Comorbid anxiety and substance use disorders associated with a lower use of mood stabilisers in patients with rapid cycling bipolar disorder: a descriptive analysis of the cross-sectional data of 566 patients

SUMMARY

Objective

To study mood stabiliser treatment in patients with bipolar disorder with or without anxiety disorders (ADs) and/or substance use disorders (SUDs).

Methods

Extensive clinical interview and Mini-International Neuropsychiatric Interview were used to ascertain DSM-IV diagnoses of rapid cycling bipolar I (RCBDI) or II (RCBDII), SUDs and ADs. Previous treatment statuses with a mood stabiliser after the first onset of mania/hypomania (unmedicated, mismedicated and correctly medicated) were retrospectively determined in patients enrolled into four similar clinical trials. T-test and chi-square/Fisher’s exact were used wherever appropriate.

Results

Of 566 patients (RCBDI n = 320, RCBDII n = 246), 46% had any lifetime AD, 67% had any lifetime SUD and 40% had any recent SUD. Overall, 12% of patients were unmedicated, 37% were mismedicated at the onset of first mania/hypomania and 51% were correctly medicated. Presence of lifetime ADs and recent SUDs was associated with fewer mood stabiliser treatments. Patients with RCBDI were more likely correctly medicated than those with RCBDII (OR = 3.64) regardless of the presence (OR = 2.6) or absence (OR = 4.2) of ADs, or the presence (OR = 2.8) or absence (OR = 3.13) of recent SUDs. Presence of lifetime ADs and recent SUDs increased the risk for mismedicated in RCBDI with odds ratios of 1.8 and 1.9, respectively, but not in RCBDII.

Conclusion

In this multi-morbid cohort of patients with RCBD, 51% of patients (64% of RCBDI and 33% with RCDBII) were correctly medicated with a mood stabiliser after the onset of first mania/hypomania. The presence of ADs and SUDs was associated with an increased risk of mismedicated in patients with RCBDI, but not with RCBDII.

Introduction

The American Psychiatric Association and Consensus Panels have recommended that mood stabilising agents be prescribed for patients with bipolar disorder regardless of the phase of illness, and antidepressants should not be used as monotherapy (1–3). However, patients with bipolar disorder continue to be treated without a mood stabilising agent (4–11). One main reason for the treatment of bipolar disorder without a mood stabiliser is because of misdiagnosis of bipolar depression as unipolar depression or another psychiatric disorder (5,8–10).

The consequences of unfocused or inappropriate treatment in patients with bipolar disorder include an increased risk of rapid cycling, direct and indirect healthcare costs and increased rates of hospitalisation and suicide (6,9–13). The issue of inappropriate treatment and incorrect diagnosis in patients with bipolar disorder was raised more than a decade ago (14). However, more recent epidemiological and treatment studies show that inappropriate treatment, such as antidepressant monotherapy, and misdiagnosis in patients with bipolar disorder are still highly prevalent (4–8). In a more recent study, using commercial claims data, Baldessarini et al. reported that even in patients diagnosed with bipolar disorder, antidepressant monotherapy was a common initial treatment (4).

Substance use disorders (SUDs) and anxiety disorders (ADs) are the two most common psychiatric comorbidities in bipolar disorder (BPD) (15–28). Previous studies have shown that comorbidities in BPD were associated with an increased risk for misdiagnosis (5,6,11,29–31). In a Canadian Community Health Survey on Mental Health and Wellbeing, Schaffer et al. reported that patients with bipolar disorder and ADs were more likely to receive anti-depressant treatment than their counterparts without an AD (32).

However, there has never been a study on the relationship between comorbid anxiety and/or SUDs and previous mood stabiliser treatments. This study was undertaken to explore the relationships between SUDs and ADs with previous history of mood stabiliser treatment (unmedicated, mismedicated and correctly medicated) by using the cross-sectional data at the time of initial assessment of four similar clinical trials. Such a study may shed light on the patterns of help-seeking behaviours and previous treatments in patients with bipolar disorder and comorbid ADs and/or SUDs.

