Mind & Body · Thursday, 17 September 2026

01 Briefing what happened

When the top of the heart quivers, blood sits still and can clot - and watches now find short episodes nobody knows whether to treat

In atrial fibrillation the heart's upper chambers quiver instead of beating, and the stroke risk depends on how long it lasts. Smartwatches find far more of it, and trials have not shown the shortest episodes are worth treating.

the risk of a stroke from a blocked blood vessel, in people with atrial fibrillation

about one person in 11 aged 65 and over has the condition [3][4]

over 4% vs under 1.68%

yearly stroke risk when device-found episodes last over a day, against under six minutes

US guidance backs blood thinners for these episodes only past 24 hours, in people at higher risk [10]

2.84x

the odds of being diagnosed with atrial fibrillation under continuous screening

blood-thinner use doubled, with no clear fall in strokes or deaths [11]

37%

fewer strokes with the blood thinner apixaban in the ARTESIA trial of device-found episodes

the same drug caused more bleeding than aspirin [10]

The lead story — what happened

A normal heartbeat starts in a small cluster of cells called the sinus node. Its signal spreads across the two upper chambers, the atria, then reaches the lower chambers, which squeeze and pump blood out. [1]
In atrial fibrillation the signals start in many places across the atria at once. They fire 300 to 500 times a minute, so the atria quiver instead of beating. [2]
A small relay between the upper and lower chambers cannot pass all those signals on. Enough get through to make the lower chambers beat unevenly, sometimes as fast as 175 beats a minute. [3]
A quivering chamber does not push blood out properly. Blood pools inside the heart, and a clot can form there. [2]
If a clot travels out of the heart it can block blood flow to the brain. Atrial fibrillation is linked to about five times the risk of that kind of stroke. [4][5]
About one person in 11 aged 65 and over has atrial fibrillation. Some feel a racing, fluttering heart, and some feel nothing at all. [3][6]
In people who have already had a stroke, finding atrial fibrillation and starting a blood thinner cuts the risk of another stroke by about 60%. [7]
About half a billion smartwatches now carry a pulse sensor. In large studies run with Apple, Fitbit and Huawei, between 0.23% and 1% of wearers were sent an irregular-rhythm alert. [8]
Pooled across studies, watches picked up atrial fibrillation in 95% of the people who had it. They correctly cleared 97% of the people who did not. [9]
Doctors count an episode as atrial fibrillation once it lasts 30 seconds on a heart tracing. Depending on that definition and the device, device-found episodes turn up in 30% to 60% of the people being monitored. [10]
Stroke risk climbs with how long the episodes last. It was under 1.68% a year for episodes under six minutes and over 4% a year for episodes longer than a day. [10]
Two trials tested blood thinners for episodes found only by a device. NOAH-AFNET 6 was stopped early because too few strokes happened, while ARTESIA found apixaban cut strokes by 37% but caused more bleeding. [10]
A 2026 analysis of screening trials found continuous monitoring nearly tripled diagnoses and doubled blood-thinner use. Neither continuous nor occasional screening clearly cut strokes, bleeds or deaths on its own. [11]
One treatment, ablation, uses heat, cold or short electrical pulses to scar the small areas of heart tissue that start the faulty signals. [12]

Who is involved

The ARTESIA trial

a study of 4,012 people whose irregular rhythm was found only by an implanted device; it compared apixaban with aspirin
NOAH-AFNET 6

a similar trial of the blood thinner edoxaban; it was stopped early because too few strokes happened to judge it
Apple, Fitbit and Huawei

smartwatch makers whose heart studies enrolled more than a million wearers between them

Where this points

Current US guidance backs blood thinners for device-found episodes only when they last over 24 hours in higher-risk people, and the review of ARTESIA argues that line may need checking again. [10]

What is pushing on the whole day

The bar and the word are our reading of how hard each one is pushing today. The arrow is where it is heading. The evidence is in the stories below.

