A survey analysis covering 50,481 adults across 10 US states found that people receiving food assistance were more likely to report using both cigarettes and cannabis than those who did not receive that assistance. After adjustment for measured differences, the relative risk ratio was 2.27, according to a study published in PLOS One on 7 October 2026.
The result does not mean food assistance caused substance use. The researchers examined a cross-sectional dataset, meaning economic circumstances and substance-use behaviour were assessed within the same survey period. The finding identifies a statistical association that warrants further investigation rather than a causal effect of public benefits.
The researchers also found associations involving food insecurity, unstable employment, housing difficulties and threatened utility shutoffs. These patterns suggest that material hardship may be relevant to understanding why some adults report cigarettes, cannabis or both, but the relationships differed across types of hardship.
The study analysed 2023 data from the US Behavioral Risk Factor Surveillance System, known as BRFSS. It included a weighted population estimate of approximately 24.9 million adults across the participating states, rather than claiming to represent every adult in the United States.
What the researchers wanted to understand
Cigarette smoking and cannabis use are often studied separately. However, some people use both products, potentially creating different patterns of exposure, dependence and healthcare needs. The researchers wanted to know whether distinct types of economic instability were associated with exclusive cigarette use, exclusive cannabis use or combined use.
The study examined five indicators of economic instability. These were employment instability, receipt of food assistance through the Supplemental Nutrition Assistance Program, food insecurity, housing insecurity and threats that utility services would be disconnected.
Each indicator captures a different form of hardship. Someone who has recently lost working hours may face different pressures from a household struggling to afford food or threatened with the loss of electricity or water. Treating all financial stress as a single measure can obscure these differences.
The researchers were particularly interested in whether economic instability was more strongly associated with using both products than with using either one alone. They also examined whether the associations remained after accounting for measured demographic, socioeconomic and health-related factors.
How the study was conducted
The researchers used data from the 2023 BRFSS telephone survey. Their analysis focused on Connecticut, Delaware, Illinois, Indiana, Maine, Mississippi, Montana, Nevada, New Mexico and Virginia because these states administered both the relevant cannabis-use and social-determinants modules.
The final analytical sample contained 50,481 adults aged at least 18 years. The authors applied survey weights to account for the BRFSS sampling design, producing estimates representing about 24,896,969 adults in the participating states.
The underlying 2023 BRFSS response rate was 44.7%. That figure is important because people who could not be reached or declined participation may differ from respondents, particularly where unstable housing or financial hardship makes telephone participation more difficult.
Cigarette use was defined using a lifetime threshold of at least 100 cigarettes and current smoking every day or on some days. Cannabis use was based on reporting use on at least one day during the previous 30 days. These definitions distinguish current cigarette smoking from any recent cannabis consumption.
The researchers divided respondents into four groups: neither product, cigarettes only, cannabis only and both products. They then used five separate multivariable multinomial logistic regression models, each centred on one economic instability indicator, with non-users serving as the comparison category.
The adjusted models accounted for factors including age, sex, race and ethnicity, marital status, education, metropolitan residence, health insurance, self-rated health, perceived stress and histories of selected chronic illnesses. Adjustment reduces some alternative explanations but cannot eliminate all confounding.
How common were the behaviours?
In the weighted sample, 8.1% reported exclusive cigarette use, 11.1% exclusive cannabis use and 3.6% reported using both. These categories were mutually exclusive in the study’s classification, so the percentages should not be added to estimates from other surveys with different definitions.
The finding that cannabis-only use exceeded cigarette-only use applies to the participating states and the researchers’ definitions. It should not be presented as a national US prevalence estimate, nor does it imply that cannabis and cigarette use carry identical health risks.
Economic instability was also common. Food insecurity was reported by 13.9%, employment instability by 12.0%, housing insecurity by 11.6%, receipt of food assistance by 10.4% and threatened utility shutoffs by 7.7%.
These percentages describe overlapping experiences rather than five mutually exclusive groups. One household may struggle with rent, food and utilities at the same time. The study’s separate models cannot fully establish which hardship is most important when several occur together.
The strongest association involved combined use
Every economic instability indicator examined was significantly associated with combined cigarette and cannabis use in the adjusted models. The largest reported relative risk ratio was 2.27 for respondents who received food assistance, with a 95% confidence interval from 1.75 to 2.94.
A relative risk ratio of 2.27 in this model is not the same as saying 2.27 times as many people receiving assistance used both substances. It compares the relative likelihood of being in the combined-use category rather than the no-use category, after adjustment for the measured covariates.
Threatened utility shutoffs were associated with a relative risk ratio of 2.19 for combined use, with a 95% confidence interval of 1.68 to 2.84. Housing insecurity was associated with a ratio of 2.03, with a confidence interval of 1.59 to 2.60.
Food insecurity was linked to a combined-use relative risk ratio of 1.97, with a confidence interval from 1.56 to 2.50. Employment instability showed a smaller but still statistically significant association, with a ratio of 1.52 and a confidence interval from 1.17 to 1.97.
