What does the research say?
In a study of 62 people with type 2 diabetes, participants were randomly assigned to five groups; for 13 weeks one group took pure CBD, one took an inactive substance (placebo), and the others took another substance found in cannabis (THCV) or combinations. No clear effect on blood sugar or "good cholesterol" (HDL) was reported in the CBD group compared with placebo. A drop in fasting blood sugar was seen only in the group taking THCV. In an 8-week study of 50 people with type 2 diabetes in Iran, an under-the-tongue spray containing very low doses of CBD and THC was reported to lower blood sugar and blood fats compared with an inactive substance. However, the numbers in the publication are inconsistent in many places (for example, in the table HbA1c goes from 7.94 to 7.8, but this is written as an 8.5% drop; the dose is given in micrograms in one place and milligrams in another), only people who completed the study were analyzed, and the result has not been tested in another study; so it is hard to rely on this result.
In a 4-week study of 16 overweight people without diabetes, CBD did not change the body's ability to handle sugar (glucose tolerance). In another study of 15 healthy people, a single dose of CBD did not affect blood sugar or fat metabolism after a meal; however, participants who took CBD ate 193 more calories at lunch. In a single-dose study of 14 overweight men, CBD also did not change blood sugar after a meal; only insulin and triglycerides were measured slightly lower in the first half hour after eating. In a 6-week study of 65 overweight people, a CBD-rich cannabis extract only increased HDL; no difference in body composition was seen compared with an inactive substance. In people with obesity, a weight loss of 2.6–4.1 kg over 3 months was reported with an oral strip containing THCV and CBD; however, only people who completed the study were analyzed, the groups were small, some numbers in the publication are inconsistent, and the result has not been confirmed. An obesity study of nabilone, a synthetic drug similar to THC, was stopped early because participants could not tolerate the drug.
In an 8-week study of 25 people with fatty liver disease, whose results are available only in the ClinicalTrials.gov record, three different doses of CBD did not clearly reduce liver fat compared with an inactive substance; more than half of those taking CBD had diarrhea. Based on these results, no well-designed human study was found that supports the claim that "CBD regulates blood sugar." Some population studies find type 2 diabetes less often in cannabis users; however, most of these studies measure a single moment in time and do not show cause and effect. The results of three large studies that followed people for years also disagree: in the US study, prediabetes was more common in cannabis users; in the Swedish study, the difference disappeared once age was taken into account; and in the UK study of 91,000 people, the risk of type 2 diabetes was slightly lower; in the same study, moderate use in people with obesity was linked to a higher risk of metabolic disease. Genetic analyses did not suggest that cannabis use changes the risk of type 2 diabetes or body weight; a genetic analysis for fatty liver also found no link, but it included only a small number of patients. The promise of weight control is also unsupported by evidence.
Strength of evidence: ○○○○ Insufficient evidence. Finding: No effect found. What does this mean?
Ongoing research on this topic
Trials testing CBD for this topic on ClinicalTrials.gov: 1 under way (1 recruiting); results are awaited from 2 that ended in the last two years. A trial being under way does not show that CBD works.
- NCT07298408: Phase 1/2, recruiting, target 20 participants
- NCT06648031: Phase 1, completed, results awaited, 148 participants
- NCT07162454: No phase (not a drug trial), completed, results awaited, 23 participants
Registries in other countries also list 1 recruiting trial on this topic: other registries.
Possible side effects and drug interactions
In the type 2 diabetes study, CBD was well tolerated. In the fatty liver study, the most common side effect in people taking CBD was diarrhea (12 of 20 people; none of the 5 people taking placebo). With the THC-like drug nabilone, tested for obesity, drowsiness and dizziness were common; all 4 people on the high dose left the study because of side effects. Among people who used a THCV-containing product for 3 months and took a urine test, 7 of 9 tested positive for THC. Because the studies on this topic are small, information about side effects is limited.
These studies did not look at drug interactions separately. CBD can slow the breakdown of some medicines in the body; this matters for people who take medicines regularly.
Interactions with medicines used for this condition
Interactions reported for these medicines in peer-reviewed publications and official prescribing information. A medicine that is not listed can still interact with CBD; tell your doctor about every medicine you take.
| Medicine | Importance | Warning | Source |
|---|---|---|---|
| Diabetes medicines | Information | Diabetes medicines: CBD did not improve blood sugar or insulin results compared with placebo. Do not replace your diabetes medicines with CBD. | PMID 27573936, PMID 41167732 |
To check interactions with the medicines you take: CBD drug interaction checker.
How was CBD used in studies?
In the studies, CBD was taken by mouth: in type 2 diabetes, 100 mg twice a day for 13 weeks (study dose; not a personal recommendation); in overweight people, 30 mg twice a day for 4 weeks (study dose; not a personal recommendation); in overweight men, a single 30 mg dose with a meal (study dose; not a personal recommendation); in healthy people, a single 298 mg dose (study dose; not a personal recommendation); and in fatty liver disease, a total of 200, 400 or 800 mg a day for 8 weeks (study dose; not a personal recommendation). None of these doses had a clear effect on blood sugar or liver fat.
