Cannabis and Depression in Women: NHANES Study Findings
By Dr. Benjamin Caplan, MD | Board-Certified Family Physician, CMO at CED Clinic | Evidence Watch
Want to apply this research to your care?
CED Clinic translates emerging research into individualized clinical care. Dr. Caplan has treated 30,000+ patients.
Book a consultation →A large analysis of national health survey data found that women who regularly used cannabis were about twice as likely to report significant depressive symptoms compared to non-regular users, while no such association was detected in men. Because this is a snapshot study rather than a longitudinal investigation, it cannot determine whether cannabis use preceded depression, depression preceded cannabis use, or both share common underlying causes.
Cannabis Use Linked to Higher Depressive Symptoms in Women, Not Men, in Large U.S. Survey
A cross-sectional analysis of over 20,000 adults from the National Health and Nutrition Examination Survey finds a gender-specific association between regular cannabis use and depressive symptoms, but the study design cannot determine whether this relationship is causal or in which direction it operates.
#72
High Relevance
Addresses a common clinical question about cannabis and mood using nationally representative data, though limited by cross-sectional design and lack of peer review.
Gender Differences
NHANES Epidemiology
Cardiovascular Risk
Mental Health Screening
Cannabis use continues to rise across the United States as legal access expands, and depression remains among the most prevalent psychiatric conditions in primary care. Clinicians urgently need clarity about whether these two problems travel together, and if so, whether the relationship differs between men and women. This nationally representative analysis offers one of the larger population-level examinations of that co-occurrence during a pivotal period in American cannabis policy, making its findings directly relevant to screening decisions and patient counseling in everyday practice.
| Study Type | Cross-sectional secondary analysis (PhD dissertation, not peer-reviewed) |
| Population | U.S. adults aged 18 to 69 years (10,293 female, 9,857 male) |
| Intervention / Focus | Self-reported cannabis use (regular vs. non-regular; daily frequency; cessation duration) |
| Comparator | Non-users or non-regular users of cannabis |
| Primary Outcomes | Depressive symptoms (PHQ-9 score of 10 or higher); antidepressant medication use; cardiovascular disease |
| Sample Size | 20,150 adults (unweighted) |
| Journal | PhD Dissertation, The Ohio State University (ProQuest #10144860); not published in a peer-reviewed journal |
| Year | 2015 |
| DOI / PMID | Not available (unpublished dissertation) |
| Funding Source | Not reported |
The relationship between cannabis use and depression remains one of the most frequently debated questions in cannabis medicine. This dissertation drew on four cycles of the National Health and Nutrition Examination Survey (NHANES), spanning 2005 through 2012, to examine whether regular cannabis use was associated with clinically significant depressive symptoms, antidepressant medication use, and cardiovascular disease in a nationally representative sample of over 20,000 American adults. The analysis employed appropriate complex survey weights to maintain national representativeness and used multiple imputation for missing data, with logistic regression models stratified by gender to examine potential sex-specific differences in the cannabis-depression association.
The most striking finding was a gender-specific pattern: women who were regular cannabis users had 80% higher odds of meeting the PHQ-9 screening threshold for clinically meaningful depressive symptoms compared to non-regular users (OR 1.80; 95% CI 1.36 to 2.34), while no significant association was observed in men (OR 1.05; 95% CI 0.06 to 1.36). A dose-response trend emerged, with higher daily joint consumption linked to progressively greater odds of depressive symptoms. Notably, women who had quit cannabis for at least one month showed a 54% reduction in odds of depressive symptoms. No significant associations were detected between cannabis use and either antidepressant medication use or cardiovascular disease. The cross-sectional design remains the principal limitation: the study cannot determine whether cannabis use contributed to depression, whether depression drove cannabis use, or whether both were products of shared underlying factors. The authors acknowledge that longitudinal and experimental studies are needed to clarify the direction of this association.
Cannabis and Depression in Women: A National Survey Finds an Association, But Not a Cause
Nearly one in five American women in this national survey who regularly used cannabis met the screening threshold for significant depressive symptoms, about twice the rate among non-regular users. That statistic demands attention. But before drawing conclusions, we need to ask the question the data cannot answer: which came first? This dissertation does several things well. It draws on a genuinely representative national dataset, applies the correct statistical machinery for complex survey data, and makes a decision that turns out to be the study’s most important methodological contribution: stratifying by gender. Without that stratification, the signal in women would have been diluted by the null finding in men, and a potentially important sex-specific pattern would have remained invisible. The dose-response analysis, showing that heavier daily consumption tracked with worse depressive symptom scores, adds another layer of plausibility to the association.
