Study archive · page 2

In a meta-analysis of 24 randomised trials, established sleep interventions such as cognitive behavioural therapy improved both sleep and pain in people with low back pain, cutting pain scores by 12.77 points on a 100-point scale. In osteoarthritis they improved sleep but not pain.

⚠What it shows: Fairly convincing for back pain: results pooled from 24 trials, and the pain benefit was consistent across the studies. The trials varied a lot in what they tested, and the authors themselves call for more rigorous work before treating this as settled.Meta-analysis · Osteoarthritis Cartilage

Pooled results from 12 randomised trials found no significant effect of increased protein intake on sleep quality, sleep time, sleep latency or sleep efficiency in adults, with very low certainty of evidence. A check excluding the weakest trials hinted at a small improvement in sleep quality with higher protein.

⚠What it shows: A useful null, but a soft one: results pooled from 12 trials in adults found no clear effect of extra protein on any sleep measure, and the overall evidence was rated very low certainty. A check that dropped the weakest trials hinted at slightly better sleep quality with more protein, so the question is not fully closed.Meta-analysis · Nutr Rev

Pooled trial data suggest supervised exercise training reduces obstructive sleep apnoea severity by 6.27 breathing interruptions per hour, alongside better fitness, less daytime sleepiness and higher sleep efficiency, with minimal change in body weight.

⚠What it shows: Promising but thin: pooled results from just 5 trials totalling 129 people, measured before and after exercise rather than against a comparison group, so some of the improvement could have other explanations.Meta-analysis · Lung

A review of 19 actigraphy studies found that total night-time sleep after childbirth was the sleep measure most strongly linked to postpartum depression, and was also related to postpartum fatigue.

⚠What it shows: An association, not proof that short nights cause postpartum depression: a sweep of 19 studies that tracked new mothers with wrist-worn monitors rather than testing an intervention. The authors say more work is needed, especially on sleep rhythms during pregnancy itself.Meta-analysis · J Sleep Res

Across 18 randomised trials in 1926 women going through the menopausal transition, digital health programmes reduced hot-flash frequency, improved sleep quality and trimmed waist circumference by 2.23 cm, but showed no clear effect on weight, blood pressure, lipids or blood sugar.

⚠What it shows: Worth taking seriously for hot flashes, less so elsewhere: results pooled from 18 trials in 1926 women, with decent confidence in the hot-flash finding but very low confidence in the sleep result. Most of the heart and metabolic measures did not move at all.Review · Maturitas

Structured exercise programmes narrowed the gap between the abdominal muscles in postpartum women with diastasis recti by 8.05 mm on average compared with no treatment or standard care, with bigger reductions when started within 3 months of giving birth. Disability scores, however, improved no more than in control groups.

⚠What it shows: Solid for the anatomy, thinner for the day-to-day payoff: results pooled from nine trials in 450 postpartum women show the gap narrows, but only two of those trials measured disability and found no benefit there. The exercise programmes also varied widely, so nobody knows the best protocol yet.Meta-analysis · Hernia

Across 12 randomised trials of 1260 women, exercise-based programmes reduced postpartum depressive symptoms compared with usual care or no therapy. Pram walking, yoga and supervised mixed exercise each helped, and pram walking ranked top, though its edge was not statistically significant.

⚠What it shows: A decent case that exercise helps, a weak case for a single winner: results pooled from 12 trials in 1260 women were rated very low to moderate quality by the authors, and pram walking's lead over yoga and mixed exercise was not statistically significant.Meta-analysis · Int J Gynaecol Obstet

Pooled results from 12 trials suggest pelvic floor muscle training after childbirth improves women's sexual function and sexual quality of life, with a smaller benefit for general quality of life.

⚠What it shows: Encouraging and consistent in direction, from results pooled across 12 trials in women after childbirth. The studies used very different training methods and disagreed with each other on size, so how big the benefit is remains uncertain and the authors call for more high-quality trials.Meta-analysis · Taiwan J Obstet Gynecol

Pooled results from seven randomised trials in 344 adults aged 60 and over found that resistance training programmes of at least eight weeks increased bioelectrical phase angle by 0.52 degrees, a marker of cellular health read from body composition devices.

