A review of 76 trials found that stopping an antidepressant slowly, with talking therapy alongside, kept depression away about as well as staying on it

The rest of the day

36 more stories on this beat.

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

1. 02

   Antipsychotics change the brain they are stopped from

   Antipsychotics are drugs for psychosis, when a person loses touch with what is real. A 2025 review says years on them change the brain: one study found about 30% more receptors for dopamine, a chemical messenger, in people who had taken them. [12] Stopping fast can bring agitation, anxiety and poor sleep, which can be mistaken for the illness coming back. [12] The authors propose cutting 5% to 10% of the latest dose each month. [12]

   Why it matters — The authors say some of the relapse seen after stopping may be caused by the stopping itself. If so, the drugs' protection may have been overstated in trials where they were stopped. [12]

2. 03

   Opioids: smaller cuts as the dose falls

   Opioids are strong painkillers, and guidelines say to stop them when the harm outweighs the help. [6] A 2025 paper, with authors at King's College London, says the guidelines disagree on how. Some cut a fixed amount each time, and others cut smaller amounts as the dose falls. [6] From the chemistry of how opioids attach to their target, it recommends cutting 1% to 10% of the most recent dose every one to two weeks. [6] For some long-term users that takes months or years. [6]

   Why it matters — It is the same rule the new antidepressant guidance uses, worked out for a different drug. The authors say it still has to be tested in randomised trials. [6]

3. 04

   Stopping denosumab needs a plan

   Denosumab is an injection given every six months to slow bone loss in osteoporosis, the thinning of bone that makes it break easily. [26][13] Unlike some other bone drugs, it is not simply stopped when the course ends. [27] Without a follow-up drug, stopping it can lead to rapid bone loss and new broken bones in the spine. [13] A 2026 review says a different bone drug, a bisphosphonate, should be given as a way out. [13]

   Why it matters — The bone loss the drug was holding back comes back quickly once it is gone. Mayo Clinic tells patients to keep taking it on schedule for that reason. [27]

4. 05

   One bone drip, protection years later

   Zoledronic acid is a bone drug of the bisphosphonate family, given as a drip into a vein. [27] In a trial of 1,054 women aged 50 to 60, a single dose did not reduce spine fractures in the first five years. It cut them by 57% in years five to ten. [28] A separate Auckland trial gave 2,000 women over 65 the drug every 18 months. Fractures outside the spine fell by 60% in years four to six, against a 17% fall in years one to three that could be chance. [29]

   Why it matters — Bisphosphonates stay in the bone after years of use, so their effect goes on after the last dose. [27] Most bone drug trials last three years or less, so they may miss the later benefit. [29]

5. 06

   Three months of romosozumab was enough

   Romosozumab is a monthly bone-building injection whose use is limited by its cost, monthly clinic visits and concern about heart risk. [26] In a US trial, 50 women at high risk of fracture got either three months of it followed by nine months of denosumab, or twelve months of romosozumab. Both built about as much hip bone. [26] Hip bone density rose 5.7% against 6.0%, inside the margin the researchers set in advance. [26]

   +5.7%3 months, then denosumab

   vsvs

   +6.0%12 months

   Why it matters — A shorter course costs less and needs fewer clinic visits, and in the US more than half of patients stop the full course early anyway. [26] The trial was small and measured bone density, not fractures. [26]

6. 07

   Bone drug trials hang on a few patients

   Researchers took 27 large trials that showed a bone drug prevented fractures and asked how many patients' results would have to change to lose that finding. [30] The median answer was 9 patients, about half of one percent of a trial. [30] For bisphosphonates it was 8.5 on average. Romosozumab's evidence was the strongest, with a median of 19.5. [30]

   Why it matters — The drugs may still work. The finding is that many of the trials were only just big enough to show that they do. [30]