Where the illness line is drawn High↑

The US diagnostic manual lowered the number of symptoms an adult needs for ADHD from six to five. [13] A bone scan score of minus 2.5 is the line between osteoporosis and the milder osteopenia. [14] England's health guidance body starts a high blood pressure diagnosis at 140/90 in the clinic. [15]

More people diagnosed and treated Building↑

Continuous heart screening doubled the number of people put on blood thinners. [11] US prescriptions for ADHD stimulants rose 30% between 2018 and 2022. [13] More than half of US adults with an ADHD diagnosis were first diagnosed as adults. [16]

Health workers worn out High→

Burnout among health workers rose from about 32% before COVID-19 to 46% to 52% at its peaks. [17] At a teaching hospital in Yola, Nigeria, 70% of doctors and nurses had burnout as colleagues emigrated. [18] In a US emergency department, staff with six or fewer days off a month had 4.7 times the odds of burnout. [19]

One break leading to the next Building↑

Of 40,391 older Israelis who broke a hip, 43% broke a bone again. [20] Nearly half of people over 85 in northern Sweden who broke a hip died within a year. [21] After a stroke with no clear cause, finding hidden atrial fibrillation is how doctors try to stop a second one. [7]

Sleep cut short Building→

People whose genes push them toward four or more caffeinated drinks a day sleep less. [22] Waking people briefly during dreaming sleep stopped their bodies calming down to upsetting pictures overnight. [23] Sleep apnoea, where breathing stops during sleep, can set off atrial fibrillation at night. [24]

The rest of the day

47 more stories on this beat.

Each with its own sources. None of these is a link to the story above.

02
A gut chemical linked to irregular heartbeat

Researchers at Cleveland Clinic, a large US hospital group, reported in July that a chemical made by gut bacteria is linked to atrial fibrillation. [25] The chemical is called TMAO. The team measured it in blood from more than 5,000 people having heart procedures, and tested it on heart cells in the lab. [25] In those cells it changed the heart's electrical environment and made abnormal rhythms more likely. [25]

Why it matters — The same team had already linked high TMAO to kidney disease, heart attack and stroke. [25] A researcher on the team said the finding may open a way to act earlier through diet, though blood samples and lab cells show a link, not proof. [25]
03
Prediabetes raises the risk of atrial fibrillation by a fifth

Prediabetes means blood sugar above the normal range but below the line for diabetes. [26] A review published in November pooled 11 studies covering more than 15 million people and 277,164 new cases of atrial fibrillation. [26] People with prediabetes had about a 20% higher risk than people with normal blood sugar. [26] The studies did not all use the same definition, because guidelines set the prediabetes line in different places. [26]

Why it matters — Diabetes was already an established risk factor for the rhythm problem, and this places some of the risk below the diabetes line. [26] Results varied a lot between studies, and the authors checked whether the definition used was one reason. [26]
04
COVID-19 set off irregular heartbeats in hospital

A review of atrial fibrillation in COVID-19 gathered hospital data from across the pandemic. [27] In one study of 30,999 patients at 120 centres, 5.4% developed atrial fibrillation for the first time while in hospital. [27] Nearly half of those patients died during that first stay. [27] Inflammation from the infection disturbs the heart's upper chambers, which the authors say helps start the rhythm and keep it going. [27]

Why it matters — A new irregular rhythm in a very sick patient marks someone at high risk of intensive care, clots and death. [27] It shows the rhythm can be set off by something outside the heart.
05
Why the rhythm often starts at night

A Cleveland Clinic heart researcher, David Van Wagoner, explained in June why atrial fibrillation often begins during sleep. [24] The heart normally slows at night, and a slower rate lets other clusters of cells start firing their own signals. [24] Sleep apnoea, where breathing stops for moments during sleep, raises pressure in the chest and strains the heart. [24] Atrial fibrillation is more common in people with sleep apnoea, high blood pressure and obesity. [3]

Why it matters — Many people have both sleep apnoea and atrial fibrillation without realising it. [24]
06
A watch caught 69% of rhythm problems after heart surgery

Doctors in a study gave a Withings Scanwatch to patients recovering from heart surgery, and checked it against hospital monitors for 24 hours. [28] Atrial fibrillation follows 20% to 30% of such operations. [28] Of 260 patients, 32 had the rhythm, and the watch picked up 69% of them. [28] When the watch raised an alarm it was right 87% of the time. [28]

Why it matters — The authors concluded the watch cannot replace hospital monitoring for these patients. [28] Its 69% is well below the 95% pooled across other smartwatch studies. [9]
07
A phone camera app tested in 236 people

FibriCheck is an app that reads the pulse through a smartphone camera and flags atrial fibrillation. [29] A study at several centres in different countries tested it on 236 people. [29] It found 96.3% of the people with the rhythm and correctly cleared 99.3% of those without it. [29] It worked less well in people with darker skin and higher body weight, until a technician checked the readings. [29]

Why it matters — A tool that works less well on darker skin gives some groups less accurate results unless a person checks. [29] Most earlier tests of the app had been small studies at single European centres. [29]
08
Energy drinks and the heart, across 37 studies

