Study archive · page 26

Pooled evidence from 178 randomised trials shows any resistance training routine builds strength and muscle in healthy adults compared with no exercise. Loads above 80% of one repetition maximum ranked best for strength, while all loads grew muscle comparably.

⚠What it shows: About as trustworthy as training evidence gets: results pooled from 178 trials with 5097 people for strength and 119 trials with 3364 people for muscle size, and the rankings survived deliberate stress testing. Most sensible routines performed similarly, so the top-ranked recipes beat the rest by small margins.Meta-analysis · Br J Sports Med

Pooled results from 20 trials covering 1,192 adults with type 2 diabetes and overweight or obesity found that combining aerobic and resistance training improved body mass index, blood sugar control, blood pressure, inflammatory markers, cardiorespiratory fitness and quality of life.

⚠What it shows: Promising but on shaky ground: results pooled from 20 trials in 1,192 people with type 2 diabetes and excess weight, yet the authors themselves rate the underlying evidence as poor quality with uncertain risk of bias. Treat the direction as credible and the sizes as unsettled.Meta-analysis · PeerJ

Pooling ten studies, training a muscle group twice a week produced greater hypertrophy than once a week when weekly volume was matched. Whether three sessions per muscle beat two remains undetermined.

⚠What it shows: Strong for a training question: results pooled from ten trials comparing muscles trained one to three days per week with weekly volume held equal. Too few studies existed to say whether three sessions beat two, so the ceiling is unknown.Meta-analysis · Sports Med

When total volume load was matched, muscle growth was similar across very low, low, moderate and high loads in pooled trials. High loads still produced superior one-rep max strength gains, beating both low and moderate loads.

⚠What it shows: Convincing: results pooled from trials where total work, meaning sets times reps times weight, was matched across loading conditions, so the comparison is fair. It cannot say what happens when lighter training adds extra sets or reps beyond that match.Meta-analysis · Appl Physiol Nutr Metab

Pooling 35 volume-equated trials, periodised programmes improved one-rep max strength more than non-periodised ones, and undulating periodisation beat linear periodisation in trained lifters. Muscle hypertrophy did not differ between any of the approaches.

⚠What it shows: Solid on strength, clear on size: results pooled from 35 trials with total training volume matched between conditions. The edge for varied, undulating plans appeared only in already trained lifters, and no form of periodisation moved muscle growth at all.Meta-analysis · Sports Med

In pooled trials of at least 10 weeks, aerobic training removed 1.06 kg more fat mass than resistance training, and concurrent training removed 1.09 kg more than resistance training alone. Body fat percentage changes did not differ between the three approaches, and with workloads matched no mode won.

⚠What it shows: Fairly robust: results pooled from 31 of 36 trials covering 1564 healthy adults. The differences only appeared in programmes lasting 10 weeks or more, and vanished entirely when workloads were matched across exercise types.Meta-analysis · J Int Soc Sports Nutr

Sleep-lab recordings show cannabis does not consistently alter sleep duration, latency, wake time, efficiency or sleep staging. Withdrawal from regular use, by contrast, was consistently linked to disturbed sleep: shorter total sleep, slower sleep onset and REM rebound.

⚠What it shows: Treat this as an early correction rather than a final verdict: 18 sleep-lab studies were reviewed and nine pooled, but many were small, doses and products varied widely, and the authors call the evidence base very limited. The withdrawal effects were the most consistent finding across studies.Meta-analysis · Sleep Med Rev

A meta-analysis of 23 randomised trials found melatonin supplementation improved sleep quality scores in adults, lowering the Pittsburgh Sleep Quality Index by 1.24 points on average. Benefits were clearest in people with respiratory diseases, metabolic disorders and primary sleep disorders, but absent in those with mental health or neurodegenerative conditions.

⚠What it shows: Solid but with a catch: results pooled from 23 trials, yet the studies disagreed with each other substantially and nearly all participants had an existing illness, so the benefit may not transfer to a healthy person with occasional bad nights. The average improvement on the sleep questionnaire was modest.Meta-analysis · J Neurol

Adults with prediabetes had 2.78 times the odds of obstructive sleep apnoea compared with people with normal blood sugar, in pooled data from eight studies covering 18,040 adults. The link was stronger in younger groups, reaching 3.44 times the odds where average age was under 55.

