These children's ears detected sound perfectly and passed none of it on - the missing piece was one protein, and a virus has put it back

The rest of the day

38 more stories on this beat.

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

1. 02

   London's clean-air zone and children's lungs

   More than 3,400 primary school children in London and Luton had their lungs measured every year, from the year before London's ultra-low emission zone began in 2019 to four years after. [36] The zone makes older, more polluting vehicles pay a daily charge to drive in the city. At the start the London children could blow out less air than the Luton children; by the end the two groups matched. The share of London children whose lung capacity counted as clinically impaired fell from 14% to 9%, against 9% to 7% in Luton. [36]

   Why it matters — Chris Griffiths of Queen Mary University of London, a senior author, said he was stunned by how fast the London children caught up. Two independent statisticians said the Covid years, and a possible switch to walking to school, cannot be ruled out as part of the reason. [36]

2. 03

   Pollution slowed lung growth to age 24

   Anna Hansell's team at the University of Leicester followed more than 5,000 people born around Bristol in the 1990s, testing their lungs at eight, at fifteen and at twenty-four, the age when lung function normally peaks. [37] They worked out each person's exposure to fine particles and to nitrogen dioxide, a gas that comes mainly from diesel cars and gas boilers, for every year of early childhood. More pollution went with slower lung growth right into early adulthood. The largest effect fell in adolescence, when lungs grow fastest. [37]

   Why it matters — An earlier study in east London found nine-year-olds' lungs holding 90 to 100 millilitres less air than expected, about the volume of two hen's eggs. Lung function tends to carry forward from childhood, so a smaller start leaves a smaller reserve for later life. [37]

3. 04

   The commonest deafness gene gets a first patient

   Skylark Bio announced on 11 August 2026 that it had given a first dose of a gene therapy to a child with a fault in GJB2. [6] That gene causes up to 30% of deafness present at birth, against otoferlin's 1% to 8%. [7][10] It is also much harder. GJB2 is switched on across supporting cells and the wall of the cochlea rather than in one cell type. Some faults are dominant and some recessive, and putting the gene into hair cells that should not carry it poisons them. [7]

   Why it matters — Joe Burns, formerly research head at Decibel Therapeutics, called otoferlin the Goldilocks case, because that ear stays intact. GJB2 is the test of whether the approach reaches most deaf children or stays with a few thousand. [6]

4. 05

   Six years from mouse to clinic

   A commentary by hearing researchers in Beijing, Gottingen and Paris tracks how fast otoferlin therapy moved: six years from the first proof in mice to children being treated. [9] The gene is large, 48 blocks of code across about 100,000 letters of DNA, and more than 200 faults in it are known to cause deafness. Chinese teams have given more doses than anyone else, 21 patients across two single-arm studies, one enrolling ages 1.5 to 23.9 and the other ages 1 to 11. The best responses came in children aged five to eight. [9]

   Why it matters — An age band where a treatment works best is the kind of detail that decides who gets offered it, and it comes from the country that has treated the most people. [9]

5. 06

   Hearing aids and dementia, in 2,777 Australians

   Researchers at Monash University in Melbourne took an existing seven-year study of older Australians and rebuilt it as if it had been a trial of hearing aids. [21] Everyone included was free of dementia and had moderate hearing trouble but had never used an aid; a median of 664 of them were newly prescribed one. Over seven years the estimated dementia risk was 5.0% among those prescribed an aid against 7.5% among those not. Their overall thinking scores, tested twice a year, were the same either way. [21]

   Why it matters — The gap between the two results is the interesting part: fewer dementia diagnoses, no measurable difference in how people scored on the tests underneath. [21]

6. 07

   The same question across 33 countries

   A pooled analysis of seven long-running studies followed 61,089 people aged 55 and over with hearing loss, across 33 countries and an average of six and a half years. [23] There were 8,911 dementia diagnoses. Hearing aid use went with a 9% lower risk overall. Split by whether people said the aid actually improved their hearing, the benefit sat entirely with those who said it did; those who said it did not showed no reduction at all. The association was strongest in middle-income countries and in people under 70. [23]