Methods

Patient population

Data from the initial assessment of a cohort of patients with rapid cycling bipolar disorder (RCBD), who were recruited for four similar randomised, double-blind, placebo-controlled clinical trials (33,34 and http://www.clinicaltrials.gov, NCT00221975 & NCT00063362) were examined. These studies were conducted to assess the efficacy of different pharmacological regimens for managing the acute and maintenance treatment of RCBD with or without a ‘recent’ history of SUD, in which all patients were treated with divalproex and lithium initially in an open-label phase. A ‘recent’ SUD was defined as having a DSM-IV diagnosis of substance dependence and continuing to meet abuse or dependence criteria for a substance(s) in the last 6 months at the time of initial assessment or having a DSM-IV diagnosis of substance abuse and continuing to abuse a substance(s) in the previous 3–6 months. Men and women at least 16 years of age were eligible for these studies if they met DSM-IV criteria for BPI or BPII as well as rapid cycling within the last 12 months (≥ 4 episodes in the last 12 months). Patients were referred from a variety of settings including specialty clinics, private and public mental health centres and advertisement. The respective institutional review board approval was obtained, and patients provided written, informed consent. The study design, inclusion and exclusion criteria, mood state at study entry and the stage of each study at the time of this analysis are summarised elsewhere (17).

Initial assessments

The procedures for the initial assessment have also been described in detail previously (16,17,33,34). Briefly, the diagnoses of RCBD, ADs including generalized anxiety disorder (GAD), panic disorder (PD), obsessive-compulsive disorder (OCD) and SUDs, abuse or dependence, were ascertained by extensive clinical interview (ECI) alone or with the Mini-International Neuropsychiatric Interview (MINI) (35) by research psychiatrists and research assistants. For the diagnosis of SUDs, the SCID-P was used instead of the MINI. The ECI consists of questions and criteria for the diagnosis of DSM-IV Axis I disorders, which is similar to the Structured Clinical Interview for the DSM-IV, Patient Edition (SCID-P) (36), but also contains items to assess mental status, demographics and other variables of interest with the mandatory presence of a patient’s significant other(s).

The ECI and the MINI combination typically included a 60–90-min initial interview carried out by a Master’s level research assistant and a 30–45 min second evaluation carried out by a research psychiatrist. A certified research assistant administered the MINI at the third visit. For the MINI certification, a minimal kappa of 0.8 on 10 patients rating in agreement with that of a leading rater who was certified by the Systematic Treatment Enhancement Program for Bipolar Disorder was required. The agreement between the ECI and the MINI diagnosis was over 90%. If any inconsistency occurred with the first and second evaluations during the MINI administration, a psychiatrist would re-evaluate the patient until a consensus was reached.

After the establishment of the diagnosis of RCBD, the criteria for a manic, hypomanic and major depressive episode were explained to patients. After they became familiar with these diagnostic criteria, the age of onset of first manic, hypomanic and major depressive episode was directly inquired, followed by the age of first treatment for mania/hypomania with a mood stabiliser(s) or electroconvulsive therapy (ECT). The mood stabiliser for mania/hypomania was defined as lithium, divalproex/valproic acid, carbamazepine and typical or atypical antipsychotics. The age of first antidepressant treatment for depression without a mood stabiliser(s) was also obtained.

Procedures

The data from four studies were used. As ADs in BPD have a chronic and relatively persistent course (18), lifetime prevalence of ADs was used. Any AD meant the presence of GAD, PD and/or OCD. In contrast to ADs, the course of SUDs is less persistent (18). Therefore, lifetime and ‘recent’ SUDs were used for the prevalence and treatment analysis. Any SUD meant the presence of alcohol or drug (legal or illegal) abuse or dependence with the exception of nicotine and caffeine.