Study details (for health professionals)
CBD did not improve blood sugar or insulin results compared with placebo; no human RCT was found that supports the claim that "CBD regulates blood sugar." In fatty liver disease, in the manufacturer's small 8-week RCT (25 people; results only in the ClinicalTrials.gov record, no peer-reviewed publication), three different doses of CBD did not significantly reduce liver fat compared with placebo. After a single dose of CBD, more food was eaten at lunch; the promise of weight control is unsupported by evidence. Pharmaceutical-grade CBD, on the other hand, can reduce appetite: in a meta-analysis combining RCTs of approved cannabinoids in different conditions, decreased appetite was more common with CBD than with placebo (OR 2.46; moderate certainty). According to the US prescribing information for the approved CBD medicine, losing at least 5% of body weight was also more common in people taking CBD in high-dose controlled epilepsy studies (16% vs 8% in Lennox–Gastaut and Dravet syndromes, 31% vs 8% in tuberous sclerosis). This is an unwanted side effect; it is not evidence supporting use for weight loss. There are three small RCTs reporting positive results with THC-containing products and other substances in cannabis (a very low-dose CBD:THC spray in type 2 diabetes, an oral strip containing THCV and CBD in obesity, and the THC-like drug nabilone); all three have serious limitations: inconsistent numbers in the publication, analysis of only those who completed the study, or early stopping because of poor tolerability. In single-dose or 4-week studies compared with placebo or a control condition, THC, nabilone or marijuana did not clearly change blood sugar regulation (in a single-dose mechanistic study in healthy men, nabilone slightly increased insulin sensitivity). In population studies, type 2 diabetes and fatty liver are seen less often in cannabis users, but most of these studies are cross-sectional; for type 2 diabetes, the results of three large studies that followed people for years disagree, and genetic analyses did not support the idea that cannabis use changes the risk of type 2 diabetes or body weight. These observational data do not show cause and effect. In type 1 diabetes, ketoacidosis was more common in cannabis users (three observational studies; the largest with 2,108 adolescents and young adults with type 1 diabetes in the US) and HbA1c was higher (observational). Delayed stomach emptying due to diabetes (diabetic gastroparesis) is covered on the site's Irritable Bowel and Stomach Problems page: in a 4-week single-center RCT of 44 patients with gastroparesis, 12 of them diabetic, pharmaceutical-grade CBD reduced symptoms more than placebo but did not speed up gastric emptying (it slowed it on some measurements); no separate result was reported for people with diabetes, and the finding has not yet been tested in another study (Zheng 2023, PMID 37482172). In people with type 1 diabetes who use cannabis daily, cannabinoid hyperemesis syndrome can be mistaken for gastroparesis (Akturk 2025, PMID 40741458). No treatment study with a cannabis or CBD product was found for diabetic eye damage (diabetic retinopathy). The only controlled study in this condition with a drug targeting cannabinoid receptors tested an experimental synthetic drug that is not a cannabis product and binds selectively to the CB2 receptor (vicasinabin): in 139 patients, after 36 weeks of treatment, the rate of at least a two-step improvement in retinopathy severity was no different from placebo (the manufacturer-funded CANBERRA study, PMID 39802207). Because selective synthetic drugs are not counted as cannabis products on this site, it is not in the table. Diabetes medicines should not be replaced with CBD. October 2026 search (PubMed, Cochrane, ClinicalTrials.gov): 14 controlled studies of cannabinoids on this topic (12 randomized, one with registry results only; between 7 and 125 people, the longest 13 weeks), 6 systematic reviews on CBD or cannabis use and 9 observational studies were found (including one meta-analysis and one observational study added in the completion round on October 28, 2026); all are in the table below.
Evidence assessment: Insufficient (negative)
CBD alone
In type 2 diabetes, in overweight people and in fatty liver disease, pure CBD did not affect blood sugar, blood lipid or liver fat results clearly differently from placebo; in a single-dose study in overweight men, only insulin and triglyceride values right after a meal were briefly measured lower (all small, short studies; the fatty liver study's results are only in the registry).
Products containing THC
Controlled studies of THC-containing products are very small and short. The numbers in the only study reporting a positive result in type 2 diabetes, with a very low-dose CBD:THC spray, are inconsistent; the obesity study of the THC-like drug nabilone was stopped early because of poor tolerability; dronabinol over 4 weeks did not change metabolic values, and a single use of marijuana or an edible product did not change insulin sensitivity. Some population studies report less type 2 diabetes in cannabis users, but these do not show cause and effect and were not supported by genetic analyses; in type 1 diabetes, ketoacidosis was more common in cannabis users (observational).
Evidence: Insufficient (small, short and conflicting studies)
Dose range in studies
In studies, 30 mg as a single dose or 2 times a day, 100 mg 2 times a day, 298 mg single dose, and 200–800 mg a day in fatty liver
Starting dose and titration
Not reported in detail in the studies.
These are doses researchers gave under medical supervision; they are not a recommended dose for you. "mg/kg/day" means milligrams per kilogram of body weight per day.
| Study | Participants | Product | Dose and titration | Duration | Result | Source |
|---|---|---|---|---|---|---|
| Jadoon 2016 Double-blind, placebo-controlled, 5-arm pilot RCT | 62 randomized (CBD 13, THCV 12, CBD:THCV 1:1 11, 20:1 12, placebo 14); 56 completed Type 2 diabetes | Pure CBD; THCV; combinations | CBD 100 mg × 2/day | 13 weeks | No difference between CBD and placebo in the primary outcome, HDL (difference −0.03 mmol/L; 95% CI −0.10 to 0.04; p = 0.41; registry results); no significant glycemic effect of CBD compared with placebo was reported either (only resistin and GIP changed from baseline); a drop in fasting glucose only in the THCV group (−1.2 mmol/L vs placebo; p < 0.05) Side effects: Well tolerated | PMID 27573936, doi:10.2337/dc16-0650, NCT01217112 |
| Ewell 2026 Double-blind, placebo-controlled pilot RCT | 16 Overweight/obese, without diabetes | Pure CBD | 30 mg × 2/day | 4 weeks | Did not affect glucose tolerance Side effects: Not reported (in the abstract) | PMID 41167732, doi:10.1177/25785125251391085, NCT05285449, investigator, institution, dose, duration and title match this record, the registry enrolled 30 people, the publication has 16) |
| Johnson 2026 Double-blind, crossover RCT | 15 Healthy | CBD isolate | 298 mg single dose | Single dose | No difference in postprandial glucose or fat metabolism; 193 kcal more eaten at lunch Side effects: Not reported (in the abstract) | PMID 41825697, doi:10.1016/j.appet.2026.108531, NCT05618756, institution, dose, design, number of participants and primary outcomes match this record) |