The consistency across five hardship indicators is noteworthy. Yet it does not show that hardship directly led individuals to begin using substances, or that removing a single hardship would automatically reduce use by the same proportion.
Different hardship indicators, different patterns
The associations were not identical for cigarette-only and cannabis-only use. Exclusive cannabis use was associated with employment instability, housing insecurity and threatened utility shutoffs, but the associations with food insecurity and food assistance did not reach statistical significance in the adjusted models.
For cannabis-only use, the adjusted relative risk ratios were 1.23 for employment instability, 1.27 for housing insecurity and 1.37 for threatened utility shutoffs. Their reported confidence intervals excluded one, supporting statistical associations within the model.
Exclusive cigarette use showed a somewhat different profile. Food insecurity was associated with a relative risk ratio of 1.57, receipt of food assistance with 1.55, housing insecurity with 1.47 and threatened utility shutoffs with 1.78.
Employment instability was not significantly associated with cigarette-only use after adjustment. That contrast matters because it suggests researchers should avoid assuming that every form of economic hardship relates to all substance-use behaviours in the same way.
The authors noted that receipt of food assistance should not be interpreted as inherently harmful. It can identify households facing economic hardship while also indicating that those households have successfully accessed support. The analysis cannot disentangle these explanations.
What the findings mean for public health
The study suggests that clinical screening for cigarette and cannabis use could be more useful when paired with questions about housing, food and utility insecurity. Healthcare teams might then identify patients who need both behavioural support and assistance with basic living conditions.
For example, a person trying to stop smoking may also be struggling to keep electricity connected or pay rent. Addressing those immediate pressures could make it easier to participate in cessation services, although this study did not test whether providing financial assistance improves cessation outcomes.
The authors advocated coordination between healthcare providers, public health programmes and social services. Their reasoning is that substance use is shaped partly by the circumstances in which people live, not only by individual decisions made in isolation.
This does not justify stigmatising people who receive food benefits or assuming that recipients use substances. Most adults in any of the hardship categories were not classified as combined users. Relative associations must be interpreted alongside the underlying prevalence and individual variation.
Why the findings cannot establish causality
The biggest limitation is timing. Because the analysis was cross-sectional, it cannot determine whether financial hardship preceded substance use, followed it or developed alongside it. The relationship may also work in both directions, and other factors may influence both.
Although the regression models adjusted for many measured characteristics, they did not include every possible confounder. The authors highlighted factors such as trauma history and family history of substance-use disorders as examples of information that could influence the observed relationships.
The data were self-reported. Participants may misremember or choose not to disclose substance use, financial hardship or other sensitive information. The measures also lacked detailed information on substance-use frequency, intensity, duration and methods of consumption.
The 10 states were selected because they included the necessary optional survey modules. They were not a random sample of all states, so the weighted estimates should not be generalised to the entire country without further evidence.
The study also did not examine whether the associations differed systematically by age, racial or ethnic group or other demographic characteristics. Such analyses could help future researchers understand where support might be most useful without relying on stereotypes.
What further research should test
Longitudinal studies could follow people over time to establish whether changes in housing, employment or food security precede changes in substance use. This would help clarify temporal relationships that a one-time survey cannot resolve.
Intervention studies could also test whether combining economic support with smoking cessation and substance-use treatment improves outcomes. That would provide more direct evidence for the policy implications discussed by the authors.
Researchers should distinguish between financial assistance as an indicator of need and financial assistance as an intervention. Finding an association among benefit recipients does not demonstrate that benefits increase substance use, and removing support could worsen hardship.
The central message is therefore more nuanced than a simple link between poverty and substance use. In a large survey across 10 states, several kinds of economic instability were associated with different patterns of cigarette and cannabis use, with the strongest relationships generally involving both products.
The findings support taking material circumstances seriously in public health research and clinical practice. They do not identify a single cause, and they do not establish that any particular policy would reduce substance use without additional testing.
Source Information
Study title: Economic instability and substance use among adults in 10 US states: Evidence from a population-based study.
Authors: Peiyuan Liu, Tania de Jesús Espinosa, Elham Samami, Bernard Mensah and Hermine Poghosyan.
Journal: PLOS One, volume 21, issue 10, article e0360066.
Publication date: 7 October 2026.
Study design: Cross-sectional population-based analysis of 2023 Behavioral Risk Factor Surveillance System data using weighted descriptive statistics and adjusted multinomial logistic regression.
Sample: 50,481 adults in 10 US states; weighted population estimate 24,896,969.
Key finding: Food assistance was associated with cigarette-cannabis co-use (adjusted relative risk ratio 2.27; 95% confidence interval 1.75 to 2.94).
Original DOI: 10.1371/journal.pone.0360066.
Funding: Authors reported no specific funding.
Competing interests: Authors declared no competing interests.
