Cannabis use in first-treatment bipolar I disorder: relations to clinical characteristics

Authors

  • Levi R. Kvitland,

    Corresponding author
    1. NORMENT, KG Jebsen Centre for Psychosis Research, Division of Mental Health and Addiction, Oslo University Hospital and Institute of Clinical Medicine, University of Oslo, Oslo
    • Corresponding author: Mr Levi R. Kvitland, NORMENT, KG Jebsen Centre for Psychosis Research, Division of Mental Health and Addiction, Oslo University Hospital and Institute of Clinical Medicine, University of Oslo, N-0424 Oslo, Norway. Email: l.r.kvitland@medisin.uio.no

    Search for more papers by this author
  • Ingrid Melle,

    1. NORMENT, KG Jebsen Centre for Psychosis Research, Division of Mental Health and Addiction, Oslo University Hospital and Institute of Clinical Medicine, University of Oslo, Oslo
    Search for more papers by this author
  • Sofie R. Aminoff,

    1. Division of Mental Health Services, Akershus University Hospital, Lørenskog, Norway
    Search for more papers by this author
  • Trine V. Lagerberg,

    1. NORMENT, KG Jebsen Centre for Psychosis Research, Division of Mental Health and Addiction, Oslo University Hospital and Institute of Clinical Medicine, University of Oslo, Oslo
    Search for more papers by this author
  • Ole A. Andreassen,

    1. NORMENT, KG Jebsen Centre for Psychosis Research, Division of Mental Health and Addiction, Oslo University Hospital and Institute of Clinical Medicine, University of Oslo, Oslo
    Search for more papers by this author
  • Petter A. Ringen

    1. Division of Mental Health and Addiction, Oslo University Hospital, Oslo
    Search for more papers by this author

Abstract

Aims

The aim of this study was to investigate the associations between recent cannabis use, current symptomatology and age at onset of first manic, depressive and psychotic episodes in a large sample with first-treatment bipolar I disorder (BD I).

Methods

One hundred one patients with first-treatment Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) bipolar I disorder were included as part of the Thematically Organized Psychosis study. The Structural Clinical Interview for DSM-IV was used for DSM-IV diagnosis and identification of episodes of illness. Earlier suicide attempts were recorded. Manic, depressive and psychotic symptoms were rated using the Young Mania Rating Scale, Inventory of Depressive Symptoms and Positive and Negative Syndrome Scale correspondingly. Cannabis use within the six last months was recorded.

Results

After controlling for confounders, recent cannabis use was significantly associated with lower age at onset of first manic and psychotic episode, but not with onset of first depressive episode (both P < 0.05). Recent use was also associated with more lifetime suicide attempts (P < 0.01). No group differences were found on symptom levels.

Conclusions

The present study confirms earlier findings of an association between cannabis use and a lower age at onset. Recent cannabis use was also associated with more lifetime suicide attempts. The current findings suggest that recent cannabis use is associated with a more severe course of illness in the early phase of BD I.

Introduction

Bipolar disorder (BD) ranks among the top 10 contributors to the global burden of disease.[1, 2] Identifying risk factors associated with the onset of BD is as a key target for research.[3] However, few risk factors have been identified apart from a family history of bipolar disorder.[4] Cannabis use is prevalent in BD,[5, 6] also before the start of first treatment.[7, 8] The evidence for cannabis as a risk factor in BD is weaker than for schizophrenia,[9] but recent research, including studies of large epidemiological and clinical samples from the United States and Europe, clearly indicate an effect of cannabis in lowering the age at onset across age cohorts.[10, 11] Cannabis use has further been shown to increase the risk for mania, prolong the duration of manic episodes[10-12] and increase suicide risk. Cannabis use also seems to increase the risk for developing BD in depressed individuals.[13]

However, previous studies showing an effect of cannabis on age at onset mainly comprise multi-episode patients. This makes it difficult to rule out the alternative explanation, for instance, that the more severe course seen in early-onset patients could increase the risk for cannabis use after onset or in other ways interacts with negative effects of cannabis use in this time period. Study samples with longer duration of illness are also more vulnerable to attrition bias and thus potentially less representative of the BD patient population than first-treatment samples. Furthermore, these samples are more vulnerable to recall bias regarding potential risk factors before onset of the illness, compared with first-treatment samples.

