Challenges in establishing a link between THC exposure and cannabis use disorder

In low-risk use guidelines [1, 2], the frequency of cannabis use and the product potency (i.e. concentration of delta-9-tetrahydrocannabinol, THC) are considered key determinants for cannabis-attributable harm, most notably cannabis use disorder (CUD). Although the link between frequency of use and risk of CUD is well established and causally plausible [3], the same cannot yet be said for potency and CUD. The study of Lees Thorne and colleagues aimed to establish THC thresholds for CUD risk [4]—a commendable effort to advance public health guidance with regards to cannabis use and CUD. Adopting the vehicle of standard THC units, participants reported the frequency of use and quantity of four types of cannabis products (sinsemilla, hash/resin, seeded herbal and/or other). Weekly THC exposure was calculated by imputing mean or median potency per product using seizure data from five locations in England in 2015/2016 [5]. Similar approaches to measuring THC exposure have been adopted in previous UK studies on this subject [6; for a systematic review, see 7]. Although we agree that THC exposure is crucial, we have concerns with potential biases introduced by the approach employed by Lees Thorne and colleagues. First, it is unclear whether seizure data from 2015/2016 is representative of the cannabis used during 2017–2021 given the increasing potency trends within [8] and outside Europe [9]. Second, the potency varied widely (e.g. 80% of sinsemilla varied between 10% and 20% THC; resin varied from 0% to 40% THC) [5]; thus, using only the median or mean omits important variation. Third, for resin and especially ‘seeded herbal’, very few samples were included in the seizure data [5], limiting their validity. For resin, which has seen a tripling of potency in European countries and averaged at 24.1% THC in 2019 [8], the imputed mean potency of 6.3% appears to be too low to be valid. Considering these limitations, the estimated individual THC exposure may not be accurate. This potential bias makes it difficult to trust the presented THC thresholds, and it appears premature to use them to inform low-risk cannabis guidance. Additionally, the publication lacks information on the reported frequency and quantities of cannabis used and their contribution to CUD risk. Thus, it remains unclear whether the imputed potency actually improved or worsened the model fit in the receiver operating characteristic (ROC) analyses. It is conceivable that similar findings would be produced by considering only the frequency of use. In fact, a single use frequency item exhibited similar if not better psychometric properties in identifying CUD among adults (area under the curve, AUC = 0.89) [10]. In conclusion, there remains a substantial research gap in characterising the link between THC exposure and cannabis-attributable harm. The study by Lees Thorne et al. is a laudable attempt to address this gap, but it does not provide convincing data about the level of THC exposure that confers risk for CUD. To advance the field, reliable self-reports [11, 12] and—more importantly—valid THC estimates in longitudinal studies are required. Jakob Manthey: Conceptualization; writing—review and editing; writing—original draft. Anna Schranz: Writing—review and editing. Jeffrey D. Wardell: Writing—review and editing. J.M. has acted as a consultant to national and international health organizations (European Union Drugs Agency, WHO) related to cannabis use. J.M. has been invited as an expert to the Health Committee of the German Bundestag on the subject of cannabis legalisation and is a principal investigator of the official evaluation of cannabis legalisation (Evaluation des Konsumcannabisgesetzes, EKOCAN, project) funded by the Federal Ministry of Health of Germany (ZMII2-2525DSM21A). J.M. and A.S. have received honoraria and travel reimbursements for presentations from various health-related organisations on the topic of cannabis legalisation. All authors declare no financial relationship to the cannabis industry: no shares, patents, consulting activities or similar.