Author: Ethan Miller

Once upon a time there was a guy called Ethan, and he loved to write.
  • NHS Mental Health Waiting Times: What the Statistics Actually Hide

    NHS Mental Health Waiting Times: What the Statistics Actually Hide

    NHS mental health waiting times are published regularly, reported in Parliament, and cited in policy documents as evidence that the system is, or is not, performing. The numbers look precise. They are not. Behind the headline figures lies a methodological tangle that obscures how long people genuinely wait, how many give up before being seen, and whether those who do reach treatment receive something that actually helps.

    I’ve spent time reading through NHS England’s own data releases alongside independent analysis from the Kings Fund and the Mental Health Foundation, and what becomes clear is that the official statistics, while not fabricated, are structured in a way that consistently flatters the picture. That matters enormously when mental health provision is already under pressure and when the people most affected rarely have the energy to interrogate the numbers on their behalf.

    Person sitting alone in a hospital corridor, reflecting the reality behind NHS mental health waiting times
    Photo by Yiğit KARAALİOĞLU on Pexels

    How NHS Talking Therapies waiting times are measured

    NHS Talking Therapies, formerly known as IAPT (Improving Access to Psychological Therapies), is the largest talking therapy service in the world by volume. The government’s stated standard is that 75% of people should begin treatment within six weeks of referral. The NHS publishes monthly data showing compliance with this target, and in most recent reporting periods that target appears to be met or closely approached nationally.

    The problem is what “referral” and “treatment” mean in this context. The waiting time clock starts when a referral is received by the service, but it only starts ticking towards the headline figure once the referral has been accepted. Referrals that are rejected or redirected, often because the person’s needs are assessed as too complex for IAPT-level support, disappear from the waiting time data entirely. People who self-refer and then wait weeks for even an initial triage call are frequently not counted in the headline metric either, depending on how local trusts log that first contact.

    NHS England’s own technical guidance acknowledges these definitional issues. What it does not do is prominently flag them in the summary statistics that end up in ministerial statements and press releases.

    CAMHS waiting times and the missing data problem

    Child and Adolescent Mental Health Services present an even murkier statistical picture. Unlike NHS Talking Therapies, CAMHS does not operate under a single nationally mandated waiting time standard. The NHS Long Term Plan set ambitions rather than enforceable targets, which means reporting is inconsistent across integrated care systems. Some trusts measure from GP referral; others from the point of specialist triage. Some include community eating disorder services; others do not.

    The result, as the Centre for Mental Health has consistently noted, is that a child waiting 22 weeks in one part of England might be recorded differently from a child with an identical wait in another. Comparing CAMHS waiting time data across regions is therefore largely meaningless without drilling into the methodology of each reporting trust, something no anxious parent with a struggling teenager has the time or technical knowledge to do.

    What CAMHS data also tends to obscure is the number of children and young people who are referred and then assessed as not meeting the threshold for service. This is not a small group. Referral rejection rates vary but can be substantial. Those young people do not vanish, they wait again, often returning to their GP and cycling through the system multiple times before reaching support. That cycle is invisible in the headline figures.

    NHS mental health waiting times data on a clipboard showing statistics and figures
    Photo by RDNE Stock project on Pexels

    Dropout rates: the statistic that rarely makes the headlines

    Even when people do enter treatment, a significant proportion leave before completing their course. NHS Talking Therapies data includes a “completed treatment” category, but the gap between people who start treatment and those who are eventually coded as completing it tells its own story. Roughly a quarter of people who begin a course of therapy do not finish it, according to figures from NHS England’s published datasets.

    Dropout is not a neutral event. It can represent recovery, certainly. But it can also represent someone who found the waiting so demoralising that motivation collapsed, someone who was offered a form of therapy that did not suit their presentation, or someone whose life circumstances, childcare, work, transport, made attending sessions impossible. The data does not distinguish between these. A person coded as “dropped out” is simply removed from the recovery rate calculation, which means the headline recovery statistics are drawn from the subset of people who both completed treatment and were assessed at the end. That is not a representative sample.

    This connects to a broader issue with how NHS mental health waiting times interact with the growing tendency for people to pursue DIY health solutions while waiting. When the system feels opaque and inaccessible, people turn elsewhere, sometimes helpfully, sometimes not.

    What the figures conceal about severity and need

    NHS Talking Therapies is designed for mild-to-moderate conditions: anxiety, depression, phobias, mild OCD. It is not designed for psychosis, personality disorders, complex trauma, or severe and enduring mental illness. Yet because IAPT is the most visible and most numerically reported part of the mental health system, its statistics dominate public perception of how the NHS is performing on mental health.

    People with more serious needs are largely invisible in published waiting time data. Secondary care psychiatric services, community mental health teams, inpatient provision, these operate under different reporting frameworks, with far less transparency. The NHS dashboard that looks relatively orderly at the IAPT level tells you almost nothing about what happens to someone who presents to their GP with symptoms that are beyond mild-to-moderate. That person may wait considerably longer, with less visibility and less accountability built into the process.

    I’d argue this is the most significant gap in the public picture. The presentable statistics we are shown mostly describe the part of the system that performs best. The parts that struggle most are also the parts measured least.

    Why the data is built this way

    It would be too simple to say this is deliberate obfuscation. The architecture of NHS mental health waiting time data reflects how the system developed: IAPT was built with reporting infrastructure from the start, precisely because it was a new national programme that needed to demonstrate value. Older services were not built that way and retrofitting consistent national data collection onto a fragmented set of legacy services is genuinely hard.

    That said, the effect of this patchwork is that ministers, commissioners, and the public receive a partial and systematically optimistic view of performance. The parallel boom in mental health apps and digital tools has partly grown in the space this leaves, people who cannot access or do not trust NHS provision looking for alternatives, many of which carry their own evidence problems.

    Reading the numbers more honestly

    If you want a more honest picture of NHS mental health waiting times, the place to start is not the dashboard headline. Look at the number of people referred but not treated, not just those waiting for treatment. Look at the proportion of completed treatment courses versus starts. Look at how your local integrated care system reports CAMHS data and what definition of “waiting time” it uses.

    For researchers and journalists, NHS England’s Mental Health Services Dataset (MHSDS) contains more granular information than the headline publications, though navigating it requires some technical patience. The Health Foundation and the Kings Fund publish accessible analyses that do much of the interpretive work.

    Data integrity in health systems matters beyond mental health, of course. Questions about how clinical information is stored, processed, and accessed are becoming increasingly pressing as more services move to cloud and AI-assisted tools, and providers like dijitul.ai are part of a growing conversation about where patient-sensitive data actually lives and who can reach it.

    The point, on NHS mental health waiting times, is not that the numbers are lies. It is that they are carefully scoped truths, and the scope excludes a lot. Until the measurement framework is reformed to include rejected referrals, dropout context, and consistent CAMHS reporting, the statistics will keep telling a story that is tidier than the reality experienced by the people waiting.

    Frequently Asked Questions

    What is the official NHS waiting time target for mental health treatment?

    For NHS Talking Therapies (formerly IAPT), the target is that 75% of patients should begin treatment within six weeks of referral, with 95% seen within 18 weeks. No equivalent enforceable national target exists for CAMHS, which is one reason CAMHS waiting time data is so inconsistent across England.

    Why do NHS mental health waiting time statistics look better than patient experience suggests?

    The main reason is how the clock is measured. Waiting times are typically counted from when a referral is accepted, not when it is received, and rejected or redirected referrals drop out of the data entirely. This means people who are turned away from services, sometimes repeatedly, are not reflected in the headline figures.

    How long are children typically waiting for CAMHS in England?

    Published CAMHS waiting times vary substantially by region and by how individual trusts define the start of the wait. NHS England’s Mental Health Services Dataset shows some children waiting over a year in certain areas. Because reporting definitions differ between trusts, direct comparisons are unreliable without examining local methodology.

    What happens to people who drop out of NHS Talking Therapies before completing treatment?

    People who do not complete their therapy course are typically excluded from the recovery rate calculations published by NHS England, meaning the headline recovery statistics only describe those who completed treatment and were assessed at the end. Dropout rates of around 25% are not uncommon nationally, but the reasons behind them are rarely broken down in published data.

  • Why the UK’s Rollout of Genomic Medicine Is Moving Slower Than the Headlines Suggest

    Why the UK’s Rollout of Genomic Medicine Is Moving Slower Than the Headlines Suggest

    The government’s pitch for genomic medicine is genuinely compelling. Sequence your DNA, understand your disease risk, target your cancer treatment precisely, and spare you years of misdiagnosis. The NHS Genomic Medicine Service (GMS), launched in 2018 and formally expanded through the 2020s, is supposed to be the engine that delivers that future to every patient in England. On paper, the UK leads the world. In practice, the picture is considerably messier, and I think most patients would be surprised by how far day-to-day clinical reality lags behind the announcements.