7. 08

   Fewer US women use hormone therapy

   Mayo Clinic, a large US hospital group, found that hormone therapy use among US women over 40 fell from 4.4% in 2007 to 1.7% in 2023. [20] Even among women aged 50 to 59, who are most likely to benefit, only about 3.5% used it. [20] Up to 75% of women get hot flushes or night sweats at menopause, and they can last years. [20] Use is at its lowest since the early 2000s, when a large trial raised fears about safety. [20]

   4.4%2007

   ↓→

   1.7%2023

   Why it matters — Stephanie Faubion, who led the study, blames misinformation and too little menopause training for doctors. [20] Use was lower among Black, Hispanic and Asian American women than among white women. [20]

8. 09

   In Denmark, users lived slightly longer

   Denmark's national registers followed 876,805 women born between 1950 and 1977 from the age of 45, for a median of 14 years. [21] The 11.9% who took hormone therapy for menopause had a 4% lower risk of dying, once health, income and other differences were accounted for. [21] The lower risk showed after one to ten years of use, not under a year or over ten. [21]

   Why it matters — Before that adjustment the users' death rate was higher, 54.9 against 35.5 per 10,000 years lived, which shows how much the answer depends on the adjustment. [21] Among 703 women whose ovaries were removed at 45 to 54, users had a 27% to 34% lower risk. [21]

9. 10

   Hormone therapy and the age of menopause

   A 2026 study used TriNetX, a network of health records covering about 111 million patients. It matched 83,670 women who started hormone therapy within a year of menopause with similar women who did not. [31] Where menopause came at 60 or younger, users had lower rates of diabetes, heart disease, spine fractures and breast cancer. [31] Where it came after 60, breast cancer was 15% more likely in users. [31]

   Why it matters — It fits the idea that timing matters. It is a records study, and the authors say women who chose the treatment may differ in ways the matching could not see. [31]

10. 11

    In Korea, more cancer with hormone therapy

    A study of South Korea's national health records matched 138,445 women who took hormone therapy for menopause with the same number who did not. It followed them for about eight years. [22] Users had about 10% more cancer overall and 57% more breast cancer. [22] Their risk of cancer of the womb was about 13% lower. [22] Tibolone, a different menopause drug, did not raise overall cancer risk. [22]

    Why it matters — The authors suggest a milder form of progestogen, the second hormone in most combined treatments, might lower the breast cancer risk. [22]

11. 12

    Hormone therapy alongside a weight-loss drug

    Mayo Clinic researchers looked at 120 women past menopause who took tirzepatide, a weight-loss drug approved by the US regulator, for at least a year. [32] Women also taking hormone therapy lost about 35% more weight than women on tirzepatide alone. [32][33] The study was not a randomised trial, so it cannot show that the hormone therapy caused the extra loss. [32]

    Why it matters — Maria Daniela Hurtado Andrade, the senior author, said women on hormone therapy may already have had healthier habits, or slept better once their symptoms eased. [32]

12. 13

    Cancer survivors' hot flushes go untreated

    Many women treated for breast cancer take hormone-blocking drugs that bring on menopause symptoms such as hot flushes. [34] Records from the US, UK and Germany showed a treatment for those symptoms in 32.7%, 20.4% and 8.3% of such women. [34] Antidepressants were the most common first choice. [34] After six months, only 42% in the US, 12% in the UK and 7% in Germany were still taking them. [34]

    US42%

    UK12%

    Germany7%

    Why it matters — Most of these women are offered drugs made for something else, and most stop within months. [34]

13. 14

    Wildfire smoke hits harder than other smog

    A 2026 study compared two kinds of fine dust in the air, from wildfires and from everything else, across 20 US states from 2006 to 2019. [14] For each small rise in wildfire smoke over two years, hospital stays rose by 10% for heart failure and 16% for asthma. [14] The same rise in other air pollution raised them by 4.7% to 8.5%. [14]

    Wildfire smoke, asthma16%

    Wildfire smoke, heart failure10%

    Other pollution, highest8.5%

    Other pollution, lowest4.7%

    Why it matters — The links were stronger in poorer communities, among people with less schooling and among ethnic minorities, so the same smoke lands harder on them. [14]

