Perceptions of Safety of Daily Cannabis vs Tobacco Smoking and Secondhand Smoke Exposure, 2017-2021 | Tobacco and e-Cigarettes | JAMA Network Open

Key Points

Decades of federal and state policies and efforts from many nonprofit organizations led to aggressive campaigns to decrease the use of tobacco and nicotine products and exposure to secondhand smoke. These have been credited with dramatically reducing the prevalence of adult cigarette smoking and creating safer smoke-free environments, which in turn, reduce secondhand smoke exposure.1 In contrast, there has been increasing legalization and use of cannabis for medicinal and recreational purposes, with rates of adult cannabis use more than doubling from 2001 to 2012.² Although other forms of cannabis are increasing in popularity, smoking is still the most common form.³ Studies of cannabis use found that it was associated with multiple negative health outcomes, including cannabis dependence, increased respiratory symptoms, worse cognitive performance, and increased incidence of psychiatric disorders.^4–7 Despite this, regulation of cannabis has tended to be less restrictive than that for tobacco, with many smoke-free laws being amended to make exclusions that allow smoking or vaping of cannabis.⁸

Although some studies have found medicinal benefits associated with cannabinoids in treatment of nausea or vomiting from chemotherapy, spasticity related to multiple sclerosis, and refractory epilepsy of childhood, several lines of evidence indicate that cannabis may be harmful and associated with negative health outcomes analogous to those associated with tobacco smoke.9^,10 Tobacco and cannabis smoke share many chemical compounds that are known carcinogens, and smoking cannabis is associated with increased risk of head and neck and other cancers.¹¹ Decades of research on cigarettes and newer research on e-cigarettes has demonstrated that they generate particulate matter that, when inhaled through primary use or secondhand smoke, is associated with increased risk of chronic lung disease and cardiovascular disease.^12,13 Although less research has been done on cannabis, studies have found that combustion of cannabis, whether through smoking or vaping, produces a greater amount of particulate matter than tobacco, raising concerns that it could be associated with similar health outcomes.^14–19

Given the different trajectories of tobacco and cannabis policy and use, it is important to understand how the perceived safety of daily smoking and secondhand exposure to tobacco and cannabis smoke may be changing. Use of tobacco and cannabis is strongly associated with risk and safety perceptions, and lower risk perception is associated with greater incident and ongoing use of tobacco20 and increased use of cannabis.^21–23 Few studies have directly compared the perceived safety of cannabis and tobacco smoke among the same respondents. Most studies assessing risk perception of tobacco and cannabis have looked at the association between use patterns and risk perception for one^7,23–25 or both substances^26,27 but have not had participants directly compare the 2 substances to each other. In 1 cohort,²⁷ cannabis was perceived to be less harmful than tobacco, and other studies have demonstrated that the perception of cannabis has become more favorable over time.^25,26 As discussed previously, these perceptions are not consistent with published data on potential risks. One study²⁴ suggested that risk perception of cannabis may in part be attributed to its relative legality and that as it becomes legalized, the risk perception may decrease further. Understanding the comparative risk perception of tobacco and cannabis is particularly important as it may influence how public health protections and laws enacted for tobacco and electronic nicotine devices are applied in the growing number of states with legal cannabis. Studies have found that many of these states have permitted smoking and vaping of cannabis in settings where tobacco would not be allowed.^8,28 In addition, a lower comparative risk perception of cannabis may be associated with substitution or increased co-use with tobacco, which could be associated with decreased success in tobacco cessation.²⁹

This survey study follows the American Association for Public Opinion Research (AAPOR) reporting guidance. The Committee of Human Subject Research of the University of California, San Francisco, exempted Ipsos’ conduct of the survey from review because it met criteria outlined in 45 CFR §46.104 (Exempt Research). As an exempt study, this research did not require signed informed consent, but participants were provided information about the study purpose and confidentiality of responses.

Our online survey was conducted in a subset of individuals in Ipsos (formerly GfK) KnowledgePanel, a probability-based, nationally representative panel of the civilian, noninstitutionalized US population30 at 3 times to allow for comparison. Ipsos KnowledgePanel uses random, address-based sampling to cover 97% of the US population. Households are provided internet connections and a tablet if needed, and the panel was not an opt-in sample, which reduces sampling and response bias. We stratified our initial sampling by state cannabis legalization status (recreational, medical, or not legal) in 2017.

