In our study of hospitalized patients with COVID-19 at two Los Angeles area hospitals, we found that cannabis use was associated with decreased disease severity as assessed by NIH severity score and was associated with improved clinical outcomes in COVID-19. Consistent with known trends, active cannabis users were overall younger than non-users. However, when adjusting for age these outcomes remained consistent. Even more, when adjusting for comorbid conditions, demographics and smoking history we found that cannabis users still had less severe disease progression compared to non-users. This remained true despite less utilization of adjunct therapies like remdesivir and corticosteroids. Consistent with our understanding of how cannabis may play a role as an immunomodulator, non-cannabis users were found to have greater elevations in inflammatory biomarkers at the time of admission and during their hospital course. Thus, based on our current understanding of cannabis’ immunomodulatory effects, the link between cannabis usage and better COVID outcomes is sensible.
In the USA in 2020, an estimated 17.9% of the population (49.6 million people) used cannabis during the past year (2020 National Survey of Drug Use and Health (NSDUH) Releases | CBHSQ Data n.d.). Given the magnitude of COVID-19 and the prevalence of cannabis use in the USA, it is important to evaluate how active cannabis usage may affect clinical outcomes in COVID-19 patients. According to a 2019 national survey, past-month cannabis usage in adults has significantly increased over the years from 7.2% in 2016 to 10.2% (Dai and Richter 2019). Important with respect to COVID-19, is a trend of increasing cannabis use in older adults. Han et al conducted a national survey including close to 15,000 respondents and found the prevalence of past-year cannabis use among adults 65 years and older increased from 2.4 to 4.2% from 2015 to 2018, a 75% relative increase (Russell et al. 2018). Given our current knowledge of COVID-19 disproportionately hospitalizing older populations, understanding the effects of cannabis use in COVID patients is increasingly relevant.
In our population of cannabis users, there was a paucity of data regarding route of administration. This is important, as the route of cannabis use can influence health outcomes. While cannabis smoking is associated with respiratory issues including bronchitis and arguably impaired lung function (Tashkin and Roth 2019), some research suggests that vaporized and edible cannabis may offer the potential for reduced health risks, although this has also been questioned (Chadi et al. 2020). Moreover, cannabis vaping has been associated with over 2800 of cases of Electronic or Vaping Associated Acute Lung Injury (EVALI) (The Lancet Respiratory Medicine 2020). The varied pharmacokinetics of cannabis depends on route of administration. Bioavailability of inhaled cannabinoid products is approximately 30% while ingesting cannabis results in 4–12% bioavailability (McGilveray 2005). Peak serum concentration of cannabinoids is also notably higher and achieved more quickly when cannabinoids are inhaled as opposed to ingested. Despite the differences in pharmacokinetics between these two main routes of administration, systematic reviews have shown that there is high user variability within cohorts that all purportedly took in cannabinoids through the same route of administration. Given the diverse ways in which cannabis can be introduced into the body, our grouping of inhaled and ingested cannabis should introduce little variability to an already highly variable cohort of cannabis users. Pooling all cannabis users, regardless of administration method, gives our study more power in analysis while minimizing the risk of overfitted data.
Previous studies have found that patients with cannabis use disorder, while younger and less comorbid, had higher risk for breakthrough infections of COVID-19 despite vaccination. Wang et al. posit that behavioral factors or adverse pulmonary and immunologic effects of cannabis may contribute to this breakthrough risk (Wang et al. 2021a). Another recent study found that COVID-19 patients with substance use orders have worse outcomes compared to general COVID-19 patients, including increased hospitalization and death (Wang et al. 2021b). This study, however, grouped several substances, including opioids and alcohol use and is not specific to cannabis use.
To our knowledge, this study is one of the first evaluations of the effect of cannabis use on outcomes in patients hospitalized with COVID-19. While previous data have determined the detrimental relationship of tobacco smoking with COVID-19, this study suggests that cannabis may actually lead to reduced disease severity and better outcomes despite a five-fold greater concomitant use of tobacco amongst cannabis users compared to non-users in our study population.
Inaccurate or incomplete documentation in the medical record may bias our findings, as we were beholden to the data contained within the clinical chart, which this retrospective analysis was based on. While the majority of data abstraction was blinded to our study purpose, our authors did evaluate each chart of reported cannabis users to ensure current use. Additionally, cannabis use is self-reported without specific focus on it during routine admission data collection. Thus, we may not have captured all current cannabis users, potentially introducing further selection bias. Additionally, we do not have complete data on the route of cannabis use, frequency or duration, and we were therefore unable to comment on dose response or durability of the potential effects of cannabis consumption. Cannabis users were defined after both automatic and manual data processing and both populations were very well characterized. Given legalized recreational cannabis use in California, our retrospective analysis is less prone to the selection bias and underreporting of cannabis use in comparison to other centers where its use is illegal. Because our focus was also on cannabis use, we were also unable to ascertain alcohol use history and usage of other substances. Therefore, we were unable to factor in substance use disorder into our inverse probability weighting process.