First-treatment BD samples are scarce and they often have a low number of participants and/or other methodological limitations. Existing large-scale studies have included only patients with psychotic BD[14, 15] or hospitalized inpatients,[12, 16] excluded patients with previous self-remitting mood episodes[17] or relied on clinical expert diagnoses rather than systematized research diagnoses.[10] To our knowledge, the current study is the largest naturalistic sample to date with a broad, representative spectrum of recent onset, first-treatment patients with bipolar I disorder (BD I), including both inpatients and outpatients with both psychotic and non-psychotic BD, and not excluding patients with comorbid illnesses. The current study aims at answering the following main research questions:

  1. Is recent cannabis use associated with younger age at onset of first manic, depressive or psychotic episode in first-treatment BD I?
  2. What are the associations between recent cannabis use and current affective and psychotic symptoms and history of suicidal behaviour in first-treatment BD I?

Methods

Sample

One hundred one patients with recent onset Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV)[18] BD I participated in the study as part of The Thematically Organized Psychosis (TOP) Study at the University of Oslo and Oslo University Hospital. The inclusion criteria were as follows: meeting the DSM-IV diagnosis of BD I, within the first year of receiving their first adequate inpatient or outpatient treatment for a manic episode, age between 17 and 65 and ability to give written informed consent. Patients entered the study in the first year after first treatment because of the ethical and practical problems of recruiting and achieving informed consent from unstable patients in the acute manic phase of BD.

The patients were recruited consecutively from 2003 until 2013 from all major hospitals in the Oslo area. All clinicians at all inpatient and outpatient units at our participating hospitals were asked to identify patients with established or suspected bipolar disorder, and motivate the patients to be referred to our study. This recruitment design has shown to give a high degree of representativity towards the general patient population in Oslo.[6] The exclusion criteria were pronounced cognitive deficits (intelligence quotient (IQ) lower than 70), moderate/severe head injury, neurological disorder, not being able to speak a Scandinavian language or not being able to give a written informed consent. All patients received a complete description of the study before giving consent to participation. The study was approved by the Regional Committee for Medical Research Ethics and the Norwegian Data Inspectorate.

Assessments

Assessments were carried out by trained medical doctors and clinical psychologists. Diagnosis and episodes of illness, including age at onset of first manic, depressive and psychotic episode in addition to previous suicide attempts were determined using the Structural Clinical Interview for DSM-IV Axis I disorders (SCID module I, chapters A-E)[19] with the aid of medical charts.

Patients were interviewed about type of substance use the past 6 months as well as age at first exposure to the specific substances. The SCID-I module E[19] was used for obtaining a diagnosis of substance abuse/dependency. Questions about substance use were part of a semistructured interview, based on a detailed manual designed for the current study.

Current depressive symptoms were rated using the Inventory of Depressive Symptoms-Clinician rated (IDS-C),[20] and current manic symptoms were rated using the Young Mania Rating Scale.[21] Global functioning was measured by the Global Assessment Functioning scale (GAF), split version.[22, 23] Current negative and positive psychotic symptoms were assessed with the Positive and Negative Syndrome Scale (PANSS).[24, 25] The PANSS was added to assess psychotic symptoms in the sample, as such symptoms are prevalent in BD I and is considered a relevant measure of illness severity.[24, 25] Sociodemographic and medication data were obtained by clinical interviews and medical charts. General intelligence level (IQ) was measured by the Wechsler Abbreviated Scale of Intelligence.[26] All assessment personnel completed a training programme based on the programme at the University of California, Los Angeles[27] and participated in regular diagnostic consensus meetings led by experienced clinical experts in the field of diagnostics of severe mental illness. Inter-rater reliability for diagnosis had an overall kappa score of 0.77 (95% CI (0.60, 0.94)). The inter-rater reliability of the assessments in the TOP study has been shown to be good, with an intraclass coefficient [28] for PANSS positive symptoms of 0.82 and for the GAF of 0.86.[6]

