Board-Certified Family Physician | CMO, CED Clinic |
Cannabis Science
A University of Colorado Boulder study of 171 anxious adults found that a person’s baseline level of systemic inflammation predicts how well cannabis works for anxiety and sleep quality. CBD-rich products reduced negative affect consistently across all inflammation levels. THC-dominant products helped only in people with average inflammation, not those at either extreme. No cannabis product of any type improved sleep in people with low baseline inflammation.
Why Your Inflammation Level May Predict How Cannabis Works for Anxiety and Sleep
A peer-reviewed study in Frontiers in Behavioral Neuroscience adds a precision-medicine dimension to cannabis care that most patients and clinicians have not yet heard about. Your body’s baseline inflammatory state may be quietly shaping whether the cannabis you use for anxiety or sleep is actually working, and which type of product is most likely to work for you specifically.
Strong Clinical Relevance
This study directly addresses one of the most common patient complaints in clinical cannabis practice: inconsistent response. The moderation finding by inflammation state offers a plausible, biologically grounded explanation and points toward CBD-rich formulations as a more reliable starting point.
Anxiety
CBD vs. THC
Sleep
Inflammation
- Why baseline inflammation may explain inconsistent cannabis results in people using it for anxiety and sleep
- How CBD-rich and THC-dominant products behave differently depending on your inflammatory state
- What the study’s 4-week design, 171 participants, and three chemovar groups actually showed
- How these findings apply to a clinical cannabis conversation in Massachusetts and beyond
- What the study cannot tell us yet, including dose specificity, route of administration, and long-term effects
- In 171 adults with anxiety, those who used CBD-rich cannabis improved on anxiety and sleep measures at every level of baseline inflammation.
- THC-dominant products only produced consistent mood improvements in people with average systemic inflammation, not at the high or low extremes.
- Cannabis use did not change cytokine concentrations over 4 weeks, but existing inflammation shaped how well it worked.
- The non-cannabis group did not improve sleep quality at any inflammation level, which strengthens the interpretation that cannabis was responsible for sleep changes in those who used it.
| Study type | Randomized observational study with ad libitum cannabis use; four-group longitudinal design |
| Sample size | 171 adults with mild-to-severe anxiety (147 cannabis users, 24 non-users) |
| Duration | 4 weeks (baseline and week-4 visits) |
| Interventions | THC+CBD flower (12%/12%), THC-dominant (24%/<1% CBD), CBD-dominant (<1% THC/24%), or no cannabis |
| Primary outcomes | DASS-21 (depression, anxiety, stress), Pittsburgh Sleep Quality Index (PSQI), plasma cytokines (IL-1a, IL-1b, IL-6, IL-8, IL-12, TNFα) |
| Key finding | Baseline cytokine levels moderated cannabis effects on anxiety (p<0.001) and sleep (p=0.04); CBD-rich products consistently improved both outcomes regardless of inflammation |
| Institution | University of Colorado Boulder, Institute of Cognitive Science and Department of Psychology and Neuroscience |
| Journal | Frontiers in Behavioral Neuroscience, Vol. 19, July 2025 |
| Limitations | No placebo control; ad libitum use without dose standardization; predominantly White Colorado sample; 4-week window only; flower products only |
Why This Matters
One of the most frustrating conversations in clinical cannabis practice is with a patient who says “it just doesn’t work for me.” Until recently, there was little scientific scaffolding to explain why two people with similar anxiety levels could have such different responses to the same product. This paper offers a biologically grounded mechanism: systemic inflammation, measured through circulating cytokine concentrations, may quietly determine whether a cannabis product does what a patient hopes it will do. CBD-rich products appear to sidestep this constraint. THC-dominant products do not. That distinction has real implications for how clinicians should approach product selection, particularly for patients who have tried cannabis and reported no benefit.
What the Study Found
Researchers at the University of Colorado Boulder enrolled 171 adults who met criteria for mild or greater anxiety (GAD-7 score of 5 or above, mean 11.8). Participants were randomly assigned to use one of three commercially available cannabis flower products for four weeks: a balanced THC+CBD formulation (12% THC / 12% CBD), a THC-dominant product (24% THC / less than 1% CBD), or a CBD-dominant product (less than 1% THC / 24% CBD). A fourth group of 24 individuals with anxiety used no cannabis during the study. Products were purchased by participants at a local Colorado dispensary, reflecting real-world use patterns. Participants were free to use as much or as little as they chose. Blood was collected at baseline and at week four to measure plasma cytokine concentrations and cannabinoid levels.