| Afshar 2022 Double-blind, placebo-controlled RCT (phase 1 according to the authors) | 50 randomized (25/25); 46 analyzed (24/22, completers only) Type 2 diabetes (HbA1c ≤ 9%), added to existing treatment (Iran) | Under-the-tongue spray, CBD:THC 10:1 (100 µg CBD + 10 µg THC per spray; CBDEX10, Yas Daru) | 2 sprays twice a day; according to the methods section, a total of about 0.4 mg CBD + 0.04 mg THC a day (the results section says '200 mg/20 mg') | 8 weeks | According to the publication, at week 8 total cholesterol (−19.7 mg/dL), triglycerides (−27.8 mg/dL), LDL (−5.4 mg/dL), fasting blood sugar (−12 mg/dL), HbA1c and insulin fell significantly compared with placebo (p < 0.05). However, the numbers in the publication are inconsistent in many places: according to the table, HbA1c fell from 7.94% to 7.8% in the spray group (about 0.14 points) and from 7.41% to 7.35% in the placebo group, but the same table gives these changes as −8.5% and +1.3%; in the spray group LDL is 98.5 mg/dL both at baseline and at week 8, yet a 4.7% drop is written; the abstract gives the HbA1c difference in 'mg/dL' units; the dose is in micrograms in the methods and milligrams in the results; the number of people analyzed differs between tables (46 and 42). Only completers were analyzed, and the result has not been confirmed in another study. Because of these inconsistencies, the result cannot be considered reliable Side effects: No serious side effects reported; 1 person in the spray group was withdrawn from the study because of dizziness | PMID 36945340, doi:10.5812/ijpr-132647, Iranian Registry of Clinical Trials IRCT20121021011192N10 |
| NCT01284634 (manufacturer's fatty liver study, registry results) Randomized, placebo-controlled, dose-ranging pilot study (phase 2; partially blinded according to the registry title) | 25 randomized (CBD 200 mg: 7, 400 mg: 6, 800 mg: 7, placebo: 5); 21 completed Adults with fatty liver disease (liver fat measured by MRI) | Purified plant-derived CBD capsule (GWP42003, GW Research) | A total of 200, 400 or 800 mg a day (in 100 mg capsules, twice a day) | 8 weeks | For the primary outcome, liver fat (liver triglyceride by MRI, percent change from baseline), none of the three doses differed significantly from placebo: differences vs placebo −6.9 (95% CI −16.4 to 2.6; p = 0.22), −9.4 (−19.7 to 1.0; p = 0.13) and −6.5 (−17.2 to 4.1; p = 0.30). For blood lipids, the registry has only mean changes, with no statistical comparison. No peer-reviewed publication; a very small study Side effects: No serious side effects; the most common was diarrhea (12 of 20 people in the CBD groups, none of 5 on placebo); stopping due to side effects 3 of 20 on CBD, 1 of 5 on placebo | NCT01284634, no publication found in PubMed in October 2026) |
| Smith 2025 Randomized (2:1:1), double-blind, placebo-controlled parallel study | 56 applied; 52 assigned to groups; 44 completed and were analyzed (low dose 24, high dose 10, placebo 10) Adults with obesity (BMI ≥ 30) without diabetes; mean age 52 (US) | An oral adhesive strip containing THCV and CBD isolate (THCV is another cannabinoid found in cannabis; no THC was detected in independent testing) | Once a day: 8 mg THCV + 10 mg CBD (1 strip) or 16 mg THCV + 20 mg CBD (2 strips) | 90 days | According to the publication, over 90 days weight fell by 2.6 kg at the low dose, 4.1 kg at the high dose and 0.1 kg with placebo (comparison with placebo p = 0.025 and 0.003); LDL fell more than with placebo at the high dose (p = 0.024); no difference in fasting blood sugar, HbA1c changes 0–0.1 points. Only completers were analyzed, a one-sided test was used, and the high-dose and placebo groups had 10 people each; some numbers in the publication are inconsistent (in the low-dose group, LDL in the table rises from 101.5 to 103.0 mg/dL while a 7.4 mg/dL drop is written). Single author; the author's affiliation is a company, the study was self-funded and no conflict of interest was declared; not confirmed in another study Side effects: No serious side effects reported; the most common was finding the taste unpleasant (38–40%); in the active groups, mild ALT elevation in 2 people and mild AST elevation in 1 (ALT also in 1 person on placebo). Of the 9 people in the high-dose group who took a urine THC test, 7 tested positive (rapid home test; no test was done before the study) | PMID 39968488, doi:10.26828/cannabis/2024/000206, NCT05574049 |
| Lopez 2020 Double-blind, placebo-controlled RCT | 65 Overweight (mean BMI 28.5) but otherwise healthy adults | Commercial hemp extract oil (PlusCBD Extra Strength; 60 mg extract a day, 15 mg hemp-derived CBD) | 60 mg extract a day (15 mg CBD) | 6 weeks | The study's main focus was stress resilience and well-being; there was no significant difference from placebo in these. Of the metabolic values, only HDL ("good cholesterol") rose significantly in the hemp group (group × time p = 0.004); no significant between-group differences were reported in body composition or other blood lipids Side effects: Well tolerated; liver, kidney and blood values stayed within normal limits (in the abstract) | PMID 32456572, doi:10.1080/19390211.2020.1765941, NCT04294706 |
| Matheson 2025 Double-blind, placebo-controlled, 3-group pilot RCT; stopped early | 18 randomized; 15 took at least one dose; 8 completed (low dose 4, placebo 4, high dose 0) Adults aged 25–45 with obesity but otherwise healthy; no cannabinoid use in the past 6 months (Canada) | Nabilone (THC-like synthetic cannabinoid, oral capsule) | 2 mg (low dose) or 6 mg (high dose) a day Dose increased gradually (registry: titrated to 2 mg or 6 mg); schedule not in the abstract | 12 weeks | The study was stopped early because the drug was poorly tolerated (reason in the registry: poor tolerability and no preliminary efficacy seen). In an analysis of only the 8 completers, the publication reported a significant reduction in weight and BMI at the low dose compared with placebo (p < 0.001); with groups of 4 people each, this result is not reliable. Blood sugar and insulin were secondary outcomes; the registry has no statistical comparison Side effects: All 4 people on the high dose stopped because of side effects; drowsiness (low dose 3 of 5, high dose 4 of 4, placebo 0 of 6), dizziness (3 of 5, 3 of 4, 0 of 6) and feeling intoxicated (1 of 5, 3 of 4, 0 of 6) were common; dry mouth was also common in the placebo group (3 of 6); no serious side effects | PMID 40353878, doi:10.1089/can.2025.0034, NCT04801641, April 9, 2026) |
| Reichenbach 2015 Double-blind, placebo-controlled pilot RCT | 13 completed (dronabinol 7, placebo 6) Adults with non-cardiac chest pain (esophageal hypersensitivity) without heart disease or serious comorbidities; this publication looks at metabolic values | Dronabinol (synthetic THC, by mouth) | 5 mg twice a day | 28 days | There was no significant change in either group after 4 weeks in BMI, HDL, triglycerides, LDL, fasting blood sugar, insulin, leptin, liver enzymes or CRP; a very small study Side effects: Not reported (in the abstract) | PMID 26283236, doi:10.1016/j.clinthera.2015.07.023, NCT01598207 |