A review published in November looked at 37 studies of energy drinks and the heart. [30] The studies covered 1,597 people with an average age of 22.6, and Red Bull and Monster were the brands most tested. [30] The review reported raised heart rate and blood pressure after drinking them. [30] It also reported a longer QT interval, a stretch of the heartbeat where changes can signal a risk of rhythm problems. [30]

Why it matters — The energy drink industry is worth about 74 billion US dollars. [30] The studies measured short-term effects in young adults, not whether regular drinking causes heart disease.
09
Hunting for hidden rhythm after an unexplained stroke

About a quarter of strokes caused by a blocked blood vessel have no clear cause after full tests. [7] A trial called WATCH AFib plans to fit 400 such patients with both a smartwatch and an implanted heart recorder for six months. [7] Short hospital monitoring finds new atrial fibrillation in only about 2% to 6% of stroke patients. [31] Implanted recorders find it in 17% to 25% within a year. [31]

Why it matters — Finding the rhythm changes treatment, because blood thinners cut repeat strokes. [7] The trial will compare how well a watch and an implanted recorder find it in the same patients. [7]
10
A computer model beat the standard stroke score

Doctors decide on blood thinners partly with a points score called CHA2DS2-VASc, which adds up age and other conditions. [32] A 2026 study built computer models using only age, other illnesses and blood tests in people newly diagnosed with atrial fibrillation. [32] The models ranked stroke risk far better than the score in outside groups, with a score of 0.88 against 0.61 on a scale where 1 is perfect. [32] A Finnish study found dropping the extra point given for being female worked slightly better. [32]

Why it matters — The points score decides who is offered a blood thinner, so a point added or removed moves people across the treatment line. [32]
11
Nearly half of over-85s died within a year of a broken hip

Researchers in Umea, Sweden, followed 1,277 people aged 85 and over through home visits between 2000 and 2017. [21] Over five years, 184 of them broke a hip, and 47.8% of those died within a year of the fracture. [21] Depression, an earlier stroke and breaks just below the hip joint raised the risk of dying. [21] Carrying extra weight went with a lower risk. [21]

Why it matters — Across older adults generally, 20% to 24% die within a year of a hip fracture and about a third need a care home. [33] Among the very old, the Swedish figure is about double that.
12
43% of Israelis who broke a hip broke a bone again

Clalit Health Services, an Israeli health provider, holds records for 40,391 members aged 65 and over who broke a hip between 2016 and 2023. [20] Of them, 17,364, or 43%, had another fracture during the study period. [20] A diagnosis of osteoporosis at the time of the first break predicted a second one. [20] So did an earlier stroke, depression and being female. [20]

Why it matters — A first broken hip is a warning that more breaks are likely. [20] The authors argue for targeted prevention in the people most at risk after that first break.
13
A bone drug cut second hip fractures in a Chinese study

Doctors in China looked back at 259 patients aged 65 and over who had surgery for a broken hip in 2020 and 2021. [34] Some were given denosumab, a drug that slows the cells that break down bone. [34] After matching similar patients, 8.8% of those on the drug had a second hip fracture, against 29.7% of those not on it. [34]

Why it matters — The study looked back at records rather than randomly assigning the drug, so other differences between the groups may explain part of the gap. [34] It fits the Israeli finding that a first hip fracture often leads to another. [20]
14
Why UK patients miss bone treatment

A review published in January gathered ten years of UK research on why people over 50 are not treated for osteoporosis. [35] Bone loss begins between ages 30 and 40, and by 80 the average person has lost more than half their bone mass. [35] The review found a large gap between who needs treatment and who gets it, despite effective drugs and national guidelines. [35] Causes included missed diagnoses, low sense of risk and people not keeping up their medicine. [35]

Why it matters — The authors call for stronger fracture liaison services, teams that check bones after a first break. [35] They also call for better access to scans and support to keep taking medicine. [35]
15
Inactive, thin men in 1970s Oslo broke hips decades later

In 1972 and 1973, 12,900 Oslo men aged 40 to 49 reported how active they were and had their height and weight measured. [36] Linking to national records, 1,542 of them, or 12%, later broke a hip, at a median age of 82. [36] Men who were inactive in midlife had a 38% higher risk than active men. [36] The highest risk was in men who were both inactive and thin. [36]

Why it matters — Research on preventing hip fractures in men is limited. [36] The link rests on one questionnaire, and the men's habits may have changed over five decades.
16
Hip fractures after falls in Korean hospitals and care homes

