 28, 2006; final revision received August 29, 2006; accepted August 30, 2006.

Financial Disclosure: Dr Hirschfeld is a consultant to or serves on the advisory board of Abbott Laboratories, AstraZeneca, Bristol-Meyers Squibb, Forest Laboratories, GlaxoSmithKline, Janssen Pharmaceutica, Eli Lilly & Co, Novartis, Organon Inc, Pfizer Inc, Shire, UCB Pharma, and Wyeth-Ayerest.

Funding/Support: The NCS-R is supported by grant U01-MH60220 from the National Institute of Mental Health (NIMH), with supplemental support from the National Institute of Drug Abuse, the Substance Abuse and Mental Health Services Administration, grant 044780 from the Robert Wood Johnson Foundation, and the John W. Alden Trust. The preparation of this article was supported by AstraZeneca.

Disclaimer: The views and opinions expressed in this article are those of the authors and should not be construed to represent the views of any of the sponsoring organizations or agencies or the US government.

Additional Information: A complete list of NCS publications and the full text of all NCS-R instruments can be found at http://www.hcp.med.harvard.edu/ncs. Send correspondence to NCS@hcp.med.harvard.edu. The NCS-R is conducted in conjunction with the World Health Organization World Mental Health (WMH) Survey Initiative. We thank the staff of the WMH Data Collection and Data Analysis Coordination Centers for assistance with instrumentation, fieldwork, and consultation on data analysis. These activities were supported by the John D. and Catherine T. MacArthur Foundation, the Pfizer Foundation, grants 1R13MH066849, R01-MH069864, and R01 DA016558 from the US Public Health Service, Eli Lilly & Co, GlaxoSmithKline, Ortho-McNeil Pharmaceutical Inc, and the Pan American Health Organization. A complete list of WMH publications and instruments can be found at http://www.hcp.med.harvard.edu/wmhcidi.

Acknowledgment: We thank the collaborating NCS-R investigators: Ronald C. Kessler, PhD (principal investigator, Harvard Medical School), Kathleen Merikangas, PhD (co-principal investigator, NIMH), James Anthony, PhD (Michigan State University), William Eaton, PhD (The Johns Hopkins University), Meyer Glantz, PhD (National Institute on Drug Abuse), Doreen Koretz, PhD (Harvard University), Jane McLeod, PhD (Indiana University), Mark Olfson, MD (Columbia University College of Physicians and Surgeons), Harold Pincus, MD (University of Pittsburgh), Greg Simon, MD (Group Health Cooperative), T. Bedirhan Ustun, MD (World Health Organization), Michael Von Korff, ScD (Group Health Cooperative), Philip Wang, MD, DrPH (Harvard Medical School), Kenneth Wells, MD (University of California at Los Angeles), Elaine Wethington, PhD (Cornell University), and Hans-Ulrich Wittchen, PhD (Max Planck Institute of Psychiatry).