| Chia 2026 Randomized, double-blind, placebo-controlled, 4-period crossover physiology study | 34 randomized; 21 completed all four visits Healthy, non-obese men with normal glucose tolerance (ages 21–55, US) | Nabilone (THC-like synthetic cannabinoid); for comparison, the experimental CB1 receptor-blocking drug CP-945,598 | Nabilone 2 mg single dose; CP-945,598 15 or 45 mg | Single dose (at each visit; visits at least 6 weeks apart) | In glucose and insulin clamp measurements, insulin secretion did not differ from placebo with any drug. Nabilone increased insulin sensitivity (glucose use) compared with placebo in the last 20 minutes of the euglycemic clamp (p = 0.039); the CB1 blocker decreased it at the high dose. A single-dose mechanistic study; it does not show a treatment effect in patients Side effects: Not reported (no side effect data found in the full text) | PMID 41869774, doi:10.1152/ajpendo.00391.2025, NCT01517100 |
| Abbotts 2022 Randomized, placebo-controlled crossover study (two visits with a meal test; per the registry, only outcome assessors were blinded) | 14 men Overweight or obese (BMI ≥ 25) adult men with no known metabolic disease (US) | Pure CBD (water-soluble powder formulation; CBD from the company funding the study) | 30 mg single dose (with a meal) | Single dose (at each visit) | In the meal test, CBD did not affect the blood sugar response differently from placebo (p = 0.31); insulin was lower at minutes 10 and 20 after the meal (p = 0.013), and triglycerides at minute 30 (p = 0.010); the meal-induced increase in energy expenditure did not change. The study's main aim was to compare how well different CBD products are absorbed; a small, single-dose study. Funded by the company that supplied the CBD (according to the authors, it had no role in design or analysis) Side effects: Side effects not reported separately; liver and kidney values stayed within normal limits | PMID 35631293, doi:10.3390/nu14102152, NCT04971837 |
| Ewell 2021 Randomized-order crossover study with a cannabis-free control product (blinding not reported; participants bought and used the products themselves) | 7 Adults who use cannabis regularly (US) | 5 commercially sold edible cannabis products (each 10 mg THC) and a cannabis-free control product | 10 mg THC single dose | Single dose (6 separate mornings, at least 1 week apart) | In a glucose tolerance test done 30 minutes after taking the product, insulin sensitivity (Matsuda index) did not differ from the control condition (p > 0.395); blood sugar and insulin (measured for two products) did not differ either. The study's main aim was to compare how the products are absorbed in the body; a 7-person study without blinding. Funded by the maker of some of the products tested Side effects: Not reported (in the abstract) | PMID 34451914, doi:10.3390/ph14080817 |
| Bryan 2026 (single-use study) Quasi-randomized (with the right to refuse the assigned product), placebo-free experimental study comparing three products; before-and-after measurement | 108 started (34 CBD-dominant, 28 THC-dominant, 46 THC + CBD); per the registry, 98 completed follow-up Healthy young adults without diabetes who had not used cannabis in the past 3 months (but had tried it at least once), of varying body weights (US; ages 21–40 in the methods, 25–45 in the abstract, mean 29.7) | Cannabis flower from the legal market: CBD-dominant, THC-dominant or containing THC + CBD (smoked or vaporized at home) | Amount chosen by each person; blood levels were measured | Single use | Insulin sensitivity (Matsuda index) after use did not differ from baseline; inflammatory markers rose as blood THC levels rose. According to the table in the registry, mean blood sugar in the glucose tolerance test was 8–13 mg/dL higher after use in all three groups (there are no statistics for this change in the registry or the publication). There was no placebo group and no true randomization (a limitation noted by the authors) Side effects: Not reported (in this publication) | PMID 42463843, doi:10.1038/s41598-026-62837-0, NCT04114903, August 6, 2026), methods publication Skrzynski 2026, PMID 41820717 |
| Bryan 2025 (4-week use study) Controlled, non-randomized comparison: users were quasi-randomly assigned to one of three products and compared with non-users; no placebo | 125 started: 97 cannabis users (9 CBD-dominant, 47 THC-dominant, 41 THC + CBD) and 28 non-users; 116 completed Healthy adults aged 21–40 without diabetes who had used cannabis at least once a week for at least a year or had not used it at all in the past year (US) | Cannabis flower from the legal market: THC-dominant (23% THC, 0% CBD), THC + CBD (10% THC, 8% CBD) or CBD-dominant (20% CBD, 1% THC) | Amount chosen by each person, 4 weeks | 4 weeks | After 4 weeks, the THC:CBD ratio of the products did not change insulin sensitivity (Matsuda index) compared with non-users; inflammatory markers were lower in users than in non-users. Because cannabis use was not randomly assigned, cause and effect cannot be established Side effects: Not reported (in the abstract) | PMID 40324550, doi:10.1016/j.amjmed.2025.05.002, NCT04114903, August 6, 2026) |
| Pinto and Martel 2022 (systematic review) Systematic review (no meta-analysis) | 11 RCTs Placebo-controlled RCTs of CBD reporting appetite or weight (mostly as a side effect); most were epilepsy studies | CBD (mostly pharmaceutical-grade pure CBD; in one study a CBD-dominant cannabis extract, in one CBD-rich inhaled cannabis) | Between 5 mg/kg and 1,500 mg a day | Search up to July 31, 2022 | In the abstract, the authors conclude that CBD reduces appetite and that this is associated with weight loss; however, according to the discussion section, only three studies measured weight (one of them a type 2 diabetes study), and none of the three found a weight difference compared with placebo. In seven studies, decreased appetite was more common in people taking CBD (as a dose-related side effect in the epilepsy studies); in one study appetite increased. Most studies had a risk of bias at the 'some concerns' level; no meta-analysis was done | PMID 36180814, doi:10.1007/s40261-022-01205-y |
| Spanagel and Bilbao 2021 (meta-analysis) Comparative systematic review and meta-analysis (RCTs only; registered in PROSPERO; Cochrane risk-of-bias tool and GRADE) | For appetite, 15 RCTs, 1,974 people (for sleep, 17 RCTs, 1,479 people); how many studies went into the CBD analysis is not in the abstract People with physical or mental illness; unwanted effects of approved cannabinoids on sleep and appetite | Approved cannabinoids: dronabinol, nabilone and CBD | Decreased appetite was more common with CBD than with placebo (OR 2.46; 95% CI 1.74–4.01; moderate certainty); THC given as a medicine (dronabinol, nabilone) did not decrease appetite, and neither had a negative effect on sleep. According to the authors, the side effect of decreased appetite with approved cannabinoids appears to come from CBD. It examines appetite, not body weight; it does not answer the question of treatment for weight loss. Fixed-effect model; because the full text could not be read, the studies and conditions included in the CBD analysis could not be confirmed | PMID 34181977, doi:10.1016/j.neuropharm.2021.108680, PROSPERO CRD42021229932. Full text could not be read (paywalled, subscription required on Europe PMC), row based on the PubMed abstract. The same study also appears in the non-cancer appetite loss table | ||
