For how we live.
Pooling 19 studies, researchers found all-cause mortality raised about 1.7-fold in people with ADHD, with suicide and injury deaths most elevated — while those on ADHD medication had lower all-cause mortality.
Across 27 trials and 88,830 people, each 10-mmHg drop in systolic pressure cut cardiovascular events whether or not the kidneys were impaired — while aggressive lowering nudged up short-term markers of kidney decline.
A meta-analysis of 15 cohorts found 45% of deaths in people with intellectual and developmental disabilities were classed as avoidable — about 3.6 times the rate in people without disabilities, on rich-country data.
Injecting one sperm into each egg was meant for severe male-factor cases, yet clinics now use it widely. Pooling six randomised trials found no live-birth advantage over standard IVF when sperm is not the problem.
A meta-analysis of 36 studies and 2.6 million women found iatrogenic menopause raised stroke and cardiovascular risk. The death-rate excess was confined to menopause before 45, with null estimates after.
A pooled analysis of four trials found depemokimab roughly halved exacerbations, yet it did not significantly improve asthma control or reach the threshold for a meaningful quality-of-life gain.
A meta-analysis of 247 studies across 41 countries pooled a 15.3% detection rate, rising to 38% in fish and seafood markets and running highest across South-East Asia.
A meta-analysis of 10 randomised trials found ultra-processed diets raised weight and eating rate, yet the effect on how much people ate mostly disappeared when diets were not more energy-dense.
Guidelines set weight-gain ranges by starting BMI, not a doubled appetite. A large meta-analysis links gaining outside those ranges to worse outcomes at both ends — smaller and larger babies alike.
Knowing you are dreaming and sometimes steering it is real and surprisingly common, and it can be trained. But deliberately inducing it may carry a mental-health cost for some people.
Ginger's strongest evidence is unglamorous: it eases nausea in pregnancy and during chemotherapy, and modestly relieves osteoarthritis pain. The trials are small and the quality is moderate at best.
Garlic supplements modestly lower blood pressure in hypertensive people, and may nudge cholesterol down a little. The cold-prevention evidence rests on a single trial, and no study shows garlic prevents heart attacks.
Beetroot juice really does lower blood pressure — the effect is one of the better-documented in food science, driven by dietary nitrate. But most trials are short and small, and no study shows it prevents heart attacks.
Across 154 studies, activities that lower arousal — slow breathing, meditation — reduced anger and aggression. Venting-style activities that raise it, like hitting a bag or going for a run, did not.
Retrospective studies keep finding bilinguals develop dementia symptoms three to four years later than monolinguals. Prospective studies that follow healthy people find no drop in who gets it at all.
Two meta-analyses of 104 trials found music lowered both felt stress and physical markers like heart rate. The effect is consistent and moderate — larger for how stressed people feel than for blood pressure.
Echinacea is a pharmacy staple every winter. But a Cochrane review of 24 trials found no reliable treatment benefit, and the largest trial shaved barely half a day off a cold — a difference too small to be sure was real.
Standing up every half hour is common desk advice. A meta-analysis of seven trials found brief standing breaks trim post-meal glucose a little, but short light-walking breaks cut both glucose and insulin far more.
Pooling 58 randomised trials, about 3.5g a day of oat beta-glucan lowered LDL cholesterol by 0.19 mmol/L. A separate meta-analysis of 28 trials found 0.25 mmol/L. Small per person, but consistent.
Trials can show cinnamon nudging fasting glucose down, but the effect is inconsistent, leaves long-term control (HbA1c) untouched, and cheap cassia cinnamon carries a coumarin safety flag.
In people who are iron-deficient without anaemia, supplements ease self-reported fatigue in trials — but do nothing for objective exercise capacity, and the benefit hinges on being iron-deficient, not just tired.
For most women without complications, activity in pregnancy is safe and beneficial. Trials link exercise to lower odds of gestational diabetes and high blood pressure, with a rough dose to aim for.
