Feelings of distress reflect a basic dimension of human emotionality that is expressed under conditions when the individual perceives a lack of control over threatening environmental pressures and/or forces (Buchanan 2000; Gallagher et al. 2014; Vigil 2009), and it is possible that Cannabis usage reduces such perceptions. The current study helps explain why many patients attempting to treat feelings of distress voluntarily substitute medical cannabis for several classes of prescription medications, including those used to treat negative affect (e.g., SSRIs, SNRIs, TCAs, MAOIs, beta blockers, atypical antipsychotics, and benzodiazepines), when given the legal opportunity to do so (Bachhuber et al. 2014; Bradford and Bradford 2016; Piper et al. 2017; Powell et al. 2018; Stith et al. 2018a; Vigil et al. 2017; Wen and Hockenberry 2018). Expanding upon a previous study (Cuttler et al. 2018), our real-time effects showed that Cannabis flower is an effective anxiolytic medication and it is relatively fast-acting, but it can also produce negative side effects that may exacerbate momentary symptoms of negative affect in a small minority of sessions. While in some sessions users reported no change in symptom intensity levels or experiencing feelings that could contribute to distress (e.g., feeling restless), in 95% of sessions, people reported an average overall symptom intensity reduction of approximately 3.8 points on a standard 0 to 10 visual analog scale.

The current observational research design maximizes the external validity and generalizability of the findings through assessments of patients’ actual medical treatment decisions, including their choice across a range of product options, and the experienced effects of those decisions, in real-time. The mobile software technology used in the study solves the significant practical, medical, and scientific challenge of monitoring and measuring therapeutic and side effects across the vast range of products available at medical and recreational cannabis dispensaries, which vary by strains, consumption method, and major cannabinoid contents. The current results suggest benefits from patient-directed cannabis therapy as a mid-level anxiolytic treatment. Thus, despite the conventional wisdom that smoking cannabis makes one paranoid, we found consumption much more likely to be associated with relaxation and sense of calm, with users most likely to report feelings of peacefulness, optimism, and happiness. One potential explanation for the disparity between our findings and popular perceptions of cannabis is that the “paranoia” users may have historically reported could have arisen in part from cannabis’ illicit status (e.g., anxiety over committing an illegal act), rather than the plant’s typical endemic pharmacodynamic effects when consumed in contexts typical of legal medicinal use. Individual factors such as user’s experience level likely also contribute to cannabis’ effects.

This study finds that the effectiveness and side effect manifestation vary with the characteristics of the Cannabis flower consumed and the specific type of distress-related symptom treated. In particular, mid to higher THC levels are statistically significant predictors of increased symptom relief, while CBD levels and inhalation method (joint, pipe, vape) are largely not. In contrast, plants labeled as C. sativa were associated with less overall symptom relief. The relationship between higher THC and increased symptom relief appears to be driven by cannabis sessions treating specifically “anxiety” and to a lesser extent “stress” rather than “agitation/irritability,” although sample sizes are small in the subgroup analyses and some of the variation could arise from anxiety being a more clinical and clearly defined term than the other two symptoms included in our analyses.

The differences in symptom relief across THC levels might arise because THC has been shown to both decrease and increase negative mood states. However, unlike in this study, smaller doses of isolated or synthetic THC have been found to be anxiolytic and higher doses (in isolated form) appeared to be anxiogenic (Childs et al. 2017). The mechanisms by which THC potency levels can produce these biphasic effects are not fully understood and likely encompass multiple brain regions and complex interactions with other chemotypic characteristics of the plant and endogenous neurotransmitters. In rats, low doses of THC microinjected into the prefrontal cortex (e.g., 10 μg) and ventral hippocampus (e.g., 5 μg) were anxiolytic, whereas higher doses appeared to be anxiogenic. By contrast, microinjections of low doses (e.g., 1 μg) of THC in the basolateral amygdala produced anxiogenic effects, while higher doses were found to be ineffective (Rubino et al. 2008). Moreover, while low doses of THC stimulate an anxiogenic signal in the amygdala, the anxiolytic signals generated in the prefrontal cortex and hippocampus override this effect by suppressing the amygdala activation (Rubino et al. 2008).

THC’s anxiolytic effects are likely mediated by CB₁ and CB₂ receptors, whose respective roles appear to be to modulate neurotransmitter and cytokine release (Pertwee and Ross 2002). For example, both CB1 and 5-HT_2A receptors are expressed in most glutamatergic neurons in the prefrontal cortex and hippocampus (Hill et al. 2007). The CB₁ and 5-HT_2A receptors have been shown to physically interact and form heteromers, and the costimulation of these CB₁R-5-HT_2AR heteromers appears to modulate cellular signaling in specific brain structures, including the prefrontal cortex and the hippocampus (Viñals et al. 2015). Other research suggests that anxiolytic effects of THC are mediated through the CB₁ receptors on cortical glutamatergic terminals (Rey et al. 2012). Hence, CB₁R-5-HT_2AR heterodimerization may play a significant role in the reduction of glutamate levels in the prefrontal cortex and hippocampus, which could be leading the reductions in visceral feelings of distress reported by app users. However, in addition to the complexity of understanding the effects of isolated THC on the brain, the synergistic effects of THC and other compounds in the cannabis plant, including CBD, are even less understood. The fact that higher THC appears to confer greater anxiolytic effects in our study at higher THC levels suggests that the whole natural Cannabis plant may act very differently on the brain as compared to synthetic or derived THC isolates.

