The Relationship between Cannabis Use and Cognition in People with Bipolar Disorder: A Systematic Scoping Review

Abstract

Bipolar disorder (BD) and cannabis use are highly comorbid and are each associated with cognitive impairment. Given the prevalence of cannabis use in people with BD, it is important to understand whether the two interact to impact cognitive function. We performed a systematic scoping review to determine what is currently known in this field. We systematically searched PubMed, Embase, CINAHL, Web of Science, and PsycINFO for studies on the relationship between cannabis use and cognition in people with BD or relevant animal models. Six observational human studies and no animal studies met inclusion criteria. Two studies found cannabis use in BD was associated with better performance in some cognitive domains, while three studies found no association. One study found cannabis use in BD was associated with worse overall cognition. Overall, most identified studies suggest cannabis use is not associated with significant cognitive impairment in BD; however, the scope of knowledge in this field is limited, and more systematic studies are clearly required. Future studies should focus on longitudinal and experimental trials, and well-controlled observational studies with rigorous quantification of the onset, frequency, quantity, duration, and type of cannabis use, as well as BD illness features.

1. INTRODUCTION

With increasing legalization and shifting attitudes regarding safety and acceptability, rates of cannabis use in the United States and other parts of the world have risen in recent years (Hasin, 2018; Kraus et al., 2018). Given these higher levels of use, understanding both the positive and negative effects of cannabis is becoming an increasingly important public health concern. Studying cannabis use in psychiatric populations is especially relevant, as mental health conditions are associated with a higher likelihood of cannabis use (Hasin et al., 2016). People with bipolar disorder (BD) are notable for their high rates of cannabis use, with 71% of people with BD reporting a lifetime history of cannabis use compared to 27% of the general population (Agrawal et al., 2011). Furthermore, 17% of people with BD have a lifetime cannabis use disorder (CUD) compared to 6% of the general population, and 7.2% have a past-year CUD compared to 1.2% of the general population (Hasin et al., 2016; Hunt et al., 2016; Lev-Ran et al., 2013). Overall, this research shows that cannabis use is highly prevalent among people with BD. It is therefore important to understand the health effects of cannabis in this population.

Previous research has typically focused on the relationship between cannabis and mood in people with BD. Many people with bipolar disorder report that acute cannabis use temporarily relieves their symptoms and this symptom relief contributes to sustained use (i.e. – the self-medication theory) (Grinspoon and Bakalar, 1998; Gruber, 2012; Weiss et al., 2004). Interestingly, in contrast to acute use, other studies find that chronic use is associated with worse affective symptoms and clinical course (Baethge et al., 2008; de la Fuente-Tomás et al., 2020; Strakowski et al., 2007; Zorrilla et al., 2015). Fewer studies, however, have investigated the effects of cannabis on cognition in BD. Many people with BD exhibit cognitive deficits, including impairments in attention, executive function, fluency, learning, memory, and working memory (Bourne et al., 2013; Kurtz and Gerraty, 2009; Robinson et al., 2006; Young et al., 2019). These cognitive deficits can persist during euthymia (Robinson et al., 2006) and are a source of significant functional impairment independent of affective symptoms (Burdick et al., 2010). Cannabis and its primary psychoactive cannabinoid, tetrahydrocannabinol (THC), have also been associated with cognitive impairment. Many studies find that, in otherwise healthy cannabis users, acute cannabis/THC intoxication impairs some of the same cognitive domains impaired in people with BD, including attention, learning, and memory (Broyd et al., 2016; Kroon et al., 2020; Theunissen et al., 2015). Multiple studies also find an association between chronic cannabis use and impaired attention, learning, and memory in otherwise healthy people (Broyd et al., 2016; Kroon et al., 2020), although recent studies suggest the effect size of these associations is small to moderate (Figueiredo et al., 2020). Alternatively, a few recent studies in otherwise healthy individuals have failed to find a negative association between cannabis and cognition (Meier et al., 2018; Ross et al., 2020), suggesting the relationship between cannabis and cognition is still not fully understood.

