Study archive · page 4

A critical appraisal of 15 published reviews concluded that the short-term association between airborne particles and cardiopulmonary mortality is valid and satisfies most standard criteria for causality, with effects observed even where air quality standards were respected.

⚠What it shows: A 2001 critical assessment of the arguments in 15 existing reviews of ecological time-series studies, not a pooled analysis of individual data; time-series designs compare days rather than people, and the exposure landscape has changed since publication.Meta-analysis · J Air Waste Manag Assoc

Across five periods in Beijing between 1990 and 2017, concentrations of SO2, NO2, PM10 and PM2.5 fell by up to 42%, 10%, 33% and 15%, and the estimated short-term mortality effect of each pollutant shrank alongside them — PM2.5's effect on cardiovascular and respiratory deaths falling by up to 0.33% and 0.13%.

⚠What it shows: Longitudinal comparative study in one city: published effect estimates for 1990-2013 pooled by random-effects meta-analysis and combined with newly analysed 2015-2017 daily data. The design is ecological, so parallel changes in healthcare, smoking or population structure cannot be excluded.Meta-analysis · Chemosphere

In a meta-analysis of cohort studies with individual-level data, each 1 µg/m3 increase in long-term PM2.5 exposure was associated with a 4% higher risk of SARS-CoV-2 infection, an 11% higher risk of COVID-19 hospitalisation and a 9% higher risk of COVID-19 death; NO2 was associated with hospitalisation and death, while ozone showed no association with any outcome.

⚠What it shows: Systematic review and random-effects meta-analysis restricted to cohort studies with individual-level data, pooled only where more than three studies existed per exposure-outcome pair. Observational throughout, so residual confounding remains possible, and country of study explained much of the heterogeneity between results.Meta-analysis · Environ Pollut

Across 12 counties in Hubei, China, covering 42 739 stroke deaths between 2009 and 2012, cold spells were associated with a pooled stroke mortality risk of 1.180 at lags of 3-14 days, with the effect appearing after a delay of 2-3 days and lasting about 10 days, while heatwaves carried a pooled risk of 1.114 concentrated in the first 0-2 days.

⚠What it shows: Observational multivariate meta-analysis pooling daily mortality and weather records from 12 counties in one Chinese province over four years; ecological time-series design with no individual data, and county-level estimates varied widely, from 0.968 to 1.523 for cold spells.Meta-analysis · Zhonghua Liu Xing Bing Xue Za Zhi

Pooling 58 observational studies, extreme cold carried the highest all-cause mortality risk at a relative risk of 1.478, above heatwaves at 1.280 and extreme heat at 1.145; heatwaves were also associated with more hospitalisations (1.060) and emergency department visits (1.047).

⚠What it shows: Systematic review and meta-analysis of 58 observational studies published between 2000 and 2024, so all estimates are associations from population records rather than controlled comparisons; definitions of extreme cold, extreme heat and heatwave vary between studies, and few studies covered length of stay or costs.Meta-analysis · Int J Hyg Environ Health

In a meta-analysis of 28 studies, days combining heat with high ozone were associated with a 17% higher overall mortality risk than days without co-exposure, with elevated cardiovascular and respiratory mortality and increased risks across age and gender groups and in urban areas.

⚠What it shows: Systematic review of 78 studies with 28 pooled by random-effects meta-analysis; all observational, with the authors flagging risk of bias from inadequate confounder adjustment and unclear handling of missing data, plus inconsistent definitions of what counts as a compound heat-ozone event.Meta-analysis · Environ Pollut

A meta-analysis of 33 randomised trials in 1,396 middle-aged and older adults with type 2 diabetes found resistance training reduced the atherogenic index of plasma (effect size -0.56) and improved glycaemic control, lowering HbA1c by 0.62% and HOMA-IR by 0.90. Systolic blood pressure fell by 3.91 mmHg and body fat by 0.54%, while VO2max rose, with stronger effects in overweight or obese participants.

⚠What it shows: Meta-analysis of 33 randomised trials with a modest combined sample (n = 1,396) of middle-aged and older adults with type 2 diabetes; heterogeneity was substantial across several outcomes, methodological quality varied, and the included studies were geographically concentrated, so pooled effect sizes are indicative rather than precise.Meta-analysis · Maturitas

Pooling five randomised trials in older adults (74 doing photobiomodulation plus resistance training, 72 doing sham), adding light therapy to periodised lifting did not significantly improve maximal strength (SMD 0.26, 95% CI -0.08 to 0.59; p = 0.14). A small benefit could not be ruled out.