Korea's emergency department injury records covered 170,095 people aged 65 and over who came in after a fall between 2011 and 2020. [37] The share with a broken hip rose with age, with odds over four times higher at 85 and over than at 65 to 74. [37] Falls inside hospitals and nursing homes were more likely to break a hip than falls at home. [37]

Why it matters — The authors say fall prevention should be aimed at nursing homes and hospitals, where falls more often broke a hip. [37]
17
More broken hips worldwide, lower rates

Two studies drew on the Global Burden of Disease 2021, a worldwide health database. [38][39] Falls cause about 684,000 deaths a year. [38] The number of hip fractures from falls in older people rose from 1990 to 2021 and is projected to keep rising to 2036, while the rate per person falls. [38] Among women around menopause, there were 300,104 cases at ages 50 to 54 in 2021, against 164,196 at ages 45 to 49. [39]

Why it matters — The count rises because there are more old people, even as each person's risk drops. [38] A rising count and a falling rate are both true at once.
18
A 92-year-old broke her spine after a 'normal' scan

Doctors described a 92-year-old woman whose bone scans put her fracture risk below the line for treatment. [40] Her spine read normal and her hip read as osteopenia, the milder zone between minus 1 and minus 2.5. [40] Soon after, she had a spontaneous fracture in her spine. [40] Wear in the spine can make a scan look healthier than the bone really is. [40]

Why it matters — About one in three adults over 50 without osteoporosis has osteopenia. [41] The case shows a person on the safe side of the treatment line can still break.
19
Older athletes and stress fractures

A review published in September looked at bone health in older people who compete in sports such as running. [42] Their training can match professional volumes. [42] High-impact exercise usually thickens bone, yet it also raises the risk of stress fractures, small cracks from repeated load. [42] Many older athletes do not know their bone density has fallen, and the review stresses scans and rest. [42]

Why it matters — The same running that protects bone in most people can crack it in someone whose bone is already thin. [42] Fitness is not a reliable sign of strong bones.
20
Rehab after a broken hip for people with dementia

A review published in October looked at recovery after hip fracture in older people with memory problems. [33] Cognitive impairment is common in these patients and raises the risk of falls, slow recovery and moving into care. [33] People with mild to moderate impairment still made real gains with intensive, tailored exercise programmes. [33] Such programmes are often blocked by untrained staff, few resources and no agreed rules on who qualifies. [33]

Why it matters — The review says people with dementia can gain from rehab that many programmes are not set up to give. [33]
21
Burnout rose in the pandemic and so did mistakes

A review published in February pooled research on more than 500,000 health workers. [17] Burnout rose from about 32% before COVID-19 to 46% to 52% at its peaks, and 24% to 34% of frontline staff had PTSD. [17] Doctors with burnout had 2.72 times the odds of reporting medical errors. [17] Being sued then deepened depression and thoughts of suicide. [17]

Why it matters — The review describes burnout, PTSD and errors each making the others worse. [17] Its authors call for workload limits, psychological support and blame-free reporting of errors. [17]
22
70% burnout at a Nigerian hospital as staff leave

Researchers surveyed 200 doctors and nurses at Modibbo Adama University Teaching Hospital in Yola, in Nigeria's Adamawa State. [18] Using the standard Maslach burnout questionnaire, they found 70% had burnout. [18] Low sense of achievement was the most common part, at 51.5%, with high exhaustion at 29%. [18] The state has severe staff shortages as doctors and nurses emigrate. [18]

Why it matters — Emigration and a growing workload fall on the staff who remain. [18] Patients in a state with severe staff shortages depend on them.
23
Doctors in southern Italy report little reward

An anonymous survey in July and August 2025 asked doctors and trainees at a large hospital in Catania, Sicily, about burnout. [43] 76.7% reported low gratification from their work and 70.8% high depersonalisation, treating patients as objects. [43] Nearly half had high emotional exhaustion. [43] Women and doctors who qualified after 2020 had lower gratification and more distress. [43]

Why it matters — The survey covered one hospital, so it cannot say whether the same holds across Italy. [43]
24
Days off tracked burnout in a US emergency department

An academic emergency department in the US surveyed its doctors and practitioners with the Maslach questionnaire. [19] Of 55 who replied, 52.7% had burnout. [19] Those with six or fewer days off a month had 4.7 times the odds of burnout. [19] So did those with fewer hobbies and less than four hours outdoors a week. [19]

Why it matters — The survey had only 55 replies, so the size of each link is uncertain. [19]
25
Doctors in Singapore clinics most exhausted