Questions on the perception of cannabis and tobacco smoke exposure in our survey were created after reviewing questions from existing surveys and the peer-reviewed literature. We also consulted with experts in substance abuse and mental health, cannabis dispensary staff, and cannabis industry professionals. All survey items were written at an eighth-grade reading level, and comprehension testing was conducted in a sample of 40 adults of different ages, educational levels, and cannabis and tobacco use statuses. Subsequent waves of the survey were piloted internally by the study team and by Ipsos on a random sample of panel participants to review and refine online administration. Full details of baseline survey design and administration were previously published.31

The first wave was administered from September 27, 2017, to October 9, 2017, to 16 280 US adults, with 9003 people completing the survey (response rate, 55.3%). The second wave of the survey, from August 5 to 31, 2020, was administered to 8529 respondents of the original survey who remained available for contact, with responses from 5979 participants (66.4% of the original cohort, 70.1% of individuals contacted). The third wave of the survey was administered from August 5, 2021, to August 31, 2021, to the 7305 respondents from the first or second wave of the survey, with responses from 5420 participants (60.2% of the original cohort, 74.2% of individuals contacted). A total of 5053 people responded to all 3 waves of the survey. Of these, 18 individuals did not respond to any tobacco or cannabis risk questions and were therefore excluded from the study, leaving an analytic sample of 5035 participants (eFigure in Supplement 1). Individuals who completed the baseline survey but did not respond to the second or third wave were significantly more likely to be younger, members of racial or ethnic minority groups (Black, Hispanic, multiracial [participants who chose >1 racial option], and other [including American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, or a different race response ]), have lower educational attainment, and report current use of tobacco or cannabis at baseline (eTable 1 in Supplement 1). Nonrespondents were significantly more likely at baseline to view cannabis smoke as being safer than tobacco smoke and to hold safer views of secondhand cannabis smoke exposure (eTable 1 in Supplement 1).

To reduce systematic error, we asked questions about secondhand smoke for tobacco vs cannabis in the same way. Finally, we attempted to reduce response bias with nonnegative, unbiased language in the survey. Many respondents in the pilot of the survey did not have a negative view of cannabis, and so we chose to use the words “safe” and “safer” rather than “risk” to avoid introducing bias into our study population.31^,32

To evaluate changes in views over time in questions that directly compared the safety of primary and secondhand cannabis vs tobacco exposure, we compared the proportion of respondents endorsing each response option in 2017 vs 2021 using multinomial logistic regression models. Survey year was entered as a covariate, and we used a variance estimator that accounts for repeated measures within individuals in the longitudinal survey cohort. We identified individuals who from 2017 to 2021 moved toward a safer view of cannabis vs tobacco across the 5 response options for each question (for example, moving from a response that secondhand smoke from cannabis was “somewhat safer” to “much safer” than tobacco). Individuals who rated cannabis as “much safer” than tobacco in 2017 were excluded from analyses given that they had already endorsed the most favorable view of cannabis. We used χ² tests and t tests as appropriate to compare baseline characteristics from Table 1 for individuals whose views increasingly favored cannabis over time vs those whose views were unchanged or who increasingly favored tobacco (eTable 2 in Supplement 1). We then developed multivariate logistic regression models with this dichotomous dependent variable (cannabis safer vs less safe or no change) that included as independent variables all factors associated with the outcome at a significance level of P ≤ .10 in univariate analyses. We used the Homer-Lemeshow test for goodness of fit, which yielded values of P > .9, indicating good fit. We repeated these analyses using multivariate linear regression models with a continuous dependent variable representing the change in views from 2017 to 2021. For this outcome, we treated Likert scale responses as values of 1 (tobacco much safer) to 5 (cannabis much safer) and calculated the change as the value in 2021 minus the value in 2017. We performed model diagnostics for linear regression models and did not identify violations of key modeling assumptions.