Given the high rates of cannabis use among people with BD, understanding the effects of cannabis use in this population is important. Since both cannabis use and BD can impair cognitive functions such as attention, learning, and memory, it is possible that combined cannabis use and BD may further impair performance in these domains. Alternatively, some studies find that cannabis use is associated with better cognition in people with schizophrenia (Schnakenberg Martin et al., 2016; Yücel et al., 2012). Therefore, it is also possible cannabis may be associated with better cognitive function in people with BD. In any case, it would be useful and important to understand the effects of cannabis use on cognition in BD, as this information could guide counseling or treatment for people with BD. The increasing availability of cannabis with legalization makes this matter timely and relevant. A previous review on the relationship between cannabis use and cognition in BD speculated that cannabis may contribute to impaired neuropsychological function in people with BD (Cahill et al., 2006); however, there was little to no research on this topic at that time. Our goal in this study was to perform a scoping review to determine how much and what is currently known about the relationship between cannabis use and cognition in people with BD. We also looked for studies on the effects of cannabis on cognition in relevant animal models, as previous studies have used animal models of bipolar mania to study cognition (Young et al., 2019). Finally, we sought to critically evaluate this literature, identify gaps in knowledge, and suggest directions for future research.

2. METHODS

2.1. Information sources and search strategy

We performed this systematic scoping review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Liberati et al., 2009). Reports were identified by searching PubMed/MEDLINE, Embase, CINAHL, Web of Science, and PsycINFO for articles published through Oct 2019. We searched for reports on the relationship between cannabis and cognition in people with BD or relevant animal models with the help of a librarian experienced in systematic reviews. The search strategies used for each database are provided in Supplemental File 1.

2.2. Eligibility criteria and study selection

Eligible studies included those examining individuals 18 years or older with a bipolar spectrum disorder and current or past cannabis use who underwent cognitive assessment. Cognitive assessments of interest included, but were not limited to, attention, concentration, decision-making, executive function, general intelligence, learning, memory, processing speed, response inhibition, and working memory. We excluded studies with subjects with potentially confounding conditions such as neurological disorders or head/brain trauma. Furthermore, if a study reported including participants with psychiatric conditions other than BD or active substance use disorders (SUD) other than CUDs, then that study was excluded. Studies with BD participants with a history of psychosis were included. We included studies that examined individual cannabinoids such as THC or cannabidiol (CBD) – a cannabinoid with no psychotomimetic effects – on cognition. Our primary interest was comparing BD individuals with current or past cannabis use to BD individuals without current or past cannabis use. To be as comprehensive as possible, we included all types of human studies examining the relationship between cannabis and cognition in BD. We also sought relevant animal studies. There were no restrictions by study setting or geographical location. We included only articles published in or translated into the English language.

2.3. Data collection process and data items

Database search results were uploaded to the internet-based EndNote software where duplicated results were deleted. All unique articles were then uploaded to Rayyan QCRI, an internet-based software that aids with systematic reviews. A level 1 assessment was performed in which two reviewers, TJW and NP, independently read the titles and abstracts of each report. They came to a consensus regarding which reports were eligible for further review based on the predetermined inclusion and exclusion criteria described above. Next, a level 2 assessment was performed in which TJW and NP independently read the full text reports and determined whether they met the inclusion and exclusion criteria. They came to a consensus regarding which reports would be included in data abstraction and risk of bias assessment. TJW and NP independently extracted data from eligible reports using standardized forms. Extracted information included the first author, publication year, diagnoses of the participants, mood status at time of study, medication status at time of study, number of subjects (n), age of onset of BD, duration of BD, number of mood episodes, history of psychosis/number of psychotic episodes, age onset cannabis use, comorbid neuropsychiatric/sleep disorders, comorbid substance use, cannabis use criteria, cognitive assessments performed, and main findings. TJW and NP compared data extraction and came to a consensus. Disagreements were resolved by discussion.

2.4. Risk of bias in individual studies

The risk of bias of individual studies was assessed using the Agency for Healthcare Research and Quality Evidence-based Practice Center Methods Guide (Viswanathan et al., 2012). Studies were rated on selection bias, performance bias, attrition bias, detection bias, and reporting bias as applicable. These ratings were carried out independently by TJW and NP using a standardized form. Consensus regarding risk of bias assessments was reached through discussion. Disagreements were resolved by discussion.

3. RESULTS

3.1. Search results

The initial search results yielded 827 reports (Fig 1). Once duplicates were removed, 668 unique articles remained. The majority of reports (n=650) did not meet the basic inclusion criteria of studying the relationship between cannabis and cognition in BD. Most articles addressed one or two, but not all three of the main concepts of cannabis, cognition, and BD. For example, many articles were excluded for examining the relationship between cannabis and cognition in people with schizophrenia rather than people with BD. Of the 668 unique articles, 18 (2.7%) were considered relevant based on a Level 1 title and abstract screening. Of these, six articles (0.9%) were selected as meeting eligibility criteria based on a Level 2 full text screening (Fig 1). The most common reasons for excluding articles at level 2 included the report being a conference abstract that duplicated a published study and the report not specifically assessing cognition.