Although we were unable to account for these in our study, terpene and terpenoid contents that contribute to the overall phytocannabinoid-terpene-terpenoid synergy or “entourage effect” from whole, natural Cannabis plants can vary from one plant to another and by inhalation method. Many terpenes and terpenoids share a direct precursor with phytocannabinoids. For example, geranyl pyrophosphate is a precursor to the phytocannabinoids found in Cannabis and to the monoterpenes and monoterpenoids. Terpenes and terpenoids may comprise as much as 10% of total cannabis trichome content, and individual concentrations of > 500 ppm are considered to be of pharmacological interest. Serum terpene levels in the single-digit ng mL⁻¹ range have been found to induce physiological effects potent enough to alter animal and human behavior, including anxiolytic and perceived negative side effects (Ross 2003; Russo 2011; Souto-Maior et al. 2011). For example, linalool and limonene have both demonstrated in several studies to possess potent anxiolytic properties (Carvalho-Freitas and Costa 2002; De Almeida et al. 2014; De Moraes Pultrini et al. 2006; Franco et al. 2016; Harada et al. 2018; Lima et al. 2013; Linck et al. 2010; Souto-Maior et al. 2011). β-Caryophyllene is a selective full agonist at the CB₂ receptor and has unique effects on negative affect, making the CB₂ receptor a prospective therapeutic target for the treatment of both anxiety and depression with cannabis (Bahi et al. 2014; Galdino et al. 2012; Kamal et al. 2018; Russo 2011). Future studies identifying strains with the most notable effects on negative affect should help elucidate why we find suggestive evidence that products labeled as “indica” and “hybrid” may be to be more anxiolytic than strains typically labeled as “sativa.” Future research will benefit from identifying and measuring the effects of particulate chemotypic profiles, including cannabinoid-terpene-flavonoid combinations, magnitudes, and ratios across varying plant strains, beyond conventional plant characteristic labeling.

The current study does have limitations, the most prominent of course being the lack of absolute experimental control (e.g., double-blinded randomization and use of a placebo intervention) and the analysis of naturalistic behaviors and dosage patterns rather than a directed and uniform regimen. Likewise, the study did not include individuals who do not use cannabis to treat their distress or any cannabis consumption sessions not tracked in the app potentially resulting in selection bias. People who choose to use cannabis to treat their distress-related symptoms may be those most likely to benefit from it or those for whom conventional treatments are less effective. The direction of the bias for app use is not as clear. Not using the app could be simply a matter of not knowing about the app or a dislike of app-based technologies, or, along with attrition, be due to dissatisfaction with cannabis or the app. Alternatively, not using the app or stopping app use could arise from satisfaction with existing cannabis use and the lack of a need to explore other product options. Within the app itself, the overt pro-cannabis language would also likely influence the type of individuals who would use the app and attract users with views aligned with the authors of the app; similar types of sample selection biases are common in large epidemiological studies where people volunteer their time to discuss or describe a discrete research topic and, in clinical trials, among people that choose to be participants in a study. Although our study extended the literature by incorporating a wider range of product characteristics than has been previously examined, we still were not able to include the full range of characteristics of products available (e.g., terpene profiles) and did not include nonflower cannabis products. We also did not account for user demographics, cannabis experience, or the concomitant use of medications other than cannabis beyond those time-invariant characteristics captured by the user fixed effects. Additional factors such as frequency of use and resultant changes in tolerance levels likely also contribute to individual differences in potential anxiogenic and anxiolytic effects, and future research may benefit by incorporating dosage and tolerance-related factors in their analyses. Finally, while improvements in testing and regulatory oversight may be reducing this issue, studies have shown that THC and CBD levels reported on product labels are often inaccurate, particularly at the higher end of the distribution, which would reduce our ability to distinguish the effects of higher versus lower potency products (Bonn-Miller et al. 2017; Vandrey et al. 2015). We attempted to mitigate this issue with our binning approach and by cutting any observations reporting THC or CBD levels exceeding 30%/dry wt. Future research would benefit from independent product testing rather than relying on user reporting based on product labels. Despite these limitations, the current finding of a dose-response effect, particularly for flower with higher THC levels, is consistent with the results of several clinical trials (Childs et al. 2017; Tambaro and Bortolato 2012) and suggests that cannabinoid contents is a major factor in Cannabis’ potential anxiolytic effects.