⚠What it shows: A consistent finding on a thin base: pooled from just seven trials with 344 older adults in total. The authors say more rigorous studies are still needed to work out which equipment and protocols get the most out of the effect.Meta-analysis · Rev Endocr Metab Disord

Pooled results from randomised trials in adults with type 2 diabetes show resistance-based exercise programmes reduced arterial stiffness and improved endothelial function compared with no exercise, with the most consistent benefits when weights were combined with aerobic training. Wave reflection measures did not change significantly.

⚠What it shows: Reasonably trustworthy: results pooled from trials where people were assigned at random, with the stiffness and vessel-lining benefits rated moderate certainty. Most programmes mixed weights with aerobic work, so the evidence for lifting on its own is thinner, and not every vascular measure improved.Meta-analysis · Front Endocrinol (Lausanne)

A pooled analysis of 17 trials found caffeine gave a small boost to functional performance during post-exercise recovery, had no clear effect on perceived or physiological recovery, and was associated with poorer sleep after exercise.

⚠What it shows: Treat this as a lean, not a verdict: results pooled from 17 trials, but the sleep harm rests on just four studies, and both the performance and sleep findings wobbled when the analysis was re-run in different ways.Systematic review · Front Physiol

Pooled data from 157 people in post-exercise cold-water immersion trials show that colder water, longer immersions and intermittent dunking each pull core temperature down further than their alternatives. The biggest gap versus no immersion appeared 60 minutes after the intervention ended.

⚠What it shows: Useful dosing detail rather than proof of benefit: the data come from 157 people pooled across one research group's own trials, and the outcome is core temperature, not whether anyone recovered faster. How that extra cooling translates to performance is a question this analysis cannot answer.Meta-analysis · Int J Sports Physiol Perform

Results pooled from 63 studies covering 265,079 people link cannabis use to roughly double the odds of perpetrating violence in the general population, rising to 3.75 times the odds for violence leading to a criminal conviction. Cannabis users also had 1.49 times the odds of becoming victims of violence, a risk more pronounced in females.

⚠What it shows: A strong and consistent association, but not proof of cause: results pooled from 63 studies of 265,079 people, many of them one-off snapshots that cannot show whether cannabis use or violence came first. The link survived in studies that followed people over time, though it was considerably smaller there.Meta-analysis · Psychol Med

Pooled results from 22 randomised trials in up to 3089 healthy adults without hypertension found that stress management techniques modestly lowered blood pressure: mind-body exercise cut systolic pressure by 2.17 mmHg, meditation by 1.49 mmHg and relaxation by 4.76 mmHg compared with controls.

⚠What it shows: Credible but modest: results pooled from 22 trials in up to 3089 healthy adults all point the same way, and the drops are a few points of pressure at most. The techniques and trial designs varied widely, so the exact size of the benefit is uncertain.Review · J Hum Hypertens

Across pooled trials in 1,401 adults with overweight or obesity and cardiometabolic disorders, exercise training lowered post-meal blood glucose and insulin, and the effect held regardless of age, BMI, baseline glucose levels or how often and how long people trained.

⚠What it shows: Consistent but modest in size: results pooled from 29 trials in 1,401 adults who were overweight or living with cardiometabolic problems, so the findings may not read across to lean, healthy people. Post-meal glucose fell with both weights and cardio, while insulin only clearly fell with cardio.Meta-analysis · Diabetes Res Clin Pract

Pooled results from 40 randomised trials in 1,460 people show sprint interval training improved blood sugar, insulin and cholesterol markers compared with no exercise, and performed broadly on par with moderate continuous cardio.

⚠What it shows: Promising but shaky: results pooled from 40 trials in 1,460 people pointed the same way, yet the trials disagreed with each other and some results shifted when single studies were removed. The authors themselves rated the certainty of the evidence as low to very low.Review · J Sports Sci

In pooled trials covering 4566 older adults, wearable activity trackers increased physical activity time and daily step counts compared with usual care, but made no measurable difference to BMI, body fat, daily sedentary time or physical function tests.

