CanJPsychiatry 2015;60(3):151–156

Original Research

Prevalence of Bipolar I and II Disorder in Canada Keltie C McDonald, BSc1; Andrew G M Bulloch, PhD2; Anne Duffy, MD, MSc3; Lauren Bresee, PhD4; Jeanne V A Williams, MSc5; Dina H Lavorato, MSc5; Scott B Patten, MD PhD2 1

MSc Student, Department of Community Health Sciences, University of Calgary, Calgary, Alberta. Correspondence: Department of Community Health Sciences, University of Calgary, 3rd Floor TRW Building, Calgary, AB T2N 4N1; [email protected].

2

Professor, Department of Community Health Sciences, University of Calgary, Mathison Centre for Mental Health Research & Education, Calgary, Alberta; Member, Hotchkiss Brain Institute, University of Calgary, Calgary, Alberta.

3

Professor, Department of Psychiatry, University of Calgary, Mathison Centre for Mental Health Research & Education, Calgary, Alberta; Member, Hotchkiss Brain Institute, University of Calgary, Calgary, Alberta.

4

Adjunct Assistant Professor, Department of Community Health Sciences, University of Calgary, Calgary, Alberta; Drug Stewardship Pharmacist, Alberta Health Services, Calgary, Alberta.

5

Research Associate, Department of Community Health Sciences, University of Calgary, Calgary, Alberta.

Key Words: prevalence, bipolar disorder, epidemiology, psychiatric illness, mood disorders, cross-sectional studies Received May 2014, revised, and accepted August 2014.

open access

Objective: Current epidemiologic knowledge about bipolar disorder (BD) in Canada is inadequate. To date, only 3 prevalence studies have been conducted: only 1 was based on a national sample, and none distinguished between BD I and II. The objective of this study was to estimate the prevalence of BD I and II in Canada in 2012. Method: Data were obtained from the 2012 Canadian Community Health Survey: Mental Health and Well-being, a cross-sectional survey of a nationally representative sample of household residents ages 15 years and older (n = 25 113). The survey response rate was 68.9%. Interviews were based on the World Health Organization Composite International Diagnostic Interview (CIDI). Prevalence was estimated using generalized linear modelling. Prevalence of self-reported diagnosis of BD and use of lithium were also estimated. Results: The estimated lifetime prevalence of BD I and II (based on the CIDI) in Canada in 2012 was 0.87% (95% CI 0.67% to 1.07%) and 0.57% (95% CI 0.44% to 0.71%), respectively. Prevalence did not differ by sex. The estimated prevalence of self-reported BD was 0.87% (95% CI 0.65% to 1.07%). There was a lack of congruence between CIDI-defined and self-reported BD, and few people taking lithium were positive for BD on the CIDI, which raises some concerns about the validity of the CIDI’s assessment of BD. Conclusions: These prevalence estimates align with those reported in prior literature. However, caution should be exercised when interpreting general population studies that use CIDI-defined BD owing to the possibility of misclassification. WWW

Prévalence du trouble bipolaire I et II au Canada Objectif : Les connaissances épidémiologiques actuelles sur le trouble bipolaire (TB) au Canada sont inadéquates. À ce jour, seulement 3 études de prévalence ont été menées : seulement 1 se basait sur un échantillon national, et aucune ne distinguait entre le TB I et II. Cette étude visait à estimer la prévalence du TB I et II au Canada, en 2012. Méthode : Les données ont été obtenues de l’Enquête sur la santé dans les collectivités canadiennes—Santé mentale et bien-être, une enquête transversale menée auprès d’un échantillon netionalement représentatif de répondants à domicile de 15 ans et plus (n = 25 113). Le taux de réponse de l’enquête était de 68,9 %. Les entrevues étaient basées sur l’entrevue diagnostique composite internationale (CIDI) de l’Organisation mondiale de la santé. La prévalence a été estimée à l’aide d’un modèle linéaire généralisé. La prévalence des diagnostics de TB auto-déclarés a également été estimée. Résultats : La prévalence de durée de vie estimée du TB I et II (d’après la CIDI) au Canada en 2012 était de 0,87 % (IC à 95 % 0,67 % à 1,07 %) et de 0,57 % (IC à 95 % 0,44 % à 0,71 %), respectivement. La prévalence ne différait pas selon le sexe. La prévalence estimée du TB auto-déclaré était de 0,87 % (IC à 95 % 0,65 % à 1,07 %). Il y avait un manque de congruence entre le TB défini par la CIDI et le TB auto-déclaré, et www.TheCJP.ca