   Why it matters — It moves the question from owning an aid to using one that works, which is a different problem for health services to solve. [23][25]

7. 08

   Forty-six studies on hearing aids and thinking

   A team at Peking University pooled 46 studies covering 231,565 older adults with hearing loss to see what hearing aids do to memory and attention. [22] Hearing aid use went with better overall thinking scores and better memory, and with a 16% lower rate of cognitive impairment. Attention and executive function moved slightly the other way. The results differed by age, sex, education and how bad the hearing loss was, which is why the authors call the evidence mixed rather than settled. [22]

   Why it matters — Two of today's three hearing-aid findings point the same way and the third is a set of caveats, which is roughly where this field is. [21][22][23]

8. 09

   The argument that the link is not cause

   An editorial in a hearing research journal sets out the ways the hearing-and-dementia link could be an artefact rather than a cause. [24] A person who cannot hear the tester does worse on a spoken memory test whatever their memory is like. Early dementia may itself make people report hearing trouble. People who buy hearing aids differ from people who do not in income, education and health. The authors call for different study designs and blood markers rather than more of the same cohorts. [24]

   Why it matters — Hearing loss is currently listed as one of the largest changeable risk factors for dementia, so whether it causes dementia decides where prevention money goes. [24][25]

9. 10

   Damage after one loud night

   Forty-two young adults were tested before and after attending large music events, wearing personal sound meters. [13] They took an average of 100 decibels over about ten hours. Only one showed a clinically significant drop on a standard hearing test. Five showed sharp falls in electrical markers of damaged nerve connections within 24 hours, and in two of them the change was still there fourteen days later, with a normal hearing test throughout. How loud the exposure was did not predict who got hurt. [13]

   Why it matters — The connections between the ear's sensor cells and the nerve are the same joins the gene therapy above restores chemically, and once noise destroys them nothing approved brings them back. [13]

10. 11

    Music students, normal tests, different response

    Researchers compared 23 young adults who practise music and are often around loud sound with 21 who are not. [14] All had normal hearing. Thresholds matched across every frequency up to 16 kilohertz, the emissions from the ear's outer cells matched, and the first wave of the brainstem response matched. What differed was the response to the second of two rapid clicks, which was significantly weaker in the exposed group. Word recognition was the same in both groups. [14]

    Why it matters — A paired click asks the ear to fire twice in quick succession, which is what a conversation in a noisy room asks of it. [14]

11. 12

    Monkeys whose hearing came back anyway

    Macaque monkeys were given a noise exposure that causes a temporary drop in hearing, then examined two and ten months later. [15] Unlike rodents, they lost no hair cells, no inner hair cell synapses and no efferent nerve endings. What did change was the ribbons inside the synapses, the structures that hold neurotransmitter ready for release: they were significantly enlarged in both inner and outer hair cells. Damage varied a lot between individual animals, as it does between people. [15]

    Why it matters — Most of what is known about hidden hearing damage comes from mice, and primates resist noise better, so this is a check on how far the mouse results carry. [15][51]

12. 13

    The neighbours that make noise damage worse

    Noise damage has long been blamed on direct harm to hair cells. A study in mice puts the supporting cells next door at the centre of it. [17] Noise causes oxidative stress in hair cells, which switches on a receptor in the supporting cells, which cuts a protein called gasdermin D into its active form. Active gasdermin D in the supporting cells then makes the oxidative injury in the hair cells worse, so the two cell types drive each other in a loop. Blocking that loop is now a drug target. [17]

    Why it matters — It moves the search for a noise-damage drug from protecting the sensor cells to interrupting a conversation between two kinds of cell. [17]