The central finding is a three-way interaction: cannabis chemovar, time, and baseline cytokine concentration all together determine the trajectory of anxiety and sleep outcomes. CBD-rich products, whether CBD-dominant or balanced THC+CBD, produced significant improvements in DASS-21 total scores at low, average, and high baseline inflammation levels. The CBD group specifically outperformed the non-cannabis group in the rate of negative-affect reduction at every inflammatory tier. THC-dominant products were a different story. They produced significant reductions in anxiety and depression only at average cytokine levels. When inflammation was very low or very high, THC-dominant use was not associated with significant improvements. For sleep, no cannabis product moved the needle in people with low baseline inflammation. At average and high inflammation levels, all three cannabis groups improved Pittsburgh Sleep Quality Index scores compared to baseline, with CBD-containing products showing a tendency to outperform THC-dominant products. Importantly, cannabis use did not change cytokine concentrations during the four weeks, ruling out a simple anti-inflammatory mechanism as the explanation for symptom improvement.
What This Paper Does Not Show
The study does not establish that measuring cytokines before starting cannabis will reliably predict outcomes in a clinical setting. The inflammation measure used here is a composite z-score across six cytokines, not a single, clinically accessible biomarker. The study also does not show which THC dose, frequency, or delivery method would produce different results. Products were flower-based; whether edibles, tinctures, or concentrates would follow the same moderation pattern is not addressed. The sample was predominantly White adults in Colorado with legal access to cannabis, which limits generalizability to populations with different baseline inflammatory profiles, health histories, or access to regulated products.
The four-week window does not speak to what happens at six months or a year. The absence of a placebo control means expectation effects cannot be ruled out, particularly given that participants knew their product type via the required label. The non-cannabis group did show some improvement in DASS total at average and high inflammation levels, which introduces some regression-to-the-mean interpretation. Finally, the study was not powered to compare specific subscales of anxiety, depression, and stress with the same confidence as the composite score.
How This Fits With the Broader Clinical Conversation
The inconsistency of cannabis outcomes across patients is one of the most persistent challenges in this field. Researchers have attempted to explain variability through dosing, frequency, delivery method, endocannabinoid system genetics, and prior cannabis experience. Inflammation as a moderator is a newer hypothesis, but it has biological plausibility. CBD has been shown in preclinical work to suppress inflammatory cytokine release through mechanisms that do not require CB1 or CB2 activation, including suppression of the NF-kappaB pathway and upregulation of STAT3 in immune cells. THC, by contrast, appears to have more receptor-dependent and dose-sensitive immunomodulatory effects. If those preclinical findings hold at real-world blood concentrations, they would predict exactly what this study found: CBD provides a more consistent anti-inflammatory signal across a wider inflammatory range, while THC’s effects are narrower.
This connects to a broader trend in cannabis research toward stratified or personalized medicine. The Lancet Psychiatry meta-analysis published in March 2026 found limited evidence for medicinal cannabis in anxiety and depression broadly. But meta-analyses pool heterogeneous populations, which can obscure the effects that exist for specific subgroups. A finding like this one suggests that a subgroup defined by inflammatory state might respond very differently from the pooled average. That doesn’t mean the Lancet analysis was wrong. It means the question of “does cannabis work for anxiety” may be less useful than “for whom, with what product, and at what baseline biological state.”
In clinical practice, many of the patients who report the most benefit from medical cannabis are those dealing with chronic pain, autoimmune conditions, or post-viral syndromes, populations often carrying elevated inflammatory burden. This study’s finding that people with higher baseline cytokines showed greater improvements in sleep quality from cannabis is consistent with what clinicians using cannabis with these patients have reported observationally for years. It does not prove causation, but it adds structure to a clinical intuition that has been hard to quantify.