Heavy loads drove significantly greater one-rep-max strength than light loads in a 21-study meta-analysis, while size gains were similar. A trial found strength maxed out at low volume, the main driver of growth.
An umbrella review of 21 meta-analyses found caffeine helps endurance, strength, power and sprinting. The effect is real but modest, and the useful dose is smaller than many gym pre-workouts pack in.
A short nap does not wreck most people's night sleep, and can help offset sleep debt. The evidence that gives pause is about long naps: naps of 60 minutes or more track with worse cardiovascular outcomes.
It does not. In a Danish cohort of 657,461 children, vaccinated children had a fully adjusted autism hazard ratio of 0.93; a meta-analysis of 1,256,407 children found no link with MMR, thimerosal or mercury.
Two meta-analyses of dozens of randomised trials agree that about 3 grams a day of oat beta-glucan lowers LDL cholesterol by a small, real amount. The effect is genuine, modest, and dose-dependent.
A 43-year meta-analysis found loneliness in young adults rising by about half a standard deviation — while its own authors call the word 'epidemic' exaggerated. The pandemic added a small further bump.
Across 23 cohort studies and 441,199 people, large weight fluctuation was linked to a 41% higher risk of death. Whether the cycling itself does harm, or reflects who cycles, the data cannot fully separate.
For people whose pancreas cannot make enzymes, replacement therapy is proven in randomised trials. For healthy people with bloating, the flagship study is a single-arm survey of customers who bought the product online.
The evidence that recessions and job loss kill is stronger than it looks. Three landmark studies quantify the mortality signal — and where it stops being causal.
In people with healthy kidneys, trials find high-protein diets do not worsen kidney function. The picture changes for those who already have reduced function, where animal protein may speed decline.
A randomised trial found ordinary ibuprofen cut acute mountain sickness from 69% to 43% on a fast ascent. A Cochrane review found most supplements and pre-acclimatisation gadgets do not hold up.
A 2006 Cochrane review found the classic ingredient, potassium, only patchily convincing. A newer network meta-analysis of 32 trials is firmer: stannous, potassium and arginine toothpastes can be recommended for pain.
A jog and a few stretches has thin evidence for cutting injuries. Structured programmes that add strength and balance work — like the FIFA 11+ — cut football injuries by roughly a third in trials.
Observational studies disagree — evening looked best in adults with obesity, morning-to-midday in the wider population. None is a trial, and total activity dwarfs timing.
Physically acting out dreams is one of the strongest known early signs of Parkinson's and related diseases. In a 1,280-patient cohort, most converted within 12 years — but it is a risk marker, not a verdict.
An independent 2026 meta-analysis of AAV micro-dystrophin trials found small, mostly uncontrolled functional benefits — largest in the youngest boys — set against serious harms including rare deaths.
Randomised trials show a Palaeolithic diet trims weight and risk markers in the short term. A two-year trial found the early edge over an ordinary healthy diet had mostly faded — and that people struggled to stick to it.
Pooled trials put the herb roughly on a par with standard antidepressants for mild-to-moderate depression, with fewer dropouts for side effects. The catches are dangerous drug interactions and wildly variable products.
A fear that shaped urology for decades held that raising testosterone feeds prostate cancer. Randomised trials and a large 2024 safety analysis find no such rise in screened men.
A 2025 meta-analysis of 10 trials found small, short-term reductions in weight and BMI — but the studies were heterogeneous, and the viral single trial that fuelled the hype is one small study, not a settled result.
A meta-analysis of four small trials found zilebesiran lowered office systolic pressure by 7.16 mmHg versus placebo, with wide confidence intervals, a hint the effect wanes by six months, and more hyperkalaemia.
Topical diclofenac and ketoprofen beat their carrier gels for knee osteoarthritis pain, with about 60% of users reporting much less pain. A large network meta-analysis now rates topical diclofenac a first-line option.