    Scientist examining NHS genomic medicine UK sequencing data on a laboratory screen
    Photo by Tima Miroshnichenko on Pexels

    This is not a story about failure. The science is real, the ambition is right, and there are places where genomic medicine is genuinely changing lives today. But the gap between government rhetoric and what most patients actually encounter is wide enough to deserve honest examination, and that gap has consequences.

    What the NHS Genomic Medicine Service actually is

    The GMS operates through seven Genomic Laboratory Hubs across England, each covering a large geographic region. These hubs process genetic tests ordered by clinicians, link into a shared NHS Genomic Medicine Service dataset, and are meant to be the infrastructure through which whole genome sequencing eventually becomes routine. Genomics England, a company owned by the Department of Health, runs the 100,000 Genomes Project data and feeds into this infrastructure. Together, these bodies represent a genuinely significant investment: the government has committed hundreds of millions of pounds to the programme since its inception.

    The GMS focuses primarily on rare diseases and cancer genomics. For a child with an undiagnosed rare condition, a whole genome sequence can end a diagnostic odyssey that might otherwise take a decade. For a patient with certain cancers, tumour profiling can identify whether a targeted therapy will work before expensive and gruelling treatment begins. These are real, meaningful applications, not theoretical ones. According to Genomics England, whole genome sequencing has now provided a diagnosis for around 25 per cent of previously undiagnosed rare disease patients who go through the programme. That is not nothing. For those families, it is everything.

    Where the access gaps are widest

    The problem starts with referral. A whole genome sequence only helps if a clinician thinks to order one, knows how to interpret the result, and has access to a genetics specialist who can act on it. Across much of England, particularly in rural areas and in the Midlands and North, that chain is broken at multiple points.

    The NHS has around 3,500 registered clinical geneticists and genetic counsellors in total, according to NHS workforce figures. For a population of 56 million in England alone, that number is inadequate. Most GPs have had minimal genomics training. A 2023 Health Education England review found that fewer than a third of NHS trusts felt their non-specialist clinicians had sufficient understanding to order or act on genomic tests appropriately. That figure has not materially improved since. What this means in practice is that access to NHS genomic medicine in the UK often depends on whether you happen to be treated by a clinician who knows enough to refer you, or who works somewhere with a strong genetics team nearby.

    Geography compounds the problem. The seven Genomic Laboratory Hubs are not evenly spread. Patients in London or Manchester are significantly better served by specialist genetics outpatient services than patients in Cornwall, Lincolnshire or large parts of Wales, where the NHS in Wales has its own separate genomics programme that lacks equivalent infrastructure investment. Health inequality, which runs through almost every part of the NHS, runs through genomics too.

    Ethnicity matters here as well. Genomic reference databases have historically been built predominantly from people of European ancestry. This is a global problem, not unique to the UK, but it means that variant interpretation is genuinely less accurate for patients from South Asian, Black African or other non-European backgrounds. A genetic variant that looks unusual against a European reference population may be entirely normal in a Bangladeshi or Nigerian context, and vice versa. The NHS has made some effort to diversify its datasets, but the 100,000 Genomes Project cohort was roughly 78 per cent white British when analysed. Clinical decisions made from skewed data carry real risk for the patients those datasets underrepresent.

    The data infrastructure problem nobody talks about

    Whole genome sequencing generates enormous quantities of data. A single genome is around 200 gigabytes of raw reads. The NHS Genomic Medicine Service is sequencing tens of thousands of patients per year and that number is rising. Storing, linking, and meaningfully interrogating that data requires infrastructure that the NHS has not historically been designed to support.

    NHS trusts still run a mixture of legacy electronic record systems. Getting genomic data to sit alongside clinical notes, pharmacy records and imaging in a way that a clinician can actually use during a consultation is technically and organisationally difficult. Interoperability between NHS systems is a longstanding problem, and genomics is merely the latest discipline to discover this the hard way. There are also legitimate data governance questions. The public’s trust in NHS data-sharing has been dented by past schemes, and the ICO’s oversight of health data remains a live policy issue, as I’ve written about elsewhere on this blog when covering how health app data privacy works under UK law. Patients are right to ask how their genomic data is stored, who can access it, and whether secondary research use requires their explicit consent.

    The clinician training gap

    Even where the infrastructure exists, many clinicians lack confidence with genomic results. Interpreting a variant of uncertain significance, explaining penetrance to a patient, or deciding whether a finding warrants cascade testing across a family are tasks that require specific training most doctors have not received. Medical schools are only beginning to embed genomics into undergraduate curricula in any meaningful way.

    This matters because NHS genomic medicine cannot scale if it relies entirely on a small group of specialist clinical geneticists. The model that will actually work at population level is one where oncologists, cardiologists, neurologists and paediatricians all have enough genomic literacy to order, interpret and act on tests within their specialty. Building that workforce takes years. The government’s Genome UK strategy, published in 2020, acknowledged this gap and committed to addressing it. The pace of change in medical education, though, has been slow.

    I find it useful to compare this with the self-diagnosis trap that many patients fall into while waiting for specialist input, which I’ve discussed in the context of NHS waiting lists and DIY health decisions. Genomics creates a version of the same pressure. Patients who can afford direct-to-consumer genetic tests from companies like Dante Labs or Genomics PLC are already getting partial information without clinical interpretation. That is not always safe, and it does not replace a conversation with a genetic counsellor.

    What patients can actually expect right now

    If you or your child has a rare undiagnosed condition, asking your GP for a referral to a clinical genetics service is the right first step. The GMS genuinely does deliver for this group, even if waiting times vary by region. For cancer, whether tumour profiling applies to your specific diagnosis depends heavily on cancer type and the capabilities of your treating trust. Haematological cancers and some solid tumours have well-established genomic pathways. Others do not yet.

    For anyone hoping that mainstream preventive genomic medicine, knowing your polygenic risk scores for heart disease, diabetes or dementia before symptoms appear, is an NHS offering today, the honest answer is that it is not. Some pilot studies are running, including work through the NIHR and NHS England’s FH (familial hypercholesterolaemia) programme, but routine preventive genomics on the NHS remains years away for most people. The science is developing faster than the health system’s capacity to deploy it equitably.

    NHS genomic medicine in the UK is genuinely world-leading in places. The ambition behind the GMS is sound. But ambition and delivery are different things, and the patients who most need genomic medicine, those with rare diseases, with harder-to-treat cancers, with family histories that warrant investigation, are often the same patients who face the longest waits, the least-informed clinicians, and the most uneven postcode lottery. Closing that gap requires honest assessment of where the system is actually falling short, not just more press releases about sequencing milestones.

    Frequently Asked Questions

    How do I get a genomic test on the NHS?

    You need a referral from a specialist, typically through a clinical genetics service, an oncologist, or in some cases a paediatrician. GPs can refer you to clinical genetics if there is a relevant family history or suspected rare condition. Not all genomic tests are available through every NHS trust, so access depends partly on where you live and what condition is being investigated.

    What conditions does the NHS Genomic Medicine Service cover?

    The GMS focuses primarily on rare and inherited diseases and cancer genomics, including whole genome sequencing for patients with undiagnosed rare conditions and tumour profiling for certain cancers. It does not currently offer routine preventive or lifestyle genomic testing. The programme is expanding, but coverage varies by region and clinical indication.

    Is NHS genomic data kept private?

    Genomic data collected through NHS programmes is held under NHS data governance rules and subject to UK GDPR, overseen by the ICO. Genomics England operates under a specific data access framework where researchers must apply for access and agree to strict conditions. Patients can request information about how their data is used, and in most cases can opt out of secondary research use.

  • Britain’s Sleep Debt Is Getting Worse, and Sleep Hygiene Advice Is Not Fixing It

    Britain’s Sleep Debt Is Getting Worse, and Sleep Hygiene Advice Is Not Fixing It

    Sleep deprivation in UK adults is not a new story, but the data keeps getting worse. The Sleep Council’s Great British Bedtime Report has tracked a steady deterioration in sleep quality across the country for years, and occupational health researchers are increasingly blunt about why: the standard public health messaging around wind-down routines, limiting caffeine after 2pm, and putting your phone face-down at 9pm is not touching the populations most affected. It was never going to.

    I am not dismissing sleep hygiene entirely. For someone whose main obstacle to sleep is an overactive mind or an ingrained habit of scrolling, it has real value. But when the underlying cause is a 4am warehouse shift, a flat above a main road, or the kind of low-grade financial anxiety that does not switch off at bedtime, advice about lavender pillow spray is almost insulting. The problem is structural. The fixes need to match.

    Person lying awake at night illustrating sleep deprivation in UK adults
    Photo by SHVETS production on Pexels

    What the survey data actually shows

    The Office for National Statistics estimated that poor sleep costs the UK economy around £40 billion a year in lost productivity, placing Britain among the worst-performing developed nations for sleep duration. Around a third of UK adults report regularly getting fewer than six hours a night, well below the seven-to-nine hour range most sleep researchers consider the minimum for healthy adult function. The NHS itself acknowledges that one in three people in the UK suffers from poor sleep.

