Association between Cannabis Use Disorder and Mental Health Disorders in the Adolescent Population: A Cohort Study

Introduction: According to the literature, early initiation to cannabis use and a dependent pattern of use are important risk factors for the development of mental health disorders. However, there are few cohort studies which look at the development of mental health disorders associated with cannabis use among young people with cannabis use disorder (CUD). The aim of the study was to determine the cumulative incidence of mental health disorders and the risk of developing mental health disorders among minors who commenced treatment for CUD in Catalonia during 2015–2019. Methods: This was a retrospective fixed cohort study, matched for confounding variables, based on data from the Catalan Health Surveillance System. The exposed cohort comprised young people who entered treatment for CUD during 2015–2019 (n = 948) and who were minors on the date of commencing treatment. Matching was done with a paired cohort (n = 4,737), according to confounding variables. Individuals with a diagnosis of a mental health disorder prior to the study period were excluded. The cumulative incidence was calculated for mental health disorders for the exposed and the paired cohorts and stratified by type of mental disorder. Incidence rate ratios were estimated using the conditional Poisson model with robust variance, stratified by sex. Results: The cumulative incidence for development of a mental health disorder was 19.6% in the exposed cohort and 3.1% in the paired cohort; with higher incidence among females (females 32.7%; males 15.8%). The exposed cohort had an 8.7 times increased risk of developing a mental health disorder than the paired cohort. The most frequent diagnoses were reaction to severe stress, adjustment disorder, and personality disorders. Conclusion: This study confirmed that the exposed cohort was at increased risk of developing mental health disorders compared to the paired cohort. To date, few studies have analyzed the association between cannabis use and the development of mental health disorders, considering cannabis dependence. Further studies should be undertaken considering CUD. In addition, more studies are needed to understand the factors that determine the development of CUD. Further research in these areas would contribute to the design of prevention strategies aimed at those young individuals with a higher risk of developing cannabis dependence and suffering its consequences.


Introduction
Cannabis is the most consumed illicit drug in the world, with particularly high rates of use among adolescents [1].The European Drug Report 2020 showed that approximately 15% of young people aged between 15 and 34 years used cannabis in the past year [2].In Catalonia, according to the 2018 Spanish Survey on Drug Use in High Schools, cannabis is the most consumed illegal drug among 14-to 18-year-old students.The survey found that 39.4% of this population had used cannabis at least once in their life, 23.1% reported using cannabis in the month prior to the survey, and 4% showed risky cannabis use according to the cannabis abuse screening test [3].Furthermore, in 2020, cannabis was the main substance of use in 94.7% of admissions to treatment for substance use disorders in minors in Catalonia [4].
Scientific evidence supports a relationship between drug use and mental health disorders, especially among adolescents and young adults [5,6].The risks of developing mental disorders among cannabis users are associated with genetic predisposition [7], higher frequency of use, and initiation to use at a young age [8].Adolescence is a stage of neural and cerebral development.Therefore, young cannabis users could be more vulnerable to the effects of tetrahydrocannabinol, the principal active ingredient in cannabis [9,10].Early initiation to cannabis use and regular use are predictors of greater risk of dependency and of developing mental health disorders such as psychosis, mood disorders, and antisocial behaviors [8][9][10][11].
Regarding the association between cannabis use and mental disorders, the association with psychotic disorders has been the most studied, with high rates of cannabis use disorder (CUD) seen among people experiencing a first psychotic episode.Substance use disorders are observed in 20-60% of people experiencing psychotic episodes, with CUD being the most common [12].In addition, numerous studies suggest a temporal association between age at initiation to cannabis use and age of onset of psychotic illness [13,14].The association between cannabis use and affective disorders is less consistent [15][16][17][18].However, various studies support an association between cannabis use and depression [19], highlighting age of initiation to use as a risk factor, observing a greater association among females [20,21].Bipolar and personality disorders may also be associated with cannabis use, although the evidence is less consistent [22].Regarding the association between cannabis use and attentiondeficit hyperactivity disorder (ADHD), the evidence is even less consistent.Some prospective studies have shown that use of cannabis increased the risk of developing ADHD in adulthood [23,24].However, the evidence is not as strong.Regarding risk of starting cannabis use, numerous studies suggest that having a mental health disorder, especially externalizing disorders, is a risk factor for starting cannabis use and for ongoing use [25,26].
To date, most research is derived from general population studies and does not analyze the differences between frequent cannabis users with and without dependence.Therefore, dependent frequent users could have magnified the observed association between cannabis use and the development of mental health disorders.According to the literature, those who use cannabis differ from the general population in terms of sex, age, nationality, education, employment status, and experience of childhood adversity.However, those differences have not been found between dependent and nondependent cannabis users.In the literature which does consider frequent and dependent use, frequent users have been found to be at higher risk for any mental disorder than the general population and dependent users have been found to have higher rates of mental disorders than nondependent users [27].
Therefore, early cannabis use and a dependent pattern of use are important risk factors in the development of mental health disorders.However, there are few cohort studies which look at the development of mental health disorders associated with cannabis use among young people with CUD.
In Catalonia, cannabis accounts for more than 90% of treatment initiations for substance use disorders in minors [4].Therefore, the objective of this study was to estimate the incidence of mental health disorders and the risk of developing these, after initiation to cannabis use, in minors who sought treatment for CUD in Catalan Drug Dependence Centres (CAS) during 2015-2019.

