Cannabis for Depression: UK Registry's 698-Patient Real-World Data

RCTs Aren’t the Only Game in Town

If you read the Cooling scoping review and the Lancet Psychiatry meta-analysis, you came away with a sober conclusion: the randomized controlled trial evidence for cannabis as a primary treatment for depression is, frankly, nearly nonexistent. One trial in the Cooling review. Zero in the Lancet meta-analysis. That’s the gold-standard literature, and it’s thin.

But RCTs aren’t the only kind of clinical evidence. Patient registries — large, prospectively collected datasets from regulated healthcare systems — tell a complementary story. They can’t prove causation, but they can show what happens to real patients, prescribed real products, by real clinicians, over real time.

In early 2026, Lillywhite, Erridge, and colleagues at Curaleaf Clinic published a two-year case series in the Journal of Affective Disorders analyzing 698 patients prescribed medical cannabis for primary depression through the UK Medical Cannabis Registry. It’s the largest dataset of its kind in the world. The headline: depression scores, anxiety scores, sleep, and quality of life all improved — and stayed improved — over 24 months. The fine print is more interesting.

Before we go further: This piece is science explanation, not medical advice. Depression is a serious, sometimes life-threatening condition. Do not start, stop, or substitute treatment based on a blog post. If you’re in crisis in the US, call or text 988 (Suicide and Crisis Lifeline). In the UK, call Samaritans on 116 123, free, any time.

What the UK Medical Cannabis Registry Is

Medical cannabis became legally prescribable in the UK in November 2018, but NHS prescribing is vanishingly rare; almost all UK medical cannabis is dispensed through private specialist clinics. To address the “prescribing without data” problem, Curaleaf Clinic (formerly Sapphire Medical Clinics) established the UK Medical Cannabis Registry in December 2019 — a prospective registry that captures formulations, doses, adverse events, and patient-reported outcomes for everyone treated through their service.

Patients consent to data collection at their initial consultation. Validated questionnaires — PHQ-9 (depression), GAD-7 (anxiety), the Single-Item Sleep Quality Scale (SQS), and the EQ-5D-5L (quality of life) — are administered remotely at baseline, 1, 3, 6 months, and every six months thereafter. Adverse events are recorded as they occur.

By January 2025, the registry contained 34,563 patients. Of those, 698 (about 2%) had primary depression as their indication for cannabis. That subset is the dataset for the new paper.

Who the 698 Were

This is not the demographic you might assume.

• Mean age: about 37 years
• Sex: 71.4% male
• Baseline depression severity: average PHQ-9 in the moderate-to-severe range
• Comorbid severe anxiety at baseline: 50.9% scored GAD-7 ≥ 15

The male skew is striking — depression in the general UK population is more common in women — and probably reflects who is willing to pay out of pocket for private cannabis prescribing (a few hundred pounds a month is typical) and who self-identifies as cannabis-receptive. In earlier registry analyses, most patients prescribed cannabis for depression had previously used it recreationally and had failed at least one prior antidepressant. This is a treatment-resistant or treatment-fatigued population that found its way to a private cannabis clinic — hard to help by conventional means, and highly motivated for cannabis to work.

What Improved

Across all four primary measures, the registry showed statistically significant improvement at every time point — 1 month, 3 months, 6 months, 12 months, 18 months, and 24 months — versus baseline (p < 0.001):

• PHQ-9 (depression): Mean scores fell from moderate-to-severe at baseline into a substantially lower range. By 24 months, about 62% of patients had achieved a “minimal clinically important difference” — a change of around five PHQ-9 points, the threshold the field uses for “this is a real change, not noise.”
• GAD-7 (anxiety): Scores fell sharply. By 24 months, about 80% of patients reported a clinically meaningful reduction in anxiety. Given that half of the cohort had severe anxiety at baseline, this is the largest signal in the dataset.
• SQS (sleep): Scores improved, though changes were smaller than for mood — a pattern consistent across multiple registry analyses.
• EQ-5D-5L (health-related quality of life): Index values improved across the board.

The temporal pattern is the most interesting feature. The largest improvements happened in the first 1–3 months. After three months, scores stabilized — they didn’t keep falling, but they didn’t bounce back either. That’s exactly what you’d predict from a real treatment effect plus partial regression to the mean: an early steep response followed by a maintained plateau.

The depression-anxiety correlation at baseline (r = 0.67, p < 0.001) is a reminder of how tightly these conditions co-travel. People with severe depression usually have severe anxiety; products and protocols that move both are valuable. These patients aren’t dealing with one symptom at a time.