    What those headline figures obscure is the distribution. Sleep deprivation in UK adults is not evenly spread. It clusters around specific populations: shift workers, those in insecure or low-paid employment, people in overcrowded housing, and those with long daily commutes. A 2023 analysis published in the journal Occupational and Environmental Medicine found that workers on rotating shifts had sleep durations averaging 90 minutes shorter than those on standard day schedules. That is not a gap you close with a consistent bedtime.

    Shift work is the single biggest overlooked factor

    Roughly 3.5 million people in the UK work shifts, many of them in logistics, healthcare, retail and manufacturing. For these workers, the circadian disruption is chronic and cumulative. The body’s sleep-wake cycle is governed by light exposure and mealtimes; rotating shifts repeatedly contradict both. I have written before about the metabolic consequences of shift work, which extend well beyond tiredness into increased risks of type 2 diabetes and cardiovascular disease. Sleep is the mechanism through which most of that damage is mediated.

    The difficulty is that most sleep hygiene advice presupposes a fixed schedule. “Go to bed at the same time every night” is biologically sensible but completely irrelevant to someone whose shifts rotate weekly. The more useful interventions for this group are tightly targeted: strategic light therapy to reset the circadian clock before a phase shift, melatonin timed to the new sleep window rather than taken out of habit, and employers scheduling shifts in a “forward rotation” pattern (moving from morning to afternoon to night rather than the reverse), which research suggests the body adapts to more easily. Very few UK employers actually implement the last of these.

    Financial stress and sleep: a relationship the NHS rarely addresses

    Separate from shift work, financial anxiety is one of the most consistent predictors of poor sleep quality in population studies. A 2024 survey by the Money and Mental Health Policy Institute found that 86% of people with problem debt reported that their mental health suffered as a result, with sleep disruption among the most commonly cited symptoms. The mechanism is straightforward: cortisol, the stress hormone, suppresses melatonin production. Chronic financial stress means chronically elevated cortisol at night.

    This is not something a screen curfew addresses. What does move the needle, according to a body of cognitive behavioural therapy research, is structured worry postponement: a technique where people set aside a specific 20-minute “worry window” earlier in the evening to write down and problem-solve financial concerns, actively refusing to engage with those thoughts at bedtime. It is a component of CBT for insomnia (CBT-I), which the NHS recommends ahead of sleep medication but which few GPs have the time or training to deliver properly. The NHS’s own Sleepio programme offers a digital version, though access remains patchy across different integrated care boards.

    Housing quality and the noise problem

    Poor housing is another cause of sleep deprivation in UK adults that receives almost no attention in public health campaigns. Around 8.4 million people in England alone live in non-decent homes, according to the English Housing Survey, and thermal discomfort, damp, noise bleed from neighbouring flats, and proximity to traffic are all independent predictors of disturbed sleep. A terraced house or purpose-built flat in a high-density urban area presents sleep challenges that no amount of white noise apps can fully compensate for.

    There is decent evidence for sound-masking through broadband noise (white or pink noise, not music) when environmental noise is intermittent rather than continuous. Blackout curtains make a genuine difference for those near streetlights. But these are coping strategies, not solutions. The solution is housing stock that meets basic acoustic and thermal standards, which in the UK remains an ongoing policy failure rather than an individual lifestyle choice.

    Long commutes and the hidden sleep tax

    The average UK commuter spends around 59 minutes travelling each day, according to the ONS. For those travelling into London or other major cities, that figure is often much higher. Every additional hour of commuting time comes directly out of either sleep duration or wind-down time at the end of the day. Research from the University of the West of England found that commutes over 45 minutes each way were associated with significantly higher rates of stress, poor sleep and reduced physical activity.

    Remote and hybrid working has helped some workers reclaim that time, but the gains have been unequal. Key workers, those in construction, care, retail, and hospitality, have seen no such flexibility. They are also disproportionately likely to be shift workers or living in lower-quality housing. Sleep deprivation in UK adults concentrates in exactly the populations least able to implement the individual-level advice they are given.

    What actually moves the needle

    CBT-I is the most robustly evidenced intervention for chronic insomnia, with effect sizes consistently larger than sleep medication and without the dependency risks. The National Institute for Health and Care Excellence recommends it as the first-line treatment. The problem is access: most people cannot get it through the NHS in a reasonable timeframe, and private providers charge £100 or more per session. Digital CBT-I programmes are a partial answer, and some integrated care boards are expanding access, but uptake remains low.

    For shift workers specifically, the evidence supports timed melatonin (0.5mg to 3mg taken 30 minutes before the intended new sleep time), forward-rotating shift schedules, and access to blackout sleeping environments. Employers have a legal duty of care under the Health and Safety at Work Act 1974 that extends to fatigue management, yet enforcement is inconsistent.

    At a population level, the levers are rent regulation, improved housing standards, flexible working rights, and transport investment. None of these show up in a sleep hygiene leaflet. My honest read of the evidence is that until public health policy treats sleep deprivation as an occupational and housing problem as much as a behavioural one, we will keep telling people to put their phones down while the actual causes go unaddressed. That is not hopeless, but it does require being honest about where the problem actually sits. The science on conditions like ME/CFS shows clearly what happens when chronic sleep disruption and fatigue are left untreated for years, and the NHS waiting list reality means most people experiencing serious sleep problems are not getting timely support anyway. The gap between need and provision is real, and pretending that individual habit change fills it does nobody any favours.

  • Data Brokers, Health Apps and Your Medical Privacy: What UK Law Says and What It Fails to Prevent

    Data Brokers, Health Apps and Your Medical Privacy: What UK Law Says and What It Fails to Prevent

    Most people assume that because a health app is on their phone, and because they vaguely clicked “I agree” at some point, their data is reasonably safe. It is not. Health app data privacy in the UK sits in a genuinely uncomfortable gap between legislation that sounds robust and enforcement that has, so far, been remarkably light. I’ve spent time digging through ICO decisions, GDPR guidance and academic research on this, and the picture is not reassuring.

    Woman reviewing health app data privacy settings on her smartphone
    Photo by Lisa Fotios on Pexels

    What counts as health data under UK GDPR?

    Under the UK GDPR, health data is classed as “special category” data, meaning it attracts stronger legal protections than ordinary personal data. The definition is broader than most people realise. It covers data relating to the physical or mental health of a natural person, including information that reveals their health status. That language matters because it pulls in inferred data, not just data you consciously entered.

    A period tracker that records your cycle dates is processing health data. A fitness app that logs your resting heart rate over time could reveal a cardiac condition. A sleep app that notices you are waking at 3am every night is generating a dataset from which mental health inferences can reasonably be drawn. The app companies know this. What their privacy policies often obscure is what happens to that data after it leaves your phone.

    How health apps actually monetise your data

    The phrase “we may share your data with trusted third parties” is doing a lot of heavy lifting. In practice, the data supply chain from a consumer health app can involve advertising technology platforms, data analytics firms, research organisations and, yes, data brokers who aggregate and resell profiled datasets. The legal mechanism that makes much of this possible is “legitimate interests” under Article 6 of the UK GDPR, combined with consent that was obtained through a consent management platform buried three taps deep in settings.

    Period and cycle tracking apps have received particular scrutiny. A 2021 investigation by Privacy International found that several popular apps were sharing intimately personal data with Facebook’s advertising SDK at the point of app launch, before users had any chance to interact with a consent screen. Some of those apps are still widely used in the UK. The ICO acknowledged concerns about advertising technology broadly in its 2019 report on real-time bidding, but substantive enforcement against health-specific apps has been sparse.

    Digital lock icon representing health app data privacy UK concerns
    Photo by Ann H on Pexels

    What the ICO has and has not enforced

    The ICO has real powers. Under the UK GDPR and the Data Protection Act 2018, it can issue fines of up to £17.5 million or 4% of global annual turnover, whichever is higher. It has used those powers against companies including British Airways and Marriott, though both fines were substantially reduced on appeal. For health app data privacy in the UK specifically, the enforcement record is thin.

    The ICO published its adtech and real-time bidding work and has issued guidance on special category data, but formal enforcement notices specifically targeting consumer health apps or data brokers handling inferred health data remain rare. The regulator has cited resource constraints and the complexity of cross-border enforcement as factors. That is an honest answer, but it leaves a real gap.

    There was a significant moment in 2023 when the ICO issued a reprimand to Snap over its My AI feature, touching on children’s data risk assessments. That reprimand, rather than a fine, illustrated the regulator’s tendency to use softer tools first. For people whose period tracking data or mental health journal entries have already been shared with third parties, a reprimand issued years later feels inadequate.