Materials and Methods
This was a retrospective fixed cohort study, paired for the following confounding variables: age, sex, socioeconomic level, and degree of urbanization of the place of residence (e.g., mountains, rural, semi-urban, urban) [5,25].

Study Population
The exposed cohort was selected from the Catalan Information System on Drug Dependence (SIDC).It comprised individuals under 18 years who commenced treatment for CUD during 2015-2019, in Catalonia's CAS (n = 1,448).Those with a mental health diagnosis prior to initiation of cannabis use were excluded (n = 500).Therefore, the final exposed cohort was comprised of 948 individuals.The paired cohort was identified from the Central Register of Insured Persons (RCA).This cohort was paired by the following confounding variables: year of birth, socioeconomic level, and degree of urbanization of their residence.The paired cohorts were first paired using a ratio of 1:8.Following exclusion of those with a mental health condition diagnosed prior to the study period, 5 people were randomly selected for each exposed person.The final paired cohort comprised 4,737 people.Therefore, there were 946 matchings using a ratio of 1:5, one matching with a ratio of 1:4, and one with a ratio of 1:3 (Fig. 1).The exposed person's year of initiation to cannabis use was assigned as the beginning of the study period.

Data Set
Study data were obtained from three sources.First, the SIDC database; from this, we obtained data related to the sex and age of individuals who commenced treatment for CUD; age at first use of cannabis; and treatment admission date.
Second, the Catalan Health Surveillance System (CHSS) comprising the RCA and the Registry of the Minimum Basic Dataset (CMBD).The RCA is an automated database that manages individual health identifiers and includes 99.6% of the population of Catalonia.From the RCA, we were able to obtain individual-level information (age and sex) related to the paired cohort.This database also records individual pharmacy co-payment information and economic benefits received through the social security system; this information was used to define the socioeconomic position (SEP) for each individual and the municipality of residence of the paired and exposed cohort [28].The CMBD contains information on diagnoses made in the health system, coded according to the International Classification of Disease (ICD-10) [29].Each type of medical service has a CMBD, and those included in the study were the CMBD of primary healthcare, the CMBD of outpatient mental health centers, the CMBD of psychiatry outpatient resources, and the CMBD of psychiatric hospitals.This allowed us to identify mental health diagnoses made before and after the start of the study period.
Finally, from the Statistical Institute of Catalonia (IDES-CAT), information was sourced on the population registered in each municipality (2015), allowing us to identify the degree of urbanization of each participant's place of residence [30].All data were linked using an anonymous individual health identifier.