When a patient tells me cannabis stopped working, or that it never worked for them at all, I used to have to say something vague about individual variability. Now I have a better starting point for the conversation. The finding that THC-dominant products lose their effect at extremes of inflammation is clinically important. It suggests that for patients who already have significant systemic inflammation, or conversely, patients who have very low baseline inflammation perhaps well-regulated, active, healthy adults, THC-only cannabis may not be the right first choice. CBD or a balanced CBD/THC product covers more ground.
I also want to name what this study does carefully: it measures flower-based ad libitum use in Colorado adults at commercially available concentrations. These are real-world doses and real-world habits, not the locked-down clinical trial doses that sometimes make cannabis research feel irrelevant to actual patients. That external validity matters. At the same time, the lack of a placebo, the four-week cap, and the mostly White Colorado sample are genuine constraints. I would not read this paper and immediately shift all anxiety patients to CBD-only products. I would read it as one more piece of evidence that CBD-rich formulations are more forgiving, and that THC-dominant options deserve more individualized assessment before recommending them as a primary approach. For the many Massachusetts patients I see who have tried cannabis and felt nothing, or felt worse, their inflammatory state is now on my checklist of things worth considering.
What a Careful Reader Should Take Away
This paper offers a genuinely new angle on a persistent clinical problem, but it does not change the evidence base overnight. CBD-rich cannabis products continue to accumulate support as a more consistent and forgiving choice across populations. THC-dominant products appear to have a narrower window of effectiveness that may depend on biological factors patients cannot easily know without lab work. The practical implication is not “get a cytokine panel before starting cannabis,” but rather “CBD-containing products are a lower-risk starting point for anxiety and sleep in most patients, and THC-heavy products warrant more individualized assessment.”
What this paper does not do is establish proof of benefit for any cannabis product in clinical anxiety disorders, revise the conclusions of large systematic reviews, or provide enough information to use cytokine levels as a clinical decision-making tool. The finding is mechanistically interesting, biologically plausible, and clinically suggestive. It is an invitation to design better-controlled, longer-term studies that account for inflammatory state at baseline. For now, it adds a useful dimension to the clinician-patient conversation, and that alone is worth reading carefully.
Read This Paper Through Eight Different Lenses
A single study can mean different things depending on who is reading it. This card separates the patient takeaway, clinical meaning, skepticism, study critique, prior research context, practical implications, future directions, and likely public misreadings.
How to use this: Choose a lens above to see how the same paper reads from a different evidence, clinical, or practical angle.
Patient Takeaway
If you have tried cannabis for anxiety or sleep and found it either hit-or-miss or entirely ineffective, this research suggests your body’s inflammatory state might be part of the explanation. In this study, people using CBD-rich products improved their anxiety scores and sleep quality at every level of baseline inflammation measured. People using THC-heavy products only improved when their inflammation was at an average level, not when it was unusually low or unusually high. The non-cannabis group did not improve sleep quality at all, which gives some weight to the idea that cannabis was genuinely doing something for sleep in those who used it. What this does not mean: you should not try to guess your inflammatory state, nor should you switch products without talking to a clinician. What it does mean: if you have found CBD-containing products more consistently helpful than THC-heavy ones, there is now biological research consistent with that experience. This study also does not mean cannabis works for everyone with anxiety. It is a four-week study of 171 adults in Colorado using specific flower products, and it cannot tell you whether a given product will help you specifically. These findings are most useful as a starting point for a conversation with a clinician about product type, your health history, and what you have already tried.
Clinician’s POV
The clinically actionable dimension of this paper is the THC moderation finding. CBD-rich and balanced products produced consistent improvements across the inflammatory spectrum, which supports what many clinicians using cannabis therapeutically have observed: CBD formulations are more predictable in their response profile. The THC-only window being restricted to average inflammation levels raises a practical question about how to identify patients who may not respond well to THC-dominant products before they try them. The study cannot answer that directly, but it gives clinicians a new domain to ask about: chronic inflammatory conditions, recent infections, immunological history, and even lifestyle factors known to affect cytokine levels. There is no validated clinical test to run before product selection, but the concept of inflammatory status as a moderating variable is now supported by a prospectively designed study. For patients who have tried THC-heavy cannabis and felt nothing or felt worse, this paper offers a framing that goes beyond “you just didn’t respond.” For product selection counseling in Massachusetts and other regulated markets where CBD-rich flower is accessible, this finding provides an additional reason to lean toward balanced or CBD-dominant formulations as first-line options for anxiety and sleep, particularly in patients with known inflammatory conditions.