13. 14

    A drug given four weeks late still worked in rats

    After loud noise the connections between hair cells and the nerve can die at once and permanently, while the nerve cell bodies survive for months or years. [18] Researchers used that gap: rats with noise-induced tinnitus were given a combination drug, NHPN-1010, at least four weeks after the injury rather than straight away. The treated rats showed fewer tinnitus-like responses, more of the brain's calming signalling in the cochlear nucleus, larger brainstem responses, and more of the ribbon structures inside the hair cells. [18]

    Why it matters — Almost every protective agent for noise damage has to be given before or immediately after the noise, which is no use to somebody who already has the problem. [18][19]

14. 15

    What decides how much noise hurts

    Rats were exposed to noise under different conditions and tested on days 1, 7 and 21, with the cochlea examined afterwards. [20] Loudness and length of exposure were by far the strongest drivers of hearing loss. Repeating the exposure and widening the band of frequencies also did harm, but less. The team characterised five distinct noise conditions as separate experimental models, each producing a different pattern of cochlear damage. [20]

    Why it matters — Workplace noise rules are written in exactly these two terms, loudness and hours, and this is the evidence underneath them. [19][20]

15. 16

    Moderate noise changed the brain's first relay

    The auditory nerve hands over to the brain at a large synapse called the endbulb of Held. [16] Mice were given noise that is loud but not damaging, and those endbulbs changed: each impulse released less, while the pool of packets ready to be released grew larger. Electron microscopy found no change in the contact area or the number of release sites, and the terminals simply looked darker, meaning more packets were packed inside. The authors read it as the synapse holding its output steady during busy periods. [16]

    Why it matters — It is the same kind of structure as the ear's own synapse, one step further in, adjusting rather than failing. [16]

16. 17

    Two tinnitus treatments beat one, barely

    Four hundred and sixty-one people with long-term tinnitus at five clinics across the European Union were randomly given either one treatment or two, for twelve weeks. [27] The options were talking therapy, hearing aids, app-based counselling and app-based sound therapy, in ten combinations. Distress scores fell 11.7 points on a 100-point scale with one treatment and 14.9 points with two. Talking therapy and hearing aids alone already produced large effects that adding a second treatment did not improve. [27]

    Why it matters — The authors say the gain came from a stronger treatment offsetting a weaker one, not from two treatments adding up, which is an argument against stacking them. [27]

17. 18

    Newcastle's modified notes quieted tinnitus

    Researchers at Newcastle University tested a sound therapy on 77 people, part-funded by the deafness charity RNID. [28] Participants listened for an hour a day for six weeks, took a three-week break, then listened for another six weeks with the sounds swapped, without being told which was which. The active sound was a synthetic musical note altered so that nerve cells tuned to pitches near the tinnitus fired at slightly different moments instead of together. Tinnitus was about 10% quieter for roughly three weeks after treatment stopped. [28]

    Why it matters — Will Sedley, the neurologist who led it, wants the effect built into ordinary listening, so people could accumulate hours without setting time aside. [28]

18. 19

    A tinnitus app against a dummy app

    Sixty people with chronic tinnitus, median age 58, were randomly given either an app carrying education and talking-therapy exercises or a sham app with the therapeutic parts stripped out. [29] Neither the participants nor the assessors knew which was which. After sixteen weeks the real app group had improved by 20.4 more points than the sham group on the 100-point distress scale. Most of that was still there at week 24, eight weeks after the therapeutic functions were switched off. One person in the sham group got worse. [29]

    Why it matters — Talking therapy for tinnitus works and is hard to get, so the question these trials are really testing is whether it survives being turned into software. [27][29]

19. 20

    A tongue-and-sound device outside a trial

    Lenire is a device that plays sound while stimulating the tongue, and it is approved in the United States for tinnitus. [30] A New York clinic reviewed the records of 140 patients it treated between May 2023 and January 2024. Among those whose tinnitus was moderate or worse, 81.8% improved by at least seven points on the distress scale and the average fall was 23.8 points after twelve weeks. Among patients whose tinnitus was slight or mild, the average change was close to zero. [30]