Statistical analyses

The analyses were conducted by using the statistical package for the social sciences (SPSS) version 19.0 (SPSS Inc, Chicago, IL, USA). Group comparisons for continuous variables were evaluated with independent sample t-tests, group comparisons for dichotomous data were evaluated with chi-squared tests or Fischer's exact tests. Level of significance was set to P < 0.05, two-sided.

The sample was divided into those who had used any cannabis at all over the past 6 months (recent use) and those without any use in this period (no cannabis use; Table 1).

Table 1. Demographic and clinical characteristics of patients with and without recent cannabis use
 No cannabis use last 6 monthsCannabis use last 6 monthsP
N = 77N = 24
NMeanSDNMeanSD
  1. aMissing data.
  2. IDS, Inventory of Depressive Symptoms; GAF, Global Assessment Functioning Scale; PANSS, Positive and Negative Syndrome Scale; SD, standard deviation; YMRS, Young Mania Rating Scale.
Age (years)7732.511.32426.76.60.002
Education (years)74a14.72.92413.12.50.011
IDS total score76a15.410.722a20.514.10.177
YMRS total score774.56.4246.46.00.208
PANSS positive total score7710.34.62411.22.60.428
PANSS negative total score7710.03.2249.42.50.177
GAF-S7758.614.12454.310.10.849
GAF-F7751.312.72450.910.80.844
Duration between first mood episode and treatment (years)775.17.7245.07.30.939
 N% N% P
Female sex5166.2 1041.7 0.054
Not currently a student or employed3646.8 1562.5 0.234
In a relationship3849.4 625.0 0.058
Daily smoking3444.7 2291.7 0.000
Of European origin6584.4 2291.7 0.509
Psychotic features5064.9 1562.5 0.813

Since several demographic and clinical characteristics were associated with both recent use and age at onset (Table 3), we conducted a series of hierarchical block-wise multiple linear regression analyses to test for possible confounders of their associations. For lifetime suicidal attempts we conducted a logistic regression analysis based on the same principles. We included the variables with the highest significant correlation (P < 0.01) with the dependent variable in step one, and the variables either correlated to recent use or with the least significant correlation (P. < 05) with the dependent variable at the next step, before adding information about recent use on the third and last step. We did not include daily smoking as this is strongly associated with recent use in this (Table 1) and previous studies,[16] nor did we include age in the analyses since age at onset in a first-episode sample is highly correlated with current age (r = 0.8 in the current sample). To explore any age or cohort effects on recent cannabis use, we instead conducted a series of hierarchical block-wise multiple linear regression analyses including the age at onset for cannabis (age at first exposure) use in the last step of the model. The main analyses were repeated including only patients experiencing their first ever episode of mania. This did not give any differences in results (data not shown).

Results

Mean age of the sample was 31.1 years (SD: 10.6 years), N = 61 (60%) were female. Lifetime cannabis use was reported in 55%. There were more males in the ‘recent use’ group (P = 0.05) than in the ‘no cannabis use’. The ‘recent use’ group was also significantly younger and had significantly less education (Table 1).

The ‘recent use’ group was significantly younger at the onset of their first manic, depressive and psychotic episodes, and thus also at the age at onset of their illness, regardless of the characteristics of their first episode. There were no significant differences between the groups regarding current symptoms and functioning, but the ‘recent use’ group used antipsychotics to a significantly lower degree compared with the ‘no cannabis use’ group. They had also significantly more often a lifetime history of suicide attempts (Table 2).