There were 5035 survey respondents (mean [SD] age, 53.4 [16.2] years; 2551 males [50.7%]; 314 Black [6.2%], 447 Hispanic [8.9%], and 3945 White [78.4%]) with data on risk perceptions of tobacco and cannabis smoke in all 3 years (2017, 2020, and 2021) (Table 1). Less than half of respondents had a bachelor’s degree or higher (2263 individuals [45.0%]), and the largest proportion of people had an income of $30 000 to $74 999 (1728 individuals [34.3%]). Most respondents were married (3111 individuals [62.0%]) and employed (3014 individuals [59.9%]).

Regarding the safety of daily smoking of cannabis vs tobacco, there was a significant shift from 2017 to 2021 toward a more favorable perception of cannabis. In 2021 compared with 2017, there were fewer people reporting that cannabis was somewhat or much less safe than tobacco (1213 participants [25.5%] vs 1601 participants [33.7%]) and more people reporting that cannabis was somewhat safer or much more safe than tobacco (2107 participants [44.3%] vs 1742 participants [36.7%]) (P < .001). (Table 2). Weighted analyses yielded similar findings (eTable 3 in Supplement ).

Findings for questions directly comparing the safety of secondhand smoke exposure to cannabis vs tobacco were similar to questions about daily smoking, with a significant shift toward a more favorable view of cannabis from 2017 to 2021. In 2021 compared with 2017, there were fewer people reporting that secondhand smoke was somewhat or much less safe for cannabis vs tobacco (1093 participants [23.0%] vs 1390 participants [29.3%]) and more individuals responding that secondhand smoke was somewhat safer or much more safe for cannabis vs tobacco (1908 participants [40.2%] vs 1668 participants [35.1%]) (P < .001) (Table 2). Findings in weighted analyses were similar (eTable 3 in Supplement 1).

Perceptions on the safety of exposure to secondhand smoke from cannabis and tobacco for children, adults, and pregnant women remained similar over the study period (eTable 4 in Supplement 1). However, at all times, more people endorsed the greater safety of secondhand smoke exposure for cannabis vs tobacco in all groups; these differences were most dramatic for exposure of adults (Figure). Participants were more likely to rate secondhand smoke exposure to cannabis vs tobacco as completely or somewhat safe in adults (629 participants [12.6%] vs 119 participants [2.4%]; P < .001), children 238 participants [4.8%] vs 90 participants [1.8%]; P < .001), and pregnant women (264 participants [5.3%] vs 69 participants [1.4%]; P < .001). Findings were not substantially changed in weighted analyses (eTable 5 in Supplement 2).

As an additional analysis, we looked at factors associated with a favorable change in cannabis vs tobacco risk perception over time (Table 3). Younger individuals (adjusted odds ratio [aOR] for ages 18-29 years vs ≥60 years, 1.4 [95% CI, 1.1-1.8]; P = .01) and those who were not married (aOR, 1.2 [95% CI, 1.0-1.4]; P = .01) were significantly more likely to move toward viewing smoking cannabis as safer than cigarettes, while those who were retired (aOR vs working, 0.8 [95% CI, 0.7-0.9]; P = .01) were less likely to move in this direction (Table 3). In analyses focused on secondhand smoke, individuals who were retired were less likely to shift toward perceiving cannabis secondhand smoke as safer compared with tobacco secondhand smoke over time (aOR, 0.8 [95% CI, 0.6-0.9]; P = .007) (Table 3). Weighted analyses yielded similar findings except that in analyses of secondhand smoke perceptions, working status was no longer associated with a change in outcome (eTable 6 in Supplement 1). Sensitivity analyses using a continuous outcome for change yielded similar findings (eTable 7 in Supplement 1), with the same factors associated with movements toward more or less safe views of cannabis vs tobacco smoke. In these linear models, female sex was also associated with moving toward a safer view of exposure to primary cannabis smoke.