Singapore's National University Polyclinics surveyed 228 doctors, nurses and other staff in 2024. [44] Doctors reported more emotional exhaustion than other staff. [44] Staff who exercised less and scored lower on resilience were more exhausted. [44] Staff who felt more recognised by patients showed less depersonalisation. [44]

Why it matters — The survey shows links at one moment, not causes. [44]
26
Two-thirds burnout in Tehran hospitals at a COVID peak

Four hospitals in Tehran, Iran's capital, surveyed 723 health workers during the country's third COVID-19 peak. [45] Using the Copenhagen Burnout Inventory with a cut-off score of 50, 67.41% had moderate to high burnout. [45] Frontline, female and married workers scored highest. [45] Workload, lack of control and poor communication from leaders were all linked. [45]

Why it matters — The Tehran and Yola figures used different questionnaires and cut-offs, so the two cannot simply be ranked. [45][18]
27
A calming app did not measurably cut nurse burnout

A US hospital tested a daily meditation app on its emergency department nurses for four weeks. [46] Twelve nurses signed up and six finished both surveys. [46] The results showed no significant change in burnout or well-being. [46] A second analysis that filled in missing answers did show a drop, which is a weaker kind of evidence. [46]

Why it matters — Six nurses is too few to show whether the app helps. [46] Other surveys today link burnout to hours and days off. [19]
28
Nursing interns in China describe unpaid work

Researchers interviewed 20 final-year nursing interns who had burnout symptoms at three hospitals in western China. [47] The interns described heavy workloads, rigid rules and no voice. [47] They also described unpaid tasks and a lack of respect treated as normal. [47] Their hopes for the job clashed with what they met on the wards. [47]

Why it matters — Nursing already faces chronic staff shortages and heavy workloads after the pandemic. [48] These interns met the same pressures before they had qualified. [47]
29
ADHD pills work on wakefulness, not attention

Scientists at Washington University in St Louis reported in the journal Cell in December that ADHD stimulants act on brain networks for wakefulness and reward. [49] They did not act on the attention networks, as had long been assumed. [49] The team used brain scans of nearly 5,800 children aged 8 to 11, of whom 337 had taken a stimulant that morning. [49] The drugs' effects looked like the effects of more sleep. [49]

Why it matters — Stimulants improved school grades in children with ADHD and in children who were short of sleep. [49] The researchers suggest that getting enough sleep could help in similar ways. [49]
30
15.5 million US adults have an ADHD diagnosis

The CDC, the US public health agency, published data in December on adults with ADHD. [16] In 2023 about 15.5 million US adults had a diagnosis. [16] More than half, 55.9%, were first diagnosed as adults. [16] Visits to health centres by adults with ADHD were highest at ages 18 to 24 and fell steeply with age. [16]

Why it matters — ADHD was long seen as a childhood condition, and adult diagnosis is now a large share of the total. [16] That feeds the debate over whether adult ADHD is being over-diagnosed. [13]
31
Why adult ADHD may be over-diagnosed

A commentary published in September set out why adult ADHD may be over-diagnosed in the US. [13] The current US diagnostic manual raised the latest age for first symptoms from 7 to 12. [13] It also cut the symptoms an adult needs from six to five. [13] Stimulant prescriptions rose 30% from 2018 to 2022, leading to shortages, and US regulators wrote to doctors in 2023. [13]

Why it matters — The authors name rushed assessments, faking, phone distraction and other illnesses that harm attention as further causes. [13] The commentary does not put a number on how many diagnoses are wrong.
32
Stimulant misuse higher where more students are prescribed

A US survey covered 224,469 undergraduates at 395 universities between 2019 and 2022. [50] 9.6% had an ADHD diagnosis, 5.1% had taken medicine for it in the past year, and 2.4% had misused a stimulant in the past three months. [50] For every one-point rise in a campus's share of medicated students, misuse was 7% higher. [50]

Why it matters — More prescriptions put more pills in circulation for students who were never prescribed them. [50] Students with ADHD who were not on medicine were 40% more likely to misuse. [50]
33
In Sweden, drug problems cut ADHD treatment short

Swedish national records identified 9,283 people taking ADHD medicine who went on to have a substance use disorder between 2006 and 2020. [51] They were compared with 46,401 similar people on ADHD medicine without one. [51] Those with substance problems were about twice as likely to stop their ADHD treatment within a year. [51] They were also less likely to restart it. [51]

Why it matters — Guidelines recommend treating ADHD in people with substance problems, yet it remains disputed, and this shows who drops out. [51]
34
ADHD medicine use rising almost everywhere