Yet the fundamental problem is one of design, not execution. A cross-sectional study captures a single moment in time. It is like photographing a parking lot and concluding that cars cause buildings, because the two always appear together. The association between cannabis use and depressive symptoms in women is real and robust in these data, but it is equally consistent with depressed women turning to cannabis for symptom relief, a phenomenon known as self-medication. It is also consistent with a third possibility: that shared vulnerabilities like childhood adversity, chronic stress, or genetic predisposition drive both cannabis use and depression independently. The author, to her credit, acknowledges this limitation in the methods section but then, in the conclusion, suggests that treating depression might reduce cannabis use. That inference requires a directionality the cross-sectional design simply cannot provide. I also note that several confidence intervals in the dose-response analysis appear to be duplicated across categories, which, without peer review to catch such discrepancies, raises a minor but real concern about the precision of reported estimates.
In my practice, I would not tell a female patient that cannabis is causing her depression based on this evidence. I would, however, treat the co-occurrence seriously and recommend screening for both conditions whenever either is present. The cessation finding, showing lower depressive symptom odds among women who had quit for at least a month, is intriguing but not definitive. It could reflect genuine benefit from stopping, or it could reflect that women with less severe depression found it easier to quit. What I would tell a colleague is that this study adds a credible epidemiological data point to an existing pattern, and that gender-stratified analysis should become standard in cannabis-mental health research. What I would tell a policymaker is that as cannabis access expands, we need prospective studies and mental health surveillance built into regulatory frameworks. A large sample and statistical significance do not compensate for a cross-sectional design when the core question is causal direction; the most important unanswered question, whether cannabis use precedes or follows depression, requires prospective data that this study, by design, cannot provide.
This dissertation sits at an early stage of the research arc: it identifies a population-level association that generates a testable hypothesis rather than confirming a causal pathway. For clinicians evaluating patients who use cannabis, the gender-specific finding is the most actionable element. The significantly elevated odds of depressive symptoms in female regular users suggest that depression screening should be routine in this population, and that cannabis use history should be explored as part of any depression workup in women. The null finding in men should not be mistaken for evidence of safety; rather, it reflects inadequate statistical precision in the male subgroup analysis.
From a pharmacological standpoint, the study does not address cannabis product type, THC concentration, CBD content, or route of administration, all of which have evolved substantially since the 2005 to 2012 data collection period and may modify the cannabis-depression relationship. Clinicians should also consider potential interactions between cannabis and antidepressant medications, particularly given that the study found no significant association between cannabis use and antidepressant use, which may reflect underreporting or prescribing patterns rather than a true absence of co-use. The most concrete recommendation from this analysis is straightforward: whenever a patient, particularly a female patient, presents with either regular cannabis use or depressive symptoms, the clinician should systematically screen for the other condition and engage in an open, non-judgmental conversation about both.
This is an unpublished PhD dissertation presenting a cross-sectional secondary analysis of NHANES survey data. Cross-sectional analyses occupy a lower tier in the evidence hierarchy because they capture associations at a single point in time and cannot establish causal relationships or temporal sequence. The most important inference constraint is that the study has not undergone external peer review, meaning its methodological choices, statistical output, and interpretive claims have been evaluated only by the dissertation committee rather than by independent experts in the field.
This analysis aligns with a growing body of epidemiological literature documenting associations between regular cannabis use and depressive symptoms, particularly in large population-based surveys. The Degenhardt et al. (2008) WHO World Mental Health Survey findings similarly noted co-occurring patterns of cannabis use and mental health symptoms across multiple countries. The gender-specific finding reported here is consistent with broader research showing that women may experience different psychiatric trajectories in association with substance use compared to men, though the specific biological or social mechanisms remain unresolved.
The null cardiovascular finding in this analysis is notable given the rising interest in cannabis-related cardiovascular risk. It does not contradict studies that have found cardiovascular effects in clinical or high-use populations, but it adds a data point suggesting that at the general population level, the association may not reach statistical significance in cross-sectional data. Future longitudinal studies, particularly those leveraging natural experiments created by state-level legalization policies, will be essential to clarify the findings reported here.
The most consequential analytic choice was the binary classification of cannabis use as “regular” versus “non-regular.” A more granular exposure measure incorporating frequency, duration, product type, and potency could have revealed different dose-response patterns or identified subgroups at particularly elevated or reduced risk. Additionally, modeling cannabis use disorder as distinct from recreational use might have produced meaningfully different associations, since individuals meeting criteria for disordered use likely differ from recreational users in both cannabis consumption patterns and psychiatric vulnerability.
The wide and potentially erroneous confidence intervals in the men’s subgroup and dose-response analyses suggest that alternative approaches to subgroup analysis, or pooling with additional NHANES cycles, might have produced more precise and interpretable estimates for these secondary findings.
The most likely overinterpretation is concluding that cannabis use causes depression in women. The odds ratio of 1.80 is statistically significant and clinically noteworthy, but it reflects an association observed at a single time point. Reverse causation, in which women experiencing depressive symptoms turn to cannabis for relief, is an equally valid interpretation of the same data. A third possibility, that shared risk factors such as trauma history, socioeconomic stress, or genetic vulnerability drive both conditions independently, cannot be ruled out.