| Porr 2020 (rapid review) Rapid systematic review (for Diabetes Canada) | 6 observational studies (2 full articles, 4 conference abstracts) People aged 13 and over with type 1 or type 2 diabetes | Recreational cannabis use | Not applicable | Search January 2008–January 2019 | In type 1 diabetes, five studies reported higher HbA1c in cannabis users; in one study, use in the past 12 months was associated with about twice the risk of diabetic ketoacidosis (OR 1.98; 95% CI 1.01–3.91). In type 2 diabetes, increases in vascular blockage, heart attack and kidney problems were reported. The methodological quality of the studies was weak or moderate | PMID 32807222, doi:10.1186/s13643-020-01411-9, PROSPERO CRD42019122829 |
| Ravi 2018 (systematic review) Systematic review (observational studies; strength of evidence graded) | 13 studies for risk factors (11 for outcomes) Adults who use marijuana in any form | Marijuana use | Not applicable | Search 1975–September 2017 | All 6 studies suggesting a metabolic benefit were cross-sectional and were not supported by prospective studies; the evidence on marijuana's effect on diabetes and blood lipids was found to be insufficient | PMID 29357394, doi:10.7326/M17-1548, PROSPERO CRD42016051297 |
| Mousavi 2023 (meta-analysis) Systematic review and meta-analysis (observational studies) | 7 publications (11 cross-sectional surveys, 4 cohorts) General population | Cannabis use | Not applicable | Search up to July 1, 2022 | The odds of type 2 diabetes were lower in cannabis users (OR 0.48; 95% CI 0.39–0.59); however, heterogeneity between studies was very high and most of the data come from cross-sectional surveys: in a cross-sectional design it cannot be known which came first, and cause and effect is not shown | PMID 37526051, doi:10.1002/ptr.7973 |
| Quinapanta Castro 2025 (meta-analysis) Systematic review and meta-analysis (observational studies) | 11 studies, about 6 million people Adults; fatty liver (MASLD/NAFLD) identified by different methods | Cannabis use | Not applicable | Search up to January 2025 | The odds of fatty liver were lower in cannabis users (OR 0.58; 95% CI 0.42–0.81), but heterogeneity between studies was very high (I² 97%) and no significant association was seen in frequent users. There was no difference in directly measured liver fat (two studies), liver enzymes or triglycerides. Most studies were cross-sectional with a high risk of bias; cause and effect cannot be established | PMID 41141207, doi:10.7759/cureus.93183, PROSPERO CRD42025025065 |
| Bancks 2015 (CARDIA) Prospective cohort (18-year follow-up) and cross-sectional analysis | 3,151 people (follow-up analysis); 3,034 people (year-25 cross-sectional analysis) US community sample followed from young adulthood to middle age (no diabetes or prediabetes at the start of follow-up) | Marijuana use (self-reported) | Not applicable | 18 years | People who had used marijuana 100 or more times in their lifetime had a higher risk of prediabetes (HR 1.39; 95% CI 1.13–1.71); no association with diabetes was found. In the cross-sectional analysis at year 25, current users also had higher odds of prediabetes (OR 1.65; 1.15–2.38) | PMID 26364621, doi:10.1007/s00125-015-3740-3 |
| Danielsson 2016 Population-based prospective cohort (Sweden) | 17,967 people; 608 new cases of type 2 diabetes Swedish women and men aged 18–84 | Cannabis use (questionnaire) | Not applicable | About 9 years (2002–2011) | In the crude analysis, the risk of type 2 diabetes appeared lower in cannabis users (OR 0.68; 95% CI 0.47–0.99), but the association disappeared once age was taken into account (OR 0.94; 0.63–1.39) | PMID 27843955, doi:10.1155/2016/6278709 |
| Zhou 2026 (UK Biobank) Prospective cohort (UK Biobank) | 91,002 people (without metabolic disease at baseline) UK general population sample (UK Biobank participants) | Cannabis use (self-reported) | Not applicable | Follow-up duration not in the abstract | After accounting for other factors, the risk of type 2 diabetes was slightly lower in cannabis users (aHR 0.82; 95% CI 0.69–0.99); the risk of obesity was lower in frequent users (aHR 0.57; 0.41–0.78); there was no association with high cholesterol or fatty liver. The association varied by body weight: the inverse association was seen mostly in people of normal weight; in people with obesity (BMI > 30), moderate use was associated with a higher risk of metabolic disease (aHR 1.26) and high blood pressure (aHR 1.40). Observational study; does not show cause and effect | PMID 42115075, doi:10.1016/j.numecd.2026.104762 |
| Baumeister 2021 (Mendelian randomization) Two-sample Mendelian randomization (an analysis that uses genetic predisposition as a natural experiment) | Genome studies: cannabis use 184,765 people; type 2 diabetes 74,124 cases, 824,006 controls Summary data from large genome studies | Genetic predisposition to cannabis use | Not applicable | Not applicable | Genetic predisposition to lifetime cannabis use (OR 1.00; 95% CI 0.93–1.09) and to cannabis use disorder (OR 1.03; 0.99–1.08) did not change the risk of type 2 diabetes; according to the authors, cannabis use does not play a causal role in type 2 diabetes | PMID 33580533, doi:10.1111/adb.13020 |
| Alshaarawy 2019 (NESARC) Nationally representative prospective study (US) | About 33,000 people US population aged 18 and over | Cannabis use (reported in interview) | Not applicable | 3 years (2001–02 → 2004–05) | Over three years, BMI rose in all groups; in cannabis users (those who quit, new users and persistent users), the increase was slightly smaller than in never-users (persistent users β = −1.26; 95% CI −1.81 to −0.72). Height and weight self-reported; cause and effect cannot be established | PMID 30879064, doi:10.1093/ije/dyz044 |
| Alayash 2021 (Mendelian randomization) Two-sample Mendelian randomization | Cannabis use genome study 184,765 people; BMI up to 322,154 people; waist circumference and waist-to-hip ratio 210,088 people Genome study participants of European ancestry | Genetic predisposition to cannabis use | Not applicable | Not applicable | Genetic predisposition to lifetime cannabis use was not associated with BMI (β −0.018; 95% CI −0.05 to 0.015); a causal relationship was not supported for waist circumference or waist-to-hip ratio either | PMID 34304124, doi:10.1016/j.drugalcdep.2021.108863 |
| Wang 2020 (Mendelian randomization) Two-sample Mendelian randomization | Cannabis use genome study 184,765 people; fatty liver in UK Biobank 1,122 cases, 399,900 controls Genome study participants of European ancestry | Genetic predisposition to cannabis use | Not applicable | Not applicable | Genetic predisposition to lifetime cannabis use, cannabis dependence or cannabis use disorder did not significantly change the risk of fatty liver (NAFLD) (p > 0.05 in all tests). Only 1,122 people had a fatty liver diagnosis; the authors also note that the study may have been underpowered, so the result cannot be read as "no effect" | PMID 32922442, doi:10.3389/fgene.2020.00949 |