Treatment statuses were determined according to the age onset of first mania/hypomania and the treatment of mania/hypomania with a mood stabiliser(s) or ECT, and the age onset of depression and the treatment with an antidepressant(s). If a patient was not treated with any pharmacological agents prior to the initial assessment, he or she would be classified as unmedicated. If a patient did not receive a mood stabiliser or ECT treatment, but received an antidepressant(s) without a mood stabiliser(s) after the onset of first mania/hypomania, he or she would be classified as mismedicatd. Patients treated with a mood stabilising agent(s) or ECT, alone or combination, at the first time for mania/hypomania were classified as correctly mediated. For patients who were treated with an antidepressant (s) and a mood stabiliser during the first mania/hypomanic phase, if a mood stabiliser was added and antidepressant(s) was tapered off or discontinued at the onset of mania/hypomania, the patient was classified as ‘correctly medicated.’ If an antidepressant was added or continued during the first mania/hypomania, the patient was classified as ‘mismedicated.’ As patients with current SUDs could be present with mood symptoms including manic/hypomanic symptoms (37), the differentiation between substance-induced mood episodes and spontaneous mood episodes was carried out by asking patients which occurred first, substance use or mood symptoms. Only the onset of first spontaneous mood episode was recorded in the initial assessment.

Statistical analysis

T-tests were used to evaluate continuous variables with standard deviation to reflect the magnitude of variance. Chi-square tests were used to evaluate categorical data with odds ratios (OR) and 95% confidence interval (CI) to reflect the magnitude of variances. Fisher’s exact tests were used for cells with n < 5. Although this was an exploratory analysis, our hypotheses were that patients with ADs and/or SUDs were less likely to be treated with a mood stabiliser after the onset of mania/hypomania regardless of bipolar subtype. Given the exploratory nature of the study, statistical significance was set at α = 0.05, two-tailed without adjustment for multiple comparisons, to detect potential associations.

Results

Demographics

Of 568 patients with RCBD, 566 (RCBDI, n = 320; RCBDII, n = 246) were eligible for analysis. The mean ages at study entry of patients with RCBDI and those with RCBDII were similar, 36.5 ± 10.2 vs. 36.3 ± 10.6 years old. Overall, there were more women than men, 53% vs. 47%. In male patients, RCBDI was significantly more common than RCB-DII, 53.3% vs. 38.9%, with an OR of 1.79 (95% CI 1.28–2.51) (Table 1).

Table 1.

Demographics and rates of substance use and anxiety disorders in patients with rapid cycling bipolar I or II disorder

Age at the study entry (years)	Bipolar I disorder N = 321		Bipolar II disorder N = 247		Full sample N = 568
	Mean 36.5 N	SD 10.2 %	Mean 36.3 N	SD 10.6 %	Mean 36.5 N	SD 10.4 %
Gender
Male	171	53.3	96	38.9	267	47.0
Female	150	46.7	151	61.1	301	53.0
Lifetime anxiety disorder
Any	174	54.4	87	35.4	261	46.1
GAD	140	43.8	59	24.0	199	35.1
PD	108	33.8	48	19.5	156	27.6
OCD	25	7.8	16	6.5	41	7.2
Lifetime substance use disorder
Any	236	73.8	140	56.7	377	66.6
Alcohol
Abuse	73	22.7	57	23.1	130	23.0
Dependence	141	43.9	57	23.1	198	35.0
Marijuana
Abuse	77	24.0	60	24.3	137	24.2
Dependence	53	16.5	11	4.5	64	11.3
Cocaine
Abuse	45	14.0	26	10.5	71	12.5
Dependence	73	22.7	15	6.1	88	15.5
Other drugs*
Abuse	50	15.6	36	14.6	96	17.0
Dependence	21	6.6	2	0.8	23	4.1
Recent substance use disorder†
Any	133	41.7	47	19.0	181	40.0
Alcohol
Abuse	31	9.7	16	6.5	47	8.3
Dependence	77	24.0	20	8.1	97	17.1
Marijuana
Abuse	29	9.0	20	8.1	49	8.7
Dependence	24	7.5	4	1.6	28	5.0
Cocaine
Abuse	15	4.7	3	1.2	18	3.2
Dependence	24	7.5	9	3.6	33	5.8

^*Other drugs included stimulants, sedatives, opiates and hallucinogens.