A Skeptical Read
The absence of a placebo control is the most significant limitation in this design, and it is one the authors acknowledge directly. Participants knew which product they were assigned to because Colorado regulations require cannabinoid content on the label. Someone who believes CBD is calming may report less anxiety because they believe they received the calming product. The magnitude of the effect size in the CBD group at low inflammation is large enough that placebo alone is unlikely to fully account for it, but the possibility cannot be excluded. The cannabis non-use group also showed improvement in DASS total at average and high inflammation, which complicates interpretation. The authors invoke regression to the mean as a possible explanation for that, which is plausible but not conclusively established. The predominantly White Colorado sample introduces demographic specificity that makes generalizing to a Boston patient population non-trivial. Many patients seeking medical cannabis in Massachusetts are older, carry more chronic disease burden, and may have inflammatory profiles that differ systematically from healthy Colorado adults in their early thirties. This study does not tell you what happens in those populations, and extrapolating directly from this sample would outrun the data.
Study Critic
The core statistical finding is a three-way generalized estimating equations interaction: time, group, and baseline cytokine z-score. The DASS moderation reached p less than 0.001 and the PSQI moderation p equals 0.04, which is statistically significant but notably weaker for the sleep outcome. In a study with several outcomes and multiple comparisons, a p-value of 0.04 for the sleep moderation warrants caution. The post hoc power analysis showed adequate power for a Cohen’s f as small as 0.17, which is a modest effect. The cytokine composite used in the analysis averages z-scores across six pro-inflammatory cytokines, which reduces individual specificity. It is possible that individual cytokines such as IL-6 or TNFa are driving the moderation while others are noise, but the study design cannot separate those contributions. The ad libitum use design, while realistic, means the statistical models could not include frequency of use or blood cannabinoid concentration because the non-use group had zero values for both, introducing a structural constraint that limits what can be attributed to dose. The 24-hour abstinence period before each visit means blood cannabinoids at the week-4 visit reflect stable trough concentrations, not acute peak exposure, which affects interpretation of the cannabinoid verification data.
Compared to Past Research
This paper extends previous work from the same research group at the University of Colorado Boulder. Prior analyses from the same longitudinal dataset (Bidwell et al. 2023 and 2024) reported improvements in sleep and anxiety from cannabis use, with CBD-containing products generally outperforming THC-dominant ones on anxiety measures. This paper adds the inflammatory moderation layer to that foundation, which is genuinely new. The broader literature on cannabis for anxiety has been mixed, with the March 2026 Lancet Psychiatry meta-analysis finding limited evidence across the pooled literature for cannabinoids in anxiety and depression. This study does not contradict those findings; rather, it suggests that pooling heterogeneous samples may obscure meaningful subgroup effects. The preclinical literature on CBD’s immunosuppressive mechanisms through the NF-kappaB pathway and STAT3 upregulation provides biological scaffolding for the moderation hypothesis, though the study did not observe cannabis-mediated changes in cytokine concentrations directly. The comparison to prior literature is constrained here because few previous studies assessed inflammatory status as a moderator of cannabis outcomes; the authors note only one prior study examined cytokines in the context of anxiety treatment and it used a non-pharmacological intervention. This paper is, to their knowledge, the first to examine this moderation in a cannabis chemovar context.
Practical Considerations
Colorado’s regulated market provided access to standardized, ISO-certified flower products, which is a far cry from what patients in many markets face. Massachusetts has a regulated medical and adult-use market, which provides some comparability, but product variability in THC-to-CBD ratios is real, and what a product label says does not always match what a patient consumes or absorbs. The ad libitum use pattern, while realistic, also means some patients may have used products in ways that differ significantly from others in the same group. In practice, a CBD-rich flower product at 24% CBD does not translate straightforwardly to a CBD tincture or an edible, and the sleep and anxiety benefits observed in this study may not replicate with other delivery methods. The practical implication that clinicians can act on today is modest but real: when a patient with anxiety asks whether to try cannabis, and the choice is between a THC-heavy product and a CBD-containing one, this paper adds evidence-based support for recommending a CBD-containing option first. For patients who have already tried THC-dominant products without benefit, this research offers a biologically coherent explanation that may help them understand their experience and stay open to trying a different formulation.