| Kinney 2020 (T1D Exchange) Cross-sectional analysis of patient registry data | 932 adults with type 1 diabetes (61 of them moderate cannabis users) Adults with type 1 diabetes (US, T1D Exchange registry) | Cannabis use (self-reported on a substance use scale) | Not applicable | Past 12 months (retrospective) | After accounting for sex, age and HbA1c, cannabis use was associated with about 2.5 times the odds of having diabetic ketoacidosis in the past 12 months (OR 2.5; 95% CI 1.0–5.9). The confidence interval is wide; the association does not show cause and effect | PMID 31628116, doi:10.2337/dc19-0365 |
| Roberts 2025 (SEARCH) Cross-sectional analysis at the SEARCH 4 follow-up visit of a multicenter cohort (SEARCH for Diabetes in Youth, US; visits 2016–2019) | 2,465 people (type 1 diabetes 2,108, type 2 diabetes 357) Adolescents and young adults aged 10–36 diagnosed with type 1 or type 2 diabetes in childhood or adolescence (US); in those with type 1 diabetes, mean age 21, diabetes duration 11 years, HbA1c 9.1% | Cannabis (marijuana) use, self-reported | Not applicable | Cannabis use lifetime and past 30 days; outcomes over the past 12 months (self-reported) | Of those with type 1 diabetes, 19.5% currently used cannabis and 19.0% had used it in the past (22.1% and 21.8% in type 2 diabetes). In type 1 diabetes, after accounting for race, income, parental education, age and sex, the odds of having diabetic ketoacidosis in the past 12 months were about twice as high in current users (aOR 2.01; 95% CI 1.49–2.70) and also higher in past users (aOR 1.72; 1.25–2.36); HbA1c was 0.61 points (95% CI 0.32–0.90) higher in current users. Hospitalization was significantly higher only in past users (aOR 1.46; 1.03–2.07; in current users 1.39; 0.99–1.95); there was no association with severe hypoglycemia. 17.3% of current users with type 1 diabetes said cannabis made it harder to remember diabetes care "sometimes or on most days." In type 2 diabetes, the number of people was too small to build a model; no association with HbA1c was seen. Cannabis use and outcomes are self-reported; a cross-sectional analysis does not show cause and effect (according to the authors, mental health problems such as depression may mediate part of the association) | PMID 39959518, doi:10.2337/ds24-0019, PMC11825399 (read from the PMC article page, the Europe PMC full-text service and NCBI efetch did not return the article body). SEARCH 4: supported by NIH (NIDDK) and the US Centers for Disease Control, the authors reported no relevant conflicts of interest |
Who should be careful
- Diabetes medicines should not be replaced with CBD
- High doses and over-the-counter doses cannot be compared: in low-dose (≤175 mg/day) studies, plasma levels were 5–100 times lower than with the 20 mg/kg Epidiolex dose (Warren 2026, PMID 42473010)
- Anyone taking long-term medication should consult a physician before starting; the studies on this page did not systematically assess interactions
- In type 1 diabetes, diabetic ketoacidosis was about twice as common in cannabis users (three observational studies, the largest with 2,108 adolescents and young adults with type 1 diabetes in the US; does not show cause and effect)
- In type 1 diabetes, recurrent episodes of nausea and vomiting during daily cannabis use may be cannabinoid hyperemesis syndrome; because gastric emptying tests can also come back delayed in this syndrome, it can be mistaken for diabetic gastroparesis. In a single-center retrospective review, all 11 patients diagnosed with gastroparesis while using cannabis daily met the criteria for this syndrome (Akturk 2025, PMID 40741458). Telling the physician about cannabis use is important for a correct diagnosis
Drug interactions
- The studies on this page did not systematically assess drug interactions; details are in the site's Drug interactions section
Abbreviations: RCT, randomized controlled trial (a study in which participants are assigned to groups by chance); placebo, an inactive substance; SMD, OR and RR, effect measures used in analyses that pool studies; CYP, liver enzymes that break down medicines.
If you are considering CBD
- Do not replace your diabetes medicines or insulin with CBD, and do not lower their dose on your own.
- Review the medicines you take with your doctor or pharmacist; checking for drug interactions is important.
- A product being sold with promises such as "balances blood sugar" or "helps you slim down" does not mean research has shown this.
- Check the product's label and certificate of analysis (COA); find out how much CBD it contains and whether it contains THC.
Choose your age group, medicines and situation to prepare a short summary you can take to your doctor. Nothing you enter is sent anywhere.
Bottom line
Current research does not show that CBD regulates blood sugar or body weight. On this topic it is important to go by evidence, not claims. Continuing your diabetes treatment together with your doctor is the safest path.
Frequently asked questions
Does CBD oil work for diabetes?
Current research does not support this. In a study of people with type 2 diabetes, CBD did not improve blood sugar results compared with an inactive substance (placebo). Small studies in overweight and healthy people also found no clear effect on blood sugar.
Does CBD lower blood sugar?
No human study showing this was found. In the type 2 diabetes study, a drop in fasting blood sugar was seen only in the group taking another substance found in cannabis (THCV), not in the CBD group. Some population studies report less diabetes in cannabis users, but these do not show cause and effect: the results of studies that followed people for years disagree, and genetic analyses did not support the idea that cannabis use changes diabetes risk.
Does CBD help with weight loss?
There is no evidence for this. In fact, in a study of healthy people, those who took a single dose of CBD ate 193 more calories at lunch. Decreased appetite is one of the common side effects of pharmaceutical-grade CBD used at high doses for epilepsy; a meta-analysis combining studies of approved cannabinoids in different conditions also shows, with moderate certainty, that CBD reduces appetite. According to the US prescribing information for this medicine, losing at least 5% of body weight was also more common in people taking CBD in controlled epilepsy studies: 16% vs 8% on placebo in Lennox–Gastaut and Dravet syndromes, and 31% vs 8% in tuberous sclerosis (study doses of 10–20 mg and 25 mg per kilogram of body weight per day; not a personal recommendation). This is an unwanted side effect seen at high doses in people with severe epilepsy; it does not show that CBD can be used to lose weight. In two small studies of CBD or a CBD-rich extract in people with type 2 diabetes and in overweight people, weight or body composition did not change differently from placebo. The only small study reporting weight loss with a product containing both THCV and CBD has serious methodological limitations. One population study also shows slightly less weight gain over the years in cannabis users; however, a genetic analysis did not support the idea that cannabis use changes body weight.
Can CBD be used together with diabetes medicines?
The diabetes studies did not look at drug interactions separately. Because CBD can slow the breakdown of some medicines in the body, talk to your doctor about it if you take medicines regularly. Do not replace your diabetes medicines with CBD.
Can CBD oil be obtained with a prescription for diabetes in Turkey?