^†A recent substance use disorder was defined as any alcohol or drug abuse/dependence in the last 6 months at the initial assessment for patients who had a DSM-IV diagnosis of substance abuse or dependence. OCD, obsessive-compulsive disorder.

Prevalence of ADs and SUDs

As demonstrated in Table 1, 46% of patients had any lifetime AD, with 35% of a lifetime GAD, 28% of PD and 7% of OCD. Sixty-seven percentage had a lifetime SUD and 40% had a protocol-defined recent SUD. Patients with RCBDI had significantly higher risks for any AD, GAD and PD than those with RCBDII, but no significant difference between the two subtypes in the prevalence of OCD (Table 1).

Patients with RCBDI also had significantly higher risk of any lifetime SUD than patients with RCBDII as well as the risks of lifetime substance dependence: alcohol dependence, marijuana dependence and cocaine dependence (Table 1). Similarly, patients with RCBDI also had significantly higher risks of any protocol-defined recent SUD and alcohol dependence compared with their RCBDII counterparts, but there were no significant differences between the two subtypes in individual substance abuse disorders measured with either lifetime or protocol-defined recent category (Table 1).

Overall previous treatment statuses

Overall, 50.9% of patients were correctly medicated with mood stabilisers or ECT after the onset of first manic/hypomanic episodes, 37.2% were mismedicated and 11.9% were unmedicated (Table 2). The probability of correctly medicated was significantly higher in patients with RCBDI than those with RCB-DII with an OR of 3.64 (95%, CI 2.56–5.17). Conversely, the mismedicated risk was significantly lower in patients with RCBDI than those with RCBDII with an OR of 0.32 (95% CI 0.23–0.46). There was also a trend towards a lower risk of patients with RCBDI being unmedicated compared with those with RCBDII with an OR of 0.62 (95% CI 0.37–1.07).

Table 2.

Treatment statuses of patients with rapid cycling bipolar disorder at the initial assessment

Treatment statuses	Bipolar I disorder		Bipolar II disorder		Full sample		BPI vs. BPII
	N	%	N	%	N	%	OR	Wald’s 95% CI	Pearson’s p
Total	320		246		566
Unmedicated	31	9.7	36	14.6	67	11.8	0.62	0.37–1.03	0.0654
Mismedicated	83	25.9	128	52.0	211	37.3	0.32	0.23–0.46	< 0.0001
Correctly medicated	206	64.4	82	33.3	288	50.9	3.64	2.56–5.17	< 0.0001

BPI, bipolar I disorder; BPII, bipolar II disorder; CI, confidence interval; OR, odds ratio.

Treatment statuses with or without a lifetime history of ADs

As a group, a lifetime history of ADs was not significantly associated with any previous treatment status (Table 3). However, patients with RCBDI were more likely to be correctly medicated compared with those with RCBDII (OR = 3.64, 95% CI 2.56–5.17) regardless of the presence (OR = 2.6, 95% CI 1.48–4.45) or absence (OR = 4.2, 95% CI 2.6–6.78) of a history of ADs. Conversely, patients with RCBDII were more likely to be mismedicated than those with RCBDI.

Table 3.

Treatment statuses of patients with rapid cycling bipolar disorder with or without a lifetime anxiety disorder at the initial assessment

Treatment status	With an anxiety disorder		Without an anxiety disorder		With vs. without an anxiety disorder
	N	%	N	%	OR	Wald’s 95% CI	Pearson’s p
Overall
Total	261		305
Unmedicated	25	9.6	42	13.8	0.7	0.39–1.11	0.117
Mismedicated	98	37.5	113	37.0	1.0	0.73–1.44	0.8962
Correctly medicated	138	52.9	150	49.2	1.2	0.83–1.62	0.3741
Bipolar I disorder
Subtotal	174		146
Unmedicated	12	7.0	19	13.2	0.5	0.23–1.06	0.0653
Mismedicated	54	31.0	29	19.7	1.8	1.08–3.05	0.0231
Correctly medicated	108	62.0	98	67.1	0.8	0.50–1.27	0.347
Bipolar II disorder
Subtotal	87		159
Unmedicated	13	15.1	23	14.3	1.0	0.49–2.14	0.9507
Mismedicated	44	50.0	84	53.1	0.9	0.54–1.54	0.7314
Correctly medicated	30	34.9	52	32.6	1.1	0.63–1.90	0.7508

CI, confidence interval; OR, odds ratio.