14. 15

    About 24,100 US deaths a year from smoke

    A 2026 study estimated how long-term smoke from wildfires affected death rates across the continental US from 2006 to 2020. [15] It put the toll at about 24,100 deaths a year and found no safe level: deaths rose steadily with exposure. [15] Deaths from brain and nerve diseases rose most. [15] As a check, deaths from falls and road accidents showed no link, as expected. [15]

    Why it matters — This is a model of whole areas, not a count of death certificates, and it covers years of breathing smoke rather than the smoky days that make the news. [15]

15. 16

    Canada's smoke toll swings with the season

    A 2026 study modelled smoke from wildfires across Canada for 2019 to 2023. [35] Long-term exposure caused between 660 and 5,400 early deaths a year, depending on how bad the fire season was. [35] The yearly cost of that harm ranged from $6.4 billion to $52 billion. [35] More than 80% of people in Canada breathed at least one microgram of smoke per cubic metre, on average, in a typical fire season. [35]

    Why it matters — The deadliest year in the study had about eight times the early deaths of the mildest, so the toll rises and falls with each fire season. [35]

16. 17

    Fire smoke sends children to hospital

    A study linked daily hospital admissions in 1,012 communities across seven countries and territories to smoke from landscape fires, such as forest fires. [16] Each rise in smoke raised children's admissions for lung problems by 1.9%, heart problems by 2.9% and diabetes by 3.7%. [16] Children aged 5 to 9 and those in poorer areas were hit hardest. [16]

    Why it matters — Admissions for lung problems, infections and brain-related illness rose even at low levels of smoke. [16]

17. 18

    Asthma worse in the smoky summer

    In summer 2023, smoke from distant wildfires reached Vermont and upstate New York, a region that rarely saw it. [36] Records from 1,217 visits by young people with asthma showed poorer asthma control that summer than in 2022, on two of three measures. [36] Control was no different from 2024, and local smoke levels within 2023 did not track it. [36]

    Why it matters — The signal is real but patchy, and the authors say it needs more watching as fire seasons get longer. [36]

18. 19

    In Wyoming, the smoke's toll comes a week late

    Researchers counted breathing-related claims to Medicaid, the US health cover for people on low incomes, in 23 Wyoming counties during wildfire seasons from 2012 to 2024. [37] Claims averaged about 1.6 a day during smoke episodes and 7.9 a day in the week after, a five-fold rise. [37] The authors say distance and cost delay care until the smoke has passed. [37]

    Why it matters — Counting only the smoky days would miss most of the harm in rural places far from a doctor. [37]

19. 20

    Night shifts and bad sleep

    UK Biobank is a study of half a million people aged 40 to 69 who joined between 2006 and 2010. [17] Among 285,175 of them who were working, those who always worked nights had 50% higher odds of both trouble sleeping and daytime sleepiness than people who never did shifts. [17] Those who sometimes worked nights had 37% higher odds, and day-shift workers 21%. [17]

    Day shifts only21%

    Nights sometimes37%

    Always nights50%

    Why it matters — The odds rose with every step towards permanent nights, and the link differed by age, ethnicity and smoking. [17]

20. 21

    Night work and thinner bones

    Another UK Biobank analysis followed 276,774 working people for osteoporosis. [18] Every group that did night shifts had more of it than day workers, and those who usually worked nights had 29% higher risk. [18] Among 75,120 people with full job histories, more than ten years of nights carried an estimated 21% higher risk, though that estimate was not statistically certain. [18]

    Why it matters — It adds bone to the body systems linked to night work, alongside sleep. [18][17]

21. 22

    Night shifts and breast cancer: not settled

    A 2026 review pooled 12 studies of night work and breast cancer among health workers, covering 12,132 cases. [38] Women who had ever worked nights showed no clear rise in risk. [38] Twenty years or more carried 25% higher risk, but that result stopped being statistically clear under checks for bias. [38] Six in ten studies were judged at high risk of error in how they measured night work. [38]