⚠What it shows: A consistent link but not proof of cause: all eight studies measured people at a single moment in time, so nobody knows whether disturbed sleep drives the blood sugar problem, the reverse, or whether something else drives both. Pooled from 18,040 adults, though the studies disagreed a fair amount on the size of the link.Meta-analysis · Front Endocrinol (Lausanne)

A pooled analysis of 65 randomised trials found that interventions which improved sleep quality also produced medium-sized improvements in mental health, including depression and anxiety. The bigger the improvement in sleep, the bigger the mental health benefit.

⚠What it shows: About as strong as sleep evidence gets: results pooled from 65 trials in 8608 people, all randomly assigned. The trials used many different ways of improving sleep, and how sleep works on the mind is still not explained.Meta-analysis · Sleep Med Rev

Over six months or longer, intermittent fasting reduced body weight by 2.84 kg, fat mass by 3.06 kg and waist circumference by 3.85 cm versus a control diet in adults with overweight or obesity; against continuous calorie restriction its edge shrank to 0.70 kg more fat lost.

⚠What it shows: Strong for a diet question: results pooled from 24 trials of 2032 adults with overweight or obesity, all lasting six months or more. The edge over ordinary calorie cutting was real but small, so neither approach clearly wins.Meta-analysis · Obes Rev

In a four-week trial of 42 adults who ate little fruit and veg, allowing one daily portion of juice or smoothie within UK 5-a-day advice did not significantly reduce whole fruit and vegetable intake: 6.6 portions a day versus 8.9 for whole produce alone, against 2.45 in the control group. The group including juice also reported lower depression symptoms than control, with no differences in anxiety, gut symptoms or metabolic markers.

⚠What it shows: Suggestive but thin: one four-week trial in 42 healthy adults who were eating two servings a day or fewer, and every group was handed weekly cash for produce, so it tells you little about ordinary shopping. The mood result was a secondary outcome and the work was paid for by a fruit juice industry body, which is worth holding in mind.Study · Br J Nutr

A meta-analysis of 16 randomised trials found protein source, animal or plant, made no difference to gains in absolute lean mass or muscle strength. Animal protein showed a small edge on percent lean mass, and adults under 50 gained 0.41 kg more absolute lean mass with it.

⚠What it shows: Reassuring for plant eaters: results pooled from 16 trials where protein intakes were already above recommended levels, so any advantage of source had room to show and mostly did not. The small animal protein edge appeared mainly in adults under 50, and doing resistance training alongside made no difference to the outcome.Meta-analysis · Nutrients

Pooled results from 7 randomised trials found aerobic exercise improved fasting glucose, systolic and diastolic blood pressure, HbA1c, HDL cholesterol and triglycerides in older adults with type 2 diabetes, with no significant adverse effects reported.

⚠What it shows: Encouraging but thin: results pooled from just 7 trials, and the certainty was rated moderate for blood pressure, low for glucose and very low for everything else. The trials tracked lab markers rather than symptoms or complications, so the real-world payoff is not yet proven.Meta-analysis · Rev Peru Med Exp Salud Publica

Pooled data from 52 studies found no association between GLP-1 receptor agonists used for weight loss in obesity and an increased risk of pancreatitis or pancreatic cancer, with findings consistent across drug type, dose, study design and follow-up duration.

⚠What it shows: About as reassuring as pooled evidence gets: 52 studies in people with obesity, and the null result held across drug type, dose, study design and follow-up length. The estimate for pancreatic cancer still carried a wide band of uncertainty, so a modest effect cannot be completely ruled out.Study · Horm Metab Res

Across 32 randomised trials in 1098 people, both cold and heat applied within an hour of exercise reduced the pain of delayed onset muscle soreness; cold helped mainly within 24 hours, heat helped both within and beyond 24 hours, hot packs showed the largest effects, and there was no significant difference between cold and heat overall.

⚠What it shows: Trustworthy on direction, fuzzy on size: results pooled from 32 trials with 1098 people, but the trials varied widely in how heat and cold were applied and the authors themselves call for better studies. Head to head, neither therapy clearly beat the other.Meta-analysis · Phys Ther Sport

In a meta-analysis of 23 trials in adults under 50, creatine combined with resistance training increased upper-body strength by 4.43 kg and lower-body strength by 11.35 kg on average compared with placebo. Men gained significantly in both, while women, who appeared in only a handful of the trials, showed no significant strength gains.