Figure 1.

PRISMA Flowchart of Study Selection

3.2. Human studies

We found six observational studies that examined the association between cannabis use and cognition in people with BD (Table 1).

Table 1.

Data extraction from studies examining the association between cannabis use and cognition in people with bipolar disorder

Author/Year	Subject Diagnosis	Mood Status at Time of Study	Medication	Cannabis Use Criteria	N	Cognitive Assessments	Main Findings
Abush et al., 2018	Psychotic BD I; diagnosis established by consensus of at least two experienced research clinicians based on history, SCID-IV, YMRS, and MADRS	BD+/C− average MADRS = 12.0 (mild depression), average YMRS = 7.6 (normal); BD+/C+ average MADRS = 8.3 (mild depression), average YMRS = 6.1 (normal)	Participants taking any psychotropic medication: BD+/C−: 92%, BD+/C+: 100%; FGA: BD+/C−: 8%, BD+/C+: 43%; SGA: BD+/C−: 75%, BD+/C+: 43%; Lithium: BD+/C−: 17%, BD+/C+: 14%; Other mood stabilizer: BD+/C−: 33%, BD+/C+: 71%; Antidepressant: BD+/C−: 42%, BD+/C+: 57%; Sedative/anxiolytic: BD+/C−: 33%, BD+/C+: 57%; Anticholinergic: BD+/C−: 8%, BD+/C+: 43%; Side effects and observance not reported	Adolescent use: 5 or more occasions of cannabis use before 18; average duration use = 20 years; 86% had more than 50 lifetime uses; quantity, type, and reason/motivation for use not specified	HC = 32. C+ = 14. BD+/C− = 12, BD+/C+ = 7	WRAT 4, BACS Composite, Verbal Memory, Digit Sequencing, Token Motor, Verbal Fluency, Symbol Coding, Tower of London	BD+/C+ equal to BD+/C− in all assessments
Braga et al., 2012	BD I; diagnosis based on consensus of at least three diagnosticians based on SCID-IV, medical records, and interviews with families	Not clearly specified; 60% BD+/C− “clinically symptomatic”; 46% BD+/C+ “clinically symptomatic”; polarity not specified	Mean number medications: BD+/C−: 2.9, BD+/C+: 3.5; Antipsychotics: BD+/C−: 74%, BD+/C+: 94%; Anticonvulsant use: BD+/C−: 57%, BD+/C+: 59; Lithium: BD+/C−: 33%, BD+/C+: 37%; Side effects and observance not reported	History of cannabis use disorder i.e. - cannabis abuse/dependence, ranging from current to full sustained remission; duration, frequency, quantity, type, and reason/ motivation for use not described	BD+/C− = 150, BD+/C+ = 50	WRAT-3, Trail Making Parts A and B, WAIS-Revised: Digit Span Forward and Backward, CVLT-Abridged, COWAT/Letter Fluency, Animal Naming/Category Fluency	BD+/C+ better than BD+/C− in Trail Making Part B (processing speed/working memory), Digits Forward (attention), and Digits Backward (attention/working memory)
Halder et al., 2016	BD; diagnosis determined by two psychiatrists based on DSM-IV criteria	All subjects acutely symptomatic: BD+/C−: 43% depressed, 10% hypomanic, 33% manic, 13% mixed; BD+/C+: 30% depressed, 13 hypomanic, 40% manic, 17% mixed	Not mentioned; Side effects and observance not reported	Cannabis dependence; duration, frequency, quantity, type, and reason/motivation for use not described	Each group = 30	Animal Naming, Fruit Naming, Word Naming, Trail Making Test Part A and B, Stroop Word, Stroop Color, Clock Drawing Test	BD+/C+ worse than BD+/C− in all assessments, except Stroop Word
Lagerberg et al., 2019	BD I, BD II, and BD NOS; process of diagnosis not explained	Not mentioned	Not mentioned; Side effects and observance not reported	Premorbid and/or current cannabis use; duration, frequency, quantity, type, and reason/motivation for use not described	BD+= 380	Not mentioned	BD+/C+ equal to BD+/C− in all assessments