⚠What it shows: Trustworthy for the headline effect but modest in strength: results pooled from 23 trials in 4566 older adults, with the review's own authors rating the evidence low to moderate. Trackers beat doing nothing rather than beating other structured programmes, and fitness and body composition did not shift.Meta-analysis · J Med Internet Res

Pooled results from 18 studies found oximetry-based devices caught 97% of obstructive sleep apnoea cases but correctly cleared only 63% of people who did not have it. Consumer wearables alone detected 93% of cases, so a negative reading is reassuring while a positive one still needs a proper sleep study.

⚠What it shows: About as solid as screening evidence gets: results pooled from 18 studies comparing these devices against full laboratory sleep tests in adults. The catch is the high false-alarm rate, which means a positive from a wearable only earns a follow-up test, never a diagnosis on its own.Meta-analysis · Sleep Med Rev

A meta-analysis of randomised trials found corn-derived dietary fibre supplements lowered total cholesterol by 7.93 mg/dL and LDL cholesterol by 8.82 mg/dL, with no significant effect on blood sugar control, blood pressure, body measurements or inflammation.

⚠What it shows: Reasonably solid on the cholesterol result: it pools trials where people were assigned at random, and the effect held across different doses and study lengths. Everything else measured, from blood sugar to blood pressure, did not change, so the benefit is narrow.Review · Diabetes Res Clin Pract

Pooling 76 trials in healthy adults, dietary nitrate supplementation gave a small to moderate boost in time-to-exhaustion tests but made no significant difference to time-trial performance.

⚠What it shows: Strong for one narrow claim: results pooled from 76 trials in healthy adults showed a benefit only in tests of how long you can keep going, not in race-style time trials. Optimal dosing and who benefits most are still unknown.Meta-analysis · Sports Med

A systematic review of randomised trials lasting longer than six months found weight loss lowered systolic blood pressure in people with hypertension by about 5.2 mmHg, more than salt restriction (2.9), stress management (1.0) or exercise (0.8); the authors caution that even these figures are probably overestimates.

⚠What it shows: Take the ranking seriously but the exact numbers loosely: this pooled trials that lasted more than six months, yet most were of low quality, bias tended to inflate the changes, and few accounted for people changing other habits at the same time. The reviewers themselves call these figures probable overestimates.Meta-analysis · J Public Health Med

In a small study of 21 healthy adults, a 20-minute treadmill walk significantly increased bowel sounds, an indirect marker of gut motility, within 1 to 2 minutes of finishing the walk.

⚠What it shows: Early days: a small lab study in 21 healthy adults that used bowel sounds as a stand-in for gut movement, so it cannot say whether walking actually relieves constipation. Nobody in the study had bowel problems, and the recording stopped 15 minutes after the walk.Study · Sci Rep

Across 109 daily time-series studies worldwide, short-term rises in air pollution were associated with about 2.0% excess all-cause deaths per typical increment in PM10 particles, with larger effects on respiratory deaths for most pollutants.

⚠What it shows: Strong for a link but still city-level pattern watching: results pooled from 109 studies that track daily death counts against daily pollution, which shows association rather than proof of cause for any one person. The effects shrank, though mostly persisted, when several pollutants were modelled at once.Meta-analysis · J Air Waste Manag Assoc

A Finnish modelling study estimated that light to moderate drinking 'prevents' around 400 coronary heart disease deaths a year among men aged 30 to 69, about 12 to 14% of observed deaths, while only around 20 are 'caused' by drinking above the estimated optimum. The figures rest on associations between drinking and heart disease, not on trials of alcohol itself.

⚠What it shows: Treat these counts as a what-if, not a measurement: the underlying link between moderate drinking and fewer heart deaths comes from studies that followed people rather than testing on them, so healthier habits among moderate drinkers could explain it. The death tallies were modelled from Finnish survey data on 4,818 people, and the authors themselves put 'prevented' in quote marks.Meta-analysis · J Stud Alcohol