Independent Variables
The individual sociodemographic variables were sex (male or female), age (continuous variable), nationality (Spanish or not Spanish), SEP (very low or others).Very low SEP was defined as people exempt from the pharmaceutical co-payment, those receiving unemployment benefits, and those receiving a noncontributory pension [31].
The contextual sociodemographic variable was the degree of urbanization of residence as defined by the Spanish National Institute of Statistics [32].This was categorized by four groups as follows: mountain (≤5,000 residents), rural (>5,000 and ≤20,000 residents), semi-urban (>20,000 and ≤100,000 residents), and urban (>100,000 residents).
Variables related with CUD were age at first use of cannabis, age at commencement of treatment, and time between first use and treatment initiation.All were expressed in years as continuous variables.
The follow-up period for the exposed cohort was from the year of first use of cannabis, reported at treatment initiation, until December 31, 2019.The follow-up period for the paired cohort was the year of first use of cannabis of their corresponding exposed person until December 31, 2019.

Dependent Variables
The dependent variable was a diagnosis of a mental health condition during the study period.The mental health diagnoses included were those conditions associated with cannabis use in the relevant literature [6,8,19] (see the online suppl.File; for all online suppl.material, see https://doi.org/10.1159/000530331),with the exception of gambling addiction and substance use disorders.Mental health diagnoses were obtained from the CMBD of outpatient mental health centers, the CMBD of psychiatry outpatient services, and the CMBD of psychiatric hospitals.The primary healthcare CMBD was not used because it is not a specialized mental health service.

Data Analysis
We performed a descriptive analysis of independent variables for both cohorts.The results are presented in absolute and relative frequencies for categorical variables and with medians and interquartile ranges (IQRs) for continuous variables.All results were stratified by cohorts and by sex.Statistically significant differences were determined by p values with the χ 2 test for categorical variables and the Mann-Whitney test for continuous variables.
To study the association between CUD and the development of mental health conditions, we calculated the cumulative incidence of developing a mental health condition in the exposed group.For both cohorts, we also calculated the cumulative incidence for each of the specific mental health conditions considered in the study.Finally, a conditional Poisson regression with robust variance was done, stratified by sex [33][34][35].This allowed estimating incidence rate ratios (IRRs).To analyze the proportion of mental health disorders in the exposed cohort associated with CUD, we calculated the attributable risk percent for the exposed cohort.The confidence interval was 95% (95% CI).The significance level was set at 5%.Stata software, version 15 was used.

Description of the Exposed Cohort
Table 1 shows the sociodemographic characteristics of those in the exposed cohort.The cohort was 77.4% male, with a median (IQR) age of 16.7 years (15.8-17.4) and a median (IQR) age at first use of cannabis of 14 years (13-15); 80% (76.6% of males; 91.5% of females) had Spanish nationality; 6.9% (7.2% of males; 5.6% of females) had a very low SEP; and 76% lived in urban or semi-urban areas.
The males who commenced treatment were slightly older than the females, with a median (IQR) age of 16.8 years (15.9-17.4) and 16.2 years (15.4-17.1),respectively.The median (IQR) number of years between first use of cannabis and entry to treatment was two (1-3).

Conditional Poisson Analysis
Analysis of the association between CUD and development of mental health disorders showed that the exposed cohort had 8.7 (95% CI: 6.8-11.0;p < 0.0001) times increased risk of developing a mental health disorder than the paired cohort (Fig. 4).Stratified for sex, a much higher risk was seen among females (IRR: 14; 95% CI: 9.4-20.7)compared to males (IRR: 6.9; 95% CI: 5.2-9.1).

Attributable Fraction
The results of the analyses of the attributable fractions showed that 88.5% (95% CI: 85. 3