Patients with RCBDI and ADs were more likely to be mismedicated compared with those without a history of AD, 31.0% vs. 19.7% with an OR of 1.8 (95% CI 1.08–3.05). On the other hand, patients with RCBDI and ADs were less likely to be unmedicated than those without ADs, 7.0% vs. 13.2% with an OR of 0.5 (95% CI 0.23–1.06). In contrast, there were no significant associations between the history of ADs and previous treatment statuses in patients with RCBDII (Table 3).

Treatment statuses with or without a recent history of SUD

Patients with a recent history of SUDs were more likely to be unmedicated than those without a lifetime history of SUDs, 17.4% vs. 10.7% with an OR of 2.3 (95% CI 1.26–4.06). Increased risk for being unmedicated in those with a recent history of SUDs compared with those without lifetime SUDs was observed in patients with RCBDI, an OR of 3.3 (95% CI 1.08–9.74) and those with RCBDII, an OR of 2.1 (95% CI 0.91–4.61). Patients with RCBDI and recent SUDs were also more likely to be mismedicated than their RCBDI counterparts without a lifetime SUD, 29.8% vs. 18.3% with an OR of 1.9 (95% CI 0.99–3.65). Conversely, those without recent SUDs were more likely to be correctly medicated (Table 4).

Table 4.

Treatment statuses of patients with rapid cycling bipolar disorder with or without a recent history of substance use disorder at the initial assessment

Treatment status	Recent SUD		Without lifetime SUD		Recent SUD vs. without lifetime SUD
	N	%	N	%	OR	Wald 95% CI	Pearson’s p
Overall
Total	213		187
Unmedicated	37	17.4	20	10.7	2.3	1.26–4.09	0.0054
Mismedicated	71	33.3	69	36.9	0.9	0.57–1.29	0.4557
Correctly medicated	105	49.3	97	51.9	0.9	0.61–1.34	0.6071
Bipolar I disorder
Subtotal	161		82
Unmedicated	23	14.3	4	4.9	3.3	1.08–9.74	0.0273
Mismedicated	48	29.8	15	18.3	1.9	0.99–3.65	0.0526
Correctly medicated	90	55.9	64	76.8	0.4	0.19–0.66	0.0007
Bipolar II disorder
Subtotal	52		106
Unmedicated	14	26.9	16	16.0	2.1	0.91–4.61	0.0796
Mismedicated	23	44.2	56	52.8	0.8	0.38–1.46	0.3954
Correctly medicated	15	28.8	34	32.1	0.9	0.43–1.83	0.7409

CI, confidence interval; OR, odds ratio; SUD, substance use disorder.

Similar to the history of AD, patients with RCBD I were more likely to be correctly medicated compared with those with RCBDII regardless of the presence (OR = 2.8, 95% CI 1.84–4.40) or absence of a lifetime SUDs (OR = 7.2, 95% CI 3.77–13.93). Patients with RCBDI and recent SUD were also more likely to be correctly medicated compared with patients with RCBDII and a recent SUD with OR of 3.13 (95% CI 1.59–6.15) (Table 4).

Patterns of previous pharmacological treatments

In a subgroup of patients who had the data of all DSM-IV ADs and SUDs (n = 218), the pattern of previous pharmacological treatments was explored. As showed in Figure 1, antidepressants were the most common agents prescribed prior to study entry. Patients with ADs had fewer mood stabiliser treatments than those without AD when a recent SUD was not considered. When both anxiety and substance disorders were considered, patients with an AD and a recent SUD not only had fewer mood stabiliser treatments, but also had fewer antidepressant treatments compared with those with a recent SUD alone (Figure 2).