A randomised trial and a 21-study meta-analysis found thyroid hormone normalised the blood test but left tiredness and symptom scores unchanged. A guideline now advises against routine treatment.
Sun-driven brown patches are hard to shift. A Cochrane review found a triple-combination cream beat hydroquinone alone; newer meta-analyses point to tranexamic acid and PRP, on thinner evidence.
The line is marketing, not trial data. A meta-analysis of 13 randomised trials found adding breakfast raised daily calories and, if anything, nudged weight up, not down. A controlled feeding trial explains why.
Meta-analyses found calcium pills raised heart-attack risk by around a quarter. A later review — funded by an osteoporosis foundation — found no such link. The disagreement is the story.
A 2019 pooling of 12 trials rated peppermint oil safe and effective for IBS, with a number-needed-to-treat of three. The largest, most rigorous trial then missed its primary endpoint against placebo.
A harmonised meta-analysis of over a million adults found 60 to 75 minutes of daily moderate activity erased the extra death risk tied to sitting more than eight hours. The catch is how much activity that is.
WHO and ILO estimate long hours caused 745,000 deaths in one year. The effect is real but modest per person, and clearest for stroke rather than heart disease.
A 17-cohort meta-analysis links each extra 1,000 daily steps to 15% lower mortality, and brisk walkers outlive slow ones. But walking builds little strength or top-end fitness.
A 2016 meta-analysis found training a muscle twice a week beat once. A larger 2019 update found the edge vanished once weekly volume was equated — leaving frequency as a scheduling choice.
Honey and table sugar are both free sugars with almost identical calories. Trials hint at small metabolic differences in honey's favour, but the certainty is low and honey is still added sugar to be limited.
Two systematic reviews point the same way: for the pill, IUDs, implants and most methods, about 8 in 10 women conceive within a year of stopping — similar to using no method at all.
A meta-regression found each extra weekly set added about 0.37% more growth, a graded but flattening curve. A separate trial found strength barely moved between one, three and five sets.
Meta-analyses of more than a million people find a U-shaped link between sleep duration and death, lowest around seven hours. But this is observational, and reverse causation is the hard part to rule out.
Guidelines settle on a limit near 200 mg a day — roughly two mugs of coffee. But the underlying studies are observational, and a vocal minority argues there is no clearly safe level.
Organic produce carries far fewer pesticide residues and modestly more of some antioxidants. But there is no strong evidence that eating organic makes you healthier, and the cancer signal is observational.
A meta-analysis of 56 trials and 134,424 children finds RTS,S and R21 cut clinical malaria by about a third overall, with efficacy strongest alongside seasonal chemoprevention and waning after 36 months.
Meta-analyses put high-intensity intervals roughly 1.2 mL/kg/min ahead of continuous training for peak oxygen uptake. Both produce large gains; the format you will keep doing wins.
Higher whole-grain intake is consistently linked to lower rates of heart disease, cancer and early death, with benefit rising up to about seven servings a day. Refined grains show no such signal.
Cohorts link both to modestly higher rates of heart disease and early death, on low-certainty evidence. A 2019 guideline then told people to keep eating them — and split the field over how to weigh nutrition science.
People who eat nuts have lower rates of heart disease and early death in large cohorts, and a Mediterranean diet with nuts cut cardiovascular events in one trial. Most of the evidence is observational.
Whole fruit is linked to lower type 2 diabetes risk, sugary drinks to clearly higher risk, and fruit juice sits closer to the drinks than to the fruit it comes from.
For heart disease and death, dairy looks roughly neutral to modestly favourable in large cohorts, and the long-standing warning against whole-fat dairy is not well supported by outcome data.
Across nearly 1.1 million people, each extra daily cup tracks with lower diabetes risk in a clear dose-response. It is consistent and large — and still only observational.