Ringen et al., 2010	BD I, BD II, and BD NOS; diagnoses established by trained clinicians based on SCID-IV	Not clearly specified; all patients were “symptomatically stable”	Not clearly specified; some subjects in both groups were taking medication; Side effects and observance not reported	Cannabis use in past 6 months; duration, frequency, quantity, type, and reason/motivation for use not described	BD+/C− = 115, BD+/C+ = 18	NART Error Score, PAS-Academic Functioning, WASI-IQ, WAIS-III: Digit Symbol Test, WAIS-III: Digits Forward and Backward, WM-MA: 2-Back, D-KEFS: Verbal Fluency and C-W Interference, WMS-III, CVFT-II	BD+/C+ better than BD+/C− in D-KEFS: Semantic Verbal Fluency
Sagar et al., 2016	BD I; process of diagnosis not clearly described	BD+/C− average MADRS = 7.2 (mild depression), average YMRS = 4.0 (normal); BD+/C+ average MADRS score = 10.6 (mild depression), average YMRS = 5.6 (normal)	Any psychotropic medication: BD+/C−: 83%; BD+/C+: 92%; Mood stabilizers: BD+/C−: 72%, BD+/C+: 75%; Antidepressants: BD+/C−: 28%, BD+/C+: 17%; Antipsychotics: BD+/C−: 56%, BD+/C+: 58%; Benzodiazepines: BD+/C−: 17%, BD+/C+: 8%; Side effects and observance not reported	Chronic users reporting smoking a minimum of 2500 times in their lives, smoking 4 out of 7 last days and testing positive for urinary cannabinoids; average duration use = 6.8 years; average frequency use = 15.5 smokes/week; average quantity of use = 5.2 grams/week; type and reason/motivation for use not described	HC = 21, MJ = 23, BD+/C− = 18, BD+/C+ = 12	WASI, WCST, Stroop Test, Trail Making Parts A and B, COWAT, WAS-R: Digit Forward and Backward, ROCF, CVFT-II, HVOT	BD+/C+ equal to BD+/C− in all assessments
Author/Year	Age Onset BD (Mean ± SD)	Duration BD (Mean ± SD in years)	Number of Mood Episodes	History of Psychosis/Number of Psychotic Episodes	Age Onset Cannabis Use	Comorbid Neuropsychiatric/Sleep Disorders	Comorbid Substance Use
Abush et al., 2018	Not described	Not described	Not described	Not described	BD+/C−: 18.7; BD+/C+: 12.6	Not described	Substance abuse within 1 month or dependence within 3 months of enrollment was exclusionary; drug free status confirmed by urine toxicology at initial appointment and at the beginning of the study
Braga et al., 2012	BD+/C−: 22.5 ± 8.4; BD+/C+: 21.3 ± 7.2	BD+/C−: 15 ± 11; BD+/C+: 13 ± 11	Not described	History of psychosis: BD+/C−: 67%; BD+/C+: 82%*; number of episodes not described	Not described	Not described	Alcohol misuse: BD+/C−: 36/150; BD+/C+: 29/50*
Halder et al., 2016	BD+/C−: 25 ± 6; BD+/C+: 22 ± 6	Not described	BD+/C−: 3.7 ± 1.3; BD+/C+: 3.1 ± 1.1;	Not described	Not described	Not described	Not described
Lagerberg et al.,2019	Not described	Not described	Not described	Not described	Not described	Not described	Not described
Ringen et al.,2010	Not described specifically for BD participants	Not described specifically for BD participants	Not described specifically for BD participants	Not described specifically for BD participants	Not described specifically for BD participants	BD+/C+ had lower anxiety ratings than BD+/C−	No substance use other than cannabis during previous 6 months
Sagar et al.. 2016	BD+/C−: 18.1 ± 4.5, BD+/C+: 15.2± 3.2	Not described	Not described	Not described	BD+/C−: 16.4 ± 2.3, BD+/C+: 16.9 ± 2.6	Any DSM-IV Axis I pathology was exclusionary	No more than 15 lifetime uses of any illicit drug; Alcohol use (days/month): BD+/C−: 3.9 ± 5.7; BD+/C+: 5.2 ± 4.7 BD+/C+ had more nicotine use than BD+/C−* (quantitative data not provided)