⚠What it shows: Strong evidence for men: results pooled from 23 trials consistently showed bigger strength gains when creatine was added to lifting. Women were nearly absent from those trials, so the lack of a clear effect in women may simply reflect too little data rather than a true sex difference.Meta-analysis · Nutrients

A meta-analysis of 38 randomised trials in 1682 middle-aged and older adults found aerobic training improved VO2max, six-minute walk distance and body mass more than resistance training, while resistance training was better for preserving lean body mass. The authors conclude a combination of both is likely optimal for ageing bodies.

⚠What it shows: A clear split from good evidence: results pooled from 38 trials with 1682 mostly middle-aged and older adults, and the aerobic advantage for fitness held across sexes, ages and trial lengths. It compares group averages, so it cannot say how to divide training time for any one person.Meta-analysis · Arch Gerontol Geriatr

An umbrella review of 101 meta-analyses found high-certainty evidence that cannabis based medicines reduce chronic pain and that cannabidiol reduces seizures in epilepsy, but convincing evidence linking cannabis to psychosis in the general population, car crashes in drivers, and low birth weight in pregnancy. The authors conclude cannabis should be avoided in adolescence, in pregnancy, before driving and by anyone prone to mental health problems.

⚠What it shows: Strong for the headline harms: this graded 101 separate pooled analyses of the evidence, and the psychosis, driving and pregnancy links reached the top credibility tier. Much of the harm evidence still comes from following users rather than experiments, so cause cannot be fully pinned down.Systematic review · BMJ

Pooled data from 26 randomised trials in 1,891 people found probiotics improved laboratory markers of gut barrier function, including lower serum zonulin, endotoxin and LPS, and also reduced inflammatory markers such as CRP, TNF-α and IL-6 compared with control groups.

⚠What it shows: Encouraging but marker-deep: results pooled from 26 trials with 1,891 people showed changes in blood and laboratory measures of gut leakiness, not in symptoms people actually notice. The authors themselves say higher-quality trials are needed before drawing a definitive conclusion.Meta-analysis · Front Immunol

Long-term exposure to traffic-related air pollution is associated with higher all-cause, circulatory, ischaemic heart disease and lung cancer mortality, as well as asthma onset in children and adults, with high or moderate-to-high confidence across 353 studies.

⚠What it shows: A strong and carefully vetted signal: an independent expert panel reviewed 353 studies and rated confidence high or moderate-to-high for these links. Almost all the research followed people rather than testing anything, so it shows association, not proof of cause.Meta-analysis · Environ Int

An overview of 21 Cochrane reviews covering 381 studies and 37,143 adults with chronic pain found exercise caused few adverse events and tended to improve pain severity and physical function, though most effects were small to moderate and the underlying evidence quality was low.

⚠What it shows: Reassuring on safety, less certain on how much relief to expect: this sweep of all the studies covered 381 trials with 37,143 adults, but most individual trials had fewer than 50 people and rarely followed anyone beyond three to six months. The clearest signal is that exercise did little harm.Meta-analysis · Cochrane Database Syst Rev

In postmenopausal women, exercise significantly improved bone mineral density at the lumbar spine, femoral neck and total hip across 84 exercise groups, and the benefit held largely regardless of whether the training was resistance work, weight-bearing exercise or a mix.

⚠What it shows: Strong evidence for the headline effect: results pooled from controlled trials covering 84 exercise groups of postmenopausal women, each lasting at least six months. The pooled data are too coarse to say which precise routine builds the most bone, a limit the authors themselves flag.Meta-analysis · Calcif Tissue Int

Across 65 studies of wrist-wearable activity trackers, the Fitbit Charge measured steps with under 25% error and the Apple Watch measured heart rate with under 10% error, but energy expenditure was off by more than 30% for every brand tested.

⚠What it shows: A trustworthy overview: a sweep of all 65 studies on wrist trackers in the general population. The studies measured accuracy so differently that results could not be combined into one number, so device comparisons are informal rather than head to head.Systematic review · J Med Internet Res