⚠What it shows: Systematic review of 11 RCTs (n = 456), but only five trials with 146 participants in total reported one-repetition maximum and were pooled; devices, wavelengths and dosing varied, and the authors say a small effect cannot be excluded.Meta-analysis · Photobiomodul Photomed Laser Surg

In a meta-analysis of seven randomised trials in adults with low bone mass, combining exercise with antiresorptive drugs increased lumbar spine bone mineral density significantly more than the drugs alone (SMD 0.55, 95% CI 0.36 to 0.75; p < 0.00001).

⚠What it shows: Meta-analysis of seven RCTs in adults with low bone mass; most trials scored 3 on the Jadad quality scale, and the authors call for longer trials with larger samples. Only lumbar spine bone density was measured — not fractures, and not other skeletal sites.Meta-analysis · Osteoporos Int

Across 20 randomised trials in older adults, probiotic supplementation showed a small positive trend on cognitive test scores and a reduction in inflammatory markers, but neither reached statistical significance. Heterogeneity between trials was high enough that no true overall effect size could be determined.

⚠What it shows: Meta-analysis of 20 randomised controlled trials in older adults; both pooled effects had confidence intervals crossing zero, and the authors state that high heterogeneity prevents determining a true overall effect size.Meta-analysis · Benef Microbes

Combined aerobic and resistance training improved global cognition in healthy adults aged 65 and over compared with doing nothing, but a meta-analysis of 12 randomised trials found it was no better than active comparison conditions.

⚠What it shows: Meta-analysis of 12 RCTs (2557 healthy adults aged 65+); the overall pooled effect was non-significant, benefit appeared only against passive controls, and heterogeneity was moderate.Meta-analysis · Exp Gerontol

In adults with depression, regular exercise reduced rumination and improved cognitive shifting in a meta-analysis of randomised trials, with benefits tied to specific doses: moderate-intensity aerobic sessions for rumination and high-intensity aerobic training for shifting. Inhibition and updating did not improve significantly.

⚠What it shows: Meta-analysis pooling few RCTs per outcome (5-9 studies, 260-832 participants); dose effects come from subgroup analyses, and the authors themselves call the findings preliminary and tentative.Meta-analysis · Psychol Sport Exerc

Pooling 54 randomised trials in 49,602 people with type 2 diabetes, GLP-1 receptor agonists were associated with a lower risk of bone fracture than placebo or other glucose-lowering drugs, and exenatide ranked as the most favourable agent (RR 0.17, 95% CI 0.03-0.67). A statistically significant inconsistency appeared in some of the indirect comparisons.

⚠What it shows: Bayesian network meta-analysis of 54 RCTs (49,602 participants, 28,353 on GLP-1 RAs) in people with type 2 diabetes, not healthy adults. Fractures were rarely the trials' main outcome, the exenatide estimate rests on very few events with a wide confidence interval, and the network showed significant inconsistency in some comparisons, so drug rankings are fragile.Meta-analysis · Osteoporos Int

Pooling 8 cardiovascular outcome trials of GLP-1 receptor agonists in type 2 diabetes (60079 patients), the trials that lowered HbA1c and body weight most also showed the largest reductions in major cardiovascular events, at an estimated 25% lower relative risk per 1% absolute HbA1c reduction and 7% per 1 kg of weight lost. Reductions in systolic blood pressure and LDL cholesterol were not associated with event reduction.

⚠What it shows: Trial-level meta-regression across 8 cardiovascular outcome trials (60079 patients, 30693 on GLP-1 RAs) in type 2 diabetes. Because it correlates average results between trials rather than tracking individuals, it cannot show that losing weight caused the event reduction, and the weight-loss association was only just statistically significant (p = 0.047).Meta-analysis · Cardiovasc Drugs Ther

Across 11 randomised trials in 546 adults with prediabetes, prebiotic supplementation did not significantly change fasting glucose, insulin, HbA1c, HOMA-IR or blood lipids, and did not shift BMI or waist circumference. The single significant result was a reduction in body fat percentage, rated very low certainty.

⚠What it shows: Systematic review and meta-analysis of 11 randomised trials totalling 546 adults with prediabetes; the authors themselves rate certainty in the pooled estimates as very low for most outcomes, and the one positive finding sits on a wide confidence interval.Meta-analysis · J Nutr

In a systematic review of ten moderate-to-high quality trials in knee osteoarthritis, quadriceps strengthening reduced pain and improved function, and outperformed proprioceptive training. No one strengthening method beat the others.

⚠What it shows: Systematic review of ten trials selected from 1,128 records, synthesised narratively rather than pooled statistically; exercise protocols, comparators and add-on treatments varied widely, and all participants had diagnosed knee osteoarthritis rather than healthy knees.Systematic review · Rheumatol Int

Pooling 183 studies of people with knee osteoarthritis, walking produced the greatest knee loading and stair ascent the most pain, while sit-to-stand was lowest on both; during walking, higher knee loading was correlated with less pain (r = -0.507).