BACS=Brief Assessment of Cognition in Schizophrenia, BD=bipolar disorder, C=cannabis, COWAT=Controlled Oral Word Association Test, CVLT=California Verbal Learning Test, C-W=Color-Word, D-KEFS=Delis-Kaplan Executive Function System, DSM-IV=Diagnostic and Statistical Manual of Mental Disorders – Edition IV, FGA= First generation antipsychotic, HC=healthy control, HVOT=Hooper Visual Organization Test, MADRS= Montgomery Asberg Depression Rating Scale, MJ=marijuana, NART=National Adult Reading Test, NOS=not otherwise specified, PAS=Premorbid Adjustment Scale, ROCF=Rey-Osterrieth Complex Figure test, SCID-IV= Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders IV, SGA=Second generation antipsychotic, WAIS=Wechsler Adult Intelligence Scale, WASI=Wechsler Abbreviated Scale of Intelligence, WCST=Wisconsin Card Sorting Test, WM=working memory, WM-MA=Working Memory-Mental Arithmetic, WMS=Wechsler Memory Scale, WRAT=Wide Range Achievement Test, YMRS= Young Mania Rating Scale,

^*= BD+/C− vs. BD+/C+ p < 0.05 for that measure

Two of these reviewed studies concluded that a history of cannabis use in people with BD was associated with better performance in certain cognitive domains. For example, Braga et al. examined the association between cognition and a CUD history in people with BD (Braga et al., 2012). This study found that a history of CUD was associated with better attention, working memory, and processing speed compared to no history of CUD in people with BD (Table 1). A different study compared people with BD who used cannabis in the past six months (meeting criteria for a CUD was not required in this study) to people with BD who did not use cannabis (Ringen et al., 2010). Results showed that cannabis use in the past six months was associated with better verbal fluency and a trend toward better verbal learning in people with BD (Table 1). Unlike Braga et al. (Braga et al., 2012), Ringen et al. did not find that cannabis-using people with BD performed better on attention, processing speed, or working memory measures compared to non-cannabis using people with BD (Ringen et al., 2010). Overall, these two studies suggest a history of cannabis use in BD may be associated with better performance in certain cognitive domains, although the results of the studies differed with respect to the specific cognitive domains affected.

Three studies concluded there was no association between cannabis use and cognition in people with BD. One study focused on adolescent cannabis use and cognition in adulthood (Abush et al., 2018). This study found that adolescent cannabis use was not associated with different cognitive performance in people with BD during adulthood. Cognitive domains assessed included general intelligence, learning, memory, attention, working memory, motor speed, verbal fluency, processing speed, and executive function (Table 1). A second study examined people with BD with chronic cannabis use (Sagar et al., 2016). In this study, the cannabis-using people with BD performed no differently than non-cannabis using people with BD in general intelligence, executive function, attention, processing speed, working memory, verbal fluency, learning, memory, and visuoperception (Table 1). Finally, we also identified an abstract reporting that individuals with BD and premorbid or current cannabis use had similar cognitive performance to non-cannabis using people with BD (Lagerberg et al., 2019) (Table 1). Overall, these reports indicated that a history of cannabis was not associated with cognitive changes in people with BD.

Finally, one study found that cannabis dependence was associated with worse performance in multiple cognitive domains in people with BD (Halder at al. 2016). This study was unique among the six, as it was the only study to examine only acutely symptomatic participants. Results showed that participants with BD and current cannabis dependence had worse verbal fluency, attention, processing speed, working memory, executive function and global cognition than non-cannabis-dependent people with BD (Table 1). Therefore, cannabis dependence in people with acutely symptomatic BD may worsen performance in certain cognitive domains.

3.3. Animal studies

No animal studies were found that examined the relationship between cannabis or cannabinoids and cognition in models relevant to BD. Overall, the scarcity of animal studies highlights the need for further animal research in this area.

3.4. Risk of bias in individual studies

Studies were rated on selection bias, performance bias, attrition bias, detection bias, and reporting bias as applicable. A thorough review of the bias in these individual studies can be found in the Supplementary File 2. The studies varied in their selection and performance bias, with Braga et al. and Ringen et al. having a low risk of these biases (Braga et al., 2012; Ringen et al., 2010). There was concern Halder et al. may have a high risk of selection bias because they discussed neither the percentage of men and women in their groups, nor the medication status of their groups (Halder et al., 2015). It was therefore unclear whether differences in these factors could have contributed to their positive results. The risk of selection and performance bias could not be determined in the other studies because it was unclear whether the analyses controlled for important confounding and modifying variables or whether the strategies for recruiting participants into the study differed across study groups. Attrition bias was not a concern across these six studies given their cross-sectional nature. Detection bias was a concern across all studies due to failure to discuss blinding. Finally, reporting bias could not be fully determined due to lack of registered protocols, although all studies examined the endpoints described in their methods.

