In this longitudinal study examining patterns of past-6-month frequency of cannabis and illicit opioid use, we found that the odds of daily illicit opioid use were lower (by about half) among those who reported daily cannabis use compared to those who reported no cannabis use. However, we observed no significant association between occasional cannabis use and daily opioid use, suggesting that there may be an intentional therapeutic element associated with frequent cannabis use. This is supported by cross-sectional data from the sample in which certain reasons for cannabis use were observed to differ according to cannabis use frequency. Specifically, daily users reported more therapeutic motivations for cannabis use (including to address pain, stress, nausea, mental health, or symptoms of HIV or antiretroviral therapy, or to improve sleep) than occasional users, and non-medical motivations—although common among all users—were not more likely to be reported by daily users. Together, our findings suggest that PWUD experiencing pain might be using cannabis as an ad hoc (i.e., improvised, self-directed) strategy to reduce the frequency of opioid use.

A recent study analyzed longitudinal data from a large US national health survey and found that cannabis use increases, rather than decreases, the risk of future non-medical prescription opioid use in the general population [40], providing important evidence to challenge the hypothesis that increasing access to cannabis facilitates reductions in opioid use. The findings of our study reveal a contrasting relationship between cannabis use and frequency of opioid use, possibly due to inherent differences in the sampled populations and their motivations for using cannabis. Within the current study population, poly-substance use is the norm; HIV and related comorbidities are common; and pain management through prescribed opioids is often denied, increasing the likelihood of non-medical opioid use for a medical condition [6,9]. Furthermore, our study is largely focused on this relationship in the context of pain (i.e., by examining individuals with self-reported pain and accounting for intensity of pain). Our findings align more closely with those of a recent study conducted among HIV-positive patients living with chronic pain, in which the authors found that patients who reported past-month cannabis use were significantly less likely to be taking prescribed opioids [41]. While this finding could have resulted from prescription denial associated with the use of cannabis (or any illicit drug), we show that daily cannabis users in this setting were slightly more likely to have been prescribed a pain medication at baseline, and adjusting for this factor in a longitudinal multivariable model did not negate the significant negative association of frequent cannabis use with frequent illicit opioid use.

The idea of cannabis as an adjunct to, or substitute for, opioids in the management of chronic pain has recently earned more serious consideration among some clinicians and scientists. A growing number of studies involving patients who use cannabis to manage pain demonstrate reductions in the use of prescription analgesics alongside favourable pain management outcomes [24–33]. For example, Boehnke et al. found that chronic pain patients reported a 64% mean reduction in the use of prescription opioids after initiating cannabis, alongside a 45% mean increase in self-reported quality of life [29]. Degenhardt et al. found that, in a cohort of Australian patients on prescribed opioids for chronic pain, those using cannabis for pain relief (6% of patients at baseline) reported better analgesia from adjunctive cannabis use (70% average pain reduction) than opioid use alone (50% average reduction) [33]. However, more recent high-quality research has presented findings to question this narrative. For example, in the 4-year follow-up analysis of the above Australian cohort of pain patients, no significant temporal associations were observed between cannabis use (occasional or frequent) and a number of outcomes including prescribed opioid dose, pain severity, opioid discontinuation, and pain interference [35]. Thus, several other explanations for our current results, aside from an opioid-sparing effect, are worthy of consideration.

We chose to include individuals with chronic pain regardless of their opioid use status to avoid exclusion of individuals who may have already ceased illicit opioid use at baseline, as these individuals may reflect an important subsample of those already engaged in cannabis substitution. On the other hand, there may be important characteristics, unrelated to pain, among regular cannabis users in this study that predispose them to engage in less frequent or no illicit opioid use at the outset. We attempted to measure and control for these factors, but we cannot rule out the possibility of a spurious connection. For example, individuals in this cohort who are consuming cannabis daily for therapeutic purposes may simply possess greater self-efficacy to manage health problems and control their opioid use [42]. However, it is notable that our finding is in line with a previous study demonstrating that cannabis use correlates with lower frequency of illicit opioid use among a sample of people who inject drugs in California, all of whom used illicit opioids [43]. Our study builds on this work by addressing chronic pain, obtaining detailed information on motivations for cannabis use, and examining longitudinal patterns.