Figure 1.

Patterns of previous treatments in patients with rapid cycling bipolar disorder with or without a lifetime history of anxiety disorder. AD, antidepressant; MS, mood stabliser; AP, antipsychotic; Benzo, benzodiazepines

Figure 2.

Patterns of previous treatments in patients with rapid cycling bipolar disorder with or without a lifetime history of anxiety disorder and/or a recent history of substance use disorder. AD, antidepressant; MS, mood stabliser; AP, antipsychotic; Benzo, benzodiazepines

Discussion

In this large sample of patients with RCBD, co-occurrence of ADs and SUDs was highly prevalent. Only half of these patients were correctly medicated after the onset of first manic/hypomanic episodes. Among the patient with bipolar II, only one-third of patients was correctly medicated. Co-occurrence of ADs and/or SUDs was associated with an increased risk for being mismedicated in bipolar I, but not in BPII.

Although the cause–effect relationship between ADs and/or SUDs and previous mood stabiliser treatments could not be established, the results are consistent with previous studies demonstrating that patients with BPD are commonly treated without a mood stabiliser. This is the first study showing that AD or SUD in BPD was associated with lower use of a mood stabiliser, especially in patients with bipolar I disorder. In a Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD) study, however, Simon et al. reported that patients with BPD and a history of current or past-ADs or SUDs had similar proportions of patients who received adequate mood stabiliser treatment (38). The discrepancy between these two studies might be caused by the difference of the timeline of treatments measured. In our study, the use of a mood stabiliser after the onset of first mania/hypomania was as a proxy for treatment statuses. In the STEP-BD study, the lifetime treatment with a mood stabiliser was used. In addition, we used the number of medications in each class to study the pattern of previous treatments. In the STEP-BD study, the proportions of patients treated with each class were used.

One possible reason for the lower use of mood stabilisers in patients with comorbid anxiety and SUDs is a misdiagnosis of BPD as a primary AD, SUD or unipolar depression. Previous studies have shown that misdiagnosis is a major factor for mistreatment (6,9,39). Increased risk for misdiagnosis in patients with BPD and co-occurring ADs has been reported (6,8,40). Increased risk for misdiagnosis in those with BPD and a history of SUDs has also been reported as well (5,8,29). It is also possible that the mismedicated might be because of the reluctance of taking a mood stabiliser by patients who misperceived the side effects or risks from taking mood stabilisers. In contrast to the misdiagnosis of bipolar disorder as unipolar depression, AD and/or SUD (5,8), over-diagnosis of bipolar disorder in patients with current SUD has also been reported (37). Clearly, a longitudinal approach for diagnostic assessments should be used to minimise misdiagnosis or over-diagnosis of bipolar disorders.

The finding of patients with RCBDI and recent SUDs with increased risk for being unmedicated prior to presentation suggests that patients with recent SUDs are less likely to seek professional help. The increased risk for being mismedicated in this group further suggests that even though they previously sought professional help, their bipolar illness was ignored or mismedicated. Increased risk for not receiving mood stabiliser treatment in patients with RCBDII and recent SUDs suggests that this subgroup of patients with RCBDII is also less likely to seek professional help as their RCBDI counterparts. This is consistent with our previous finding that patients with RCBDI or RCBDII with recent SUDs have more similarities than differences (16).