4. DISCUSSION

4.1. Human studies

We identified six observational human studies comparing cognition in BD individuals with and without cannabis use. Results of the six studies varied, with two finding cannabis was associated with better performance in certain cognitive domains, three finding no association, and one finding cannabis was associated with worse cognition. The different results between the studies may be due to multiple factors. Perhaps the most striking difference between the studies is their different criteria for cannabis use. Average duration, frequency, quantity, and type of cannabis use were not specified for most of the studies. It is possible differences in these parameters may be associated with discrepant effects on cognition. Also, some studies examined people with a CUD, while others simply examined participants with previous, current, or chronic cannabis use. There may be different effects of cannabis use versus a CUD on cognition. Indeed, progressive neuroadaptations are thought to occur in the transition from substance use to a SUD (Koob and Volkow, 2016). Therefore, different underlying neurobiologies may contribute to different effects on cognition; however, in our review, the two studies examining people with CUDs found opposite associations. In one, a CUD was associated with better cognition (Braga et al., 2012), while in the other, a CUD was associated with worse cognition (Halder et al., 2015). Furthermore, the studies focusing on cannabis use also found different results. Overall, these findings suggest the impact of cannabis use versus a CUD on cognition in people with BD is complicated and will require further research to understand. Furthermore and importantly, many studies allowed for varying abstinence periods between last cannabis use and cognitive testing that may have contributed to contrasting outcomes. Abstinence from cannabis use is associated with normalized cognition in healthy controls (Schreiner and Dunn, 2012; Scott et al., 2018). Therefore, prolonged abstinence in these BD studies may have allowed for normalized cognition. Also, because many studies did not verify the last time cannabis was used before cognitive testing, it is possible that withdrawal or even intoxication effects contributed to the variation in results. Furthermore, the type and potency of the cannabis used is also not reported. The primary cannabinoids found in cannabis, such as THC and CBD, are known to have different effects on cognition (Morgan et al., 2018), therefore, use of different cannabis strains with varying levels of cannabinoids may have diverse effects on cognition in BD. Overall, the marked differences in definitions of cannabis use likely contributed to the contrasting results of these six studies.

Other variations between the studies may have contributed to the differing results. Firstly, the studies included different subtypes of BD, such as bipolar disorder I (BDI), bipolar disorder II (BDII), bipolar disorder not-otherwise-specified (BD NOS), and psychotic BD. A recent meta-analysis found cognitive differences between subtypes of BD, with BDI being more impairing than BDII, and with psychotic BD being more impairing than non-psychotic BD (Bora, 2018); The differences, however, were subtle and the authors suggested other factors, such as age of onset of BD, may account for these results. The relationship between cannabis and cognition in psychotic BD is particularly interesting. Cannabis use is associated with psychotic illness (Hasan et al., 2020), and psychotic disorders are associated with cognitive impairment (McCleery and Nuechterlein, 2019); however, cannabis use is often associated with better cognition in people with schizophrenia (Schnakenberg Martin et al., 2016; Yücel et al., 2012). One of our reviewed studies examined only participants with psychotic BD and, unlike studies in schizophrenia, found no association between cannabis use and cognition in this population (Abush et al., 2018). Another study included many people with a history of psychosis and found that cannabis use was associated with better cognition in certain domains, but controlled for psychosis history in its analysis (Braga et al., 2012). None of the six studies reviewed here directly compared the association between cannabis use and cognition in people with psychotic versus non-psychotic BD. Overall, how a history of psychosis, or BD subtype, may impact the relationship between cannabis use and cognition in BD is likely complex and warrants further study, especially given the high prevalence of psychotic symptoms in people with BD.

Studies also differed on the mood status of the participants included. Mood status is a relevant factor, as previous research finds different mood states are associated with different patterns of cognitive dysfunction (Kurtz and Gerraty, 2009; Martínez-Arán et al., 2004). These six reviewed studies mostly examined stable patients, with one study including some patients who were symptomatic (Braga et al., 2012), and one study examining only acutely symptomatic participants (Halder et al., 2016). Halder et al. was also the only study to find that cannabis was associated with worse cognition in people with BD (Halder et al., 2015). It is possible cannabis dependence interacted with acutely symptomatic BD to worsen cognition, an effect not seen in stable patients.

Also, some studies did not report the number or types of medications used. The medications used to treat BD have varying effects on cognition (Xu et al., 2020), making it possible that differences in medications may have contributed to different study outcomes. Also, these six reviewed studies included both men and women, but most did not examine sex differences. Only Braga et al. examined sex differences and found no effect of sex on any of the neurocognitive measures (Braga et al., 2012). Differences in the proportion of men versus women in these studies may have contributed to the varying outcomes, as previous studies find that cannabis has different effects in people with BD depending on sex (de la Fuente-Tomás et al., 2020). Furthermore, previous research finds that disease characteristics such as duration of BD illness and number of mood episodes may be related to cognitive impairment (although not all studies find such associations) (Cullen et al., 2016). Most studies did not report these disease characteristics, making it possible that differences in these measures could have contributed to varying results. Many studies also did not report whether there were differences in comorbid neuropsychiatric disorders or comorbid substance use between groups. Future studies should rigorously control for these variables.