Discussion
Of the 948 minors who commenced treatment for CUD, 19.6% (15.8% of males; 32.7% of females) developed a mental health disorder in the years following the onset of cannabis use.This was significantly higher than in the paired group where the cumulative incidence was 3.1% (2.8% of males; 4.2% of females).In Spain, the prevalence of mental health disorders in the general population aged 15-24 years is 3.2% in males and 4.8% in females [36].
This study shows that young people with CUD have an 8.7 times higher risk of developing a mental health disorder in their youth compared with those without a diagnosis of CUD.As in other studies, the observed prevalence of CUD was higher among males [37,38].However, in both cohorts, there was a higher incidence of mental health disorders among females.The magnitude of the ratio of females to males was higher in the exposed cohort than in the paired cohort (2.1 vs. 1.5).
There is increasing evidence that among females with CUD, the time between first use of cannabis and development of dependency is shorter than among males and that females are more susceptible to the effects of cannabis [39].In the present study, differences between the sexes were not observed in the median number of years between first use and commencing treatment.Nonetheless, the differences between the sexes seen in the development of mental health disorders among young people with CUD are important.The evidence suggests that these differences could be explained by neurochemical differences in the endocannabinoid system, pharmacokinetics, pharmacodynamics, and the influence of sex hormones.However, more studies are needed in humans to explain these differences [40].
Moreover, females face greater barriers to accessing drug treatment centers than males, due to greater stigma and discrimination [41,42].Delayed access to treatment could explain the increased severity of the CUD at treatment initiation and therefore greater risk of developing mental health disorders in a shorter time period.
Recent studies estimate that 50% of mental health disorders present before 18 years of age [43] and that age and sex are factors which determine the onset of symptoms [44].The present study had a median of follow-up of only 4.5 years, and the median (IQR) age of the study population at the end of the study period was 19.1 (17.9-20.6)years.
Therefore, this could affect the observed results.The disorders most seen in the exposed cohort, in both sexes, were reaction to severe stress and adjustment disorders.Regarding ADHD, 4.3% of the exposed cohort developed ADHD compared to only 0.8% of the paired cohort, which may suggest that CUD is also associated with ADHD [24].Regarding psychotic disorders, only 2% of people in the exposed cohort and 0.3% of the paired cohort developed a  psychotic disorder.It should also be noted that some studies show that the association between cannabis use and psychosis is observed from 23 years in males and from 26 years in females [44].
This study has certain limitations that cannot be avoided.Minors with CUD who did not enter treatment were not included.Therefore, possible selection bias must be considered, and these results should only be extrapolated to people with CUD who are in treatment.The follow-up period varied according to the year of first cannabis use, and some mental health conditions may take a longer time to present associated symptoms.Therefore, some conditions may not have been diagnosed during the study period.The paired cohort was constructed considering sociodemographic confounders; however, this study only included those diagnosed with CUD in the public health system.Therefore, we do not know if those in the paired cohort were diagnosed for CUD in private health centers or if they use cannabis.Another limitation is that it was not possible to control both cohorts for genetic predisposition to developing a mental health disorder [45,46].
Regarding strengths, the large sample size and matched design are strengths of the study, as is the exclusion of people with a mental health disorder diagnosed prior to the study period.This allowed studying the relationship between CUD and the development of mental health disorders.This is also one of the first studies using data from the CMBD, which are exhaustive, valid registries of morbidity reported by all the public health centers in Catalonia.
The results of the present study are consistent with previous studies [16,20].Although the results cannot be extrapolated to the general population who use cannabis or to minors with CUD who are not in treatment, this study confirms that minors with CUD have an increased risk of developing a mental health disorder.Therefore, it highlights the need to focus prevention actions on those most at risk of developing CUD.It also confirms the high prevalence of dual pathology among minors with CUD and raises the need to address both conditions simultaneously to be effective.Dual diagnosis adds complexity to assessment, diagnosis, treatment, and recovery and can be associated with increased incidences of relapse and poorer health.Moreover, during the last decade, the potency of the tetrahydrocannabinol consumed has increased, which could be a determining factor in the association with developing a mental health disorder [47,48].Future research should also further study the differences in outcomes seen between the sexes and the impact of cannabis use within a socioeconomic and cultural context [49] and consider if the individuals have CUD.

Table 1 .
SociodemographicVery low, people exempt from the pharmaceutical co-payment, those receiving unemployment subsidies, and those receiving a noncontributory pension.p values for categorical variables χ 2 test and for continuous variables Mann-Whitney test.α = 0.05.IQR, interquartile range; SEP, socioeconomic position.