Future Directions (Expected)
The most important next step is a placebo-controlled design. Federal scheduling constraints have made this difficult in the United States, but the April 2026 rescheduling of state-licensed medical cannabis to Schedule III may open new research pathways that were previously blocked. A follow-up study should measure individual cytokines rather than a composite z-score to identify which inflammatory markers carry the most predictive weight. A study that stratifies participants by clinically accessible proxies of inflammation, such as high-sensitivity CRP or standard metabolic markers, would be far more translatable to clinical practice than a research-grade cytokine panel. Longer follow-up is needed: four weeks is enough to see an initial signal but not enough to determine whether the inflammatory moderation persists or reverses at six months or a year. Expansion beyond flower products to edibles, oils, and sublingual preparations would address the route-of-administration gap. A replication in a clinically defined population with DSM-5 generalized anxiety disorder, rather than the broader screening threshold used here, would sharpen the clinical relevance considerably. And a racially and geographically diverse sample is essential to assess whether the finding holds across different baseline inflammatory profiles.
Misreadings & Bad-Faith Takes
One likely misreading: “Study proves CBD cures anxiety.” This is a distortion. The paper found that CBD-rich products were associated with improvements in anxiety scores over four weeks in adults with mild-to-severe self-reported anxiety. Correlation across four weeks is not proof of cure, and anxiety scores improving in a study does not translate to clinical remission of an anxiety disorder. Another predictable distortion: “Cannabis does not work for anxiety unless your inflammation is just right.” This misreads the THC finding as a finding about all cannabis. CBD-rich products worked across every inflammation level in this study; the inflammation sensitivity was specific to THC-dominant products. A third misreading, from the opposite direction: “This study shows inflammation testing should be part of cannabis consultations.” The study does not support that conclusion. The cytokine panel used is a research tool, not a clinical one, and there is no validated protocol for translating composite cytokine z-scores into product recommendations. Finally: “CBD is now proven to be better than THC for anxiety.” The study found CBD-rich products to be more consistent, not categorically superior in all patients for all outcomes. THC-containing products, including the balanced THC+CBD group, also showed significant improvements in this study. The distinction is about consistency and range, not a binary verdict.
? Join the Conversation
Have a question about how this applies to your situation? Ask Dr. Caplan →
Want to discuss this topic with other patients and caregivers? Join the forum discussion →
Have thoughts on this? Share it:
Lisano JK, Skrzynski CJ, Giordano G, Bryan AD, Bidwell LC. Inflammatory state moderates response to cannabis on negative affect and sleep quality in individuals with anxiety. Frontiers in Behavioral Neuroscience. 2025;19:1549311. doi: 10.3389/fnbeh.2025.1549311 | PubMed: 40667474
Frequently Asked Questions
What does this study actually mean for someone using cannabis for anxiety?
The study suggests that your baseline level of systemic inflammation may shape how well cannabis works for anxiety. CBD-rich products showed consistent reductions in anxiety scores across all inflammation levels tested. THC-dominant products only produced significant improvements in people with average inflammation, not at low or high levels. If you have tried a THC-heavy product and felt no benefit, your inflammatory state could be part of the explanation. The practical takeaway is not to order a cytokine panel, but to consider whether a CBD-containing formulation might be worth trying if a THC-dominant one has not helped.
Does this study prove that CBD is better than THC for anxiety?
Not categorically. The study found that CBD-rich products produced more consistent improvements across the range of inflammatory states studied. The balanced THC+CBD product also showed significant improvements in anxiety scores at all inflammation levels. THC-dominant products worked at average inflammation but not at the extremes. This suggests CBD-containing formulations are more predictable, not that THC has no value for anxiety. The paper does not establish superiority in a clinical treatment sense; it identifies a pattern in a four-week observational study of 171 adults.
Can inflammation testing tell me in advance which cannabis product will work for me?