The Canadian Journal of Psychiatry, Vol 60, No 3, March 2015 W 151

Original Research

peu de personnes qui prenaient du lithium étaient positives à la CIDI, ce qui soulève des préoccupations sur la validité de l’évaluation du TB par la CIDI. Conclusions : Ces estimations de la prévalence sont conformes à celles de la littérature antérieure. Cependant, la prudence est de mise lorsqu’on interprète des études dans la population générale qui utilisent le TB défini par la CIDI, étant donné la possibilité d’une classification erronée.

P

revalence estimates serve as a foundation for planning population health interventions, monitoring health system progress, and setting research priorities. To date, only 3 general population studies of BD prevalence have been conducted in Canada, and only 1 of these used a nationally representative sample. Two community-based studies, conducted in Edmonton1 and Ontario,2 obtained prevalence estimates for a manic episode of 0.4% (lifetime) and 0.4% to 0.6% (6 to 12 month), respectively. More recently, a study using data from the 2002 CCHS—Mental Health and Well-being3 estimated a lifetime prevalence of mania at 2.2%.4 However, it has since been suspected that the diagnostic tool used to diagnose BD I in the latter study may have overestimated the prevalence.5 Also, a substantial limitation of the previous studies is their inability to distinguish between BD I and BD II—no available Canadian estimates have made this distinction. The goal of our study was to employ newly calibrated algorithms to produce national estimates of BD I and BD II prevalence.

Method Data Source

The 2012 CCHS was a cross-sectional survey conducted by Statistics Canada. Detailed information about the survey is available from Statistics Canada.6 Information was collected from a nationally representative sample of household residents aged 15 years and older living in the 10 provinces (this target population covers about 97% of the Canadian general population). People excluded from the survey were living on reserve settlements, institutionalized, and full-time members of the Canadian Armed Forces. The survey employed a multi-stage probability-sampling technique. One person aged 15 years or older per sampled household was randomly selected to participate. Interviews were conducted by trained lay interviewers. The majority of interviews (87%) were conducted in person.

CIDI Classification

Interviews were based on the World Mental Health CIDI,7 a fully structured interview intended to derive diagnoses

Abbreviations BD

bipolar disorder

CCHS

Canadian Community Health Survey

CIDI

Composite International Diagnostic Interview

DSM

Diagnostic and Statistical Manual of Mental Disorders

MDE

major depressive episode

152 W La Revue canadienne de psychiatrie, vol 60, no 3, mars 2015

Clinical Implications •

The prevalence of CIDI-assessed BD in Canada resembles that of other countries.

•

The frequency of lithium use was low relative to the apparent prevalence of BD I and II.

•

Major deficits remain in the understanding of BD epidemiology in Canada.

Limitations •

The estimated prevalence of BD I and II may have been underestimated owing to recall bias, as suggested by declining age-specific prevalence.

•

The limited overlap between CIDI-defined diagnoses, self-reported diagnoses, and lithium use raises concern about the validity of the CIDI module for BD.

of numerous mental disorders. Respondents answered a series of questions about possible symptoms of mania, hypomania, and MDE. Algorithms determined the presence or absence of BD I and BD II based on the answers received. The algorithms were updated since the 2002 CCHS through calibration adjustments based on clinical reappraisal studies intended to increase the accuracy of estimation.7 These recalibrated algorithms were used to derive a diagnosis according to operationalizations of DSM-IV8 definitions and criteria. Respondents were classified as having BD I if they had ever experienced 1) at least 6 symptoms of mania; and 2) at least 2 super-symptoms (becoming overly friendly, behaving inappropriately, getting involved in foolish schemes, getting into financial trouble, and believing that they were someone else or connected to a famous person). Respondents were classified as having BD II if they ever experienced 1) distinctly elevated mood lasting 7 days or longer, at least 3 symptoms of mania, euphoria, or racing thoughts, and marked impairment in social or occupational functioning (or symptoms causing less severe impairment, but lasting at least 14 days); 2) at least 1 lifetime MDE; and 3) did not meet the criteria for lifetime manic epdisode.5 Notably, the recalibrations do not completely align at the level of face validity with the DSM-IV diagnostic criteria. For example, the concept of super-symptoms and a 14-day duration are not present in DSM-IV (nor are they present in www.LaRCP.ca