    The specific risks from period trackers, mental health apps and symptom diaries

    Period trackers carry risks that go beyond embarrassment. Inferred fertility status, pregnancy history or menstrual irregularities can be of interest to insurance underwriters, employers and, in some jurisdictions, law enforcement. UK law offers some protection here: the Equality Act 2010 prohibits discrimination on grounds of pregnancy and maternity, and using health data to discriminate in insurance pricing is tightly restricted by the FCA. But data shared with a broker in a third country, processed under a different legal framework, is much harder to protect.

    Mental health apps sit in a similarly fraught position, and I’d argue they carry the highest reputational risk for the sector. Someone using a mood diary, anxiety tracker or cognitive behavioural therapy app is generating a longitudinal record of their psychological state. If that record is accessible to a data broker, it can be used to build a profile that follows a person across the web. As I covered in an earlier piece on the UK’s mental health app regulation gap, many of these apps operate without any meaningful clinical oversight, which compounds the data problem: there is no regulatory body with clear authority over both the therapeutic claims and the data practices simultaneously.

    Symptom diary apps, the kind people use to track chronic pain, digestive flares, fatigue levels and medication responses, present a third category of risk. These are often used by people with conditions like IBD, ME/CFS or long-term post-viral illness, populations that are already under-served by the NHS and more likely to turn to digital tools to fill the gap. The data generated is clinically detailed. Whether it is being treated with the care that implies is a question most app privacy policies answer evasively.

    What you can actually do right now

    Audit the permissions your health apps hold. On an iPhone, go to Settings, then Privacy and Security; on Android, go to Settings, then Apps, then Permissions. Revoke location and advertising ID access from any health app that does not have an obvious clinical reason to need it. Check whether the app offers an opt-out from data sharing with third parties, and assume the default is opt-in unless you can confirm otherwise.

    Look at where the company is based. An app with a UK company registration and a data protection officer listed on its website is meaningfully more accountable under UK GDPR than one incorporated in a jurisdiction with no equivalent law. You can check company registrations at Companies House in under two minutes.

    Some people navigating health monitoring, especially around recovery, wellness and longevity goals, are also shifting towards hardware and offline supplementation rather than app-based tracking. Based in Nottinghamshire, HealthPod Mansfield supplies hyperbaric oxygen tanks, red light therapy beds and supplements to people who want to actively support their health and live longer without necessarily feeding their data into a subscription app ecosystem. The core appeal for wellness-conscious users is straightforward: you can pursue recovery and health goals using physical equipment at healthpodonline.co.uk without generating a behavioural dataset that a third party can monetise. That is a real consideration, not a minor one.

    The gap that UK law has not closed

    The fundamental problem is that UK GDPR requires lawful basis and transparency, but it does not require that consent be genuinely informed in any meaningful cognitive sense. A 4,000-word privacy policy that mentions data sharing in paragraph 23 is technically transparent. It is not practically transparent. The ICO’s guidance on consent emphasises that it must be freely given, specific, informed and unambiguous, but auditing whether those conditions are met across thousands of consumer apps is not something the regulator currently has the capacity to do systematically.

    There is also the inferred data problem. UK GDPR protects health data you actively input. The question of whether inferred health data, a score calculated from your sleep patterns and heart rate variability that predicts your likelihood of depression, is equally protected is not yet definitively settled in UK case law. The ICO’s position is that inferred data can be special category data if it reveals health information, but that position has not been tested in a major enforcement case specifically targeting consumer health apps.

    People who use apps to manage their health deserve clarity. The current framework provides a legal structure that looks protective but has meaningful gaps in practice. Pushing for stronger enforcement, and being more selective about which apps get access to the most sensitive data you generate, are the two most practical responses available right now. For anyone interested in how digital health tools interact with your personal data more broadly, the NHS waiting list and the self-diagnosis trap piece covers some of the downstream risks when people turn to unregulated tools to fill care gaps, and it is worth reading alongside this one.

    HealthPod Mansfield, known in Nottinghamshire for supplying red light beds and recovery-focused supplements alongside hyperbaric oxygen equipment, represents one end of the spectrum: people choosing be healthy through tangible, offline means rather than through apps that ask for extensive data permissions. Whether or not that approach appeals to you, the underlying instinct to question what health tools actually do with what they learn about you is one more people should develop.

  • The UK’s Mental Health App Boom Has a Regulation Problem Nobody Is Talking About

    The UK’s Mental Health App Boom Has a Regulation Problem Nobody Is Talking About

    There are now thousands of mental health apps available to UK consumers. Calm, Wysa, Headspace, Kooth, Woebot, Silvercloud, the list keeps growing. The App Store and Google Play are flooded with products promising to reduce anxiety, improve sleep, treat depression, or help users manage their mood. Some of them are genuinely useful. Some of them are not. And the uncomfortable truth is that most UK consumers have no reliable way to tell the difference, because the framework governing mental health apps regulation UK-wide has serious gaps in it.

    I’ve spent time looking at what oversight actually exists here, and the picture is more fragmented than most people assume. This is not a case of regulators being asleep at the wheel, it is a structural problem built into the way digital health products are classified.

    Person using a mental health app on their smartphone, illustrating the mental health apps regulation UK debate
    Photo by cottonbro studio on Pexels

    What qualifies as a regulated medical device?

    The MHRA (Medicines and Healthcare products Regulatory Agency) oversees software as a medical device (SaMD) in the UK. Since Brexit, the UK has moved away from CE marking for medical devices towards its own UKCA marking, though the MHRA has repeatedly extended transition periods for software. The current framework uses a risk-based classification system: Class I devices are low risk, Class IIa and IIb are moderate risk, and Class III are the highest risk.

    A mental health app that makes a specific diagnostic or therapeutic claim, for example, one that says it is a clinically validated treatment for generalised anxiety disorder, should, in theory, qualify as a SaMD and require MHRA registration. But the classification hinges entirely on the claim the app makes. An app that describes itself as a “wellness tool” or a “mood journal” sits outside that definition, even if its actual functionality is nearly identical to something making clinical claims. Developers are well aware of this, which is why the language used in app store listings is often conspicuously careful.

    The MHRA published updated guidance on software as a medical device in 2023, and it is genuinely thorough. The problem is enforcement capacity and the sheer pace at which new products arrive. Registering a device is not the same as independently verifying the evidence behind it.

    What NICE does, and does not, cover

    The National Institute for Health and Care Excellence has a programme called the Evidence Standards Framework for Digital Health Technologies (DHT), developed in partnership with NHS England. It sets out what evidence digital health tools should provide at different tiers, from simple patient information tools through to products that claim to replace clinical interventions.

    The framework is sound in principle. Tier 3b products, those claiming to treat or diagnose, are expected to provide evidence from randomised controlled trials or equivalent. But participation in the framework is voluntary for commercial products not commissioned through the NHS. An app sold directly to consumers on the App Store has no obligation to submit to NICE review, publish its trial data, or demonstrate that its claimed outcomes hold up in a general UK population. The NHS App Library, which previously listed vetted apps, was quietly wound down. NICE now points users towards a curated but limited set of guidance pages, and the broader consumer market continues largely unvetted.

    Where the CQC fits in, and where it stops

    The Care Quality Commission regulates health and social care services in England, including some digital services. If a mental health app employs regulated professionals who provide clinical advice, a psychiatrist, a psychologist, a counsellor operating under a professional duty of care, the service may fall within CQC registration requirements. But an app that offers automated chatbot therapy, AI-driven cognitive behavioural therapy exercises, or algorithm-generated mood analysis does not automatically require CQC registration, because the service is not being delivered by a regulated professional in a traditional sense.

    This is the gap where a lot of the most popular mental health apps actually sit. The AI element is significant here. As I’ve written before when covering the MHRA’s evolving approach to AI medical devices, the regulatory question of when an algorithm becomes a medical device is genuinely contested. A chatbot that listens, reflects, and suggests breathing exercises exists in a grey zone that neither the MHRA, the CQC, nor any professional regulator definitively owns.

    The evidence problem in practice

    A 2019 review published in npj Digital Medicine analysed 73 depression and anxiety apps available to UK consumers and found that fewer than 4% had been tested in a randomised controlled trial. The situation has improved since, Wysa, for example, has published peer-reviewed studies, and Silvercloud (now Brightside Health) has a reasonable evidence base built from NHS-commissioned research. But the majority of apps in the mental health and wellbeing category make claims that rest on small, industry-funded pilot studies, or no published evidence at all.

    The user reading a five-star review and a vague reference to being “evidence-based” cannot assess the quality of that evidence. And this matters clinically. For someone with mild-to-moderate anxiety who cannot get a GP referral quickly, a reality explored in detail in what I’ve written about the NHS waiting list and the DIY health trap, an app might feel like a reasonable bridge. If the app is genuinely effective, that bridge holds. If it is not, the user may spend months investing time and hope in something that delays them seeking help through other routes.