⚠What it shows: Systematic review and meta-analysis of 183 studies, but the loading differences across activities were mostly non-significant apart from sit-to-stand, the populations and methods were highly heterogeneous, and many studies measured only one activity — usually walking. The loading-pain correlation is cross-study association, not cause.Meta-analysis · Clin Biomech (Bristol)

An umbrella review pooling meta-analyses of observational studies linked higher dietary fibre intake to a 22% lower cancer risk and 17% lower cancer mortality, with the strongest inverse association for digestive tract cancers and a smaller one for breast cancer. Observational data cannot establish cause.

⚠What it shows: Umbrella review and meta-meta-analysis of 25 papers and 28 reports, all built on observational epidemiology — people who eat more fibre differ in many other ways, and no randomised trial evidence is included, so causation is not established.Meta-analysis · Crit Rev Food Sci Nutr

A meta-analysis of 13 randomised trials found oropharyngeal (mouth-and-throat) muscle training and respiratory muscle training were associated with reductions in obstructive sleep apnoea severity, while conventional resistance training was not. Daytime sleepiness showed no consistent, clinically meaningful improvement with any modality.

⚠What it shows: Meta-analysis of 13 small RCTs totalling 429 participants; substantial heterogeneity in protocols, durations and participants, and daytime sleepiness did not consistently improve.Meta-analysis · Respir Med

Pooling randomised trials, probiotic supplementation produced small increases in lumbar spine areal bone mineral density (+0.010) and hip aBMD (+0.022 g/cm2), with no significant effect on femoral neck density or on the bone turnover markers P1NP and CTX. Most included trials were in postmenopausal women.

⚠What it shows: Systematic review and meta-analysis of randomised trials (16 comparisons for lumbar spine, 9 for hip, 7 for femoral neck), mostly in postmenopausal women, with moderate-to-high heterogeneity throughout; the hip effect became non-significant when key trials were removed, and no fracture outcomes were pooled.Meta-analysis · Osteoporos Int

A Bayesian network meta-analysis of 93 randomised trials in 4,446 adults with overweight or obesity found that all six exercise types tested improved VO2max, with high-intensity interval training ranked highest and resistance training alone ranked least effective.

⚠What it shows: Network meta-analysis of 93 randomised trials with 4,446 adults living with overweight or obesity; rankings come from indirect statistical comparison rather than head-to-head trials, and the pooled effect sizes are unusually large, which suggests variation in how VO2max was measured across studies.Meta-analysis · J Sports Sci

Pooling 8 randomised trials in women with polycystic ovary syndrome, probiotic or synbiotic supplements produced small reductions in fasting blood sugar (-2.52 mg/dl), insulin (-2.27 uIU/mL), HOMA-IR (-0.69), C-reactive protein and total testosterone (-0.12 ng/mL). Weight and body mass index did not change significantly.

⚠What it shows: Meta-analysis of only 8 RCTs (nine treatment arms) in women with polycystic ovary syndrome, searched to September 2018. The individual trials were small and short, effect sizes are modest, strains and doses varied between studies, and the findings do not transfer to women without PCOS.Meta-analysis · Eur J Clin Nutr

A Cochrane review of six randomised trials with 599 adults found that exercising shortly before an influenza vaccination was neither beneficial nor harmful, with no significant difference in outcomes between people who exercised and those who did not. The one trial reporting actual influenza cases also found no significant difference.

⚠What it shows: Cochrane review of six small RCTs (599 adults aged 18-80, study size 46-158), rated low quality: five of six trials had at least four unclear risk-of-bias domains, most measured biochemical rather than participant-centred outcomes, and no trial reported influenza complications.Meta-analysis · Cochrane Database Syst Rev

A meta-analysis of 30 randomised trials in 2,124 overweight or obese postmenopausal women found exercise training reduced TNF-alpha (SMD -0.47) and CRP (SMD -0.36), while IL-6 and adiponectin did not change significantly. Neither total weekly exercise volume nor programme duration showed a clear dose-response, and higher intensity was only borderline associated with larger TNF-alpha reductions (p = 0.063).

⚠What it shows: Meta-analysis of 30 randomised trials (N = 2,124) restricted to overweight and obese postmenopausal women, so it does not generalise to other groups. The intensity signal is a borderline meta-regression trend (p = 0.063), not a demonstrated threshold, and the authors' 'intensity threshold' framing goes beyond what that statistic establishes.Meta-analysis · Front Immunol