A review of 31 studies and five national drug registers traced ADHD medicine use since 2000. [52] Use rose steadily in most countries, and the Netherlands was the one place where it recently fell. [52] Boys were medicated more than girls in childhood. [52] Faster rises among women reversed that gap in several adult populations. [52]

Why it matters — The review links the differences between countries to different guidelines and diagnostic rules, not only to different rates of the condition. [52]
35
Half of military clients improved in community PTSD care

A US community mental health network treated 2,717 veterans, service members and their families for PTSD. [53] They received one of three trauma therapies: cognitive processing therapy, EMDR or prolonged exposure. [53] Among those attending at least four sessions, about half of each group cut their symptom score by at least 18 points. [53] All three beat other treatments. [53]

Why it matters — Most evidence came from trials and military hospitals, and this shows the therapies working in ordinary clinics. [53]
36
Packing PTSD therapy into fewer weeks

The US Department of Veterans Affairs piloted massed therapy, three or more sessions a week, in six live-in PTSD programmes. [54] Data from 466 veterans at one site in 2022 and 2023 were compared with usual weekly care. [54] Veterans on the massed schedule had shorter stays and finished treatment more often. [54] Symptom improvement was similar. [54]

Why it matters — High dropout is one of the main problems with PTSD therapy. [55] A schedule that keeps more people to the end without losing effect addresses it directly. [54]
37
Dutch veterans with PTSD and addiction in a 12-week programme

A Dutch pilot combined trauma therapy with dialectical behaviour therapy, which teaches ways to handle intense emotions. [56] It was for veterans whose PTSD came with emotional control or substance problems, groups where standard trauma therapy works less well. [56] Self-reported PTSD symptoms fell sharply over the 12 weeks. [56] Borderline personality symptoms overall did not fall significantly. [56]

Why it matters — It was a pilot with no comparison group, so it shows promise rather than proof. [56]
38
Resilience before PTSD treatment did not change the gain

A Chicago programme treated 727 veterans with two weeks of intensive cognitive processing therapy. [57] Veterans who started with higher resilience had lower PTSD and depression scores throughout. [57] But their scores did not improve any faster or further than those of less resilient veterans. [57]

Why it matters — A low resilience score was not a reason to expect less benefit from the treatment. [57]
39
Trauma therapies still lead, and new ones lack tests

A review published in September summed up PTSD research from 2019 to 2024, since NICE's 2018 treatment guidelines for England. [55] Trauma-focused cognitive behavioural therapy and EMDR still had the strongest evidence and were cost-effective. [55] Dropout from therapy remained high. [55] Much research on new treatments was low quality and rarely compared them with the established ones. [55]

Why it matters — Virtual reality, intensive courses and remote delivery are growing, but mostly as new ways to deliver old therapies. [55]
40
Four caffeinated drinks a day shorten sleep

Researchers used genes linked to caffeine habits in two groups: the UK Biobank, 485,511 people, and the HypnoLaus sleep study, 1,702. [22] This method, Mendelian randomisation, is built to test cause rather than simple association. [22] Four or more caffeinated drinks a day shortened total sleep in every model, by estimates from 11 to 229 minutes. [22] People did not rate their sleep as worse. [22]

Why it matters — Sleep recordings caught a loss that people's own ratings missed. [22]
41
Caffeine pushed blood pressure higher in a sleepless night

In a study, 41 volunteers stayed awake for 38 hours twice, once with caffeine and once with a dummy pill. [58] Staying awake raised blood pressure and heart rate. [58] Caffeine at 9am and 2pm made the rise in blood pressure bigger. [58] It also raised IL-6, a marker of inflammation. [58]

Why it matters — About 16.3% of workers in France work nights and go without sleep again and again. [58] The study tested the body's short-term response, not long-term heart disease.
42
The coffee and depression link may run backwards

A review published in March re-examined 25 years of studies on coffee, caffeine and depression. [59] Studies that follow people over time often find coffee drinkers are less depressed. [59] The review says that may partly reflect cause running the other way, other hidden differences, and relief from caffeine withdrawal. [59] Caffeine can briefly ease tiredness, but it also disrupts sleep and raises anxiety. [59]

Why it matters — A link between who drinks coffee and who is depressed does not show that coffee changes mood. [59]
43
Dreaming of a puzzle helped some people solve it

Researchers at Northwestern University in the US recruited 20 people who often have lucid dreams, where they know they are dreaming. [60] Each tried and failed to solve puzzles, each with its own soundtrack. [60] During dreaming sleep, some soundtracks were played to steer dreams toward those puzzles. [60] People whose dreams did include the cued puzzles solved more of them the next day. [60]

