Health Reports
Using the Severity of Dependence Scale to examine cannabis consumers with impaired control in Canada

Abstract

Background

The 2018 Cannabis Act legalizing the production, sale, and use of cannabis for non-medical purposes renewed interest in the importance of ongoing and more detailed monitoring of cannabis consumption and consequences. Some cannabis users will experience impaired control over their use of cannabis, putting them at risk for cannabis use disorder (CUD, sometimes called addiction) and other harms. Including the Severity of Dependence Scale (SDS) in the annual Canadian Community Health Survey (CCHS) would allow for monitoring of one of the more harmful consequences of cannabis use in the post-legalization period.

Data and methods

Data from the nationally representative 2019-2020 CCHS were used to examine cannabis consumers with and without impaired control. Respondents who used cannabis in the past year were categorized according to their SDS scores: those with impaired control (SDS >=4) versus those without impaired control (SDS < 4). Cross-tabulations were used to examine the sociodemographic, mental health, health behaviour and cannabis exposure characteristics of those with impaired control. Multivariable logistic regression models assessed associations between these characteristics and the risk of impaired control. The prevalence of self-reported cannabis-related problems experienced by consumers—with and without impaired control—is also presented.

Results

In 2019-2020, 4.7% of past-year cannabis consumers scored >=4 on the SDS and were considered to have impaired control. Multivariable logistic regression suggested that the odds of having impaired control remained higher for people who were male, were aged 18 to 24 years, were single or never married, were from lower-income households, were diagnosed with an anxiety or a mood disorder, started consuming cannabis at age <= 15, and consumed at least monthly.

Interpretation

A better understanding of the characteristics of cannabis consumers experiencing impaired control (a correlate of future CUD or addiction) could help with the development of more effective education, prevention and treatment strategies.

Keywords

marijuana, controlled and illegal drugs, Cannabis Act, C-45, problematic use, dependence, population survey, illicit drug use, cannabis-related problems

Author

Michelle Rotermann is with the Health Analysis Division at Statistics Canada.

 

What is already known on this subject?

• Cannabis remains one of the most widely used substances in Canada.
• Most people who consume cannabis do so without harm. However, a minority will be adversely affected, including by the development of a cannabis use disorder, by cannabis abuse or by cannabis dependence.

What does this study add?

• According to combined data from the 2019-2020 Canadian Community Health Survey (CCHS), 4.7% of people aged 15 years or older who reported using cannabis in the past year experienced impaired control over their use.
• Few nationally representative Canadian surveys include screening tools to measure cannabis dependency or related harms. The inclusion of the Severity of Dependence Scale in the CCHS enables the monitoring of one of the more harmful consequences of cannabis use in the post-legalization period.
• People who were male, were aged 18 to 24 years, were single or never married, were from lower-income households, were diagnosed with an anxiety or a mood disorder, started consuming cannabis at age 15 or younger, and consumed monthly or more often were at an increased risk of experiencing impaired control.
• Quantifying the number of past-year consumers who experienced impaired control has many applications, including for cannabis-related education and for health services planning, because some users will need treatment.

Introduction

Cannabis is one of the most widely used substances in Canada, with more than 4 in 10 Canadians reporting having tried it.^Note 1 The use of cannabis in Canada has been rising for several decades,^Note 2 including after legalization.^Note 3,^Note 4 Some data also show that higher frequency use, such as daily or almost daily use, was more common in 2020 than in the previous two years.^Note 4