    There is also a data angle worth taking seriously. Mental health apps collect sensitive personal data at scale: mood logs, sleep patterns, journal entries, sometimes audio. ICO guidance on special category data under UK GDPR applies, and several apps have faced criticism for opaque privacy policies and data sharing arrangements. The ICO and the MHRA technically have overlapping but distinct jurisdictions here, and there is no single point where all of this gets checked together.

    What would actually help

    The honest answer is not more regulation for its own sake. A blanket requirement for every mindfulness app to run a phase three clinical trial would kill useful low-risk tools and drive developers offshore. The more practical solution is clearer labelling and a functional public registry.

    Some version of a tiered trust mark, one that consumers can actually look up and verify, backed by the MHRA or NHS England, would give the market a credibility signal that is currently absent. NHS Digital, before it was folded into NHS England, attempted something like this with the App Library. The logic was right; the execution and resourcing were not.

    For now, my practical advice to anyone considering a mental health app is this: look for published peer-reviewed trials, check whether the app is listed in current NHS England pathways, and treat any app claiming to “treat” a named condition with the same scepticism you would apply to any other unverified therapeutic claim. Apps that are upfront about being wellness tools rather than clinical interventions are being more honest, not less useful.

    The mental health apps regulation UK landscape is not broken in an obvious, dramatic way. It is just porous in ways that are invisible to most users, and that invisibility is the actual problem.

    Frequently Asked Questions

    Are mental health apps regulated in the UK?

    Some are. Apps that make specific clinical or diagnostic claims may need to register with the MHRA as software as a medical device. However, apps marketed as wellness tools fall outside this requirement, which means a large portion of the market operates without formal regulatory oversight of their claimed benefits.

    What does UKCA marking mean for a health app?

    UKCA marking is the post-Brexit UK conformity assessment mark that replaces CE marking for medical devices sold in Britain. For a mental health app, it would signal the product has been assessed against MHRA standards as a software medical device. In practice, transition periods have been extended repeatedly and the mark is rare on consumer-facing apps.

    Does NICE endorse specific mental health apps?

    NICE has an Evidence Standards Framework for Digital Health Technologies, which sets out what evidence apps should provide. However, commercial apps sold directly to consumers are not required to go through this process. Only apps commissioned by the NHS through formal procurement pathways face meaningful scrutiny against NICE standards.

    What data do mental health apps collect and who regulates that?

    Mental health apps typically collect sensitive personal data including mood logs, journal entries, and behavioural patterns. This falls under special category data rules in UK GDPR, regulated by the ICO. However, the ICO and MHRA operate separately, so there is no single point of oversight that covers both clinical claims and data handling together.

    How can I tell if a mental health app is actually evidence-based?

    Look for published peer-reviewed studies in indexed journals rather than vague references to being ‘clinically validated’. Check whether the app appears in NHS England clinical pathways or has been reviewed under the NICE DHT framework. Treat industry-funded pilot studies with caution, and prefer apps that are transparent about the limits of what they can and cannot do.

  • Shift Work and Metabolic Health: The Silent Risk Millions of UK Workers Are Largely Unaware Of

    Shift Work and Metabolic Health: The Silent Risk Millions of UK Workers Are Largely Unaware Of

    Around 3.5 million people in the UK work shifts, according to the Office for National Statistics. That covers NHS nurses working nights, warehouse staff on rotating rotas, lorry drivers, hotel and restaurant workers, and a significant chunk of emergency services. Most of them are aware that shift work is tiring. Fewer are aware that it is doing measurable damage to their metabolic health, sometimes long before any symptom appears.

    NHS nurse on night shift, illustrating shift work metabolic health risks for healthcare workers
    Photo by Daniil Kondrashin on Pexels

    I’ve spent some time looking at what the circadian biology research actually says here, because the gap between the science and public awareness is striking. This is not a case of vague risk; the mechanisms are reasonably well understood, and they play out in ways that affect insulin sensitivity, cardiovascular risk, and even cancer incidence. The workers bearing the largest burden are disproportionately in lower-paid roles with the least flexibility to change their schedules.

    What shift work does to your body clock

    Human physiology is built around a roughly 24-hour cycle, governed by the suprachiasmatic nucleus in the hypothalamus and synchronised primarily by light. When you eat, sleep, and wake at irregular times, that internal clock and the external environment fall out of step. Researchers call this circadian misalignment, and its downstream effects are not trivial.

    Insulin sensitivity drops meaningfully during biological night. A 2019 study published in Science Advances by researchers at the Brigham and Women’s Hospital found that people in a simulated night-shift schedule showed a 17% reduction in insulin sensitivity within days. The pancreas partially compensates, but not always fully, and over years of shift work the cumulative exposure appears to push people towards type 2 diabetes at a notably higher rate than day workers. A meta-analysis in Occupational and Environmental Medicine put the elevated risk for shift workers at approximately 9% higher than those working standard hours, with rotating shifts carrying more risk than fixed nights.

    The cardiovascular picture is similarly concerning. Cortisol and inflammatory markers behave differently when sleep is displaced. A review in the European Heart Journal found shift workers have a roughly 23% higher risk of myocardial infarction compared to non-shift workers. Blood pressure regulation is also disrupted; the normal nocturnal dip in blood pressure, which is protective, does not occur reliably in people sleeping during the day.

    The cancer link is real, though nuanced

    Since 2007, the International Agency for Research on Cancer has classified night shift work as a probable human carcinogen (Group 2A). The biological mechanism centres on melatonin suppression. Melatonin, produced by the pineal gland in darkness, has anti-tumour properties. When artificial light suppresses it during a night shift, that protective output is reduced. Breast cancer is the most studied outcome, with several large cohort studies finding modestly elevated risk in women doing long-term night work, though effect sizes vary and the epidemiology is complicated by lifestyle factors.

    The evidence is strong enough that the WHO and the NHS recognise night shift work as an occupational exposure. What is less clear is the exact dose-response relationship, i.e., how many years of shift work, how many nights per week, and what kind of light exposure tips someone into meaningfully elevated risk. The honest answer is that researchers are still working this out.

    Warehouse worker on night shift, representing the shift work metabolic health risks in UK logistics
    Photo by IAN on Pexels

    Who in the UK is most exposed

    NHS data from NHS England suggests roughly 350,000 nurses work some form of night or rotating shift. The logistics sector, expanded significantly since 2020, employs hundreds of thousands of warehouse and distribution workers on 24-hour rota patterns. Hospitality workers on late and split shifts face a different but related problem: irregular sleeping windows, high caffeine intake, and compressed recovery time between shifts.

    What links these groups is that most cannot choose their hours, and many are not given adequate information about the health implications. If you spend time looking at what research says about sedentary office work and cardiovascular risk, you quickly notice that desk workers at least have some agency over their movement patterns. Shift workers often have less.

    There is also an inequality dimension. Night and rotating shift work is more common in ethnic minority communities in the UK, partly due to sector concentration in healthcare, logistics, and manufacturing. Since cardiovascular risk and type 2 diabetes already have elevated prevalence in some of these communities, circadian disruption can compound an existing burden.

    What mitigation actually looks like

    I want to be clear that there is no perfect solution here short of ending shift work, which is obviously not realistic. But the evidence does point to some practical levers worth taking seriously.

    Light exposure matters more than most workers know. Getting bright light during the first half of a night shift and then minimising light exposure during the commute home (blue-light-blocking glasses genuinely help here, despite how they look) can reduce circadian disruption. Blackout curtains for daytime sleep are not optional; they are a genuine physiological intervention.

    Meal timing is underrated. Eating during biological night amplifies insulin resistance. Workers who can shift their main caloric intake to biological day hours, even imperfectly, appear to have better metabolic outcomes. This is not about dieting; it is about when food lands relative to your body clock. If you are looking at evidence on time-restricted eating, the research on how meal timing affects weight and metabolic markers is directly relevant here.

    Rotating shifts vs. fixed nights. The evidence consistently shows that fixed night shifts, while socially disruptive, produce less metabolic damage than rotating patterns. The body can achieve a degree of adaptation to a fixed schedule; it cannot adapt to one that keeps changing. Where employers have the flexibility to offer fixed patterns, there is a reasonable case for doing so on health grounds.

    Caffeine management. Many shift workers use caffeine heavily to manage alertness, which is understandable. But consuming it in the second half of a shift significantly worsens daytime sleep quality. The research on caffeine dependence and its clinical significance applies with particular force to people already fighting a compromised sleep window.

    Some NHS trusts are beginning to build structured health checks specifically for long-term shift workers into occupational health programmes, including HbA1c testing and blood pressure monitoring more frequently than for day workers. That is the right direction. Tools that help workers and their employers manage scheduling and health data more intelligently, from workforce planning software to health-monitoring integrations, are part of how this problem gets addressed at scale; dijitul.ai is one example of how AI-assisted administration tools are starting to appear in operational contexts like this.