Patients with bipolar II have been speculated to more likely be misdiagnosed/mismedicated than those with bipolar I (41). Depression has been described as a hallmark of RCBD (42). A recent review suggests that patients with rapid cycling usually suffer from bipolar II disorder (43). Unipolar depression is the most common misdiagnosis of bipolar disorders (8). It is very likely that the higher risk of mismedication in patients with RCBDII than those with RCBDI in our study was because of a higher rate of misdiagnosis as unipolar depression. The lack of associations of ADs and SUDs with treatment statuses in RCBDII and a higher rate of mismedication of RCBDII in our study suggests that the mismedication in BPDI and BPDII has different aetiologies. One such aetiology may be because of the difficulty in uncovering a hypomanic episode in BPDII. The reliability for diagnosing hypomania was initially challenged (44,45); however, good reliability for diagnosing hypomania by experienced psychiatrists was later demonstrated (46,47), suggesting that experience may be an important factor explaining the high rates of misdiagnosis. Patients with BPDII had a more severe deficit in self-awareness of illness compared with patients with BPDI (48), so that it is possible that patients with BPDII may underrecognise and underreport their hypomanic symptoms or might be unwilling to take mood stabilisers. However, in the DMDA 2000 survey, the majority of patients with a history of misdiagnosis believed that the lack of understanding about BPD among health professionals was the primary barrier to a more timely correct diagnosis (8).

Undoubtedly, the burden of reducing the risk for mismedication of BPD is on health professionals instead of patients. Patients with bipolar disorder may present with chief complaint of depression, anxiety, drug and/or alcohol problem, and their initial encounters are primary care providers, not mental health professionals. This further poses challenges for an accurate diagnosis and correct treatment in this population. More recently, the International Society for Bipolar Disorder has developed guidelines for diagnosis of BPD (49–51). However, ADs and SUDs were not on the list of variables, which may be used to differentiate bipolar depression from unipolar depression (49, 50). Future studies should focus on the importance of these comorbidities in differentiating BPD from other psychiatric disorders. Before these data become available, a thorough assessment for BPD in those who present with mood and ADs and/or SUDs is essential to avoid misdiagnosis. Clearly, a systematic effort should be made to include all health professionals, patients and family members to reduce mismedication in bipolar disorder.

Limitations

First, data were retrospective, cross-sectional and involved four studies. Second, not all ADs were assessed, so some individuals with other ADs (e.g. posttraumatic stress disorder (PTSD) or social phobia alone) could be misclassified as having no AD. Third, as with most psychiatric cross-sectional phenomenological studies relying upon the recall of history by patients, the diagnoses and historical variables might not be accurate. Fourth, the information on disorders previously misdiagnosed was not collected. Fifth, our sample only included outpatients with RCBD; therefore, our results might not be generalisable to other bipolar populations. Sixth, intentionally not adjusting for multiple comparisons may increase the chance of type I error. More importantly, the cause–effect relationship between the diagnosis and mismedication could not be established because of the retrospective nature of the data. Similarly, the distinction between the first spontaneous mania/hypomania and substance-induced mania/hypomania could not be verified. However, in a recent study, Frank et al. reported that only 10% (n = 101) of patient with bipolar I disorder had substance use immediately preceding (within 3 weeks) the first onset of mania (52). Substance use preceded more than 3 weeks or followed the first onset of mania/hypomania in the majority of patients. This suggests that the first onset of mania/hypomania in patients with bipolar disorder and SUDs are more likely to be spontaneous instead of substance induced. Clearly, prospective studies are warranted to distinguish the spontaneous mania/hypomania and substance-induced mania/hypomania in the subgroup of patients with bipolar disorder.

Conclusions

In this highly comorbid cohort of patients with RCBD, about half of those with RCBDI and two-third of those with RCBDII were mismedicated without a mood stabiliser after the onset of first mania/hypomania. The presence of ADs and recent SUDs were associated with increased risk for being mismedicated in those with RCBDI, but not in those with RCBDII. Clinicians should pay more attention to those with depression, anxiety and/or SUDs to avoid misdiagnosis and mismedication in patients with bipolar disorder.

What’s known

• Patients with bipolar disorder are commonly misdiagnosed and mistreated.

• Anxiety and substance use disorders are the two most common comorbidities in bipolar disorder.

What’s new

• This is the first study showing that comorbid anxiety and/or substance use disorders was associated with a lower use of mood stabilisers in patients with bipolar disorder.

• Patients with bipolar disorder and a recent history of substance use disorders were less likely to seek professional help.