A kilogram of muscle burns about 13 calories a day at rest — far less than the fitness-marketing figure implies. But resistance training does nudge resting metabolism up, by roughly 96 calories a day in a meta-analysis.
Pooled trials show a small boost to oxygen uptake and exercise tolerance from dietary nitrate. The effect shrinks toward zero in the most trained athletes.
Training light loads with a limb cuff grows muscle about as well as heavy lifting, trials show. For maximal strength the evidence is mixed, and heavy weights may still edge it.
An individual-participant meta-analysis found metformin given in pregnancy did not lower the odds of gestational diabetes, though it modestly lengthened gestation and reduced preterm birth.
A 2024 meta-regression found strength gains barely track how close you stop to failure, while muscle growth rises the nearer you get. Two trials put the size gap close to zero.
A meta-analysis of 8,459 athletes found programmes with the exercise cut hamstring injuries by about half. In elite football, where the injury keeps rising, most squads leave it out.
A cohort of 5.24 million UK adults linked higher BMI to 17 of 22 cancers — strongly for the uterus, weakly or not at all for others, and inversely for a couple. The pattern is specific, not uniform.
A meta-analysis of 28 studies and 684,065 people links intimate partner violence to raised odds of neonatal, infant and under-5 death — the association strongest for sexual violence.
A network meta-analysis of 163 trials found EPA, vitamin D and a St John's wort extract beat placebo by margins so large the researchers treat them as a warning about the evidence, not a green light.
A meta-analysis of more than 155,000 children finds mortality is elevated even at oxygen saturations of 90–94%, above the threshold at which WHO advises hospital oxygen.
Cochrane reviews rate anti-inflammatory painkillers a very effective treatment for menstrual cramps, roughly doubling the odds of good pain relief. Exercise helps too, but on lower-quality evidence.
A network meta-analysis of 53 studies and more than 7 million admissions finds machine learning out-discriminates traditional sepsis scores — and would raise roughly two false alarms for every real one.
A meta-analysis of 14 studies and 7,465 men found little for the conventional semen parameters, and something more consistent for sperm DNA fragmentation.
Across 34 trials, mindfulness apps produced small effects on stress, anxiety and depression, and none on general distress. A 2026 review confirms they can help without a facilitator — at low certainty.
A Cochrane review found women doing pelvic floor muscle training were eight times more likely to report their stress incontinence cured than untreated women — with the strongest evidence for that one condition.
The gap between those two numbers is the whole argument. An individual-patient meta-analysis of 20,827 people found real effects on both comparisons, and found that the choice of sham changed the answer.
A review screening 14,125 articles found evidence neither for nor against morning or evening training — only that testing and training at the same hour helps. Evening sessions may cost some sleep continuity.
A Cochrane review of 50 trials found cranberry products lower UTI risk in women with recurrent infections and in children, but no benefit in the elderly, in pregnancy, or in bladder-emptying problems.
A meta-analysis of 51 studies found no effect on race time or time to exhaustion. The 2026 review of post-exercise compression for soreness rated its own evidence very low certainty.
A 2026 meta-analysis of 86 randomised trials found body mass, waist circumference and maximal oxygen uptake still better than baseline more than six months after training stopped. Not everything held.
A meta-regression of 23 hypertrophy studies found total intake the strongest predictor and no timing effect after adjustment. A later network meta-analysis of 116 trials still favours post-exercise, narrowly.
A 2026 meta-analysis of 19 trials found training at maximal fat-oxidation intensity cut body weight by 3.66 kg. Head-to-head trials against interval training find no consistent winner on body fat.
Two 2026 meta-analyses of type 2 diabetes apps found HbA1c reductions of 0.39 and 0.58 points versus usual care — modest, clinically meaningful, and consistent, though the trials vary widely in quality.
In a meta-analysis of 25 randomised trials and 26,610 participants, stretching produced no reduction in sports injuries. Strength training cut them to under a third — on a much smaller evidence base.