Why it matters — That last result came from an analysis done after the main test, so it needs repeating. [60] It is still a rare experiment that changes dreams on purpose.
44
Nightmares may be a deep-sleep problem too

Nightmare disorder has long been treated as a problem of dreaming, or REM, sleep. [61] A study published this year compared overnight brain recordings of 26 adults with the disorder and 32 without. [61] Those with nightmares showed weaker linking between two brain rhythms in deep, non-dreaming sleep. [61] They also had fewer K-complexes, a type of brain wave seen in deeper sleep. [61]

Why it matters — Nightmares are linked to PTSD and depression. [62] If deep sleep is involved, treatment may need to look beyond dreams, though the study was small and looked back at records. [61]
45
Broken dreaming sleep kept upsetting images upsetting

Seventeen young adults slept in a lab twice. [23] On one night a wrist vibration briefly woke their brains during dreaming sleep. [23] After the broken night, their hearts did not calm down to upsetting images as they normally would overnight. [23] The more the vibration disturbed the brain, the less the calming. [23]

Why it matters — It is rare direct evidence, from a small group, that dreaming sleep helps the body calm down about emotional events. [23]
46
England's blood pressure lines, updated in February

NICE, which writes health guidance for England, updated its high blood pressure guideline in February. [15] A diagnosis starts when clinic readings reach 140/90 and are confirmed by readings at home or across a day. [15] Readings outside the clinic are held to a line 5 points lower, 135/85. [15] Whether people with the mildest stage are offered drugs depends on their age and other risks. [15]

Why it matters — The line decides who is told they have high blood pressure and who is offered drugs. [15]
47
After cancer surgery, other causes now kill more

A US study tracked 3.42 million people with 21 solid cancers from 1992 to 2021, using a cancer registry covering about 28% of the US population. [63] Deaths from the original cancer fell sharply over those decades. [63] In 14 of the 21 cancers, including prostate, breast and bowel, deaths from other causes now outnumber deaths from the cancer. [63] By the 2020s other causes led in those cancers. [63]

Why it matters — Many survivors now live long enough for other illnesses to become the bigger danger. [63] The authors say care after surgery needs to cover those illnesses, not only the cancer. [63]
48
Social support did not clearly extend patients' lives

A meta-analysis pooled 14 non-randomised trials covering 39,493 patients with chronic illness. [64] It asked whether programmes that add social support help them live longer. [64] The pooled results did not reach statistical significance. [64] Studies that followed patients longer did show longer survival. [64]

Why it matters — The authors say benefits may take years to show. [64] The trials were not randomised, so the groups may have differed in other ways.

02 Lesson why it matters

Move the cut-off and more people have the illness

Many illnesses are a line drawn across a smooth scale, so moving the line changes how many people have them while nobody's body changes.

The twist

A lower cut-off or a more sensitive test can raise the number of sick people without anyone's body changing. The extra people it finds are mostly the mildest cases, so treatment helps each of them less.

How it works

Blood pressure, bone density and attention run on a smooth scale with no natural gap
Experts pick a number on that scale and call everything past it an illness
They set it by weighing the harm of missing sick people against the harm of treating people who would have been fine
Lower the number, or measure more often, and more people cross it
The extra people found are mostly milder cases, so each gains less from treatment while side effects stay the same

The same force, elsewhere today

Where this chain is also running, in today's other stories.

Watches that catch short heart-rhythm episodes

Episodes under six minutes carry little stroke risk, yet continuous screening nearly tripled diagnoses and doubled blood-thinner use with no clear fall in strokes.
Why adult ADHD may be over-diagnosed

The US manual cut the symptoms an adult needs from six to five and raised the age limit from 7 to 12, so more adults crossed the line.
The 92-year-old with a 'normal' scan

Her bone scores sat on the safe side of the treatment line, and she still broke a bone in her spine soon after.
Burnout surveys in Tehran and Yola

Each used its own questionnaire and cut-off score, so a share of 67% in one and 70% in the other cannot simply be compared.

Where you've seen this

School exams

moving the pass mark by a few points changes how many students fail, with the same answers on the page

Poverty lines

raising the income line puts more families below it overnight

Storm warnings

lowering the wind speed for a warning means more warnings, most for milder storms

The catch

A line is still needed to decide who gets a scan, a drug or a place in a trial, and for some people a lower line catches real illness early.

And the whole of it

The panel that sets a cut-off, the company that builds a watch and the person who reads an alert each see a different part of the same scale. None of them can tell, from where they stand, which of the people who crossed the line would never have come to harm.