Table 2. Age at onset, medication use and suicide attempts in patients with and without recent cannabis use
No cannabis use last 6 monthsCannabis use last 6 monthsP
N = 77N = 24
NMeanSDNMeanSD
  1. aLower N because not all patients experienced this episode.
  2. bPsychotic, manic, depressive or mixed onset.
  3. BD I, bipolar I disorder; SD, standard deviation.
Age at onset first psychotic episode50a29.510.415a22.13.50.000
Age at onset first manic episode74a28.911.221a23.37.00.008
Age at onset first depressive episode57a24.911.021a19.86.20.013
Age at onset first psychotic episode50a29.510.415a22.13.50.000
Age at onset BD I regardless of episode typeb7724.510.42420.16.60.018
 N% N% P
Current use of antipsychotics5774.0 1041.7 0.006
Current use of mood stabilizers1924.7 416.7 0.579
One or more suicide attempts, lifetime1628.1 1263.2 0.012
Table 3. Bivariate correlations (Pearson's R) between cannabis use, clinical and sociodemographic variables
 1234567891011121314
  1. aCorrelation is significant at the 0.05 level (two-tailed).
  2. bCorrelation is significant at the 0.01 level (two-tailed).
Cannabis use last 6 months (1)1−0.219a−0.222−0.322b−0.191−0.236a−0.245a−0.214a−0.1340.209a0.0890.1400.352b−0.264
Age at onset first manic episode (2)−0.219a10.806b0.843b0.827b0.792b0.390b−0.211a−0.101−0.408b0.0690.071−0.231a0.566b
Age at onset first depressive episode (3)−0.2220.806b10.832b0.865b0.740b0.252a−0.080−0.092−0.356b0.0640.053−0.2170.483b
Age at onset first psychotic episode (4)−0.322b0.843b0.832b10.820b0.841b0.299a−0.027−0.116−0.408b0.2110.152−0.1450.618b
Age of onset of BD regardless of episode type (5)−0.1910.827b0.865b0.820b10.666b0.235a−0.214a0.017−0.407b0.0840.052−0.236a0.429b
Age (6)−0.236a0.792b0.740b0.841b0.666b10.394b−0.181−0.163−0.429b0.1900.153−0.1930.555b
Education (7)−0.245a0.390b0.252a0.299a0.235a0.394b10.0380.051−0.1780.204a−0.211a−0.230a0.008
Sex (8)−0.214a−0.211a−0.080−0.027−0.214a−0.1810.0381−0.089−0.0580.027−0.1480.090−0.246
A student or employed (9)−0.134−0.101−0.092−0.1160.017−0.1630.051−0.0891−0.009−0.0040.004−0.142−0.074
In a relationship (10)0.209a−0.408b−0.356b−0.408b−0.407b−0.429b−0.178−0.058−0.0091−0.1210.1080.1870.080
Of European origin (11)0.0890.0690.0640.2110.0840.1900.204a0.027−0.004−0.12110.0960.013−0.014
Alcohol abuse/dependency (12)0.1400.0710.0530.1520.0520.153−0.211a−0.1480.0040.1080.09610.0720.252
Cocaine or amphetamine use last 6 months (13)0.352b−0.231a−0.217−0.145−0.236a−0.193−0.230a0.090−0.1420.1870.0130.0721−0.246
Age at first exposure−0.2640.566b0.483b0.618b0.429b0.555b0.008−0.246−0.0740.080−0.0140.252−0.2461

After controlling for possible confounders of the relationship between age at onset and recent use (Table 3), we found that recent use contributed significantly to age at onset of first manic episode, age at onset of first psychotic episode and to age at onset of first depressive episode. Education and sex also contributed significantly to the age at onset of first manic episode in the final model. We found an even stronger association between lifetime suicide attempts and recent use after controlling for possible confounders (Table 4a-4d). In the follow-up analyses, age at first exposure to cannabis contributed significantly (beta 0.956, P < 0.001) to age at onset of the first manic episode (final model (F 8.232 P < 0.001, adj. r2 0.430). Likewise, age at first exposure use contributed significantly (beta 1.131, P < 0.001) to age at onset of first psychotic episode (final model F 6.858 P < 0.001 adj. r2 0.442). Likewise, age at first exposure contributed significantly (beta 0.777, P < 0.01) to age at onset of first depressive episode (final model F 2.938 P < 0.05 adj. r2 0.177).