    The policy gap

    The UK has a Working Time Regulations framework that caps hours and mandates rest periods, but it does not specifically address circadian health. There is no requirement for employers to inform workers of the biological risks of long-term night work, and occupational health access is patchy, particularly in logistics and hospitality where smaller employers dominate.

    Given what we now know about shift work metabolic health outcomes, the case for updated guidance from NICE and more systematic screening in at-risk occupational groups is strong. A worker who has done 15 years of rotating night shifts is not in the same cardiovascular risk category as a day worker of equivalent age and BMI. The clinical framework should reflect that.

    Millions of UK workers are doing essential work under conditions that quietly increase their risk of diabetes, heart disease, and potentially cancer. That is worth knowing about clearly, not buried in occupational health small print.

    Frequently Asked Questions

    How does shift work affect insulin sensitivity?

    Circadian misalignment caused by working at night disrupts the body’s normal insulin response cycle. Research shows that insulin sensitivity can drop significantly within days of night-shift schedules, and long-term rotating shift work is associated with around a 9% higher risk of developing type 2 diabetes compared to standard day workers.

    Is night shift work linked to cancer?

    The International Agency for Research on Cancer classifies night shift work as a probable human carcinogen (Group 2A). The main proposed mechanism is suppression of melatonin, which has anti-tumour properties, during exposure to artificial light at night. Breast cancer is the most studied outcome, with modest elevated risk found in women doing long-term night work across several large cohort studies.

    Which UK workers are most at risk from shift work metabolic health problems?

    NHS nurses, logistics and warehouse workers, hospitality staff, and emergency services personnel make up the majority of the UK’s estimated 3.5 million shift workers. Rotating shift patterns carry more metabolic risk than fixed night schedules, and workers with limited schedule flexibility face the greatest cumulative exposure.

  • The Real-World Accuracy of AI Coding Assistants in 2026: What UK Developers Are Actually Finding

    The Real-World Accuracy of AI Coding Assistants in 2026: What UK Developers Are Actually Finding

    There is a lot of noise around AI coding assistants right now, and most of it comes from two camps: developers who think these tools are transformative, and developers who are quietly cleaning up the mess they leave behind. The truth, as ever, sits somewhere between the two. I’ve spent time with the benchmarks, the independent surveys, and the candid developer forums, and the picture that emerges is more nuanced than the marketing suggests. For UK developers in particular, the question of AI coding assistant accuracy in 2026 deserves a straight answer.

    UK developer reviewing AI coding assistant accuracy on a laptop screen in 2026
    Photo by Lukas Blazek on Pexels

    What the benchmarks actually measure

    Most headline figures around tools like GitHub Copilot, Amazon CodeWhisperer, and Cursor come from vendor-commissioned tests or cherry-picked HumanEval scores. HumanEval, developed by OpenAI, tests whether a model can write short Python functions from a docstring. It is a useful starting point, but it bears little resemblance to what a developer at a UK fintech or NHS digital team actually deals with day-to-day.

    The Stack Overflow Developer Survey 2025 found that roughly 76% of respondents were using or planning to use AI coding tools, but satisfaction scores told a different story: just 43% said the output was reliable enough to merge without careful review. That gap matters. Accepting code suggestions is easy. Trusting them is another thing.

    Independent benchmarks like SWE-bench, which tests whether models can actually resolve real GitHub issues in established codebases, show much lower success rates than HumanEval implies. The best-performing models in early 2026 resolve around 40-50% of SWE-bench tasks. That is progress, but it also means that in roughly half of real-world scenarios, the model either fails outright or introduces a solution that compiles but behaves incorrectly under edge cases.

    Where AI coding assistants genuinely save time

    This is not a dismissal. These tools do save time, and I’d be dishonest not to say so clearly. The categories where the gains are real and consistent are fairly specific.

    Boilerplate code is the obvious one. Writing repetitive scaffold code, setting up test files, generating getter and setter methods, or producing the skeleton of a REST endpoint: these tasks are well-suited to autocomplete-style suggestions, and the time savings are measurable. Developers in the Stack Overflow survey reported saving between one and four hours per week on routine tasks, with the higher estimates coming from those working in strongly typed languages like TypeScript where the context is clearer.

    Documentation generation is another genuine win. Getting a model to produce a first-draft docstring or README section is quicker than writing one from scratch, and even if you edit it afterwards, you are starting from something rather than a blank page.

    Learning unfamiliar syntax is where I personally find these tools most useful. If you are moving between languages, or working with an API you have not touched before, having an assistant that can surface working examples in context is faster than cycling between documentation tabs.

    Where the errors creep in and why they are costly

    The errors that AI coding assistants introduce are rarely catastrophic and obvious. They tend to be subtle. That is precisely what makes them expensive.

    Security vulnerabilities are the most serious concern. A 2025 study from Stanford University found that developers using AI assistants were significantly more likely to introduce security flaws than those coding without them, partly because the suggestions feel authoritative and get merged without the scrutiny a human-written block would receive. Common patterns include outdated cryptographic practices, SQL injection vectors in generated query strings, and improper input validation. For UK developers working in regulated sectors, whether financial services under FCA rules or health data environments governed by the ICO, these are not abstract risks.

    Logic errors in conditional branches are another consistent problem. The model may generate code that handles the happy path correctly but silently fails on null inputs, empty arrays, or timezone edge cases. I have seen this described again and again in developer forums: the code runs, the tests pass, and the bug surfaces three weeks later in production.

    Hallucinated library references also remain a genuine issue. Tools occasionally suggest method calls that do not exist in the version of a library the project actually uses, which wastes debugging time and can confuse junior developers who assume the suggestion is reliable.

    There is also a subtler risk around over-reliance. Developers who lean heavily on AI suggestions for code they do not fully understand are producing code they cannot confidently maintain. The growing role of AI in regulated environments makes this a genuine governance issue, not just a craft one. For teams building tools in health or finance, code ownership and auditability matter.

    The skills and habits that change the outcome

    The developers who report the best outcomes with AI coding assistants share a few consistent habits. They treat suggestions as a starting point rather than a final draft. They run the generated code through their existing test suites before accepting it. They stay especially sceptical with any suggestion that touches authentication, data persistence, or external API calls.

    Prompt quality matters more than most vendors acknowledge. A vague instruction produces a vague suggestion. Developers who invest time in writing precise, contextual prompts, including relevant type signatures, examples of existing patterns in the codebase, and explicit constraints, consistently get better output. This is a skill, and it takes time to develop.

    Team culture plays a role too. Organisations where code review is taken seriously, where junior developers are encouraged to question AI-generated suggestions rather than defer to them, tend to catch errors earlier. The AI assistant does not remove the need for good engineering practices; it actually raises the stakes on having them.

    What this means for UK developers practically

    If you are a UK developer or engineering manager deciding how to integrate these tools into your workflow, the honest advice is to use them, but with explicit guardrails. Establish a team norm that AI-generated code requires the same review rigour as code from any other source. Consider adding AI-specific items to your pull request checklist: has this been tested on edge cases, does it reference a real and current API, has anyone checked for security patterns the model might have assumed incorrectly.

    The productivity gains are real enough to justify adoption. The accuracy gaps are real enough to justify caution. Those two things are not in contradiction. And for anyone thinking about the broader pattern here, the same principle applies whether you are assessing an AI coding assistant or thinking about workplace tools that promise more than the evidence supports: the question is never whether a tool does something useful, but whether the specific claims match the specific evidence.

    UK developers are, by most accounts, adopting these tools at roughly the same pace as their counterparts in the US and Germany. The constraint is not enthusiasm; it is informed scepticism, and that is worth holding onto. The tools will improve. The discipline of reviewing what they produce should improve alongside them.

  • Seed Oils, Inflammation and the British Diet: Separating the Science from the Culture War

    Seed Oils, Inflammation and the British Diet: Separating the Science from the Culture War

    If you spend any time in British health and wellness spaces online, you have probably encountered the seed oil debate. On one side, influencers warn that vegetable oils are quietly destroying your gut lining, triggering systemic inflammation and driving chronic disease. On the other side, the instinct is to dismiss all of that as paranoid nonsense from people who eat too much red meat and distrust mainstream medicine. I think both camps are getting it wrong, and the actual evidence is more interesting than either side admits.

    Various cooking oils on a kitchen counter, relevant to the seed oils inflammation UK debate
    Photo by Андрей on Pexels

    What are seed oils, and how much do UK adults actually consume?

    The term “seed oils” covers refined oils extracted from seeds rather than fruit or animal tissue: sunflower, rapeseed, corn, soybean, safflower and cottonseed oils are the ones usually named. In practice, rapeseed oil is by far the most common in the UK, used heavily in food manufacturing and sold widely as “vegetable oil”. Sunflower oil is the second most visible on supermarket shelves.

