Study archive · page 5

A review of five randomised trials found no across-the-board benefit of cannabis-based products for back pain or headache: a single 400-mg oral dose of CBD was no better than placebo for acute low back pain. Vaporised THC plus CBD improved acute migraine pain relief at 2 hours, with selected benefits sustained through 24 to 48 hours.

⚠What it shows: Take the split seriously but not as settled: a sweep of all the studies found only five trials with 1,072 people in total, and the back pain result rests on a single dose in one trial. Products containing THC may also have let participants guess what they were taking, which can inflate apparent benefits.Review · Cannabis Cannabinoid Res

Pooled results from 10 randomised trials suggest purslane supplementation lowered C-reactive protein by 0.33 mg/dl on average, but it did not significantly change malondialdehyde, total antioxidant capacity or superoxide dismutase, three markers of oxidative stress.

⚠What it shows: Treat the inflammation result lightly: pooled from 10 trials, the drop in C-reactive protein only just reached significance and the trials disagreed with each other heavily. Meanwhile all three antioxidant markers stayed exactly where they were.Meta-analysis · Phytother Res

Across 63 trials in 1489 people, adding creatine to resistance training produced greater gains in upper-body strength, lower-body strength and lean mass than training alone, with no significant change in fat mass. The small rise in total body weight was mostly lean tissue, and body-fat percentage dipped slightly.

⚠What it shows: About as solid as sports nutrition evidence gets: results pooled from 63 trials in 1489 people, most showing the same direction of benefit. The included trials varied in diets and training backgrounds, and the hint that younger lifters gain more strength is exploratory rather than proven.Meta-analysis · J Int Soc Sports Nutr

Pooled data from 24 randomised trials found that resveratrol supplements lowered the inflammatory markers C-reactive protein and TNF-α in people with metabolic syndrome and related conditions. Two other markers, IL-6 and the antioxidant enzyme SOD, did not change.

⚠What it shows: A real but partial signal: results pooled from 24 trials showed lower levels on two inflammation markers and no movement on two others, and the trials disagreed with each other considerably. Everyone studied had metabolic syndrome or a related condition, so healthy adults may respond differently.Meta-analysis · Food Funct

Pooled results from 32 randomised trials in 5174 adults with chronic pain show non-inhaled medical cannabis gives a small improvement in pain relief, with 10% more patients reaching a meaningful reduction than on placebo, plus very small gains in physical functioning and sleep quality. It also raised the risk of transient side effects including dizziness, drowsiness, nausea, vomiting and impaired attention.

⚠What it shows: About as solid as pain evidence gets: results pooled from 32 trials in 5174 adults, rated moderate to high certainty. The benefits are genuinely small though, follow-up lasted at most 5.5 months, and dizziness became far more common in trials running 3 months or longer.Meta-analysis · BMJ

In a network comparison of 75 randomised trials with 5,254 participants, tai chi, yoga, Pilates and sling exercise eased chronic low back pain more than conventional rehabilitation, while yoga and core or stabilisation exercises did most for physical function.

⚠What it shows: Encouraging but not settled: results pooled and ranked across 75 trials in 5,254 people, yet many of the underlying trials were small and of mixed quality, so the authors stop short of crowning a single best regimen.Meta-analysis · Front Public Health

Across 10 randomised trials in 510 people with nonspecific low back pain, Pilates reduced pain and disability more than minimal intervention, mostly with medium-sized effects, but there is no conclusive evidence it beats other forms of exercise.

⚠What it shows: Fairly trustworthy as far as it goes: a sweep of all the studies on this pulled together 10 trials with 510 people, rating the evidence low to moderate quality. It shows Pilates beats doing very little, not that it beats other exercise.Systematic review · Spine (Phila Pa 1976)

Pooled results from 6 randomised trials suggest repetitive peripheral magnetic stimulation reduces pain intensity and functional disability in people with low back pain compared with sham or other therapy, though fear of movement did not improve and the evidence was rated very low to low quality.