Prevalence of Bipolar I and II Disorder in Canada

Figure 1 Plot of the fitted values for the generalized linear model by age for each of Composite International Diagnostic Interview (CIDI)–classified bipolar disorder (BD) I and BD II, self-reported BD, and lithium use in the sample 1.6 CIDI Bipolar I

1.4

CIDI Bipolar II 1.2

Self‐reported Bipolar Lithium use

Prevalence, %

1.0 0.8 0.6 0.4 0.2 0.0 15

20

25

30

35

40

45

50

55

60

65

Age, years

DSM-59). They represent adjustments intended to improve accuracy of prevalence estimation based on clinical calibration work. The interview used in the 2012 CCHS is the same version as the one used in the clinical calibration study and was not modified by Statistics Canada.

Self-Reported Bipolar Disorder

Respondents were asked about a list of conditions diagnosed by a health professional that had lasted or were expected to last more than 6 months. For self-reported BD, respondents were asked, “Do you have a mood disorder such as depression, bipolar disorder, mania or dysthymia?” People who answered yes were then asked what disorder they were diagnosed with (more than one response was permitted).

Lithium Use

Medication use in the sample was collected as drug identification numbers for all medications taken within the 2 days preceding the interview. Lithium was of particular interest in this study because lithium is a first-line treatment for prophylaxis of BD, as recommended by current Canadian guidelines,10 and is likely to be used more exclusively to treat BD than other mood stabilizers.

Statistical Analysis

Data analysis was carried out at the Prairie Regional Data Centre at the University of Calgary using the statistical software, Stata, version 13.0.11 Crude and sex-specific prevalence were estimated as frequencies with associated 95% confidence intervals. The age-specific pattern was examined using generalized linear models of the binomial family with the log link function. The age-squared term was included in the model owing to a better fit, based on comparing Akaike Information Criterion and Bayesian Information Criterion11 between models, with and without the quadratic age term. Models were also assessed for an age–sex interaction using likelihood ratio tests and www.TheCJP.ca

Wald tests. Analyses were carried out separately for CIDI classified BD I and BD II, self-reported BD, and lithium use. All analyses were weighted using the recommended procedures and replicate bootstrap weights provided by Statistics Canada to produce valid confidence intervals and results representative of the target population. For the remainder of the study, prevalence estimates for BD (CIDIclassified and self-reported) refer to lifetime prevalence. In this case, lifetime prevalence is the most appropriate measurement choice because diagnostic criteria for BD relies on major mood episodes experienced at any time during the life course. In addition, the dynamic nature of the illness (people may experience any of manic, hypomanic, mixed and [or] depressive episodes in a given year) makes it difficult to define past-year or point prevalence in such a way that effectively captures all people with BD.

Results

The national response rate for the survey was 68.9% (n = 25 113). Incomplete data prevented classification of some people into 1 or more subgroups of interest, the proportions of missing data were as follows: CIDI BD I (0.39%), CIDI BD II (0.25%), self-reported BD (0.06%), and lithium use (0.82%). The estimated prevalence of CIDI-classified BD I and BD II in Canada in 2012 was 0.87% (95% CI 0.67% to 1.07%) and 0.57% (95% CI 0.44% to 0.71%), respectively. The estimated prevalence of self-reported BD was 0.87% (95% CI 0.65% to 1.07%), and the prevalence of lithium use in the sample was 0.14% (95% CI 0.08% to 0.20%). Sex-specific prevalence estimates were similar within each subgroup. Generalized linear models, including age as a quadratic term, provided a better fit than those that assumed a linear association between age and log prevalence; therefore, analyses included the quadratic age function. Likelihood The Canadian Journal of Psychiatry, Vol 60, No 3, March 2015 W 153