    According to the National Diet and Nutrition Survey (NDNS), UK adults consume a significant proportion of their fat intake from vegetable oils and spreads. The key point for any honest discussion is that these oils are not a fringe ingredient; they are embedded in the processed food supply. Ready meals, crisps, biscuits, takeaway food and most manufactured sauces contain them. So if you are eating an average British diet, you are consuming seed oils in meaningful quantities whether you think about it or not.

    The linoleic acid argument: is there anything to it?

    The core claim from seed oil sceptics centres on linoleic acid, an omega-6 polyunsaturated fatty acid (PUFA) that makes up a large share of most seed oils. The argument runs like this: linoleic acid is metabolised into arachidonic acid, which is itself a precursor to pro-inflammatory signalling molecules called eicosanoids. Higher linoleic acid intake therefore equals more inflammation. It sounds mechanistically plausible, and that is exactly why it has traction.

    The problem is that the leap from metabolic pathway to clinical outcome is where the evidence does not hold up cleanly. A 2020 systematic review published in Circulation examined randomised controlled trials and found that replacing saturated fat with linoleic-acid-rich oils reduced cardiovascular events. A 2021 meta-analysis in the British Journal of Nutrition found no consistent evidence that higher linoleic acid intake raises circulating arachidonic acid or inflammatory markers in humans to a clinically significant degree. The body appears to regulate this conversion tightly under normal dietary conditions.

    I am not saying those findings close the debate entirely. The quality of available trials is variable, follow-up periods differ, and some researchers make a credible case that the omega-6 to omega-3 ratio matters more than absolute linoleic acid intake. But “plausible mechanism” is not the same as “demonstrated harm”, and that distinction matters enormously. You can find the same pattern in a lot of nutrition debates, including the one around fibre intake in the UK, where mechanistic logic and real-world outcomes sometimes point in different directions.

    What does NICE guidance actually say about dietary fats?

    NICE does not single out seed oils for restriction. The current lipid guidance (CG181, updated most recently in 2023) recommends reducing saturated fat intake and suggests that replacing saturated fats with unsaturated fats, including PUFAs, is associated with better cardiovascular outcomes. That is broadly consistent with the position of the British Heart Foundation and the British Dietetic Association.

    NICE does acknowledge that dietary fat quality matters and that ultra-processed food consumption is a concern, but those two things are not the same as condemning seed oils specifically. If your diet is heavy in ultra-processed food containing seed oils, that is a problem. But the evidence suggests the ultra-processing, the excess energy, the refined carbohydrates and the low fibre are doing most of the heavy lifting in terms of harm, not the oil itself.

    Rapeseed oil specifically: a nuance the debate usually skips

    Most online content about seed oils focuses on American dietary patterns and American products: soybean oil is the dominant culprit in US-centric discussions. In the UK, the picture is different. Rapeseed oil has a much more favourable fatty acid profile than sunflower oil, with a lower omega-6 to omega-3 ratio and a reasonable monounsaturated fat content. Cold-pressed rapeseed oil, in particular, retains some polyphenols with mild antioxidant properties.

    This matters because a lot of the inflammatory claims circulating online are lifted from US sources and applied wholesale to the British context without adjustment. The seed oils in your average British supermarket trolley are not identical in composition to what American commentators are warning about. That does not make rapeseed oil a health food, but it makes the blanket condemnation less well-founded here than the content suggests.

    Where the anti-seed oil crowd has a point

    I do not want to dismiss every concern, because some of them are worth taking seriously. Heat stability is a real issue: seed oils high in PUFAs are more prone to oxidation at high cooking temperatures than olive oil or butter, producing degradation products including aldehydes. The degree to which this is clinically significant under normal home cooking conditions is debated, but the chemistry is real. If you are deep-frying regularly in sunflower oil and reusing it multiple times, switching to a more stable fat is not irrational.

    The omega-6 to omega-3 ratio argument also has some support. The concern is not that linoleic acid is inherently toxic but that a diet very high in omega-6 relative to omega-3 may shift the inflammatory balance in a direction that is not ideal. The practical answer to this is less about cutting seed oils and more about increasing omega-3 intake through oily fish or a supplement. This connects to what I covered in the blog’s earlier piece on whether supplement use is justified for common UK dietary shortfalls; the answer depends heavily on what the rest of your diet looks like.

    What should UK adults actually do?

    The practical answer is less dramatic than either camp wants it to be. Seed oils in moderate amounts, used for everyday cooking, are unlikely to be meaningfully harming you based on the current evidence. The inflammation crisis being attributed to sunflower oil in your stir-fry is probably more accurately attributed to a diet low in vegetables, fibre and oily fish, and high in ultra-processed food overall.

    That said, there is no strong reason to maximise seed oil consumption either. Using olive oil or cold-pressed rapeseed oil for everyday cooking is a reasonable preference, especially for flavour. Keeping omega-3 intake adequate, whether through two portions of oily fish per week (as recommended by the NHS) or a quality supplement, matters more than obsessing over which oil is in the pan.

    The culture war framing around seed oils is ultimately a distraction. The British diet has real problems, including chronically low fibre intake, insufficient protein for many older adults, and widespread omega-3 deficiency. Fixating on a single ingredient, framed as either poison or perfectly safe, usually means missing the bigger picture. Evidence-based dietary thinking is less satisfying than a clear villain, but it is more likely to actually help you. And if you have concerns about inflammation, chronic fatigue or systemic symptoms that do not resolve, that conversation belongs with your GP, not an algorithm. You might also find it useful to read about how the NHS currently handles complex fatigue conditions, where the gap between online certainty and clinical reality is similarly stark.

    Frequently Asked Questions

    Are seed oils causing inflammation in people eating a typical British diet?

    The current systematic review evidence does not support the claim that moderate seed oil consumption causes clinically significant inflammation in otherwise healthy adults. The stronger risk factors in the British diet are low fibre, low omega-3 intake and high ultra-processed food consumption. Seed oils are a component of that processed food problem, but the evidence does not single them out as the primary driver.

    Is rapeseed oil (vegetable oil) as bad as sunflower or soybean oil?

    No, and this distinction matters for UK readers. Cold-pressed rapeseed oil has a better omega-6 to omega-3 ratio than sunflower oil and a reasonable monounsaturated fat content, making it closer to olive oil in profile than to high-linoleic oils. Much of the anti-seed oil content online is based on American dietary patterns dominated by soybean oil, which does not translate directly to the UK context.

    Should I switch from vegetable oil to butter or olive oil for cooking?

    Using olive oil or cold-pressed rapeseed oil for everyday cooking is a reasonable choice, particularly for moderate-heat cooking. Butter is fine in moderate amounts for those without cardiovascular risk factors. What matters more than the specific oil choice is the overall dietary pattern: adequate fibre, vegetables, protein and omega-3 intake will have a far greater effect on health outcomes than swapping one cooking fat for another.

  • Magnesium Deficiency in the UK: Who Is Actually at Risk and Whether the Supplement Boom Is Justified

    Magnesium Deficiency in the UK: Who Is Actually at Risk and Whether the Supplement Boom Is Justified

    The supplement aisle has changed. Where once you’d find a modest row of multivitamins, there are now entire sections dedicated to magnesium alone, offered in a bewildering range of forms: glycinate, threonate, malate, citrate, oxide. Social media has done a lot of the marketing work here, with creators attributing poor sleep, anxiety, muscle cramps, and fatigue to a single missing mineral. The question worth asking, though, is whether magnesium deficiency UK health data actually supports this level of concern, or whether most of us are buying supplements we don’t need.

    Supplement capsules on a wooden table illustrating the growing discussion around magnesium deficiency UK
    Photo by ready made on Pexels

    What the National Diet and Nutrition Survey actually shows

    The National Diet and Nutrition Survey (NDNS), run jointly by the FSA and the Department of Health and Social Care, provides the most reliable picture of what British adults are actually eating. The most recent rolling programme data consistently shows that a meaningful proportion of UK adults fall below the Reference Nutrient Intake (RNI) for magnesium. For adult men, the RNI sits at 300mg per day; for adult women, 270mg per day.

    The NDNS figures show that teenage girls and women aged 19 to 64 are the groups most likely to have intakes below the lower reference nutrient intake, which is the threshold at which deficiency becomes a genuine physiological concern. Men over 65 also appear in the data as a higher-risk group. But here’s the nuance that most supplement marketing glosses over: falling below the RNI is not the same as being clinically deficient. The RNI is set deliberately high, covering the needs of around 97% of the population. Many people eating slightly below it are fine.