A second common misreading is interpreting the null cardiovascular result as evidence that cannabis is safe for the heart. A non-significant finding in a cross-sectional survey is not the same as evidence of absence of risk, particularly given the limitations of self-reported cannabis exposure and the relatively young age of the study population.
This dissertation contributes a well-powered, nationally representative signal that regular cannabis use and depressive symptoms co-occur at significantly elevated rates in American women, with a plausible dose-response pattern. It does not establish causation, temporal direction, or mechanism. For clinical practice today, the most appropriate response is routine co-screening for cannabis use and depression, particularly in female patients, while awaiting the prospective longitudinal studies that will be necessary to clarify whether this association reflects cause, consequence, or coincidence.
Does this study prove that cannabis causes depression?
No. This study found that women who regularly used cannabis were more likely to report depressive symptoms, but because it only captured a single snapshot in time, it cannot determine whether cannabis use led to depression, whether depression led to cannabis use, or whether both conditions were driven by shared underlying factors. Proving causation would require a different type of study that follows people over time.
Why was the association found in women but not men?
The study cannot answer this definitively. Possible explanations include biological sex differences in the endocannabinoid system, different patterns of cannabis use between genders, differences in how men and women report depressive symptoms, or different social and environmental factors that influence both cannabis use and depression. It is also worth noting that the men’s result was statistically imprecise, meaning the study may simply have lacked the power to detect an association in men even if one exists.
Should I stop using cannabis if I have depression?
This study alone does not provide a basis for a blanket recommendation to stop. However, the finding that women who had quit for at least one month had lower odds of depressive symptoms is worth discussing with your physician. A trial period of cessation, combined with close monitoring of mood symptoms, can be a reasonable and informative step. Any changes to cannabis use or depression treatment should be made in partnership with a healthcare provider who understands your full medical history.
Does this study say cannabis is safe for the heart?
Not exactly. The study found no statistically significant association between cannabis use and cardiovascular disease in this sample, but the absence of a significant finding is not the same as proof of safety. The study population was relatively young, the cardiovascular outcome relied on self-report, and a cross-sectional snapshot may miss risks that develop over longer periods of use. Other research has identified potential cardiovascular concerns with cannabis use, so this single null finding should not be interpreted as a clean bill of health.
References
- Ali OH. (2015). Association of Cannabis Use with Depressive Symptoms and Cardiovascular Diseases: A Cross Sectional Analysis. PhD Dissertation, The Ohio State University. ProQuest #10144860.
- Leggett T. (2006). A review of the world cannabis situation. Bulletin on Narcotics, 58(1-2), 1-155.
- Degenhardt L, et al. (2008). Toward a global view of alcohol, tobacco, cannabis, and cocaine use: findings from the WHO World Mental Health Surveys. PLoS Medicine, 5(7), e141. doi:10.1371/journal.pmed.0050141
- Kandel DB, et al. (1997). Prevalence and demographic correlates of symptoms of last year dependence on alcohol, nicotine, marijuana and cocaine in the U.S. population. Drug and Alcohol Dependence, 44(1), 11-29. doi:10.1016/S0376-8716(96)01315-4
- Cascini F, Aiello C, Di Tanna G. (2012). Increasing delta-9-tetrahydrocannabinol content in herbal cannabis over time: systematic review and meta-analysis. Current Drug Abuse Reviews, 5(1), 32-40. doi:10.2174/1874473711205010032
- Pacher P, Mechoulam R. (2011). Is lipid signaling through cannabinoid 2 receptors part of a protective system? Progress in Lipid Research, 50(2), 193-211. doi:10.1016/j.plipres.2011.01.001
- Rajesh M, et al. (2007). CB2-receptor stimulation attenuates TNF-alpha-induced human endothelial cell activation, transendothelial migration of monocytes, and monocyte-endothelial adhesion. American Journal of Physiology – Heart and Circulatory Physiology, 293(4), H2210-H2218. doi:10.1152/ajpheart.00688.2007
- Glass M, Dragunow M, Faull RL. (1997). Cannabinoid receptors in the human brain: a detailed anatomical and quantitative autoradiographic study in the fetal, neonatal and adult human brain. Neuroscience, 77(2), 299-318. doi:10.1016/S0306-4522(96)00428-9
- Mechoulam R, et al. (1995). Identification of an endogenous 2-monoglyceride, present in canine gut, that binds to cannabinoid receptors. Biochemical Pharmacology, 50(1), 83-90. doi:10.1016/0006-2952(95)00109-D
- Gunduz-Cinar O, et al. (2013). Convergent translational evidence of a role for anandamide in amygdala-mediated fear extinction, threat processing and stress-reactivity. Molecular Psychiatry, 18(7), 813-823. doi:10.1038/mp.2012.72
Have thoughts on this? Share it:


