Study archive · page 14

Meta-analyses indicate that combining creatine supplementation with resistance training improves ageing muscle mass and upper body strength beyond resistance training alone, and creatine also shows promise for bone mineral density and indices of bone biology.

⚠What it shows: Reasonably solid for muscle and upper body strength: the conclusion rests on pooled trials of creatine combined with resistance training in the context of ageing. The bone findings are promising rather than proven, and the abstract does not report how large the added benefits are.Meta-analysis · Endocrine

Pomegranate supplementation modestly reduced body weight, BMI, systolic and diastolic blood pressure, fasting glucose and total cholesterol, and raised HDL, across 53 randomised trials in 2306 people; LDL, triglycerides, HbA1c, waist measures and insulin resistance did not change.

⚠What it shows: Real but small: results pooled from 53 trials in 2306 people show statistically clear yet modest shifts, and several key markers, including LDL and triglycerides, did not move at all. The authors call for longer and larger trials before firm conclusions.Meta-analysis · Nutr Metab Cardiovasc Dis

In people with prediabetes or type 2 diabetes, CoQ10 supplementation significantly reduced triglycerides, fasting blood sugar, HbA1c, insulin resistance and the inflammation marker C-reactive protein, and raised HDL, across 20 randomised trials; total cholesterol, LDL and blood pressure did not change.

⚠What it shows: Encouraging but not the final word: results pooled from 20 trials in people with prediabetes or type 2 diabetes, and the trials disagreed with each other on how big the benefits were. The authors want longer studies across a range of doses before firm conclusions.Meta-analysis · BMC Pharmacol Toxicol

A meta-analysis of ten randomised trials found green coffee bean extract lowered fasting blood sugar by 1.791 mg/dl on average, a statistically significant but very small change. Insulin did not improve overall, and only fell significantly in trials using doses of at least 400 mg per day.

⚠What it shows: Real but trivial: pooled from ten trials, the blood sugar drop was under two points, far too small for anyone to feel. On insulin the trials disagreed sharply, and the promising result came only from a subgroup taking at least 400 mg a day.Meta-analysis · Phytother Res

A bidirectional Mendelian randomisation study using genetic data found evidence that greater education (odds ratio 0.54 per 4.2 years of schooling), ever smoking (1.27), greater alcohol consumption (1.29 per category increase) and major depressive disorder (1.41) causally increase the risk of chronic back pain. In the reverse direction, chronic back pain increased alcohol consumption (1.19) and smoking (1.21), while physical activity and sleep did not meet the study's criteria for a causal link.

⚠What it shows: Stronger than a simple correlation, weaker than a trial: this method uses inherited genetic differences as stand-ins for lifestyle exposures, which sidesteps some confounding but rests on assumptions that can fail. The genetic data came from people of European ancestry, so it may not carry over to other populations.Meta-analysis · Eur Spine J

Across 44 trials, aerobic exercise, traditional Chinese exercises and meditation each produced moderate reductions in college students' depressive symptoms, with broadly similar effects across the three approaches.

⚠What it shows: Reasonably solid on the broad conclusion: results pooled from 44 trials all pointed the same way, but the individual studies disagreed with each other a lot, so the true size of the benefit is uncertain. Every participant was a college student, so older adults may respond differently.Meta-analysis · Medicine (Baltimore)

Pooling 12 randomised trials with 1193 participants, exercise programmes were linked to better cognitive function in people living with diabetes, with the strongest results from multi-component exercise done more than 4 times a week, in sessions over 40 minutes, sustained for more than 6 months.

⚠What it shows: Promising direction, uncertain size: the pooled benefit comes from 12 trials in 1193 people, and the trials disagreed with each other a great deal. The optimal dose findings come from comparing across studies rather than trials designed to test one dose against another.Meta-analysis · PLoS One

Adding cognitive training to an exercise programme gave older adults a small extra improvement in cognitive function compared with exercise alone, across 24 studies of 1079 people; blood levels of the brain-plasticity marker BDNF did not change.

⚠What it shows: Credible but modest: pooled from 24 studies of 1079 older adults, some with mild cognitive impairment, and the extra benefit over exercise alone was small. The authors themselves flag limitations in the underlying studies, so treat the size of the effect loosely.Review · Healthcare (Basel)

Pooled data from 21 trials in 1638 people with fibromyalgia, mostly women, found mind-body exercise and resistance training gave the largest reductions in pain, while mind-body exercise also improved six-minute walking distance.

⚠What it shows: Useful but loosely stitched: results pooled from 21 trials in 1638 people, mostly women, and the different exercise types were often compared indirectly rather than head to head. Take the rankings as a guide rather than a verdict.Meta-analysis · Pain Res Manag

Pooled results from 27 randomised trials in adults aged 50 and over found that resistance training increased voluntary activation, a measure of how fully the nervous system recruits a muscle, with the clearest signal in the plantar flexors and in healthy participants. Muscle electrical activity during maximal effort and antagonist coactivation did not change.

⚠What it shows: Reasonably solid: results pooled from 27 randomised trials in adults aged 50 and over, and the activation finding survived when only the best-quality trials were counted. Most participants were healthy older adults, so it is unclear whether the same holds for people with sarcopenia or age-related frailty.Meta-analysis · Exp Gerontol

Pooled results from 32 randomised trials in 1722 adults show calorie restriction lasting one to four weeks lowered systolic blood pressure by 5.5 mmHg, an effect the authors say is similar to what medication achieves. Restriction lasting up to six months also slowed resting heart rate and improved VO2peak.

⚠What it shows: Strong for the short term: pooled from 32 trials in 1722 adults, the blood pressure drop looks real, but most interventions lasted weeks to months, so nobody knows whether the benefit survives once normal eating resumes. The literature search also stopped in June 2017, so newer trials are missing.Meta-analysis · Clin Nutr

A meta-analysis of 32 randomised trials with 1491 participants found anthocyanin supplementation significantly reduced fasting glucose, post-meal glucose, glycated haemoglobin, total cholesterol and LDL cholesterol in both healthy people and those with cardiometabolic disease.

⚠What it shows: Fairly convincing on direction, less so on size: results pooled from 32 trials in 1491 people held up even after removing the weaker studies, but the trials disagreed with each other considerably and the best dose and formulation remain unknown.Meta-analysis · Adv Nutr

High-intensity interval training improved 14 cardiometabolic health markers in a meta-analysis of 97 randomised trials with 3,399 participants, including peak aerobic capacity, systolic and diastolic blood pressure, resting heart rate, waist circumference, body fat and blood fats.

⚠What it shows: About as broad as exercise evidence gets: results pooled from 97 trials in 3,399 people from the general population, all pointing the same way. Each individual improvement was modest, and HIIT was compared against doing nothing rather than against other forms of exercise.Meta-analysis · Sports Med

In endurance athletes with more than 2 years of training experience, pooled results from 3 randomised trials showed a moderate advantage for polarised training, mostly easy work with a small share very hard, over threshold training for time-trial performance. Only 4 trials met the inclusion criteria in total.

⚠What it shows: Suggestive rather than settled: only 4 randomised trials qualified and just 3 had data that could be pooled, all in trained endurance athletes. Those trials scored 4 or 5 out of 10 for methodological quality, which is modest, and time-trial performance was the only outcome with enough data to combine.Meta-analysis · J Strength Cond Res

In a dose-response network meta-analysis of 74 trials with 3,573 mostly postmenopausal women, exercise was associated with non-linear blood pressure reductions, with the largest predicted drops of 6.70 mmHg systolic near 1,400 METs-min per week and 4.20 mmHg diastolic near 1,500 METs-min per week.

⚠What it shows: Treat the sweet spot as a rough guide, not a target: it comes from pooling 74 trials in 3,573 women, and the authors rate the evidence as very low to low certainty. The apparent levelling off at higher doses also depended on which statistical model was chosen.Systematic review · Front Med (Lausanne)

Pooled results from 17 trials in 2,865 postpartum women show aerobic exercise reduced depressive symptoms and improved quality of life, with modelled benefits peaking near 570 MET minutes per week for depression and 420 for quality of life. Anxiety did not significantly improve.

⚠What it shows: Solid pooled evidence from 17 trials in 2,865 women, with the biggest mood gains in those who were depressed to begin with. The ideal dose figures come from statistical models rather than head to head comparisons, and anxiety did not improve at all.Meta-analysis · Front Public Health

Weight-loss programmes delivered through mobile phones helped overweight and obese adults lose an average of 1.44 kg more than control groups, according to pooled results from 14 randomised trials involving 1,337 participants.

⚠What it shows: Solid but modest: results pooled from 14 trials in 1,337 overweight or obese adults all pointed the same way, yet the average extra loss was under two kilos. The search stopped in 2014, so today's apps and wearables were never part of the evidence.Meta-analysis · Am J Epidemiol

Across 137 randomised trials in 18,443 people with chronic conditions, digital programmes combining exercise with self-management support increased physical activity more than usual care, and delivery via mixed technologies such as apps with phone calls or devices worked best.

⚠What it shows: Fairly trustworthy: results pooled from 137 trials in 18,443 people, with moderate confidence in most comparisons. Gains were biggest in type 2 diabetes and smallest in muscle and joint conditions, so the payoff depends on what you are managing.Meta-analysis · J Phys Act Health

A network meta-analysis of 30 trials in 595 healthy adults found that tart cherry and pomegranate reduced muscle soreness at 48 hours after damaging exercise, but no polyphenol supplement showed clear benefits for muscle strength or blood markers of damage and inflammation.

⚠What it shows: Worth a measured read: the soreness relief looks plausible but the authors rate the overall evidence as low certainty, pooled from 30 trials totalling 595 healthy adults. Objective recovery measures did not improve for any supplement tested.Systematic review · Front Nutr

Restricting sleep reduced insulin sensitivity in a meta-analysis of randomised trials, and both shifting sleep timing against the body clock and suppressing deep sleep also impaired it, while fragmented sleep and disturbed REM sleep had no measurable effect.

⚠What it shows: Strong for a short-term effect: results pooled from 21 trials where sleep was deliberately cut show insulin sensitivity dropping on several different measures. These were brief laboratory-style manipulations, so they cannot say how years of habitually short sleep play out.Meta-analysis · Sleep Med Rev

A dose-response meta-analysis of 21 prospective studies covering 997,464 people and 121,915 deaths found coffee drinking was associated with lower mortality, with the largest reductions at 4 cups a day for all-cause death (16%) and 3 cups a day for cardiovascular death (21%). Coffee showed no association with cancer mortality.

⚠What it shows: A strong pattern but still an association: these studies followed coffee drinkers over time rather than assigning coffee at random, so coffee is not proven to extend life. The pooled data cover 997,464 people across 21 studies, which makes the pattern hard to dismiss but easy to over-read.Meta-analysis · Am J Epidemiol

Pooled results from seven trials in 821 people with obesity show GLP-1 drugs at weight-loss doses reduced absolute lean mass by an average of 1.74 kg, with semaglutide showing the biggest loss at 5.44 kg. Even so, lean mass as a proportion of total body weight improved by 1.81%, since total weight fell faster still.

⚠What it shows: Reasonably solid on the headline numbers: results pooled from seven trials covering 821 people with obesity, all tested against placebo. The trials disagreed a lot on the size of the muscle loss, and none compared the drugs against diet and exercise plans, so it cannot say how much muscle a person who trains would keep.Meta-analysis · Int J Obes (Lond)

Pooled results from nine trials in 696 people with knee osteoarthritis found isokinetic strength training, done on machines that control movement speed, produced a large improvement in pain and a moderate one in disability, and it was well tolerated.

⚠What it shows: Encouraging but not settled: results pooled from nine trials covering 696 people with knee osteoarthritis, all judged low to moderate quality. The programmes and outcome measures varied so much that the authors say firm guidance on how to deliver this training is not yet possible.Meta-analysis · Ann Phys Rehabil Med

Across one population study and seven clinical studies, lower blood vitamin D3 levels correlated with higher COVID-19 mortality, and a regression through the pooled data pointed to a theoretical zero-mortality level near 50 ng/mL. That is a correlation and an extrapolation, not proof that raising vitamin D prevents deaths.

⚠What it shows: Treat this as a pattern, not proof: it pools one population study and seven hospital studies where lower vitamin D went with more deaths, and following people like this cannot show that raising vitamin D would have saved anyone. The zero-deaths level at 50 ng/mL is a line extended beyond the data, never tested in a trial.Meta-analysis · Nutrients