While the direct harms to cannabis consumers and to others, tend to be lower than for many other drugs,^Note 5 cannabis use is not always benign.^Note 6,^Note 7 In fact, cannabis is recognized as a dependence-producing substance in the Diagnostic and Statistical Manual of Mental Disorders (DSM),fourth (IV) and fifth (V) editions,^Note 8,^Note 9 and under the International Classification of Diseases, Tenth Revision.^Note 10 According to both diagnostic systems, cannabis dependence (CD) is defined by symptoms including tolerance, inability to cut down or stop consumption, preoccupation with cannabis in terms of time spent using it or giving up other activities, and continued use despite wide-ranging problems that are likely caused by it. The majority of people who use cannabis do so without harm.^Note 11 However, a minority will be adversely affected, including by the development of a cannabis use disorder (CUD), by cannabis abuse (CA) or by CD.^Note 7, ^Note 12, ^Note 13 Since more people are using cannabis now, it is anticipated that more people will develop a CUD.^Note 12 Some patterns of cannabis use, such as regular, heavy or persistent (chronic) use, especially among younger users, can be more harmful and risky than other patterns.^Note 11, ^Note 14, ^Note 15, ^Note 16, ^Note 17 A CUD (and the related diagnoses of CA and CD) is associated with psychiatric and physical problems, as well as higher rates of unemployment, educational underachievement and welfare dependence,^Note 16, ^Note 18, ^Note 19, ^Note 20 although competing or co-occurring factors tend to make it difficult to identify effects that unequivocally can be linked to cannabis.  

There is limited Canadian research on the problems and harms associated with cannabis use,^Note 11,^Note 19, ^Note 21, ^Note 22, ^Note 23, ^Note 24, ^Note 25 and even less about harmful cannabis use, CD, and cannabis use-related harms using Canadian data collected after legalization.^Note 26, ^Note 27, ^Note 28, ^Note 29

The five-item Severity of Dependence Scale (SDS) screening tool, which measures the degree of impaired control, also called psychological dependence, was added for the first time to the annual CCHS in 2019.^Note 30 Previously, the annual CCHS did not collect national data about cannabis use, nor did it include a measure of cannabis-related harms. These changes provide an opportunity to estimate the prevalence of impaired control over cannabis, using Canadian data and enable an examination of the characteristics of cannabis consumers with and without impaired control.   

The SDS has been used internationally^Note 31, ^Note 32, ^Note 33, ^{Note 34 Note 35}, ^Note 36, ^Note 37, ^Note 38, ^Note 39 and in Ontario^Note 24 for over two decades^Note 31 on both general population surveys^Note 24, ^Note 33, ^Note 35, ^Note 37, ^Note 39 and clinical or convenience samples.^Note 34, ^Note 36, ^Note 38, ^Note 40

The two main objectives of this analysis are

1. to classify cannabis consumers aged 15 years or older as having impaired cannabis control or not, based on the SDS, and
2. to calculate the prevalence of, and identify the factors associated with, cannabis consumers who experienced impaired control or not, in addition to examining the frequency of other reported problems experienced by the two groups.   

Methods

Data sources

The CCHS collects information related to the health status, diagnoses, health care use and health determinants for the Canadian population aged 12 years and older living in the provinces and territories.^Note 30 Excluded from the study are people living on reserves and other Indigenous settlements, people in two health regions in northern Quebec, full-time members of the Canadian Forces, the institutionalized population, and youth aged 12 to 17 years living in foster homes. These groups represent about 3% of the target population.^Note 30 Details about the CCHS, including survey frames, sampling strategy, weighting and questionnaires, are available elsewhere.^Note 30 The 2019 data were collected continuously from January to December 2019 by computer-assisted telephone and in-person interviews. Most interviews were conducted exclusively by telephone (86%). The overall 2019 response rate was 54.4%, corresponding to a final sample of 65,970 respondents.^Note 30

The COVID-19 pandemic had major impacts on the collection of CCHS data in 2020, including preventing the use of in-person interviews, cancelling data collection from mid-March to September and exceptionally limiting data collection in the territories to the capitals (Whitehorse, Yellowknife, and Iqaluit) for 9 of the 12 months of collection. Consequently, the 2020 response rate (28.9%) and sample size (42,634 respondents) were lower than in 2019.  The 2019-2020 combined-year file included 108,604 respondents.

