In the context of the rapidly expanding legal cannabis market in the U.S., there is growing evidence that various contextual factors beyond frequency and quantity of use may contribute to adverse cannabis-related outcomes. Previous studies have demonstrated that factors such as solitary versus social use, the form of cannabis and its route of administration, and the source of acquisition can influence cannabis use disorder and negative outcomes, while revealing that many of these contextual factors are interrelated. However, prior studies have integrated contextual factors into a single construct using variable-centered or person-centered approaches, thereby limiting the ability to disentangle and analyze the unique impact of specific contextual factors on cannabis use outcomes. Therefore, the current study aims to identify key contextual predictors associated with cannabis protective behavioral strategies, adverse outcomes, and cannabis use disorder symptoms by employing an exploratory machine learning technique.
The study utilized data from 5,700 college students who use cannabis, recruited from three research initiatives conducted between Fall 2016 and Fall 2020. The data were gathered from states with varying cannabis legal statuses (recreational, medical, or prohibited). Participants completed the survey that assessed various factors, including amount spent, frequency of use, subjective and peak intoxication levels, duration of use, usage settings, product type, route of administration, and method of purchase. Additionally, measures of cannabis protective behavioral strategies, negative consequences, and cannabis use disorder symptoms were obtained. An exploratory machine learning model (random forest) was applied to evaluate how effectively these factors predict cannabis use outcomes.
According to the study, various contextual factors related to cannabis use among college students were examined. Results indicated that, on average, participants spent approximately $46 on cannabis over the past 30 days, reported a typical intoxication level of about 61.8%, and a peak intoxication level of around 73.2%. Although the predominant form of use was cannabis flower, some samples exhibited a trend toward increased use of concentrates. Additionally, while most participants used cannabis with friends, a significantly higher rate of solitary use was observed in certain samples. Random forest analysis revealed that factors such as solitary use, black-market acquisition, and use at home were key predictors of adverse outcomes, showing a strong association with negative reinforcement mechanisms and higher levels of cannabis use disorder symptoms. Additionally, bong use emerged as an important predictor of negative outcomes due to increased THC exposure. In contrast, using cannabis with friends was linked to the adoption of protective behavioral strategies (PBS) and relatively positive outcomes. These findings underscore the significant impact of usage contexts on adverse cannabis-related outcomes and highlight the necessity of considering social and environmental factors in the development of prevention and intervention strategies.
Takeaway: The outcomes of cannabis use among college students are largely determined by the context of use, with factors such as using alone or with friends, black-market purchasing, and use at home acting as key predictors.