    Why it matters — The device does nothing for people who are only slightly bothered, which is a clearer finding than most trials of it produce. [30]

20. 21

    Four sound therapies, nine months, China

    Four hundred and forty people with long-term tinnitus at three Chinese hospitals were randomly given two hours a day of one of four sounds. [31] The options were plain music, plain music plus a narrow band of noise centred on their tinnitus pitch, music with the high frequencies boosted, and a sound customised to each person. Everyone was followed for nine months. Distress scores fell in all four groups, from a median of 50 to 35, and the customised sound did best. One participant out of 440 reached zero. [31]

    Why it matters — A single complete remission in 440 people is the honest measure of what sound therapy currently does. [31]

21. 22

    What the remedies on the shelf do

    A review searched five databases to March 2025 for randomised trials of oral over-the-counter tinnitus treatments, and found nine covering 390 adults. [32] The agents included melatonin, ginkgo, alpha-lipoic acid, acai extract and mixed antioxidant formulas. Several trials reported improvement within the treated group, but the results were inconsistent and reported too incompletely to pool. Separately, an open-label study gave 170 people ginkgo extract twice daily for 24 weeks and reported improvement on every measure, with no comparison group at all. [32][34]

    Why it matters — Without a control arm, a tinnitus study measures the passage of time and the effect of being looked after, both of which move these scores. [32][34]

22. 23

    Music therapy for tinnitus, reviewed

    A systematic review searched six databases up to January 2024 for studies using music to treat tinnitus, screening 552 titles down to 23 studies published between 2005 and 2022 for pooling. [33] The idea is to mask the phantom sound using rhythm, melody, harmony and tempo. The authors conclude music therapy stimulates both the hearing pathways and the pathways handling attention, memory and emotion. [33]

    Why it matters — Attention and emotion are what the distress questionnaires actually measure, which is part of why so many different sounds all produce a similar-sized improvement. [27][33]

23. 24

    Few clinics have a tinnitus service

    Ninety-six audiologists in Saudi Arabia were surveyed about how they assess and manage tinnitus. [35] Only 14.6% of their workplaces had a specialised tinnitus clinic, and appointments were generally short. Just 32.3% reported working alongside other professions, group therapy was not used at all, and fewer than half involved a patient's family. [35]

    Why it matters — The trials above compare treatments; this counts the places able to deliver any of them. [35]

24. 25

    A non-hormonal pill for hot flushes on the NHS

    Fezolinetant, sold as Veoza, became available on England's health service for hot flushes and night sweats where hormone replacement is not suitable. [46] It is a single daily 45 milligram tablet that blocks a receptor in the brain, called neurokinin 3, on the nerve pathway that triggers the flush. Britain's medicines regulator cleared it in December 2023; the national body that decides what the health service pays for has now backed it, on the Daylight, Skylight 1 and Skylight 2 trials. Around 500,000 women could be offered it. [46][47]

    Why it matters — Helen Knight of that body said options had historically been limited for women who cannot take hormones, most often those with a history of breast cancer or blood clots. [46]

25. 26

    A second non-hormonal drug cleared in the US

    The US regulator approved elinzanetant, sold as Lynkuet, for hot flushes and night sweats in menopause. [48] It blocks two receptors on the same brain pathway rather than one, and its trials showed the flushes becoming both less frequent and less severe. Doctors quoted said it may also help menopause-related sleep problems better than earlier non-hormonal options, though no trial has compared them head to head. [48]

    Why it matters — Two drugs aimed at the same nerve pathway reached two different countries' health systems within a year, after decades in which hormones were nearly the only answer. [46][48]