The 9% Adverse Event Rate

This is where registry data earn their keep. RCTs are too small and too short to characterize a real-world adverse-event profile; registries do that almost by accident.

Of the 698 patients, 63 (9.0%) reported at least one adverse event during treatment. Of the 411+ events recorded, 87.3% were mild or moderate. There were no life-threatening or disabling adverse events reported.

The most common adverse events in the depression cohort and prior registry analyses included:

• Fatigue and lethargy
• Insomnia or disrupted sleep (paradoxical, but real — particularly with stimulating chemovars)
• Dry mouth (xerostomia)
• Headache
• Dizziness
• Somnolence (drowsiness during the day)

What’s notably not prominent in the registry: psychotic episodes, mania, or severe new-onset anxiety. Those events do happen with cannabis — the acute psychosis literature is real — but in a screened, prescribed population at therapeutic doses, they are rare.

In an earlier registry analysis, adverse-event incidence was not significantly associated with THC or CBD dose. That’s a useful clinical signal: side effects aren’t a simple dose-response story, and patient-specific variation (chemotype, prior tolerance, comorbidities, concurrent medications) matters more than any single dose threshold.

A 9% adverse-event rate is comparable to or lower than the rate seen in trials of standard SSRIs for depression. This isn’t a “cannabis is safer than antidepressants” claim — different populations, different durations, different reporting standards — but it is a “the safety profile in this regulated context is reassuring.”

Why This Doesn’t Replace RCTs

A 698-patient case series with two-year follow-up sounds like a lot of evidence — and in some ways it is. But it cannot do the one thing RCTs can: establish causation.

There’s no placebo arm, no control group, no blinding. Patients knew they were getting cannabis; so did their clinicians and the analysts. Every classical bias is on the table:

• Placebo and expectancy effects. Patients who paid privately for a novel medicine they hoped would help are primed to feel better. Depression has notorious placebo response rates — often 30–40% in RCTs.
• Regression to the mean. Patients enroll when they’re at their worst and tend to drift toward their mean even without treatment.
• Selection bias. Patients who responded poorly or had side effects probably stopped attending and dropped out of follow-up. The 24-month cohort is enriched for responders.
• Concurrent care. Many patients were on antidepressants, in therapy, or making other life changes alongside cannabis.

The authors are honest. Their conclusion is careful: “Limitations of the study mean that no causal relationship can be ascertained.” This isn’t proof. It’s a strong signal to take into properly controlled trials.

What It DOES Add

What the registry does uniquely contribute — what no RCT yet has — is real-world durability and tolerability data:

• Two years of follow-up in a chronic condition where six weeks is the typical RCT length
• A sustained plateau after early improvement, not a wash-out
• Adverse-event rates and types at therapeutic doses in a treatment-resistant population
• A safety profile that does not feature the dramatic psychiatric harms cannabis is sometimes accused of producing in clinical contexts

For clinicians, this is genuinely useful. For patients considering whether cannabis can play a role in their care, it answers a different question than “does cannabis cure depression” (RCT territory) — it answers “what tends to happen to people like me who try this in a regulated way over months and years” (registry territory).

The Practical Picture: What Was Prescribed

In earlier registry analyses of depression patients, the prescribing pattern was roughly:

• About 47% on vaporized dried flower (standardized cultivar, documented THC and CBD content)
• About 16% on oils only (typically sublingual, balanced or CBD-leaning)
• About 26% on both flower and oils, often with oil for daytime/baseline coverage and flower for breakthrough symptoms

Median initial THC daily doses ran around 100–200 mg of total THC across the day (this is total drug content, not the dose felt at any one moment), with CBD typically around 5–100 mg. Dosing varied widely. The common protocol is start low, titrate slowly, monitor with the same questionnaires used in the registry.

What’s worth noting for our audience: this is closer to the balanced and relax-leaning end of the chemotype space than the high-THC stimulating end. Limited registry detail means we can’t tell you exact terpene profiles, but the supportive evidence for linalool, myrcene, and limonene in mood and sleep contexts is well-trodden ground in the broader literature.

In strain terms, the products closest to what UK clinic patients are prescribed — moderate THC, real CBD presence, calming-but-not-sedating — map to chemovars like Harlequin, Cannatonic, and ACDC. For more THC-forward but mood-supportive profiles, look to the uplift family or balance family depending on time of day. Our guides on cannabis and depression, stress relief, and evening wind-down are organized around exactly these chemotype principles.