⚠What it shows: A hint, not proof: results pooled from just 6 small trials totalling 139 people, and the reviewers themselves rated the evidence very low to low quality. Promising on paper, a long way from settled.Meta-analysis · Arch Phys Med Rehabil

Internet-delivered cognitive behavioural therapy for insomnia improved insomnia severity, sleep efficiency, sleep quality and total sleep time across 11 randomised trials totalling 1460 people. The effects were comparable to face-to-face therapy and were generally maintained 4 to 48 weeks later.

⚠What it shows: Strong for a digital treatment: results pooled from 11 trials with 1460 people showed benefits across every sleep measure, though gains were larger when programmes ran longer and included more human support, and shrank in trials where many people dropped out.Meta-analysis · Sleep Med Rev

Pooled results from 11 randomised trials in 1,066 adults with long-term pain found that non-drug sleep treatments produced a large improvement in sleep quality, a moderate improvement in fatigue and a small reduction in pain. The sleep and fatigue gains held at follow-up of up to 1 year, when a moderate reduction in depression also appeared.

⚠What it shows: Strong for the sleep effect, modest for the rest: results pooled from 11 trials in 1,066 adults with long-term pain, with no sign that flattering results were cherry-picked for publication. The body of evidence is still small, and face-to-face treatment beat phone or internet delivery.Meta-analysis · Sleep

Across 48 randomised trials in 5,037 women in China, traditional Chinese medicine outperformed Western medicine for perimenopausal insomnia, improving sleep quality scores and hormone levels with fewer reported adverse reactions. The reviewers flag methodological limitations across the evidence.

⚠What it shows: Take the headline with real caution: results pooled from 48 trials in 5,037 women, all in China, but the reviewers flag weak methods and studies that disagreed with each other, so the size of the advantage is unreliable. Better-designed trials are needed before this counts as settled.Systematic review · Front Neurol

A sweep of randomised trials found box breathing offers short-term relief for perceived stress, anxiety and mood, but only eight trials qualified and its effects on heart rate variability and cortisol were absent or inconsistent.

⚠What it shows: Worth a measured read: a sweep of all the studies on box breathing turned up just eight trials, in groups ranging from healthy adults to postpartum women, using different routines. The calm it delivered was short-term and the body's stress chemistry mostly did not follow.Review · Appl Psychophysiol Biofeedback

A meta-analysis of 31 randomised trials found virtual reality-based training produced small but significant improvements in older adults' global cognition and executive function compared with conventional training or information-based treatment, though the evidence was graded low to very low.

⚠What it shows: Encouraging but low certainty: results pooled from up to 20 trials covering 876 older adults show small gains, yet the studies disagreed with each other considerably and the authors grade the evidence low to very low, so better trials could shrink the effect.Meta-analysis · J Aging Phys Act

In a three-month randomised trial of 52 healthy middle-aged adults, an instructor-led live online exercise programme reduced depressive symptoms but did not improve pattern separation, the memory measure it was primarily designed to shift. Participants' blood also changed in ways that influenced hippocampal progenitor cells in a lab assay.

⚠What it shows: Promising but small: one three-month trial in 52 healthy middle-aged adults, and the brain-cell findings came from a laboratory dish assay using participants' blood, not from measuring anyone's brain. The main memory test did not improve, so any cognitive benefit remains unproven.Study · NPJ Aging

Across 12 randomised trials in 707 older adults with mild cognitive impairment or Alzheimer's disease, ketogenic dietary strategies improved overall cognition, memory and language, but not attention or executive function, and they raised total and LDL cholesterol.

⚠What it shows: Modest and real, but with a trade-off: results pooled from 12 trials in 707 older adults who already had cognitive impairment, so it says nothing about healthy brains. The same diets pushed up total and LDL cholesterol, which the authors flag as a cost to weigh against the benefits.Review · Arch Gerontol Geriatr

A meta-analysis of 14 randomised placebo-controlled trials in 702 people found probiotic fermented milk lowered systolic blood pressure by 3.10 mmHg and diastolic by 1.09 mmHg, with a slightly greater effect in people whose blood pressure was already raised.