26. 27

    Which pollutant does the damage in Fresno

    Air pollution studies usually cannot say which pollutant is responsible, because they rise and fall together. [38] A study in Fresno, California, recruited 299 children and tested their lungs at about seven and again at about nine, measuring residential exposure to eight pollutants over windows from one week to two years. Seven of the eight moved together; ozone moved the opposite way. Coarse particles, the bigger PM10 dust, came out as the pollutant most strongly tied to weaker lungs, and the three-month window mattered most. [38]

    Why it matters — Fresno's air already breaches the regulatory limits, and naming one pollutant is what turns a health finding into a rule someone can write. [38]

27. 28

    The weeks of pregnancy that mattered

    Researchers followed 429 mother and child pairs, estimating daily pollution and temperature at the mother's address through pregnancy from satellite models, then testing the children's lungs between eight and fourteen. [39] Fine particles in roughly the first half of pregnancy, weeks 1 to 21, went with lower lung volumes years later. Nitrogen dioxide showed the same pattern over weeks 1 to 16. Girls were affected more than boys. Warm and cold spells showed no combined effect with pollution. [39]

    Why it matters — It puts the damage in specific weeks rather than across a pregnancy, which is what a warning to an expectant mother would have to be built on. [39]

28. 29

    Cleaner cookstoves cut symptoms, not pneumonia

    A review pooled 18 studies from ten sub-Saharan African countries on smoke from cooking fires inside homes. [40] Carbon monoxide exposure tracked respiratory symptoms, nitrogen dioxide tracked tuberculosis of the lung, and both sizes of particle tracked acute respiratory infections. Exposure before and after birth both raised the risk of pneumonia and lowered lung function. Improved cookstoves cut general respiratory symptoms by about a fifth, and had little effect on severe outcomes such as pneumonia. The authors rate the overall evidence as moderate. [40]

    Why it matters — Around 2.1 billion people still cook on wood, crop waste, charcoal, coal or dung, and in a poorly ventilated home indoor smoke can carry a hundred times more fine particles than is considered safe. [45]

29. 30

    Hospital admissions after polluted days

    Daily hospital admissions for schoolchildren aged 7 to 12 in urban Peninsular Malaysia from 2018 to 2020 were matched against readings from five air monitoring stations. [42] At Seberang Jaya, a 10 microgram rise in coarse particles raised admissions the same day, sulphur dioxide raised them six days later, and benzene three days later. At Shah Alam, toluene and both particle sizes acted the same day. Adding up the lag effects, coarse particles nearly doubled the relative risk at one station. [42]

    Why it matters — The delays differ by pollutant, so a study that only looks at the day itself will miss what sulphur dioxide is doing. [41][42]

30. 31

    Northern Thailand's haze season, measured

    Chiang Mai in northern Thailand has an annual haze season when fine particles exceed the standards. [43] In 2024, researchers ran breathing tests on primary school children aged 8 to 12 near a monitoring station. Among 93 children with usable tests, 52% showed a restrictive pattern, where the lungs hold less air than they should, 18% showed an obstructive pattern, and 30% were normal. Girls were far more likely than boys to show any abnormality. [43]

    Why it matters — Asthma narrows the airways, and these children's airways were normal while their lungs simply held less air. That is what months of haze do, not what a single attack does. [41][43]

31. 32

    Brain scans of Polish children

    Researchers scanned 425 children aged 10 to 13 in Poland, 116 of them diagnosed with attention deficit hyperactivity disorder, and matched the scans to their pollution exposure before birth and in early life. [44] Nitrogen dioxide exposure went with lower structural measures across the brain's white matter, the wiring between regions. Using a newer method that separates individual fibre bundles inside each voxel, they found nothing, despite the large sample. They also found no sign that children with ADHD were more vulnerable than the rest. [44]

    Why it matters — The two null results are the useful part: a global signal that survives and a fine-grained one that does not. [44]

32. 33

    The US regulator moves to one trial

    Marty Makary, the commissioner of the US Food and Drug Administration, told a medical news site the agency will begin requiring one clinical study rather than the historical two before considering a product for approval. [49] Two trials still stand in some cases, but one becomes the default. Makary's argument is that a single well-designed and controlled trial can reach the same statistical strength as two. Many drug makers already submit one pivotal trial. [49]