For US Readers

A note on context. The UK access model is specialist-only, private-pay, and regulated — closer to a pharmaceutical model than a US state medical program. UK patients receive a single product with documented cannabinoid content, dosed by a clinician, with mandatory questionnaires and follow-up. That’s a meaningfully different exposure than what most US patients buy at a dispensary.

This doesn’t make the registry data useless for US readers — pharmacology is pharmacology — but the safety and efficacy signal applies most cleanly to standardized, monitored, lower-to-moderate-dose use under clinical supervision. It is not evidence for “smoke whatever’s at the dispensary for depression and you’ll be fine.” Concentrate use, high-dose daily THC without medical context, and cannabis use disorder are not what the registry is studying. The safest version of cannabinoid-assisted mood care looks more like the UK protocol: lower THC, real CBD, careful titration, validated check-ins, and a clinician who knows your full medical history.

The Patient Conversation

If you’re considering cannabinoids alongside depression treatment, the responsible version of that conversation includes things the registry implicitly captures and your prescriber explicitly should:

• Your full psychiatric history — including personal or family history of psychosis, bipolar disorder, or severe panic. A 9% adverse-event rate in a screened population isn’t the rate in an unscreened one.
• Every medication you’re taking. SSRIs, MAOIs, benzodiazepines, sleep medications, blood thinners — cannabis interacts with several.
• What you’ve tried already. Most registry patients had failed prior antidepressants. That context helps any clinician calibrate.
• What success looks like for you. Symptom relief, functional improvement, reduced reliance on something else — different goals lead to different protocols.
• A monitoring plan. PHQ-9 and GAD-7 take five minutes each. If your provider isn’t tracking your scores, ask.

Tracking Your Own Response

Here is the deepest takeaway from a registry study, and the reason we built TIWIH the way we did. Aggregate registry findings tell you what tends to happen across 698 people. They do not tell you what will happen to you.

The way you find out is the way the registry found out: log what you used, when, how much, and how you felt — across PHQ-9 and GAD-7 windows of two weeks at minimum, with a single product held reasonably constant. That’s the basic experiment. Run it for 8–12 weeks. Bring the data to your prescriber. The conversation that follows is qualitatively different from the one that begins with “I think it might be helping.”

That structured self-tracking is exactly what the TIWIH app is for. Same logic as the registry. Smaller n. But in the only n that ultimately matters to you.

Sources

• Lillywhite E, Erridge S, Clarke E, et al. UK Medical Cannabis Registry: A two-year case series of clinical outcomes in depression. Journal of Affective Disorders. 2025. DOI: 10.1016/j.jad.2025.121130.
• Mangoo S, Erridge S, Holvey C, et al. Assessment of clinical outcomes of medicinal cannabis therapy for depression: analysis from the UK Medical Cannabis Registry. Expert Review of Neurotherapeutics. 2022;22(11–12):995–1008. DOI: 10.1080/14737175.2022.2161894.
• Erridge S, Salazar O, Kawka M, et al. An initial analysis of the UK Medical Cannabis Registry: Outcomes analysis of first 129 patients. Neuropsychopharmacology Reports. 2021;41(3):362–370. DOI: 10.1002/npr2.12183.
• Ergisi M, Erridge S, Harris M, et al. An updated analysis of clinical outcome measures across patients from the UK Medical Cannabis Registry. Cannabis and Cannabinoid Research. 2023.
• Curaleaf Clinic plain-language summary: Real-world data on medical cannabis for depression.
• UK Medical Cannabis Registry: ukmedicalcannabisregistry.com.
• Companion analysis on RCT-level evidence: Cannabis and Mental Health: What 28 Clinical Trials Actually Show and The Lancet Cannabis Mental Health Study: What It Actually Found.

---

Crisis support: US — call or text 988 (Suicide and Crisis Lifeline). UK — call Samaritans on 116 123 (free, 24/7) or text SHOUT to 85258. If you or someone you know is in immediate danger, call your local emergency number.

Disclosure of interest: The Lillywhite et al. study was conducted by clinicians at Curaleaf Clinic using data from a registry that Curaleaf Clinic owns and operates; Curaleaf Clinic also dispensed the products studied. The authors disclose this in the paper, and it’s a reason to read the data with eyes open. It’s also a reason the underlying data still matters — you don’t get a 698-patient prospective cohort without somebody running the clinic.

This article is general science writing, not medical advice. It should not be used to diagnose, treat, or self-manage any condition. Talk to a qualified clinician before changing any psychiatric treatment.