In Turkey, products made from cannabis are regulated by the Kenevirden Elde Edilen Ürünlere Dair Yönetmelik (Regulation on Products Derived from Hemp) dated January 31, 2026. These products, including CBD, can be sold only in pharmacies. Those classed as medicines or health products are dispensed with a prescription; the regulation does not state a prescription requirement for supplement products and cosmetics with a THC content of at most 0.3%. As of October 2026, there is no approved (licensed) CBD medicine in Turkey. If you are thinking of using it, talk to your doctor first.
References
Show 29 references
- Jadoon 2016. Double-blind, placebo-controlled, 5-arm pilot RCT. PMID 27573936, doi:10.2337/dc16-0650, NCT01217112
- Ewell 2026. Double-blind, placebo-controlled pilot RCT. PMID 41167732, doi:10.1177/25785125251391085, NCT05285449, investigator, institution, dose, duration and title match this record, the registry enrolled 30 people, the publication has 16)
- Johnson 2026. Double-blind, crossover RCT. PMID 41825697, doi:10.1016/j.appet.2026.108531, NCT05618756, institution, dose, design, number of participants and primary outcomes match this record)
- Afshar 2022. Double-blind, placebo-controlled RCT (phase 1 according to the authors). PMID 36945340, doi:10.5812/ijpr-132647, Iranian Registry of Clinical Trials IRCT20121021011192N10
- NCT01284634 (manufacturer's fatty liver study, registry results). Randomized, placebo-controlled, dose-ranging pilot study (phase 2; partially blinded according to the registry title). NCT01284634, no publication found in PubMed in October 2026)
- Smith 2025. Randomized (2:1:1), double-blind, placebo-controlled parallel study. PMID 39968488, doi:10.26828/cannabis/2024/000206, NCT05574049
- Lopez 2020. Double-blind, placebo-controlled RCT. PMID 32456572, doi:10.1080/19390211.2020.1765941, NCT04294706
- Matheson 2025. Double-blind, placebo-controlled, 3-group pilot RCT; stopped early. PMID 40353878, doi:10.1089/can.2025.0034, NCT04801641, April 9, 2026)
- Reichenbach 2015. Double-blind, placebo-controlled pilot RCT. PMID 26283236, doi:10.1016/j.clinthera.2015.07.023, NCT01598207
- Chia 2026. Randomized, double-blind, placebo-controlled, 4-period crossover physiology study. PMID 41869774, doi:10.1152/ajpendo.00391.2025, NCT01517100
- Abbotts 2022. Randomized, placebo-controlled crossover study (two visits with a meal test; per the registry, only outcome assessors were blinded). PMID 35631293, doi:10.3390/nu14102152, NCT04971837
- Ewell 2021. Randomized-order crossover study with a cannabis-free control product (blinding not reported; participants bought and used the products themselves). PMID 34451914, doi:10.3390/ph14080817
- Bryan 2026 (single-use study). Quasi-randomized (with the right to refuse the assigned product), placebo-free experimental study comparing three products; before-and-after measurement. PMID 42463843, doi:10.1038/s41598-026-62837-0, NCT04114903, August 6, 2026), methods publication Skrzynski 2026, PMID 41820717
- Bryan 2025 (4-week use study). Controlled, non-randomized comparison: users were quasi-randomly assigned to one of three products and compared with non-users; no placebo. PMID 40324550, doi:10.1016/j.amjmed.2025.05.002, NCT04114903, August 6, 2026)
- Pinto and Martel 2022 (systematic review). Systematic review (no meta-analysis). PMID 36180814, doi:10.1007/s40261-022-01205-y
- Spanagel and Bilbao 2021 (meta-analysis). Comparative systematic review and meta-analysis (RCTs only; registered in PROSPERO; Cochrane risk-of-bias tool and GRADE). PMID 34181977, doi:10.1016/j.neuropharm.2021.108680, PROSPERO CRD42021229932. Full text could not be read (paywalled, subscription required on Europe PMC), row based on the PubMed abstract. The same study also appears in the non-cancer appetite loss table
- Porr 2020 (rapid review). Rapid systematic review (for Diabetes Canada). PMID 32807222, doi:10.1186/s13643-020-01411-9, PROSPERO CRD42019122829
- Ravi 2018 (systematic review). Systematic review (observational studies; strength of evidence graded). PMID 29357394, doi:10.7326/M17-1548, PROSPERO CRD42016051297
- Mousavi 2023 (meta-analysis). Systematic review and meta-analysis (observational studies). PMID 37526051, doi:10.1002/ptr.7973
- Quinapanta Castro 2025 (meta-analysis). Systematic review and meta-analysis (observational studies). PMID 41141207, doi:10.7759/cureus.93183, PROSPERO CRD42025025065
- Bancks 2015 (CARDIA). Prospective cohort (18-year follow-up) and cross-sectional analysis. PMID 26364621, doi:10.1007/s00125-015-3740-3
- Danielsson 2016. Population-based prospective cohort (Sweden). PMID 27843955, doi:10.1155/2016/6278709
- Zhou 2026 (UK Biobank). Prospective cohort (UK Biobank). PMID 42115075, doi:10.1016/j.numecd.2026.104762
- Baumeister 2021 (Mendelian randomization). Two-sample Mendelian randomization (an analysis that uses genetic predisposition as a natural experiment). PMID 33580533, doi:10.1111/adb.13020
- Alshaarawy 2019 (NESARC). Nationally representative prospective study (US). PMID 30879064, doi:10.1093/ije/dyz044
- Alayash 2021 (Mendelian randomization). Two-sample Mendelian randomization. PMID 34304124, doi:10.1016/j.drugalcdep.2021.108863
- Wang 2020 (Mendelian randomization). Two-sample Mendelian randomization. PMID 32922442, doi:10.3389/fgene.2020.00949
- Kinney 2020 (T1D Exchange). Cross-sectional analysis of patient registry data. PMID 31628116, doi:10.2337/dc19-0365
- Roberts 2025 (SEARCH). Cross-sectional analysis at the SEARCH 4 follow-up visit of a multicenter cohort (SEARCH for Diabetes in Youth, US; visits 2016–2019). PMID 39959518, doi:10.2337/ds24-0019, PMC11825399 (read from the PMC article page, the Europe PMC full-text service and NCBI efetch did not return the article body). SEARCH 4: supported by NIH (NIDDK) and the US Centers for Disease Control, the authors reported no relevant conflicts of interest
Points that could not be verified from the full text
- A meta-analysis reporting that cannabis and its "by-products" reduced body weight by an average of 1.87 kg (Reis 2024, PMID 37935909) was not included in the table: in addition to a few studies of cannabis products, the publication's reference list includes 8 publications on rimonabant, a weight-loss drug that is not a cannabis product and, on the contrary, blocks the CB1 receptor that THC binds to. The European Medicines Agency recommended suspending the sale of rimonabant in 2008 because it roughly doubled the risk of psychiatric disorders; the drug's marketing authorization was withdrawn in 2009. Because the full text could not be accessed, it could not be confirmed which studies went into this result; therefore the result cannot be counted as evidence that cannabis products cause weight loss.