Study sample

The 2019-2020 combined-year study sample consisted of 101,228 respondents aged 15 years or older, of whom 790 were past-year cannabis users with impaired control, 17,158 were past-year users without impaired control, and 83,280 reported not using cannabis in the past year. Because cannabis use questions were not asked in proxy interviews (that is, where a knowledgeable family member answers on another’s behalf), 3,410 records of proxy respondents were dropped from the analysis. Another 580 records were dropped from the study because some or all of the SDS questions were left unanswered.

Analytical techniques

Weighted cross-tabulations and percentage distributions were used to examine differences and similarities between past-year cannabis consumers with and without impaired control, cannabis use frequency by SDS score, and endorsement of the five SDS questions.

Unadjusted and adjusted multivariable logistic regression (odds ratios) was used to identify factors independently associated with impaired cannabis control, while controlling for sociodemographic (sex, age, marital status, sexual orientation, Indigenous identity, and income level), mental health (diagnosed with an anxiety or a mood disorder), health behaviour (daily cigarette smoking and heavy drinking) and cannabis exposure (age of initiation, medical or non-medical use, and frequency of use) characteristics.   

To account for survey design effects, coefficients of variation and 95% confidence intervals were estimated with the bootstrap technique^Note 41 using SAS 9.4 and SAS-callable SUDAAN 11.0.3. Differences between weighted frequencies and cross-tabulations were calculated with t-tests, and results at the p < 0.05 level were considered statistically significant. All reported differences in the text are statistically significant (meaning that they were not likely to have occurred by chance alone). To improve readability, the words “statistically significant” are not always repeated. Similarly, at times the text states that an estimate is comparable or not different between groups, even though the numbers are not identical. This occurs when the difference between the numbers is not statistically significant.

Definitions

Impaired control based on the Severity of Dependence Scale

The one- to two-minute, easy-to-administer, five-item SDS measured the degree of psychological dependence based on an individual’s feelings of impaired control, preoccupation and anxiety toward their cannabis use (Table 2).^Note 31, ^Note 39 The SDS has moderate to high internal consistency with Cronbach’s alpha values of 0.73 to 0.83.^Note 32, ^Note 37, ^Note 42 The total score is obtained by summing the scores from the five four-point (0 to 3) items. Higher scores denote greater impairment and problems.

Past-year cannabis consumers scoring at least 4 on the SDS will be considered to have impaired control. Past-year consumers with SDS scores of less than 4 will be considered not to have experienced impaired control. The SDS score of 4 or more was chosen because it was the cutpoint most used.^Note 25, ^Note 31, ^Note 32, ^Note 37, ^Note 38, ^Note 39, ^Note 40 There is no universally accepted cutpoint. 

Cannabis users

Past-year cannabis use with and without impaired control was based on the following question: “Have you used cannabis in the past 12 months?”

Covariates

The selection of covariates was guided by the cannabis and substance use disorder and harms literature, as well as their availability and sample size in the CCHS. Sociodemographic variables included sexat birth, whereby all respondents were coded as either male or female. Five age groups were defined: 15 to 17 years, 18 to 24 years, 25 to 44 years, 45 to 64 years, and 65 years or older. The categories for marital status were married or living common law, previously married (divorced, separated or widowed), and single or never married.

Sexual minorities can be at an increased risk for substance use disorders.^Note 43 Sexual orientation (coded as heterosexual versus sexual minorities, including gay or lesbian, bisexual, or sexual orientation not elsewhere classified) was assessed by asking respondents whether they considered themselves to be heterosexual, homosexual, or bisexual, or to have a sexual orientation, not elsewhere classified.