Table 4a. Multiple regression analysis with age at first manic episode as dependent variable
Block no. variableBlock model summary for each stepContribution of separate variables for last step
R2 changeF changeBetatP value95% CI of B
LowerUpper
  1. Total model F 8.232 P < 0.001.
  2. Adj. r2 0.430.
  3. CI, confidence interval; ND, no data.
ConstantNDNDND2.4700.0173.01329.566
10.12130.154     
Education  0.9372.0780.0430.0291.845
In a relationship  −4.128−1.4340.158−9.9231.668
20.1223.554     
Female sex  −5.870−2.1600.036−11.346−0.394
Cocaine or amphetamine use last 6 months  −5.481−1.2500.218−14.3183.356
30.24620.716     
Cannabis use last 6 months  0.9564.5520.0000.5321.379
Table 4b. Multiple regression analysis with age at first depressive episode as dependent variable
Block no. variableBlock model summary for each stepContribution of separate variables for last step
R2 changeF changeBetatP value95% CI of B
LowerUpper
  1. Total model F 2.938 P < 0.05.
  2. Adj r2 0.177.
  3. CI, confidence interval; ND, no data.
ConstantNDNDND9.4610.00018.39628.354
10.0190.843     
In a relationship  −2.808−0.9180.364−8.9743.357
20.0580.863     
Education  0.0770.1590.875−0.9001.054
Cocaine or amphetamine use last 6 months  −4.140−1.0000.323−12.5004.214
Female sex  −3.031−1.0010.323−9.1463.085
30.19110.473     
Cannabis use last 6 months  0.7773.2360.0020.2921.262
Table 4c. Multiple regression analysis with age at first psychotic episode as dependent variable
Block no. variableBlock model summary for each stepContribution of separate variables for last step
R2 changeF changeBetatP value95% CI of B
LowerUpper
  1. Total model F 6.858 P < 0.001.
  2. Adj r2 0.442.
  3. CI, confidence interval; ND, no data.
ConstantNDNDND12.1410.00024.72834.647
10.0381.416     
In a relationship  −3.824−1.1900.822−10.3422.694
20.0440.527     
Cocaine or amphetamine use last 6 months  −7.505−1.0080.321−22.6597.649
Education  0.3120.4830.632−1.0011.624
Female sex  −0.957−0.2860.777−7.7695.856
30.43528.865     
Cannabis use last 6 months  1.1315.3730.0000.7021.560
Table 4d. Multiple logistic regression analysis with one or more suicide attempts lifetime as dependent variable
VariableSuicide attempts lifetime
CoefficientPOR95% CI of OR
LowerUpper
  1. Nagelkerke r2 0.31.
  2. CI, confidence interval; OR, odds ratio.
Step 1     
Age−0.0200.490.980.9261.037
Female sex1.3950.014.041.36511.929
In a relationship−0.1600.780.850.2782.615
Step 2     
Age−0.0210.490.980.9921.039
Female sex1.3790.013.971.31811.973
In a relationship−0.0360.950.970.3053.052
Alcohol abuse/dependency0.3580.621.430.3515.836
Cocaine or amphetamine use last 6 months−0.9050.350.410.0622.661
Step 3     
Age−0.0080.810.990.9271.061
Female sex2.1340.018.452.01235.504
In a relationship0.1560.801.170.3433.979
Alcohol abuse/dependency0.0510.9491.050.2224.975
Cocaine or amphetamine use last 6 months0.3560.7541.430.15513.175
Cannabis use last 6 months2.2590.019.571.95246.921

Discussion

There are two main results of the current study. Firstly, that recent cannabis use had a significant association with the age at onset of the first episodes of mania, psychosis and depression in a highly representative sample of patients with first-treatment BD, also after controlling for relevant confounders. Secondly, that any lifetime suicidal attempt was more prevalent in the recent use group. There were no significant differences in other clinical characteristics between recent cannabis users and the rest of the patients.

