To our knowledge, this is the first study to investigate the impact of COVID-19 lockdown measures on a large sample of daily CU in France. In a context where all cannabis use was prohibited in the country at the time of the study, our findings reveal that the majority of respondents changed their cannabis use patterns. Specifically, over a third increased their daily number of intakes, while 36% decreased or completely stopped their use. These results are in line with a previous study performed in France from day 8 to day 13 (i.e., 24 March to 29 March 2020) of the first lockdown among 620 CU (frequency of intakes unknown) with 39.5%, 31.2% and 29.3% reporting no change, an increase, and decrease/cessation, respectively [4]. In the Netherlands, where cannabis-vending coffee shops were open for takeaway purchases during that country’s first lockdown a transversal study was performed among 1563 CU, 67.9% of whom were daily or almost daily users. That study showed that 41.3% of respondents reported increased cannabis use, 49.4% no change, and 6.6% a decrease [30]. Similarly, 38.4% of a sample of 1202 medical CU in the United States reported an increase in cannabis use since the start of the pandemic, 47.9% no change, and only 8.8% a decrease [25]. This suggests that the availability of cannabis is an important predictor of change in patterns of use. In our model, using home-grown cannabis and having stocked-up cannabis were both negatively associated with a decrease or cessation of cannabis use. Furthermore, almost everyone in our sample who reported stopping cannabis use during France’s first lockdown did not stock up on the drug beforehand. Their cessation might therefore be explained by supply shortage. Conversely, those who increased their cannabis use were more likely to have stocked up. Interestingly, respondents in the ‘unchanged’ category were more likely to report home-grown cannabis for self-use, which suggests that self-supply may help when faced with difficulties in accessing cannabis. It must be remembered however that in France, growing one’s own supply means risking legal ramifications. Other countries have recently decided to legalize its recreational use. One example is Canada, where the effects of this policy are currently being evaluated [31].

We found other correlates of changes in cannabis use. Female gender and younger age were significantly associated with increased cannabis use, which is consistent with previous studies [4, 30]. In France, universities were closed during the first lockdown, and we found that being a student was significantly associated with an increase in cannabis use, but also a decrease and even cessation, indicating the differential impact of the lockdown on students. Working practices were also abruptly impacted by the lockdown, specifically an increase in teleworking and in unemployment [32]. In our study, persons who became unemployed during the lockdown were more likely to report an increase in cannabis use than those whose working practices remained unchanged. Using cannabis exclusively to self-medicate was a protective factor against increased or decreased cannabis use, suggesting that such CU have more stable consumption patterns and were less impacted by the lockdown and supply problems.

With regard to health outcomes in our study sample during the first lockdown, unchanged cannabis use was associated with lower prevalences of depression, sleep disorder intensification, and increased pain. One hypothesis for this is that for those who increased their cannabis use, this increase was a response to adverse health outcomes participants experienced, as cannabis is commonly used to therapeutically cope with these issues [33, 34]. Conversely, stopping cannabis use was associated with a higher prevalence of these three health outcomes, even after adjustment for various potential confounders. This is not surprising as these symptoms are typical of cannabis withdrawal [35]. This association suggests that cannabis cessation was more forced than desired for a proportion of the respondents. In addition, increased use of other psychoactive substances—specifically tobacco, alcohol and benzodiazepines—was more frequent in the group which stopped cannabis use. More research is needed to evaluate whether these changes in consumption continued after the first lockdown ended. In the general population, mental health was significantly affected by the pandemic, especially during lockdown periods [36]. Thus, links between mental health outcomes and cannabis use changes might have been influenced by the crisis itself. This may explain the meaningful increase of alcohol use across all groups of cannabis changes. Overall, our results reflect a complex interplay between an unprecedented health crisis, drastic changes in life habits and cannabis supply shortage on cannabis use. It is difficult to discern whether changes in cannabis use were forced or deliberate. As designed, our cross-sectional study does not address causality and repeated studies would help to understand these relationships.

In the context of the ongoing COVID-19 pandemic, lessons can be drawn from our study as well as several public health and social implications. First, implementation of mental health care and counselling towards people with cannabis use disorder could have been beneficial to mitigate deleterious outcomes. Moreover, clinicians should now be aware of the impact of COVID-19-related lockdowns and be encouraged to tailor their responses to people who present mental and/or physical health problems associated with changes in cannabis use. Moreover, given that other studies have reported difficulties for patients to physically attend healthcare facilities during COVID-19 lockdowns, and the familiarity of CU with digital technologies, the development of telehealth could be advocated [37, 38]. Socio-environmental changes are also needed as a great deal of stigma still surrounds cannabis use and users in France, and this represents a barrier to healthcare [39]. Second, we may question whether making cannabis available as an “essential good” could have prevented abrupt cessation of cannabis use and further impact on health and other drug use, including alcohol and benzodiazepines.

Our study has limitations. First, the online recruitment method used may have led to selection bias, in particular underrepresentation of people not proficient in computing skills and those without internet access. The representativeness of our sample can only be partially assessed by comparing our results with those from the only other French study to include daily CU to date [12]. Although that study—based on a representative sample of the general population—only detailed gender and age distribution, the values for socio-demographic characteristics found were similar to ours (i.e., mostly men and a younger population). Second, to ensure the length of the questionnaire was acceptable, we chose only a small number validated scales and used only one question to measure pain increase and one for intensification of sleep disorders. Third, given that cannabis use is illegal in France, individuals may have been reluctant to participate due to fear of legal actions. However, we guaranteed complete anonymity through our secured platform. Finally, the cross-sectional nature of our data prevented us from making assumptions about causal relationships. Nevertheless, our findings provide useful information about the impact of the first COVID-19 lockdown in France on daily CU.