Indigenous identity variables were included because cannabis use may vary by Indigenous and non-Indigenous identity^Note 44 or between Indigenous groups. Many factors likely contribute to the consumption of cannabis and other substances among Indigenous people, including the harmful effects of colonialism that led to intergenerational trauma, discrimination, and socioeconomic marginalization.^Note 44 For some parts of the analysis, the number of Indigenous respondents was insufficient to support disaggregation by Indigenous group and therefore the dichotomous pan-Indigenous (Indigenous or non-Indigenous) variable was used. The other Indigenous identityvariable indicates whether the respondent reported being First Nations, Métis, Inuit or of more than one Indigenous group. Because of small cell sizes, estimates about Indigenous respondents who reported more than one Indigenous group could not be released. 

The CCHS income variable represented a combination of tax data (~84%), self-reports (~3%) and imputed values (~13%). Household income deciles were then derived by calculating the adjusted ratio of a respondent’s total household income to the low-income cut-off corresponding to their household and community size. Income quintiles are not assigned to records for respondents from the territories. A missing income variable was included in the models (not shown) to ensure records with missing income were included in the regression analyses. 

The variable “diagnosed with an anxiety or a mood disorder” identified respondents who reported having conditions diagnosed by a health professional and that had lasted, or were expected to last, six months or more, including depression, mania, phobias, OCD, panic, dysthymia, and bipolar disorder. 

Current daily smokers(cigarettes)were compared with all others (occasional and former smokers, and individuals who had never smoked). Heavy drinking (alcohol) refers to having consumed four or more (females) or five or more (males) drinks, per occasion, at least once per month during the previous year.

To be consistent with previous research, the age of initiation included a category for trying cannabis at age 15 or younger (considered early onset of cannabis use)^Note 45 in addition to other categories for those who reported first trying cannabis at age 16 or 17 compared with those starting at age 18 or older. The categories for frequency of cannabis use in the past year were less than once a month, weekly or one to three times a month, and daily or almost daily. 

Persons reporting using cannabis in the past 12 months were further separated into three groups according to the purpose: 1) non-medical only, 2) medical only with or without document, and 3) both medical and non-medical. 

Criteria used to differentiate problematic or risky cannabis use from other use are when use impedes other obligations or when cannabis consumers continue using despite experiencing problems.^Note 8

The 2019-2020 CCHS collected some information about other cannabis-related problems or experiences taken from the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST).^Note 46

1. In the past 12 months, has your use of cannabis led to health, social, legal or financial problems?
2. At any time in the past 12 months, have you failed to do what was normally expected of you because of your use of cannabis?
3. (In the past 12 months), has a relative, friend, doctor or other health professional been concerned about your use of cannabis or suggested you cut down?

Results

Severity of Dependence Scale

More than three-quarters (77.3%, or 4.96 million) of those who reported using cannabis in the past year scored 0 on the SDS, meaning they reported that they would not have any difficulty stopping and never or almost never experienced any of the other four SDS dependency symptoms (Table 1). Those who had total SDS scores from 1 to 3 represented nearly one-fifth (18.1% or 1.2 million) of past-year consumers whereas about 1 in 20 past-year consumers (4.7% or 299,543) scored 4 or more, indicating impaired control (Table 1 and Table 3).



The most frequently experienced SDS item was about how difficult it would be to stop or go without cannabis, with 12.6% of past-year consumers reporting it would be quite difficult, very difficult or impossible (Table 2).



Impaired control

According to the 2019-2020 CCHS, a higher percentage of males (5.2%) than females (4.0%) experienced impaired control over their use (Table 3). At 7.9%, the prevalence of impaired control over cannabis use in the past year was higher among 18- to 24-year-olds than among people in older age groups (4.8% among those aged 25 to 44 years, 2.7% among those aged 45 to 64 years, and 1.6% among those aged 65 years or older) and comparable to the impaired control estimate for 15- to 17-year-olds (5.5%).



The percentage of cannabis users meeting the criteria for impaired control was approximately double among people who identified as gay, lesbian, or bisexual, or whose sexual orientation was not elsewhere classified, compared with those identifying as heterosexual (8.5% compared with 4.3%, respectively). The percentage of cannabis users with impaired control was also about twice as high when people were single or never married (compared with those who were married or living common law), were from the lowest-income-quintile households, or reported being a daily cigarette smoker.