We observed that daily cannabis users endorsed intentional use of cannabis for a range of therapeutic purposes that may influence pain and pain interference. After pain, insomnia (43%) and stress (42%) were the second and third most commonly reported motivations for therapeutic cannabis use among daily cannabis users. The inability to fall asleep and the inability stay asleep are common symptoms of pain-causing conditions [44], and experiencing these symptoms increases the likelihood of opioid misuse among chronic pain patients [45]. The relationship between sleep deprivation and pain is thought to be bidirectional [44,46], suggesting that improved sleep management may improve pain outcomes. Similarly, psychological stress (particularly in developmental years) is a well-established predictor of chronic pain [47] and is also likely to result from chronic pain. Thus, another possible explanation for our finding is that cannabis use substitutes for certain higher-risk substance use practices in addressing these pain-associated issues without necessarily addressing the pain itself.

Notably, our findings are consistent with emerging knowledge of the form and function of the human endocannabinoid and opioid receptor systems. The endogenous cannabinoid system, consisting of receptors (cannabinoid type 1 [CB1] and type 2 [CB2]) and modulators (the endocannabinoids anandamide and 2-arachidonoylglycerol), is involved in key pain processing pathways [48]. The co-localization of endocannabinoid and μ-opioid receptors in brain and spinal regions involved in antinociception [49], and the modification of one system’s nociceptive response via modulation of the other [50,51], has raised the possibility that the phytocannabinoid tetrahydrocannabinol (THC) might interact synergistically with opioids to improve pain management. A recent systematic review and meta-analysis found strong evidence of an opioid-sparing effect for cannabis in animal pain models, but little evidence from 9 studies in humans [23]. However, the authors of the meta-analysis identified several important limitations potentially preventing these studies in humans from detecting an effect, including low sample sizes, single doses, sub-therapeutic opioid doses, and lack of placebo [23]. Since then, Cooper and colleagues have published the results of a double-blind, placebo-controlled, within-subject study among humans in which they found that pain threshold and tolerance were improved significantly when a non-analgesic dose of an opioid was co-administered with a non-analgesic dose of cannabis [52]. Suggestive of a synergistic effect, these findings provide evidence for cannabis’s potential to lower the opioid dose needed to achieve pain relief [52].

Finally, there is pre-clinical and pilot clinical research to suggest that cannabinoids, particularly cannabidiol (CBD), may play a role in reducing heroin cue-induced anxiety and cravings [53] and symptoms of withdrawal [54]. Although preliminary, this research supports the idea that cannabis may also be used to stabilize individuals undergoing opioid withdrawal, as an adjunct to prescribed opioids to manage opioid use disorder, or as a harm reduction strategy. Although this evidence extends beyond chronic pain patients, it warrants consideration here given the shared history of illicit substance use amongst the study sample. It is not clear what role harm reduction or treatment motivations may have played in the current study since daily and occasional users did not differ significantly in reporting cannabis use as a strategy to reduce or treat other substance use. The phenomenon of using cannabis as a tool to reduce frequency of opioid injection has been highlighted through qualitative work in other settings [55], but further research is needed to determine whether this pattern is widespread enough to produce an observable effect. Clinical trials that can randomize participants to a cannabis intervention will be critical for establishing the effectiveness of cannabis both for pain management and as an adjunctive therapy for the management of opioid use disorder. Such trials would begin to shed light on whether the current finding could be causal, what the underlying mechanisms might be, and how to optimize cannabis-based interventions in clinical or community settings.

There are several important limitations to this study that should be taken into consideration. First, the cohorts are not random samples of PWUD, limiting the ability to generalize these findings to the entire community or to other settings. The older median age of the sample should especially be taken into consideration when interpreting these findings against those from other settings. Second, as discussed above, we cannot rule out the possibility of residual confounding. Third, aside from HIV serostatus, we relied on self-report for all variables, including substance use patterns. Previous work shows PWUD self-report to be reliable and valid against biochemical verification [56], and we have no reason to suspect that responses about the outcome would differ by cannabis use status, especially since this study was nested within a much larger cohort study on general substance use and health patterns within the community. Major or persistent pain, which qualified respondents for inclusion in this study, was also self-reported. Our definition for chronic pain is likely to be more sensitive than other assessments of chronic pain (e.g., clinical diagnoses or assessments that capture length of time with pain). Although more than half (60%) of the sample reported ever having been diagnosed with a pain condition, it is possible that some of the included respondents would not have met criteria for a formal chronic pain diagnosis. Finally, we did not collect information on the type of cannabis, mode of administration, cannabinoid content (e.g., percent THC:percent CBD), or dose during the study period. Future research will need to address these gaps to provide a more detailed picture of the instrumental use of cannabis for pain and other health concerns among PWUD.