- A study in which a drug company tested THCV added to metformin for 12 weeks in 207 people with type 2 diabetes (ClinicalTrials.gov NCT02053272) was completed in February 2016; as of October 2026, its results are neither in the registry nor in a published article. The result of the largest THCV study on this topic is therefore unknown.
Research history
Here we record what changed on this page, when and why, with dates. The newest change is at the top.
October 2026 additional review: sources found for this page during the searches for other topics were read and added to the page. A US cohort examining cannabis use in adolescents and young adults with type 1 and type 2 diabetes (SEARCH; its full text could not be accessed in the first search) and a meta-analysis combining the effect of approved cannabinoids on appetite were added to the table. The weight-loss figures from the US prescribing information for the approved CBD medicine, a note that diabetic gastroparesis and cannabinoid hyperemesis syndrome can be confused, and the fact that there is no treatment study with a cannabis product in diabetic retinopathy were written into the text; the FAQ answer about weight was corrected based on these figures. The conclusion and the strength of evidence did not change.
Related studies: Spanagel and Bilbao 2021 (meta-analysis), published September 15, 2021 (online June 26, 2021) (PMID 34181977, doi:10.1016/j.neuropharm.2021.108680, PROSPERO CRD42021229932. Full text could not be read (paywalled, subscription required on Europe PMC), row based on the PubMed abstract. The same study also appears in the non-cancer appetite loss table); Roberts 2025 (SEARCH), published Winter 2025 (online September 4, 2024) (PMID 39959518, doi:10.2337/ds24-0019, PMC11825399 (read from the PMC article page, the Europe PMC full-text service and NCBI efetch did not return the article body). SEARCH 4: supported by NIH (NIDDK) and the US Centers for Disease Control, the authors reported no relevant conflicts of interest)
Display fix: internal working notes left over from the previous cleanup were removed; content and evidence did not change.
Display fix: internal working file names left in the text were removed; content and evidence did not change.
Older changes (4)
In October 2026, PubMed, Cochrane and ClinicalTrials.gov were searched systematically; 24 studies were added: 11 controlled studies (9 randomized; registry results of the manufacturer's unpublished CBD study in fatty liver, a spray containing CBD and THC in type 2 diabetes, a strip containing THCV and CBD and nabilone in obesity, CBD and CBD-rich cannabis extract in overweight people, and controlled studies of THC, nabilone and marijuana on blood sugar and insulin), 5 systematic reviews and 8 observational studies (cannabis use and type 2 diabetes, body weight, fatty liver, and ketoacidosis in type 1 diabetes). Group numbers and the primary outcome from the registry results were added to the type 2 diabetes study's row; the topic text was updated for the new studies; the conclusion and strength of evidence for THC-containing products were written for the first time ('Insufficient'). The conclusion and the strength of evidence for CBD did not change.
Related studies: Afshar 2022, published December 2022 (PMID 36945340, doi:10.5812/ijpr-132647, Iranian Registry of Clinical Trials IRCT20121021011192N10); NCT01284634 (manufacturer's fatty liver study, registry results), published January 20, 2014 (registry results) (NCT01284634, no publication found in PubMed in October 2026)); Smith 2025, published 2025 (PMID 39968488, doi:10.26828/cannabis/2024/000206, NCT05574049); Lopez 2020, published 2020 (PMID 32456572, doi:10.1080/19390211.2020.1765941, NCT04294706); Matheson 2025, published October 2025 (PMID 40353878, doi:10.1089/can.2025.0034, NCT04801641, April 9, 2026)); Reichenbach 2015, published October 1, 2015 (PMID 26283236, doi:10.1016/j.clinthera.2015.07.023, NCT01598207); Chia 2026, published May 1, 2026 (PMID 41869774, doi:10.1152/ajpendo.00391.2025, NCT01517100); Ewell 2021, published August 20, 2021 (PMID 34451914, doi:10.3390/ph14080817); Bryan 2026 (single-use study), published July 16, 2026 (PMID 42463843, doi:10.1038/s41598-026-62837-0, NCT04114903, August 6, 2026), methods publication Skrzynski 2026, PMID 41820717); Bryan 2025 (4-week use study), published September 2025 (PMID 40324550, doi:10.1016/j.amjmed.2025.05.002, NCT04114903, August 6, 2026)); Pinto and Martel 2022 (systematic review), published November 2022 (PMID 36180814, doi:10.1007/s40261-022-01205-y); Porr 2020 (rapid review), published August 17, 2020 (PMID 32807222, doi:10.1186/s13643-020-01411-9, PROSPERO CRD42019122829); Ravi 2018 (systematic review), published February 6, 2018 (PMID 29357394, doi:10.7326/M17-1548, PROSPERO CRD42016051297); Mousavi 2023 (meta-analysis), published November 2023 (PMID 37526051, doi:10.1002/ptr.7973); Quinapanta Castro 2025 (meta-analysis), published September 2025 (PMID 41141207, doi:10.7759/cureus.93183, PROSPERO CRD42025025065); Bancks 2015 (CARDIA), published December 2015 (PMID 26364621, doi:10.1007/s00125-015-3740-3); Danielsson 2016, published 2016 (PMID 27843955, doi:10.1155/2016/6278709); Baumeister 2021 (Mendelian randomization), published November 2021 (PMID 33580533, doi:10.1111/adb.13020); Alshaarawy 2019 (NESARC), published October 1, 2019 (PMID 30879064, doi:10.1093/ije/dyz044); Alayash 2021 (Mendelian randomization), published September 1, 2021 (PMID 34304124, doi:10.1016/j.drugalcdep.2021.108863); Wang 2020 (Mendelian randomization), published 2020 (PMID 32922442, doi:10.3389/fgene.2020.00949); Kinney 2020 (T1D Exchange), published January 2020 (PMID 31628116, doi:10.2337/dc19-0365); Abbotts 2022, published May 2022 (online May 21, 2022) (PMID 35631293, doi:10.3390/nu14102152, NCT04971837); Zhou 2026 (UK Biobank), published September 2026 (online April 17, 2026) (PMID 42115075, doi:10.1016/j.numecd.2026.104762)
The names of the finding labels were simplified: the "No benefit seen" label on this page became "No effect found". The assessment did not change.
The year of a publication named on the page was corrected to the journal's print year instead of the online publication year. The conclusion and the strength of evidence did not change.
Page published. This first review looked at 3 studies.