The finding of an association between recent use and earlier age at onset for bipolar disorder confirms findings from previous research comprising multi-episode patients.[4, 8, 10, 29-33] It supports the hypothesis that cannabis may increase the risk for bipolar disorder, in particular manic and psychotic episodes, interacting with individual genetic vulnerability and other environmental stressors.[33-38] However, since this is a cross-sectional study, we do not know in the case of individual patients if the cannabis use started before, at the same time or after the onset of the first manic and/or psychotic episode. Since this is a first-treatment sample, with a short time period from the first mood episode to time of first treatment, we can, to a large extent, rule out the hypothesis that the more severe course seen in early-onset BD increases the risk for later cannabis use.[39] Furthermore, it is of interest to notice that a recent review did not find support for a role for self-medication in these associations.[40] Another explanation could be that there are common underlying risk factors for both mood instability, psychosis and substance use, in line with what is hypothesized for schizophrenia.[41]

As we did not control for current age in the analyses for statistical reasons, it is not possible to completely rule out that our findings are reflections of higher rates of cannabis use in the 15–24 contra the 25–34 age groups as indicated by European epidemiological data.[42] However, recent Norwegian epidemiological data[43] show smaller age effects. In the current sample, we find no differences in recent cannabis use between the 18–24 and 25–34 age groups.

The findings of an association between cannabis use and suicide attempts may be taken as a support of the view that cannabis negatively affects the course of illness.[33, 44] The findings are in line with studies showing more suicidal behaviour in BD patients with early onset,[45] in particular in the context of cannabis use.[46] The alternative explanation, that more depressive symptoms may provoke cannabis craving, cannot be ruled out. However, this interpretation appears to be less plausible in the current sample as there were no significant differences in depressive symptoms between groups. There was also a lack of difference in the prevalence of psychotic symptoms and other clinical measures between groups. However, there was less use of anti-psychotic medication in the recent use group. This could be due to the known problem of low medication adherence in younger patients in general and in BD patients using cannabis specifically.[47]

Male sex contributed significantly to a lower age at onset for the first manic episode, supporting previous findings of a poorer course of illness in males,[14] and to the association between early onset and more severe symptoms.[48] Cannabis users and BD patients in general are less likely to be in a relationship compared with the population at large.[49-51] Patients currently in a relationship had a later age at onset for manic, depressive and psychotic episodes and also lower levels of cannabis use. The cross-sectional nature of the study makes it difficult to know whether there are common underlying factors to both relational capacities and an early onset, or if patients with early onset have problems with taking on age-appropriate social roles such as entering relationships.

Strengths and limitations

The main strength of the study is the large and well-documented sample of patients in the early course of bipolar disorder, avoiding many of the methodological limitations of previous studies in patients with longer course of illness. The thorough protocol allowed for a systematic approach, discerning the role of possible confounders for the association between recent use and significant patient characteristics and outcomes. The sampling procedure including both inpatients and outpatients with consecutive recruitment from a catchment area-based health service gives the sample a high degree of representativity.[6] However, since Norwegian law and the Regional Committee for Medical Research Ethics do not allow us to access the medical charts of patients that have not consented to research participation, we are unable to give a full account of eligible patients that might not have been referred.

The main weakness is the cross-sectional design that makes it impossible to answer questions about causality. The ascertainments of age at onset and illness history were partly retrospective, based on patients' reports and thus suffer from potential recall bias. The proximity to events in this recent onset sample is still a comparative strength of the study. The patients were included in close proximity to their manic episode, having a mildly depressed mood on group level at point of inclusion, as reflected in the IDS-C scores.

Conclusion

In conclusion, the current study shows that patients with cannabis use at the start of the first treatment for a BD I disorder have an earlier age at onset for their first manic or psychotic episode, and more often have attempted suicide. Longitudinal studies are warranted to questions about causative factors.

Ancillary