Our study expands upon prior work on public opinions about the safety of cannabis and tobacco. The National Survey on Drug Use and Health found that the proportion of respondents who felt that regular cannabis use was associated with great risk decreased from 51.3% in 2002 to 40.3% in 2012.33 Our study found that these trends continued in the era of recreational cannabis legalization. Few studies have simultaneously assessed tobacco and cannabis risk perception, and risk has typically been measured separately^26,27 rather than by asking participants to directly compare substances. Studies that have looked at both tobacco and cannabis assessed risk perception and use in samples from Australia (1046 participants)²⁷ and Germany (318 participants)²⁶ and found that lower risk perception of tobacco and cannabis was associated with concurrent and future use.

As cannabis becomes legalized in more states, risk perception may decrease further, which may be associated with increased consumption of cannabis and exposure to secondhand cannabis smoke. These cultural shifts toward increasingly safer views of cannabis may be partly attributable to the $17.5 billion legal cannabis industry’s marketing, which often highlights unsubstantiated safety claims and health benefits34^,35 without noting existing data on potential harms associated with smoking and vaping, many of which are similar to those associated with tobacco. In 1 study,³⁶ internet claims about the health benefits of cannabis were found to have evidence of support in 5% of cases. Although there are some moderate-quality data to support the use of cannabis in specific health circumstances (eg, spasticity), there are many other conditions that social media and the cannabis industry claim cannabis can improve despite a lack of scientific evidence.^6,36

More research is needed to investigate the potential medicinal benefits associated with cannabis, but this must be balanced by investigation of potential negative health outcomes, particularly for smoked or vaped cannabis products. Animal studies suggest that even 1 minute of secondhand smoke from cannabis may be associated with impeded endothelial function and therefore the same cardiovascular outcomes as tobacco.19 Smoking cannabis is associated with increased risk of head or neck and other cancers.¹¹ Tobacco and cannabis smoke share many chemical compounds that are characterized as carcinogens.¹¹ There is also moderate-quality evidence to suggest that cannabis use, especially heavy use, is associated with a higher risk of psychiatric conditions, including psychosis,³⁷ depression,³⁸ and anxiety³⁹. Lower-quality data suggest there is a higher risk of respiratory side effects with cannabis use,⁴⁰ but more studies are needed to fully understand associations with cardiovascular health⁴¹ and lung, oral, and other cancers.⁴²

Understanding changing views on tobacco and cannabis risk is important given that increases in social acceptance and decreases in risk perception may be directly associated with public health and policies. For example, there is a movement to make exceptions for cannabis in smoke-free bans in public housing, bars, and restaurants, although there are no data that suggest that secondhand exposure is safe.8^,43 In fact, emerging data suggest significant potential risks associated with exposure to secondhand cannabis smoke.¹⁴ Such smoke-free air laws are important for protecting public health, especially among already high-risk communities, and have been shown to be associated with reduced youth smoking initiation.^1,44 Exceptions for cannabis use may also erode the effectiveness of tobacco smoking bans and still leave the public at risk from harmful compounds present in cannabis.¹¹ Finally, many products combine cannabis with tobacco, and the substances are used in similar or identical devices, making enforcement of tobacco bans challenging.⁴⁵

Published: August 11, 2023. doi:10.1001/jamanetworkopen.2023.28691

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2023 Chambers J et al. JAMA Network Open.

Corresponding Author: Beth Cohen, MD, MAS, Department of Medicine, University of California, San Francisco, 4150 Clement St, San Francisco, CA 94121 (beth.cohen@ucsf.edu).

Author Contributions: Dr Cohen had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Keyhani, Ling, Ryder, Cohen.

Acquisition, analysis, or interpretation of data: Chambers, Keyhani, Hoggatt, Hasin, Nguyen, Woods, Cohen.

Drafting of the manuscript: Chambers.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: Hoggatt, Nguyen, Woods.

Obtained funding: Cohen.

Administrative, technical, or material support: Keyhani, Ryder.

Supervision: Keyhani, Cohen.

Conflict of Interest Disclosures: Dr Hoggatt reported receiving grants from the Veterans Health Administration during the conduct of the study and grants from the National Institutes of Health, Rubin Family Foundation, and Veterans Health Administration outside the submitted work. No other disclosures were reported.

Funding/Support: The study was supported by grant T30IR0988 from the California Tobacco-Related Disease Research Program.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2.