03 Truth what's really going on

What is really going on

Smartwatches and wider diagnostic rules are finding more people with atrial fibrillation and adult ADHD than before. [11][13] The trials so far have not shown that most of the newly found heart patients, especially those with the shortest episodes, end up with fewer strokes for being treated. [11][10]

Why it works on us — An alert about a person's own heartbeat feels urgent and personal, which makes it hard to keep in mind that the same alert, sent to a million wearers, mostly reaches people at low risk.

Who gains

Smartwatch makers such as Apple, Samsung and Withings — Pooled accuracy studies of 89% to 97% for spotting atrial fibrillation support selling heart-rhythm alerts as a health feature. [9]
Makers of blood thinners such as apixaban — Screening that nearly triples diagnoses doubled blood-thinner use, and ARTESIA's result could widen who is offered the drug. [11][10]
Makers of ADHD stimulants — US stimulant prescriptions rose 30% from 2018 to 2022, after diagnostic rules let more adults qualify. [13]
Older hip-fracture patients given denosumab in the Chinese study — 8.8% had a second hip fracture, against 29.7% of similar patients without the drug. [34]
Veterans in US live-in PTSD programmes using massed therapy — Sessions three or more times a week meant shorter stays and more veterans finishing treatment. [54]

Who pays

People put on blood thinners for short device-found episodes — In ARTESIA the drug that cut strokes also caused more bleeding than aspirin. [10]
Doctors and nurses at the teaching hospital in Yola, Nigeria — 70% had burnout while emigrating colleagues left the work to them. [18]
Patients of doctors with burnout — Burnt-out doctors had 2.72 times the odds of reporting a medical error. [17]
US college students on campuses with more ADHD prescriptions — Stimulant misuse was 7% higher for every one-point rise in the share of students on medication. [50]
People over 85 who break a hip — In northern Sweden 47.8% died within a year of the fracture. [21]
People with darker skin using a phone pulse app — FibriCheck spotted atrial fibrillation less reliably in them until a technician checked the reading. [29]

What nobody knows yet

Open questions from across today’s stories — ours included.

01

Whether blood thinners do more good than harm for heart-rhythm episodes lasting between six minutes and a day.

ARTESIA found apixaban cut strokes by 37% but raised bleeding, while NOAH-AFNET 6 was stopped early with too few strokes to judge, and larger trials are still running. [10]
02

How many people in the world have atrial fibrillation.

One review puts it at about 37.5 million. [4] Another puts it at 50 million in 2020. [65] Neither explains the gap, and much of the condition causes no symptoms. [6]
03

How many adult ADHD diagnoses in the US are wrong.

A commentary lists rushed assessments, faking and changed diagnostic rules as causes, but gives no count. [13] More than half of the 15.5 million US adults with a diagnosis got it as adults. [16]
04

Whether stimulants for ADHD are partly treating lack of sleep.

Brain scans showed the drugs act on wakefulness networks and look like extra sleep, but the study was of children aged 8 to 11 and did not test adults. [49]
05

Whether lowering TMAO, the gut chemical, would prevent atrial fibrillation.

The link comes from blood samples and lab cells, not a trial in people. [25]
06

Why burnout rates differ so much between hospitals.

Tehran's 67% used one questionnaire with a cut-off of 50, and Yola's 70% used another, so part of the gap may be the measuring tool. [45][18]
07

Whether denosumab really cut second hip fractures by two-thirds.

The Chinese study looked back at 259 records rather than assigning the drug at random. [34]
08

Whether coffee protects against depression at all.

The link in long-term studies may reflect cause running the other way or relief from caffeine withdrawal. [59]
09

Whether dreaming about a problem helps solve it.

Only people whose dreams actually featured the puzzle solved more, and that result came from an analysis added after the main test in 20 people. [60]

04 Hope carry this

In a US network of community clinics, about half of 2,717 veterans, service members and their families treated for PTSD cut their symptom scores by at least 18 points after four or more sessions.

Also true today

In US veterans' live-in PTSD programmes, packing therapy into three or more sessions a week meant shorter stays and more veterans finishing treatment, with the same improvement in symptoms.
Across 21 solid cancers in US records from 1992 to 2021, deaths from the original cancer after surgery fell so far that in 14 of them, other causes of death now outnumber the cancer.
In a Chinese study of older people who had surgery for a broken hip, 8.8% of those given the bone drug denosumab had a second hip fracture, against 29.7% of similar patients who were not.