Impaired control over one’s cannabis use was also more commonly found among people reporting a diagnosis of an anxiety or a mood disorder (9.2%) and among Inuit (11.8%), compared with those who had not been diagnosed with an anxiety or a mood disorder (3.3%) and non-Indigenous people (4.5%). Among Indigenous people, impaired control over cannabis use was also higher when compared with the corresponding estimate for non-Indigenous people (7.2% compared with 4.5%, respectively). The age at cannabis use initiation and the reason for consumption were also related to having experienced impaired control. For example, 7.6% of past-year consumers who first tried cannabis at the age 15 or younger scored 4 or higher on the SDS, signifying impaired control, above the 3.1% and 4.1% estimates for people who first tried it when they were older. The corresponding impaired control estimates for consumers reporting dual-purpose use (that is, for both medical and non-medical reasons) versus exclusive non-medical use showed about a two-fold difference (7.9% compared with 3.6%, respectively). Additionally, impaired control increased with the frequency of use, from less than 1% (0.6%) among those using cannabis less than monthly to a high of 13.1% for those who consumed cannabis daily or almost daily during the previous year. Percentage distributions of cannabis-use frequency by SDS score (from 0 to 15) further demonstrate this association (Table 1). For example, more than 7 in 10 cannabis consumers with an SDS score greater than or equal to 4 (72.4%) reported daily or almost daily use, about four and a half times the percentage (16.1%) of those with an SDS score of 0. Conversely, infrequent consumers (that is, a few times per month or less) accounted for 62.5% of consumers with an SDS score of 0, compared with 13.5% of those with an SDS score of 4 or more.

Of course, sociodemographics, mental health, health behaviours and variables related to cannabis exposure may not be independently associated with the risk of impaired control over one’s use. When potential confounding was considered using multivariable logistic regression, some bivariate associations were no longer statistically significant (Table 4). Nonetheless, the odds of having impaired control remained elevated for people who were male (adjusted odds ratio [AOR]=1.5), were single or never married (AOR=1.5), were from the lowest-income households (AOR=1.4), had been diagnosed with an anxiety or a mood disorder (AOR=2.1), started using cannabis at age 15 or younger (AOR=1.6), and consumed cannabis at least monthly (AOR=3.2 to 28.3). By contrast, being aged 25 years or older remained protective against the risk of experiencing impaired control (AOR=0.2 to 0.5). 



The multivariable analysis also investigated the association with another well-known risk factor for cannabis use and dependence—daily (cigarette) smoking. Once other factors were considered, this association was no longer statistically significant. Additionally, differences by Indigenous identity, cannabis user (non-medical or medical) type, and sexual orientation were no longer significant.

Experiencing cannabis-related problems is more common among consumers with impaired control

The percentage of past-year cannabis users who reported having experienced harmful consequences because of their use, or having had someone tell them they were concerned about their use was up to 10 times higher among people with impaired control compared with those without (Table 5). This pattern also generally held for both sexes, for most age groups (i.e., cannabis users aged younger than 65 years) and for those who reported consuming cannabis at least monthly.



Discussion

This study uses nationally representative Canadian data collected in 2019 and 2020 to examine the overall prevalence of impaired control over one’s cannabis use and to identify the sociodemographic, mental health, health behaviour, and cannabis use exposure variables independently associated with an increased risk of impaired control. 