Among all the outcomes of these studies, the possibility that cannabis is associated with better cognition in BD is particularly interesting and is reminiscent of studies finding cannabis use is associated with better cognition in people with schizophrenia (Schnakenberg Martin et al., 2016; Yücel et al., 2012). Braga et al. suggest cannabis may be associated with improved cognition because greater cognitive abilities are necessary to acquire and continuing using cannabis (Braga et al., 2012). Ringen et al. speculate that cannabis use may decrease anxiety, thereby improving cognitive function, noting that the BP+/C+ group in their study had lower anxiety ratings than the BP+/C− group (Ringen et al., 2010). Additional research finds that acute cannabis use temporarily improves other BD symptoms such as tension and racing thoughts (Sagar et al., 2016; Weiss et al., 2004), which may contribute to improved cognition. Finally, cannabinoids possess anti-inflammatory properties (Atalay et al., 2019; Nagarkatti et al., 2009). Since cognitive impairment in BD is associated with inflammation (Millett et al., 2019), it is possible cannabinoids have a neuroprotective effect by reducing inflammation; This hypothesis, however, remains to be tested.

4.2. Animal studies

There were no reports examining the relationship between cannabis or cannabinoids and cognition in animal models relevant to BD, showing that more work in this area is needed. Previous studies have used animal models of bipolar mania, such as dopamine transporter (DAT) knockdown mice, to assess cognition (van Enkhuizen et al., 2014; Young et al., 2019). Other animal models, such as the ClockΔ19 mouse (Kristensen et al., 2018) or the amphetamine-sensitization model (Fries et al., 2015) may also be useful in studying the relationship between cannabis and cognition in BD. Experiments in relevant animal models would provide an important complement to human studies. For example, the frequency, duration, dose, and type of cannabis or cannabinoid administration could be precisely controlled in an animal study and allow for causal inferences to be drawn. Animal studies would also enable testing of mechanistic hypotheses that could not be done in humans, and provide the opportunity to test novel therapeutics to treat cognitive deficits. Alternatively, human studies are also needed to complement the animal studies. No current animal model is a true model of bipolar disorder and cannot capture the complexity of the disease in humans. Furthermore, human studies would be important for investigating how motivation for use (i.e. – recreational vs. medical) is related to cognition with cannabis use.

4.3. Risk of bias assessment

The reviewed studies varied in their levels of bias. Braga et al. and Ringen et al. were relatively low in overall bias, suggesting we can view their results with a higher level of confidence (Braga et al., 2012; Ringen et al., 2010). Interestingly, both Braga at el. and Ringen et al. found that cannabis use was associated with better function in certain cognitive domains in people with BD. Halder et al. had a higher risk of bias due to concerns regarding selection bias (Halder et al., 2015). The sex and medication status of the groups in this study were not reported, leaving it unclear whether these factors could have contributed to the positive findings. Notably, this was the only study finding cannabis was associated with worse cognition in BD. All other studies had an unclear overall risk of bias and found no association between cannabis use and cognition in BD. Therefore, our risk of bias assessment indicates that the literature currently supports cannabis not being associated with markedly worse cognition in BD.