An umbrella review of 41 systematic reviews found no consensus that posture causes back pain. And correcting posture reliably changes alignment — but not, consistently, pain.
A 223-study meta-regression reports sperm concentration falling 51.6% since 1973. Two published critiques argue the method cannot support that reading. Both sides are peer-reviewed.
A review of 110 studies found that 52% of trialled self-guided apps were publicly available — and that evidence quality had no bearing on which ones made it to an app store.
A meta-analysis of six studies found a moderate effect on sleep duration and a small one on depressive symptoms — with high heterogeneity on both.
The meta-analysis behind ketamine's widening off-label use for suicidal crisis quantifies what 1,166 patients across two decades of trials actually showed — and where the evidence runs out.
Pooling 11 cohort studies, researchers found a consistent cardiovascular signal in Asian and North American cohorts. European cohorts showed no elevated risk at all, a geographic split the analysis can't explain.
A meta-analysis of fully automated programmes found a large effect on insomnia severity. A review of 68 studies found the field has barely measured whether recipients find them acceptable.
A double-blind trial found a lidocaine paracervical block lowered pain during IUD placement, most in women who had never given birth. A separate meta-analysis found the benefit hinges on the drug and route.
Pooled across 48 trials, chatbots cut depression, anxiety and stress by small amounts. But a review of purpose-built mental health chatbots found crisis-referral protocols underdeveloped and missed suicidal ideation.
A review pooling 23 studies found risk concentrated among people with bipolar I disorder, family history, or unsupervised use. About 4% later transitioned to bipolar disorder.
A network meta-analysis compared the ways athletes split training across intensity zones. Every credible interval crossed zero — and the model ranked first for aerobic capacity was not the popular one.
Pooled trials find forest therapy lowers stress, anxiety, and blood pressure short-term. But heterogeneity is high and reviewers grade the certainty as low to very low.
The largest syntheses find apps beat doing nothing on depression and anxiety by about a third of a standard deviation — but the edge shrinks against real treatment, and one umbrella review found no convincing proof.
A meta-regression of 169 trials coded 34 therapeutic ingredients. The most common — psychoeducation, relaxation, mindfulness — were not the ones that best predicted getting better.
Pooling decades of research, the combined estrogen-progestin regimen showed a 44% higher breast cancer risk. Estrogen-only therapy showed none in observational studies — and trials suggested it lowered risk.
Pooling 94 cohort studies, grip strength tracked with later dementia, depression and disability. The authors rate their own confidence in the numbers as limited to moderate.
An Oxford meta-analysis of 37 studies puts average regain at 0.4 kg a month across all weight-management drugs, and 0.8 kg a month for semaglutide and tirzepatide. Almost none of the newer-drug data runs past a year.
Pooling ten studies, researchers found menopause hormone therapy neither raised nor lowered dementia risk. The certainty of that evidence ranged from moderate to very low.
An individual-patient meta-analysis of 17,801 patients found a hazard ratio of 0.97 for death, reinfarction or heart failure — a result precise enough to make the question look settled.
A meta-analysis of 22 crossover trials with overnight polysomnography found consistent losses in total sleep time and efficiency. The age and dose subgroup differences everyone wants were not statistically significant.
Pooling individual data from 23,340 people in four trials, empagliflozin cut kidney failure by 34% and slowed eGFR decline by 64% — regardless of how big an initial creatinine rise was expected.
Pooling 31 studies of 556 patients, cortical 5-HT2A binding was lower in unmedicated depression than in controls. The post-mortem studies mostly found nothing, and the tracers used shape the result.
A meta-analysis of pedometer, tracker and smartwatch interventions lasting 12 to 48 weeks found no significant change in systolic or diastolic pressure — nor in weight, glucose or HbA1c.
Collagen, flavanols and polyphenols beat their comparators on skin elasticity and sun tolerance. The pooled estimates carry heterogeneity of 95% to 98%, and some comparators were low doses of the same ingredient.