Original Research

Table 1 Cross-tabulations of the Composite International Diagnostic Interview (CIDI) bipolar disorder (BD) I and BD II, self-reported BD, and lithium use subgroups Proportion (95% CI)a Denominator BD I CIDI BD II CIDI

BD I CIDI

BD II CIDI

Self-reported BD

Lithium use

—

—

0.19 (0.10 to 0.28)

0.03 (0.00 to 0.07)

—

—

0.13 (0.04 to 0.21)

0.03 (0.00 to 0.07)

Self-reported BD

0.19 (0.09 to 0.29)

0.08 (0.03 to 0.14)

—

0.14 (0.07 to 0.21)

Lithium use

0.20 (0.03 to 0.37)

0.11 (0.00 to 0.26)

0.74 (0.57 to 0.92)

—

a

Estimates are weighted.

— = Cells with repeated information or that do not exist

ratio and Wald tests showed no evidence of interactions between sex and age or sex and age-squared for any of the CIDI-classified BD I or BD II, self-reported BD, or lithium models. Sex did not appear to modify or confound the association between age and prevalence, and, therefore, was not included in the final model. Age-specific prevalence for CIDI-classified BD I and BD II and self-reported BD appear to increase with age until mid-30s, then subsequently decline. The prevalence of lithium use appears to increase steadily with age, then plateau at about 55 years of age. Figure 1 shows a plot of the fitted values with age for the generalized linear models estimating prevalence of each of CIDI-classified BD I and BD II, self-reported BD, and lithium use. Table 1 shows cross-tabulations of the CIDI-defined BD I and BD II, self-reported BD, and lithium use subgroups. Among people using lithium, most (74%) reported that they had been diagnosed with BD, consistent with expectation. However, the overlap of the other categories was low. For example, only 19% of people with CIDI-defined BD I reported a diagnosis of BD, and of those who self-reported a BD diagnosis, only 19% were positive for BD I on the CIDI.

Discussion

The estimated prevalence of CIDI-classified BD I and BD II in Canada in 2012 was 0.87% and 0.57%, respectively. The estimated prevalence of self-reported BD was 0.87%. Our study provides the first prevalence estimates for BD I and BD II, separately, in a nationally representative sample in Canada. Although CIDI-classified and self-reported BD appear to provide prevalence estimates that align with that in the literature,12 the 2 measurement methods do not appear to capture the same group of people. Among those who selfreported BD, very few met the CIDI criteria for BD I (19%) or BD II (8%). This suggests that misclassification of people may have occurred. Lithium was used in our study as an additional referent for BD classifications. Lithium is likely to be prescribed primarily for treatment of recurrent mood disorders, particularly BD. The proportion of lithium users that selfreported BD (74%) is consistent with this expectation, but 154 W La Revue canadienne de psychiatrie, vol 60, no 3, mars 2015

this frequency was considerably higher than the proportion of lithium users that met the CIDI criteria for either BD I (20%) or BD II (11%). Of course, the lack of agreement between these different measures may also be due to poor recognition and undertreatment of BD. In the absence of a gold standard, it is unclear to what extent misclassification occurred in each subgroup. An additional indication that misclassification has occurred arises from the observation that both CIDI-defined subgroups showed a parabolic pattern of change in prevalence with age, with a distinct decline in prevalence from mid- to late-adulthood. This is not concordant with the trend that would be expected for a lifelong, recurrent condition in the absence of an extremely strong effect of BD on mortality. This is also in contrast to results of a systematic review that found age-specific lifetime prevalence remained relatively steady across ages 18 to 64 years.13 The age-specific trends observed in our study may be occurring for numerous reasons. Theoretically, it could be explained, to some extent, by a reduced life expectancy among people with BD,14 but it is unlikely that early mortality could fully explain the marked decline. Another potential explanation could be a cohort effect, with diagnoses made more frequently in more recent birth cohorts,15 but this explanation is also unlikely because prevalence was not unexpectedly high, even in younger ages. Finally, the trend may be explained by recall bias. Studies of other mood disorders, such as depression, have observed similar declines with age when using lifetime prevalence,16 which has led to a greater emphasis on 12-month prevalence in the more recent literature. Unfortunately, types of prevalence other than lifetime to study BD may be unsuitable because definitions of BD depend on major mood episodes that occur during the life course. Specifically, the lifetime occurrence of a manic (or hypomanic) episode determines the definitive diagnosis of BD. However, in any given year, there may only be episodes of manic and (or) hypomanic or depressive polarity. For example, people may only have an MDE during the past year, and yet have a BD I diagnosis owing to a previous manic episode. The apparent flaws of lifetime prevalence, and its vulnerability to recall bias, coupled with the need to focus on lifetime prevalence may contribute to the discrepancy in CIDI classifications. www.LaRCP.ca