4.4. Future research

Our systematic review suggests many areas for future research. All of the studies identified were cross-sectional studies. An experimental study would provide insight into causality regarding the effects of cannabis on cognition in BD. Future studies should also consider the effects of cannabis or cannabinoids on neurotransmitter systems involved in cognition in people with BD, such as the acetylcholine or dopamine system. Such a direction would provide insight into the underlying mechanisms of the impact of cannabis on cognition in BD. Another area open for research is a well-controlled observational study with more rigorous quantification of the onset, frequency, quantity, duration, and type of cannabis used, as well as the reasons for use. Indeed, recent research finds that the reasons for cannabis use may have an important impact on cognition. A recent review notes that chronic, heavy, recreational use (i.e. – using cannabis solely to achieve an altered state of consciousness, usually euphoric or pleasurable) is associated with cognitive impairment. Alternatively, medical use (i.e. – using cannabis to alleviate symptoms of a disease, such as pain) may actually be associated with improved cognition (Sagar and Gruber, 2018). This may be because cannabis used for recreational purposes tends to be high in THC, while cannabis used for medical purposes is often higher in CBD, and previous studies find THC impairs certain cognitive functions, while CBD does not (Morgan et al., 2018). Therefore, investigating the reason or motivation for use will be important in future studies. It was also unclear from the identified studies how much time elapsed between last cannabis use and cognitive testing. Therefore, a study that defines a standardized time post-use for cognitive testing that avoided both intoxication and withdrawal effects is important. Studying the effects of individual cannabinoids such as THC and CBD on cognition in BD would provide more detailed insights into the effects of cannabis in this population. Also, the study by Halder et al. suggests cannabis use may have different effects on cognition in BD depending on the affective state (Halder et al., 2015). Therefore, another interesting study could compare the relationship between cannabis use and cognition in states of mania and depression. It is worth noting, however, that acute mood episodes, in addition to impairing cognition, may motivate cannabis use (Cahill et al., 2006). Therefore, the negative association between cannabis use and cognition during mood episodes may not be causal, but rather due to their common relationship to the mood episode itself. Future studies should also rigorously control for age of onset of BD, BD duration, number of mood and psychotic episodes, and comorbid neuropsychiatric disease and substance use. Finally, we sought studies that focused on cannabis use among people with BD. However, people with BD often use a variety of substances (Hayley et al., 2017; Hunt et al., 2016). A study examining the effects of polysubstance use on cognition in BD would be worthwhile future direction. Focusing solely on cannabis use in BD likely restricted our study to a narrow segment of the BD population and excluded many real-life cases of polysubstance abuse.

4.5. Limitations

The primary limitation of this review is the number of studies that met inclusion and exclusion criteria. Only six studies were identified, which weakens the conclusions that can be drawn regarding the association between cannabis use and cognition in BD. Also, given that each of the identified studies was observational, it is impossible to make causal inferences. The significant heterogeneity of cannabis use patterns among the participants in these six studies also makes it difficult to draw clear conclusions. Several other differences between the studies, such as the subtypes of BD included, mood status of the participants, medication status, and BD disease characteristics contribute to difficulty in determining the association between cannabis and cognition in BD. Importantly, some studies may have included participants with psychiatric conditions other than BD or CUD and not reported it. We only excluded studies if they explicitly mentioned participants with a neuropsychiatric condition other than BD (e.g. – one study included people with major depressive disorder) or an active SUD other than a CUD (e.g. – one study mentioned a woman with an alcohol use disorder). Some studies did not explicitly mention excluding participants with other psychiatric conditions. We included such studies in order to be as comprehensive as possible; however, this means that participants of these studies may have had other psychiatric conditions or SUDs that influenced their results. Also, four of the six studies were judged to have an unclear or high risk of bias, which further limits the strength of the review. Since only two of the reviewed studies were judged to have a low risk of bias, this narrows the amount of quality evidence being considered from a set of already limited data. Finally, no relevant animal studies were identified.

4.6. Conclusion

This systematic scoping review sought to determine how much and what is known about the relationship between cannabis and cognition in people with BD. We identified a small number of studies, most of which suggest cannabis use was not associated with significantly worse cognition in BD. However, given the relative lack of studies in this field, the clinical implications of this research currently remain unclear. The lack of knowledge in this area highlights the clear need for further research. Such studies should focus on longitudinal or experimental designs that will enable conclusions to be drawn regarding causality. Well-controlled observational studies with rigorous quantification of the onset, duration, amount, type, and frequency of cannabis use, as well as the reasons for use are also essential. Given the extensive cannabis use among individuals with BD and the increasing availability of cannabis, studying the relationship between cannabis and cognition in BD is becoming increasingly important. There are many gaps in our knowledge of this area, with ample opportunities for future research.

Supplementary Material

1

File 1. Search strategies used in the various databases for this study.

2

File 2. Detailed risk of bias assessment for identified studies.

HIGHLIGHTS:

• Cannabis use is highly prevalent among people with BD

• Much remains unknown about the health effects of cannabis use in BD

• This study reviewed the relationship between cannabis use and cognition in BD

• We found six observational human studies on cannabis use and cognition in BD

• Most studies found cannabis is not associated with markedly worse cognition in BD

• Given mixed findings and unanswered questions, additional studies are needed

Abbreviations:

BD

bipolar disorder

BDI

Bipolar I Disorder

BDII

Bipolar II Disorder

BD NOS

Bipolar Disorder Not Otherwise Specified

CBD

cannabidiol

CUD

cannabis use disorder

DAT

dopamine transporter

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

SUD

substance use disorder

THC

tetrahydrocannabinol