    Why it matters — If the link is real, night work would cause a substantial number of breast cancers in health workers, the authors say. [38]

22. 23

    After night work ends, telomeres grow back

    Telomeres are protective caps on the ends of chromosomes that tend to shorten with age. In 262 workers over 50, mostly men, those still on night shifts had shorter telomeres than the rest. [19] Among former night workers, more years of nights meant shorter telomeres. [19] But for each year since they stopped, telomeres were longer and a DNA-based measure of body age was lower. [19]

    Why it matters — The study measured people once rather than following them. It suggests some of what night work does to the body reverses after it stops. [19]

23. 24

    A study built to watch night workers

    EPHOR-NIGHT follows 937 workers aged 20 to 65, nearly all in health care, in Spain, Sweden, Denmark and the Netherlands. [39] They wore sensors for light, movement and heart rate, and gave saliva at five times of day. [39] Permanent night workers slept least and worst. [39] Many already reported anxiety (27%) or depression (18%), and a two-year follow-up was finished in June 2025. [39]

    Why it matters — Most night-work studies rely on people remembering their shifts, which is where a 2026 review found them weakest. [38]

24. 25

    Three more countries free of malaria

    The WHO, the UN's health agency, certified Suriname, Timor-Leste and Georgia free of malaria in 2025, after Cabo Verde and Egypt in 2024. [40] That makes 47 countries and one territory. [40] A 2026 review found the three shared steady political backing, strong tracking of cases, community involvement and steady funding. [41] Suriname worked with indigenous communities and across its borders; Timor-Leste did it after a conflict. [41]

    Why it matters — Malaria still killed an estimated 610,000 people in 2024, most of them young children in Africa, and resistance to the main drug is growing. [40]

25. 26

    Malaria by plane in Libya

    In November 2024 a 63-year-old man in Benghazi, Libya, who had not travelled, died of malaria, and three of his children were infected. [42] Libya was certified malaria-free in 1973. [42] Investigators, writing in February 2026, ruled out local spread and blamed infected mosquitoes arriving at Benina airport, about 450 metres away. [42] Sri Lanka, free for ten years, warns that imported cases can restart the disease. [43]

    Why it matters — Staying free of malaria means watching airports and travellers for decades. Sri Lanka has had no local case since 2012. [43]

26. 27

    Five habits and 13 cancers

    A 2026 study followed 287,745 UK Biobank members for a median of 11.5 years, and 12,947 developed cancer. [44] It scored five habits: smoking, alcohol, diet, exercise and body weight. [44] People with four or five healthy habits had lower risk of 13 of 27 cancer types, including lung, bowel, liver and kidney, than people with none or one. [44] Smoking and excess weight were linked to the most types. [44]

    Why it matters — The list includes less common cancers, such as liver and mouth cancer, whose rates have been rising in the UK. [44]

27. 28

    Obesity judged by illness, not only weight

    BMI, weight measured against height, cannot tell fat from muscle or say whether a person is ill. [23] A Lancet commission proposed judging obesity by body measurements and by whether the extra fat is already causing illness. [23] The WHO's cancer agency tested it on more than 290,000 UK Biobank adults over nearly 14 years. [23] Extra fat with illness carried the highest risk of death, and extra fat without illness was already riskier. [23]

    Why it matters — It could change who is counted as obese, and so who is offered treatment first. [23]

28. 29

    Higher weight causes skin infections

    Researchers used a method that compares people by the genes they were born with, which works like a natural experiment, in UK Biobank and a Finnish study of 500,348 people. [24] Every 5 extra BMI points raised the odds of a bacterial skin infection seen by a family doctor by 37%. [24] The odds of going to hospital with one nearly doubled. [24]

    Why it matters — The same method was less certain about chest infections, which shows it can separate a cause from a coincidence. [24]