    Why it matters — The deafness therapy above was approved on a single-arm study of twelve children with no comparison group, which is what the far end of this looks like. [2][49]

33. 34

    How often drug trials actually succeed

    A team built a method for tracking the share of drugs that make it through clinical trials as it changes over time, and published it as a live site. [50] The success rate has been falling since the early 2000s, has recently flattened and has started to rise again. Repurposed drugs, old medicines aimed at new diseases, succeed less often than drugs overall, which is the opposite of what the field assumes. Anti-COVID drugs had an extremely low rate. [50]

    Why it matters — How much a single trial is worth depends on how often trials are wrong, and that number moves. [49][50]

34. 35

    A deafness mutation traced to one community

    Over 200 different faults in the otoferlin gene are known to cause hearing loss. [10] An Israeli study across several centres found a new one, a splicing fault that makes the cell skip part of the gene, in seven Moroccan Jewish families. It appears in no global database and in no other Jewish community, and about 1.5% of people of that ancestry carry it. The authors confirmed the splicing effect in the laboratory and name it as a candidate for gene therapy. [10]

    Why it matters — A therapy that fixes one gene needs to know which families carry which fault, and that is a mapping job community by community. [10]

35. 36

    A second deafness gene, rescued in mice

    MYO7A was the first gene ever found to cause inherited deafness in people, in 1995, and it accounts for about 1.4% of deafness cases in Japan. [11] It makes a motor protein inside the bristles on top of the ear's sensor cells. Faults in it cause Usher syndrome type 1B: profound deafness, poor balance, and sight that fades later. A team has now rescued hearing in mice born deaf from a MYO7A fault, the first time that has worked. The gene is too big for one virus shell, so they split it across two. [11]

    Why it matters — The hearing gain in the mice was modest, and an earlier attempt on the same mouse strain produced none at all. Each deafness gene brings its own obstacle, and for this one it is size. [8][11]

36. 37

    Stem cells for deafness are not close

    A review assesses whether stem cells could regrow the damaged parts of the inner ear. [12] Embryonic stem cells, reprogrammed adult cells and cells drawn from bone marrow have all been shown in the laboratory to turn into the cell types the ear needs. Getting them to survive, sit in the right place in the coiled cochlea and wire into the nerve is where it stops. Immune rejection, poor cell survival and regulatory questions all remain unsolved. The authors are more optimistic about reprogrammed cells combined with gene therapy. [12]

    Why it matters — Gene therapy works because the cells are still there; replacing cells that have died is the harder problem and the one that would reach far more people. [8][12]

37. 38

    Ageing and noise, followed for two years

    Mice were given noise that immediately destroyed half of the connections between inner hair cells and the hearing nerve, then kept for one month, six months, one year or two years. [51] The 50% loss did not recover at any point, with a little more lost by two years. The nerve fibres themselves took far longer to die back than the synapses did, and at six months the loss was not detectable at all. Ageing killed hair cells at much the same rate whether or not the animals had been exposed. [51]

    Why it matters — The long delay before the nerve fibres go is the window any future repair treatment would have to work inside. [18][51]

38. 39

    Budgerigars and the start of a sound

    Damage to the connections between hair cells and the nerve cannot be seen on a standard hearing test, and may explain why some people cannot follow speech in a noisy room. [52] The evidence for that has been mixed. Researchers induced the damage chemically in trained budgerigars, parakeets that are good at telling complex sounds apart, then asked them to hear tones buried in noise. Long tones came through normally. So did short tones that started after the noise. Short tones beginning at the same instant as the noise were the ones the birds missed. [52]

    Why it matters — It narrows a vague complaint into one testable ability: hearing where a sound starts, which is how speech carries most of its consonants. [13][52]