⚠What it shows: Credible but modest: results pooled from 14 placebo-controlled trials in 702 people, and the drop in blood pressure is small. There were hints that flattering trials were likelier to reach publication, though excluding the small extreme ones barely moved the result.Meta-analysis · Br J Nutr

People with acne showed a significant association with both depression and anxiety in results pooled from 42 studies, with the strength of the link shifting by age, study setting and geographic region.

⚠What it shows: A consistent link, not proof that acne causes low mood: results pooled from 42 studies comparing people with and without acne. The underlying studies measured acne and mental health inconsistently, and some had no comparison group at all.Meta-analysis · J Am Acad Dermatol

Probiotic supplements reduced eczema severity scores in adults in pooled results from nine trials, with the largest benefit from Lactobacillus salivarius and in moderate-to-severe disease; L. fermentum showed no benefit, and quality of life scores did not meaningfully improve.

⚠What it shows: Real but strain-specific and incomplete: results pooled from nine trials show severity scores fell, especially in worse eczema, yet people's quality of life did not meaningfully improve. Nobody yet knows the right strain dose or how long to take it.Meta-analysis · Clin Exp Dermatol

Across 38 pooled studies, wearable AI detected sleep apnoea with a mean accuracy of 0.869, but its accuracy at grading severity was only 0.651, and the reviewers judged its performance still short of routine clinical use.

⚠What it shows: Promising but not clinic-ready: results pooled from 38 studies of wearable devices, and performance shifted depending on the algorithm, the type of data and where the device sat on the body. Sleep-lab testing remains the benchmark for now.Meta-analysis · J Med Internet Res

Across 118 randomised trials in 21,662 people with type 2 diabetes, digital tools such as apps, text messages and online platforms lowered HbA1c by 0.32 to 0.54 percentage points versus usual care, but did not increase physical activity or improve insulin resistance.

⚠What it shows: Strong on glucose, hollow on behaviour: the pooled evidence covers 118 trials in 21,662 people, though the trials disagreed with each other a fair amount on how large the benefit was. And none of it translated into people actually moving more.Systematic review · Front Digit Health

Pooled data from 106 studies covering 20,063 adults show physical function is still reduced 3 months after COVID-19 infection: walking distance averaged 84.3% of predicted at 3 to 5 months, climbing to 92.5% of predicted at 11 months or later.

⚠What it shows: A broad pooled picture rather than one trial: 106 studies following 20,063 adults for up to two years, though many had been hospitalised and the studies disagreed with each other considerably. Your own recovery curve could look quite different from these averages.Meta-analysis · Phys Ther

Pooled results from nine placebo-controlled trials in 649 people with type 2 diabetes found that probiotic supplements taken alongside standard treatment significantly reduced fasting glucose, cut HbA1c by 0.42 points on average, and lowered insulin levels.

⚠What it shows: Promising and unusually consistent: results pooled from nine placebo-controlled trials in 649 people with type 2 diabetes, and the trials agreed closely with one another. The evidence base is still small, and probiotics were only ever tested as an add-on to standard care, never a replacement.Meta-analysis · Endocr Metab Immune Disord Drug Targets

Vitamin D supplementation reduced the risk of acute respiratory infections in pooled individual data from 25 trials covering 11,321 people. The protection appeared only with daily or weekly dosing, not with large bolus doses, and was strongest in people who started out severely deficient.

⚠What it shows: About as solid as supplement evidence gets: individual records pooled from 25 trials in 11,321 people aged 0 to 95, and the evidence was rated high quality. The catch is that the benefit vanished with occasional large-dose regimens and was concentrated in people who were deficient to begin with.Meta-analysis · Health Technol Assess

Even a low dose of alcohol, roughly two standard drinks, delays and reduces REM sleep, and the disruption worsens as the dose rises, according to a meta-analysis of 27 studies in healthy adults. High doses, around five standard drinks, may shorten the time it takes to fall asleep, but this likely deepens the REM disruption that follows.

⚠What it shows: Solid as sleep evidence goes: results pooled from 27 studies of healthy adults agree that REM sleep suffers after drinking, and that the damage grows with the dose. The picture on total sleep time and night-time waking stayed too uncertain to call either way.Meta-analysis · Sleep Med Rev