One of the three main objectives of the Cannabis Act is the protection of public health and safety by allowing adults access to legal cannabis.^Note 47 Therefore, monitoring the number of people experiencing impaired control according to the SDS in the years following legalization could contribute to assessing the impact of this ACT objective. Quantifying the number of past-year consumers who have experienced impaired control could also be used for cannabis-related education and health services planning, because some users will need treatment.^Note 12, ^Note 21

Many of the sociodemographic characteristics associated with an increased risk of impaired control in this study, including being male, aged 18 to 24 years, single (never married) and lower income, have been observed previously.^Note 12, ^Note 17, ^Note 21, ^Note 48, ^Note 49 The strongest predictor of impaired control in this study was the frequency of use, with consumers who used more often having the highest odds. This is a well-recognized risk factor for CD, problematic use and related harms.^Note 45, ^Note 48, ^Note 50

Having been diagnosed with an anxiety or a mood disorder also increased the odds of experiencing impaired control. Studies conducted in other countries also found this association.^Note 12, ^Note 17, ^Note 49, ^Note 51, ^Note 52 The question as to whether cannabis use contributes to poorer mental health or whether people with poorer mental health use cannabis for symptom management remains unresolved.

Because this was the first time the SDS has been included on a large, nationally representative health survey in Canada, assessing this study’s SDS-based impaired control estimate of 4.7% is imperfect.

Ideally, the CCHS would have included the SDS questions in addition to another complete and validated tool so that CA, CD and CUD prevalence could be estimated, allowing for differences across tools, surveys, and countries to be assessed more systematically.

According to this study, higher percentages of cannabis users with impaired control (compared with those without) reported failing to meet other expectations or obligations; experiencing various health, social, legal or financial problems because of this use; and being told by others that they were concerned about their cannabis use. Given the considerable overlap between the different screening tools used to identify cannabis users who are likely to be at risk for or currently experiencing use-related problems and the diagnostic criteria for diagnosing a CUD, CA, and CD, this finding was to be expected.^Note 8, ^Note 9, ^Note 10, ^Note 13,^Note 46 

Strengths and limitations

This study has several strengths, including the use of nationally representative, recent Canadian data from the provinces and territories; a broad range of covariates; and survey administration after non-medical cannabis use was legalized. Nevertheless, results of this study should be interpreted in light of several limitations.

Although the 2019-2020 CCHS is large when compared with many other Canadian drug use monitoring or convenience-sampled surveys, the problem of small sample sizes was not entirely avoided and may have reduced the ability to detect some statistically significant differences. Similarly, this analysis occasionally used more general covariate definitions than desirable because of limited numbers of respondents in some cells.

The CCHS is a rich source of information about health, mental health, health determinants and sociodemographics; however, some relevant covariates,—particularly related to cannabis and drug exposures—were unavailable, including the quantity consumed, which products or what amounts of tetrahydrocannabinol (THC) were consumed (potency), the use for coping, and the use of other drugs. Increasing potency^Note 53, ^Note 54 and factors such as more frequent use or use of higher-risk products^Note 55 have the potential to increase CUD.^Note 12 

Although survey weights ensured that the sample is representative of the target population, bias may exist if the use of cannabis by respondents and non-respondents differed systematically.

Because the COVID-19 pandemic may have affected cannabis use patterns beyond the 2019-2020 timeframe, and as the legal retail industry continues to evolve in Canada, ongoing monitoring of cannabis-use behaviours may be necessary to identify ongoing changes. Also, the results pertain to those living in private households; people experiencing homelessness, a group particularly susceptible to substance use and mental health challenges, are not represented.

Information from the survey was self-reported and has not been verified although quality mechanisms were applied at all stages of data collection, processing, analysis and reporting. Willingness to disclose cannabis use may have increased with legalization.^Note 56 The cross-sectional nature of the data does not allow for causal inferences.

Concluding remarks

This is the first national, population-based study to provide detailed sociodemographic and other risk information about Canadian cannabis users experiencing impaired control using data collected after legalization. It is an important time for cannabis-use monitoring and research given that legal non-medical use by adults in Canada is a recent policy change, the legal, retail industry is still developing, and the Cannabis Act regulations are under review and at risk of being reduced. An improved understanding of the patterns of use and characteristics of people who are vulnerable to a CUD, CD and CA, as well as those who are currently experiencing some problems because of their use, could help with the development of more effective policy, prevention, and education initiatives.