Prevalence of Bipolar I and II Disorder in Canada

These findings appear to contrast with a study of validity of the CIDI in relation to standardized clinical assessments in the World Health Organization World Mental Health Survey Initiative. The study showed good concordance between the CIDI and Structured Clinical Interview for DSM-IV diagnoses of BD I and BD II in clinical reappraisal sample, with sensitivity and specificity of about 87% an 97%, respectively.17 However, a specificity of 97% implies a false-positive rate of 3%, and this is not consistent with the observed prevalence of less than 1%, even if the sensitivity is zero. Even much higher specificity would result in a high proportion of CIDI-positive subjects being false positive, which, combined with suboptimal sensitivity could explain these results. It is possible that fully structured interviews are inadequate for accurate diagnosis of BD in the general population. Notably, earlier versions of the CIDI included diagnostic modules for psychotic disorders,18 but the modules were excluded from later versions of the instrument because their outputs were invalid. For example, false-positive rates for psychotic symptoms, such as delusions and hallucinations, were found to be quite high, possibly owing to misunderstanding of the interview questions, lack of insight, drug-induced states, or religious beliefs,.19,20 Crossvalidation studies repeatedly showed poor agreement with clinical diagnoses within the National Comorbidity Survey and Epidemiologic Catchment Area studies.21 This may be due to the inability of a rigid, fully structured interview to assess such subtle disturbances and to collect the extensive contextual information required for accurate diagnosis.22 Similar problems may apply to the measurement of BD I and BD II in the general population using fully structured interviews. For example, it may be difficult for a fully structured diagnostic interview to distinguish a normative adolescent experience from hypomanic symptoms.23 In Canada, mental health policy and planning still relies heavily on information collected from population surveys. If the measurement tools used in these surveys are not valid information sources, then more resources must be allocated to developing infrastructure for surveillance through administrative data. In 2010, the Public Health Agency of Canada expanded the Canadian Chronic Disease surveillance system to monitor mental illnesses, including BD. As this is a relatively new initiative, the quality of information collected is yet to be determined. However, additional surveillance using health administration data does not solve all problems, as BD is believed to be an undertreated condition and physician billing and hospital discharge data reflect only the treated prevalence.

Conclusion

Prevalence research lays the framework for understanding burden of disease in the population. Many countries still rely heavily on population surveys for collecting public health information and surveillance. The CIDI is one of the most widely used instruments internationally. Findings in the current study raise questions about the validity of www.TheCJP.ca

the CIDI for measurement of BD I and BD II, and, by extension, our understanding of the epidemiology of BD to date. In light of the results of our study, it is recommended that future and past studies that use the CIDI be interpreted with caution.

Acknowledgements

Keltie McDonald is supported by a grant from Hotchkiss Brain Institute at the University of Calgary and a Canadian Institutes of Health Research–Queen Elizabeth II Scholarship. Dr Patten is a Senior Health Scholar with Alberta Innovates, Health Solutions. While the research and analysis are based on data from Statistics Canada, the opinions expressed do not represent the views of Statistics Canada.

