The aim of our study was to help GPs how to screen cannabis users and to adapt their prevention advice for each patient. By analyzing the baseline data of a randomized controlled trial in primary care, we showed that when GPs systematically screened cannabis use among their patients from 15 to 25 years, we can’t identified only one profile of users, but 5 different profiles: Low risk Adolescent users, Risky Adolescent users, Low risk student users, Risky young worker users and Risky young unemployed user. This means that GPs must interview all adolescents to identify cannabis users, without the need for a priori assumptions. On the other hand, among the 5 groups, the level of risk was not the same, nor the motivations to consume, which can allow the GPs to identify more easily risky users and choose the care to propose to the patient according to his risk profile. To date, this study is the first to describe young cannabis users in primary care who are not being monitored for a substance use disorder.

By screening their patients in routine care, a representative sample of GPs [14] identified 262 cannabis users, among whom 46.2% were daily users (more than 30 joints per month), 25.6% were regular users (from 10 to 29 joints per month), and 28.2% were recent users., and 85.9% (n = 182) had a moderate-to-high risk related to their use and therefore needed either a BI or an addiction consultation [26] according to the DSM-V criteria [27]. These users smoked 20[6–60] joints per month on average, while the impact of cannabis use on cognitive functioning can be observed at 17.5 joints per month [28]. GPs believe that adolescents are not necessarily going to answer their questions truthfully [29], but Lin’s concordance correlation coefficient showed that the answers were the same at the GP’s surgery or via anonymous self-administered questionnaire. Even if they doubt their role [29], the prevalence of cannabis use among these patients and the impact on health must convince GPs of the need to question patients on their cannabis habit.

The MCA revealed that rather than there being one sole type of cannabis user, there were five different clusters. Age, professional status, and psychotropic medication varied between groups. Analysis of a sample of young Americans in an outpatient substance abuse treatment program for cannabis problems and included in a randomized trial also showed heterogeneity among cannabis users, particularly in terms of increased problems with social functioning, more mental health issues, and a requirement for individually tailored care [30]. An Australian study showed differences between boys and girls in the level of severity and method of use [31]. Following a consultation an adolescent can be placed in one of the five clusters. This classification does not make it any easier for GPs to identify users versus non-users: GPs must continue to interview all adolescents to identify cannabis users, without the need for a priori assumptions. This screening is important because primary care health professionals may have a negative view of substance users, which can affect the quality of care [32]. On the other hand, three groups showed high risk in their CAST score (38 Risky Adolescent users, 37 Risky young unemployed users and 41 Risky young worker users) and had common characteristics: they had smoked before they were 16 years old, they smoked more than 30 joints per month, alone, during the week, in the day, and take psychoactive treatments. This classification makes it easier for GPs to identify risky users when meeting a teenager who belongs to one of these 3 profiles.

Depending on the level of use or the cluster, there were differences in motivation, perception of risk, and medical history. Regular users showed better perception of the repercussions of their use in their professional life than daily users, who may have felt less concern. The GP could, therefore, use a brief intervention [15] to inform regular users of the consequences and repercussions of their substance use. To address this topic for daily users probably has less impact on their behaviors because they are already aware of the risks involved. However, motivations for use differed among clusters: Risky Adolescents users (C1) and Risky young unemployed users (C2) mostly smoked for self-treatment (relaxation and sleep), and also more often received psychotropic treatment. Given the interplay between substance use and vulnerability to mental health disorders [33], identification of symptoms of anxiety or depression in those clusters may be an issue for the care of these individuals. Finally, risk-taking was also different: the first 3 groups drove more often after using cannabis. We found that cocaine was the preferred drug of Risky young unemployed users. These elements show that the identification of risk can allow the GP to provide appropriate care for each level of use or for a profile of a given user.

The problem for GPs is one of how and when to screen young patients [29]. The average duration of study visits (inclusion and follow-up) was 20 minutes. Given the multiplicity of reasons for consultation during the inclusion process, it can be concluded that all possible moments are good for interviewing teenagers about their cannabis use and that this does not take much time. One of the distinguishing features of the profiles is the CAST, which is high in 3 clusters. The CAST score and the number of joints per month were strongly correlated. However, it should be noted that one-third of the lowest users had high CAST scores. It is more difficult to identify an at-risk user if (s)he consumes little in terms of quantity and frequency. This difficulty particularly applies because only 2% of GPs used questionnaires to assess risky cannabis use [12]. This study provides encouragement to GPs to use the CAST score, which can be used to screen risky users regardless of the amount of cannabis use.