No, not yet. The inflammation measure used in this study is a composite research score averaging six plasma cytokine levels measured through specialized lab work. It is not the same as a standard clinical blood test. There is currently no validated protocol for translating a cytokine level into a cannabis product recommendation. What this study contributes is a proof-of-concept that baseline inflammation plays a role in cannabis response, which could eventually lead to clinically useful biomarker tools, but that research has not yet been done.
Why did the non-cannabis group also improve in some mood measures?
Participants in the non-use group showed improvements in DASS total scores at average and high inflammation levels. The researchers suggest this may reflect regression to the mean, meaning that people with higher initial anxiety scores are statistically likely to score somewhat lower on a follow-up measure regardless of treatment, simply because extreme scores at one time point tend to be less extreme at the next. The cannabis groups, particularly the CBD group, showed significantly greater rates of improvement than the non-use group, which gives the cannabis finding additional weight beyond what natural fluctuation would explain.
Does this study apply to edibles, tinctures, or vape products?
The study used only commercially available cannabis flower products. Whether the same relationship between inflammation and cannabis response would hold for edibles, tinctures, concentrates, or other delivery methods is not established by this research. Edibles have a different pharmacokinetic profile than inhaled flower, with slower onset, delayed peak, and different bioavailability. The moderation by inflammation might look different across delivery methods, or it might not apply at all. This is a genuine gap that future research should address.
How is systemic inflammation measured in everyday terms?
In this study, researchers measured six specific proteins in the blood called cytokines: IL-1a, IL-1b, IL-6, IL-8, IL-12, and tumor necrosis factor-alpha. These are secreted by immune cells in response to injury, infection, or chronic inflammatory conditions. In clinical medicine, inflammation is more commonly estimated using a simpler marker called high-sensitivity C-reactive protein (hsCRP). Whether hsCRP or other accessible biomarkers would perform similarly as moderators of cannabis response is not yet known, though it is a reasonable hypothesis for future research to test.
Is there a connection between chronic conditions and how well cannabis works for sleep?
This study’s sleep findings are consistent with that possibility. Cannabis improved sleep quality most reliably in people with average or high baseline inflammation. Many people with chronic inflammatory conditions, including autoimmune disorders, chronic pain, or post-viral syndromes, often carry elevated cytokine levels. The observation that cannabis improved sleep more consistently in higher-inflammation participants aligns with what clinicians have reported observationally in patients with those conditions. The study was not designed to test this directly in a clinical population, but the finding points toward a biologically coherent reason to expect that some of the strongest cannabis responders for sleep may be people with significant baseline inflammatory burden.
Were there any safety concerns or adverse events in the study?
The study used a licensed clinical psychologist to monitor participants for clinically significant events throughout the four weeks, including worsening anxiety, paranoia, and increased heart rate. No clinically significant adverse events occurred during the study period. All cannabis users had prior experience with the substance, which is consistent with the inclusion criteria requiring that participants be of legal age and prior users. This does not mean the products are without risk for all people, but within this particular study population, no serious adverse outcomes were observed over four weeks.
What does it mean that cannabis did not change cytokine levels over four weeks?
One hypothesis the researchers tested was whether cannabis might improve anxiety and sleep by reducing inflammation directly. If that were the mechanism, you would expect to see cytokine levels fall in cannabis users over the study period. They did not. Cannabis use was not associated with any significant change in cytokine concentrations over four weeks. This means the improvement in mood and sleep was not mediated by measurable anti-inflammatory changes in the blood. Instead, the research supports a different picture: whatever your inflammation level is when you start cannabis, that level shapes how well it works, rather than being changed by it.
How should Massachusetts patients and clinicians think about these findings?
Massachusetts has a regulated medical and adult-use cannabis market with access to products spanning a wide range of THC-to-CBD ratios. For clinicians counseling patients on product selection for anxiety or sleep, this study adds a layer of evidence-based reasoning for recommending CBD-containing formulations as a first-line starting point, particularly when the patient’s inflammatory status is unknown. For patients who have already tried THC-heavy products without benefit, this research provides a plausible biological explanation and an invitation to revisit the conversation about product type. It does not change the fundamental clinical principle that cannabis care should be individualized, dosed carefully, and revisited over time.