29. 30

    Frailty and heart disease

    Frailty means a body that has become weak, slow and easily tired. In 408,163 UK Biobank members without heart disease, followed for 15 years, 30.5% of the frail developed it, against 13.1% of the robust. [45] After adjustment, frailty carried 85% higher risk, and early frailty 24%. [45] The link held at every stage of a scale of heart, kidney and blood-sugar health used by the American Heart Association. [45]

    Why it matters — The authors say frailty picks up risk that the usual heart, kidney and blood-sugar measures miss. [45]

30. 31

    Lung disease and lung cancer

    A UK Biobank analysis compared 107,007 people with a long-term lung disease with 351,876 people without one. [46] Over nearly 4.8 million years of follow-up, 1,322 lung cancers appeared in the first group and 867 in the second. [46] After adjustment, lung disease went with about three times the risk of lung cancer. [46] The raised risk showed in both main types of the cancer. [46]

    Why it matters — It marks out a large group of people at about three times the usual risk of lung cancer. [46]

31. 32

    What goes with new depression, in women and men

    A 2026 study followed 87,648 UK Biobank members for a median of 13.7 years. [25] In that time 5.65% of women and 4.21% of men developed depression, and 6.02% and 3.60% developed anxiety. [25] A lasting tendency to worry and feel low had the largest share: about half of new depression in women and 60% in men was statistically tied to it. [25] Among physical factors, obesity counted most. [25]

    Why it matters — Long-lasting inflammation in the body was linked to both depression and anxiety, in both sexes. [25]

32. 33

    Sweet foods and chronic pancreatitis

    The pancreas makes digestive juices and insulin. A UK Biobank study followed 136,367 people aged 40 to 69 for about 11.6 years. [47] Those who ate the most sweet foods had 87% higher risk of chronic pancreatitis, a lasting inflammation of the pancreas, than those who ate the least. [47] Sweet foods showed no link with pancreatic cancer or with the sudden form of the illness. [47]

    Why it matters — Diet was reported on a questionnaire, so the link is a lead to test rather than a proven cause. [47]

33. 34

    Back pain: doing beat receiving

    Physiotherapy is active when the patient moves and exercises, and passive when treatment is done to them, such as heat, taping or hands-on therapy. [48] A 2025 analysis pooled randomised trials with 2,768 people with long-term low back pain. [48] Active physiotherapy gave the lowest pain and disability scores, better than passive care or a mix of both. [48] The ranges around the estimates were wide. [48]

    Why it matters — Long-term low back pain is one of the main causes of disability worldwide. [49]

34. 35

    Acupuncture for back pain, on weak evidence

    A 2025 review pooled eight randomised trials with 1,123 adults comparing acupuncture with usual care for long-term low back pain. [49] Acupuncture reduced pain and disability both within two weeks and up to six months later. [49] But the certainty of the evidence was rated low to very low, because of bias, mixed results and signs that small negative trials went unpublished. [49]

    Why it matters — The trials compared acupuncture with usual care, such as physiotherapy, advice or medicines, not with fake needles. [49]

35. 36

    A waiting list as the comparison

    A Dutch rehabilitation clinic tested its team treatment for long-term low back pain without a full trial, by comparing 123 treated patients with 88 still on its waiting list. [50] Pain scores fell from about 54 to 35 out of 100, and fear of moving also dropped. [50] When 51 of the waiting patients were treated later, they improved by a similar amount. [50]

    Why it matters — Randomised trials are too costly for most small clinics, and a waiting list gives them a fairer comparison than none. [50]

36. 37

    Therapy by phone app for young people

    Cognitive behavioural therapy teaches people to notice and change unhelpful thoughts and habits. A review found nine randomised trials of it delivered by mobile phone app, with 2,479 people aged 5 to 25. [51] Seven of eight trials that measured depression found it fell more than in comparison groups. [51] Only two of six that measured anxiety found a clear effect. [51]

    Why it matters — Apps can reach young people where therapists are scarce, and the evidence so far is stronger for low mood than for worry. [51]