References

1. Bland RC, Orn H, Newman SC. Lifetime prevalence of psychiatric disorders in Edmonton. Acta Psychiatr Scand. 1988;77(Suppl 338):24–32. 2. Parikh SV, Wasylenki D, Goering P, et al. Mood disorders: rural/ urban differences in prevalence, health care utilization, and disability in Ontario. J Affect Disord. 1996;38(1):57–65. 3. Statistics Canada. Canadian Community Health Survey, mental health and well-being. Ottawa (ON): Statistics Canada; 2002. 4. Schaffer A, Cairney J, Cheung A, et al. Community survey of bipolar disorder in Canada: lifetime prevalence and illness characteristics. Can J Psychiatry. 2006;51(1):9–16. 5. Statistics Canada. Derived variables (DV) specifications. Canadian Community Health Survey (CCHS)— mental health 2012. Ottawa (ON): Statistics Canada; 2013. 6. Statistics Canada. Canadian Community Health Survey—mental health (CCHS) 2012 [Internet]. Ottawa (ON): Statistics Canada; 2012. Available from: http://www23.statcan.gc.ca:81/imdb/ p2SV.pl?Function=getSurvey&SDDS=5015&lang=en&db= imdb&adm=8&dis=2. 7. Kessler RC, Ustun TB. The World Mental Health (WMH) Survey Initiative Version of the World Health Organization (WHO) Composite International Diagnostic Interview (CIDI). Int J Methods Psychiatr Res. 2004;13(2):93–121. 8. American Psychiatric Association (APA). Diagnostic and statistical manual of mental disorders. 4th ed; text revision. Arlington (VA): APA; 2000. 9. American Psychiatric Association (APA). Diagnostic and statistical manual of mental disorders. 5th ed. Arlington (VA): APA; 2013. 10. Yatham LN, Kennedy SH, Parikh SV, et al. Canadian Network for Mood and Anxiety Treatments (CANMAT) and International Society for Bipolar Disorders (ISBD) collaborative update of CANMAT guidelines for the management of patients with bipolar disorder: update 2013. Bipolar Disord. 2013;15(1):1–44. 11. StataCorp. Stata Statistical Software: release 13. College Station (TX): StataCorp LP; 2013. 12. Ferrari AJ, Baxter AJ, Whiteford HA. A systematic review of the global distribution and availability of prevalence data for bipolar disorder. J Affect Disord. 2011;134(1–3):1–13. 13. Waraich P, Goldner EM, Somers JM, et al. Prevalence and incidence studies of mood disorders: a systematic review of the literature. Can J Psychiatry. 2004;49(2):124–138. 14. Crump C, Sundquist K, Winkleby MA, et al. Comorbidities and mortality in bipolar disorder: a Swedish national cohort study. JAMA Psychiatry. 2013;70(9):931–939. 15. Blader JC, Carlson GA. Increased rates of bipolar disorder diagnoses among US child, adolescent, and adult inpatients, 1996–2004. Biol Psychiatry. 2007;62(2):107–114. 16. Patten SB, Williams JV, Lavorato DH, et al. Recall of recent and more remote depressive episodes in a prospective cohort study. Soc Psychiatry Psychiatr Epidemiol. 2012;47(5):691–696. 17. Haro JM, Arbabzadeh-Bouchez S, Brugha TS, et al. Concordance of the Composite International Diagnostic Interview version 3.0 The Canadian Journal of Psychiatry, Vol 60, No 3, March 2015 W 155

Original Research (CIDI 3.0) with standardized clinical assessments in the WHO World Mental Health Surveys. Int J Method Psychiatr Res. 2006;15(4):167–180. 18. Robins LN, Wing J, Wittchen H, et al. The Composite International Diagnostic Interview. Arch Gen Psychiatry. 1988;45(12):1069–1077. 19. Anthony JC, Folstein M, Romanoski AJ, et al. Comparison of the lay Diagnostic Interview Schedule and a standardized psychiatric diagnosis. Experience in eastern Baltimore. Arc Gen Psychiatry. 1985;42(7):667–675. 20. Eaton WW, Romanoski A, Anthony JC, et al. Screening for psychosis in the general population with a self-report interview. J Nerv Ment Dis. 1991;179(11):689–693.

156 W La Revue canadienne de psychiatrie, vol 60, no 3, mars 2015

21. 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(11):1022–1031. 22. Brugha TS, Bebbington PE, Jenkins R. A difference that matters: comparisons of structured and semi-structured psychiatric diagnostic interviews in the general population. Psychol Med. 1999;29(5):1013–1020. 23. Tijssen MJA, van Os J, Wittchen HU, et al. Evidence that bipolar disorder is the poor outcome fraction of a common developmental phenotype: an 8-year cohort study in young people. Psychol Med. 2010;40(2):289–299.

www.LaRCP.ca