    What clinical magnesium deficiency actually looks like

    True clinical deficiency, called hypomagnesaemia, is detectable via a serum blood test and produces symptoms including muscle cramps, tremor, cardiac arrhythmias, and in severe cases, seizures. It is relatively rare in otherwise healthy adults. The people genuinely at risk are those with type 2 diabetes (where urinary magnesium loss is higher), those with inflammatory bowel conditions like Crohn’s disease (where absorption is compromised), people taking certain medications including proton pump inhibitors and diuretics, and heavy alcohol users. I’d also add people who have had significant sections of their small intestine removed for any reason.

    If you have one of those conditions, the conversation about magnesium is legitimate and worth having with a GP. For the rest of the population, the honest answer is that most people eating a varied diet are not clinically deficient, even if their intake doesn’t look ideal on paper. If you’re interested in how dietary absorption and gut health interact with nutrient levels more broadly, it’s worth reading about how fibre intake affects far more than just digestion, since gut integrity plays a role in how efficiently any mineral is absorbed.

    Natural food sources of magnesium including pumpkin seeds and dark chocolate relevant to magnesium deficiency UK dietary intake
    Photo by Vie Studio on Pexels

    Are glycinate and threonate actually better?

    The premium supplement market leans heavily on two forms: magnesium glycinate and magnesium L-threonate. Glycinate is marketed as being gentler on the stomach and better absorbed than oxide. Threonate is sold with claims around cognitive function and brain health. Both are considerably more expensive than magnesium citrate or oxide.

    On the bioavailability question, the evidence does suggest that oxide is poorly absorbed relative to other forms, so if you’re going to supplement, glycinate and citrate are reasonable choices. The threonate claims are more specific. The research most cited comes from animal studies and a small number of human trials, some of which were funded by the manufacturer. The 2016 trial published in the Journal of Alzheimer’s Disease by Slutsky et al. showed promising results in older adults with cognitive complaints, but the sample sizes were small. I’d characterise the brain health evidence as genuinely interesting but nowhere near established enough to justify the price premium most UK retailers charge.

    Retailers including HealthPod Mansfield stock a range of magnesium forms, which is useful if you’re trying to compare options and understand what you’re actually buying rather than just defaulting to whatever’s trending on a given week.

    The sleep and anxiety claims

    This is where the online conversation runs furthest ahead of the science. Magnesium does have a role in regulating GABA receptors, which are involved in calming neural activity, and it interacts with melatonin pathways. That’s a real mechanism. The clinical evidence for supplementation improving sleep in people who are already replete, however, is thin. A 2021 systematic review in BMC Complementary Medicine and Therapies found modest improvements in subjective sleep quality in older adults, but the effect sizes were small and the studies had notable methodological weaknesses.

    For anxiety, the picture is similarly cautious. There’s observational data suggesting lower dietary magnesium correlates with higher anxiety scores, but correlation between dietary intake and mental health is notoriously difficult to interpret causally. If sleep and anxiety are a concern, addressing the structural causes, including screen exposure, irregular schedules, and chronic stress, is likely to do more than any supplement. If you’re curious about how one common habit may be affecting your sleep more than you realise, this piece on how smartphones disrupt sleep is a useful companion read.

    Who should actually consider supplementing

    My read of the evidence is this: if you have a condition that raises your risk of genuine deficiency (type 2 diabetes, IBD, long-term PPI use), it’s worth discussing a blood test with your GP before buying anything. If your test comes back showing hypomagnesaemia, supplementing under guidance makes sense. If you’re a generally healthy adult eating a diet that includes green vegetables, nuts, wholegrains, and legumes, you’re probably meeting your needs from food.

    The food sources of magnesium are worth knowing. Per 100g, pumpkin seeds offer around 550mg, dark chocolate around 230mg, cooked spinach around 80mg, and wholemeal bread around 60mg. These aren’t exotic or expensive foods. A diet that’s consistently low in these things is more likely to be a broader dietary quality problem than a single-mineral problem, and no supplement corrects a poor diet pattern.

    There’s a broader issue worth naming. The wellness supplement market, now worth over £500 million annually in the UK according to industry figures, runs on the anxiety that we’re missing something essential. Sometimes that anxiety is justified, as we’ve seen with vitamin D in the UK where deficiency is genuinely widespread and the evidence for supplementation is solid. With magnesium, the picture is more complicated. Certain populations need to take it seriously. Most don’t need to spend £30 a month on threonate capsules. Getting that distinction right matters, and I’d argue the first step is looking at the NDNS data rather than a YouTube thumbnail. If dietary patterns and their downstream health consequences interest you, the evidence on how ultra-processed food affects brain function is another thread worth pulling.

    Frequently Asked Questions

    How common is magnesium deficiency in the UK?

    Subclinical low intake is fairly common, particularly among teenage girls and older adults, based on NDNS data. However, true clinical deficiency (hypomagnesaemia) is relatively rare in otherwise healthy British adults and is most likely in people with specific conditions such as type 2 diabetes, Crohn’s disease, or long-term use of certain medications.

    What is the difference between magnesium glycinate and magnesium oxide?

    Magnesium oxide has poor bioavailability, meaning much of it passes through without being absorbed. Glycinate and citrate forms are better absorbed and less likely to cause digestive discomfort. For most people who do need to supplement, glycinate or citrate are the more practical choices, though they tend to cost more.

    Can magnesium supplements improve sleep?

    Magnesium has a role in regulating GABA receptors, which support calm neural activity. However, the clinical evidence that supplementing improves sleep in people who are not actually deficient is weak. A 2021 systematic review found modest benefits mainly in older adults, with small effect sizes and significant study limitations.

    Is magnesium L-threonate worth the extra cost for brain health?

    The brain health evidence for threonate is based largely on animal studies and a small number of industry-funded human trials. Results are promising but not yet robust enough to justify its significant price premium over standard forms. I’d treat the cognitive claims as interesting rather than established.

  • Workplace Wellbeing Programmes in the UK: Why the Evidence for Most of Them Is Surprisingly Weak

    British employers spent an estimated £83 billion on lost productivity due to poor employee health in 2023, according to the government’s own Health is Everyone’s Business review. The response, across countless HR departments, has been to roll out wellbeing programmes: lunchtime yoga, mental health days, employee assistance programme (EAP) hotlines, mindfulness apps, and the occasional fruit bowl. I’d argue the instinct is good. The execution, in most cases, is not backed by anything close to solid evidence.

    The uncomfortable truth is that when you dig into the research, particularly the Cochrane systematic reviews and NICE’s own workplace health guidance, the returns on most of these interventions are either marginal, inconsistent, or impossible to measure reliably. That does not mean organisations should abandon the effort. It means they should direct it somewhere it actually works.

    What Cochrane reviews actually say about common workplace wellbeing interventions

    Cochrane is as rigorous as health research gets. Its reviews synthesise findings across multiple randomised controlled trials and are treated as the gold standard by NICE, the NHS, and most public health bodies. On workplace wellbeing, the picture is sobering.

    A 2020 Cochrane review on workplace physical activity interventions found that while some programmes modestly reduced absenteeism, the quality of evidence was generally low to very low. Effect sizes were small and often failed to persist beyond the intervention period. Yoga and relaxation programmes showed some short-term benefit on mental wellbeing scores, but the authors were clear: there is insufficient evidence to conclude these translate into sustained health improvements or meaningful reductions in presenteeism.

    EAP hotlines are perhaps the most widely adopted feature of UK workplace wellbeing programmes, offered by the majority of organisations with more than 250 employees. Yet a 2021 analysis published in the Journal of Occupational Health Psychology found uptake rates often sit below 5% of the workforce, and the evidence that EAPs reduce clinical-level anxiety or depression in employees who do use them remains limited. Self-reported satisfaction is reasonably high, but self-reported satisfaction is not the same as improved health.

    Why NICE guidance points in a different direction

    NICE’s guidance on mental health at work (NG212) and its broader workplace health framework do not recommend yoga sessions or mental health days as primary interventions. They recommend structural changes: job redesign to reduce excessive demands, giving employees genuine autonomy over how they work, addressing management behaviour, and ensuring access to occupational health services for those who need them.

    The distinction matters. A mental health day might give an individual a brief respite from a stressful environment, but it does nothing about the stressful environment itself. NICE’s position, broadly, is that interventions targeting the individual whilst leaving the work context unchanged are unlikely to produce lasting improvement. I think that is exactly right, and it is frustrating how rarely employer communications acknowledge it.

    There is also a socioeconomic dimension that most wellbeing programme reviews miss entirely. UK employer surveys conducted by the CIPD (Chartered Institute of Personnel and Development) consistently show that the employees who participate in wellbeing activities tend to be those already in better health. Employees dealing with serious stress, financial insecurity, or chronic illness are the least likely to join a yoga class or call a helpline. The people who most need support are systematically the least likely to access what is on offer.

    The specific problem with mental health days

    Mental health days have received enormous media coverage over the past few years, and several large UK employers have introduced them as a formal entitlement. The intention is reasonable: to reduce stigma and give employees space to recover. The evidence that they achieve this is thin.

    A single day off does not treat clinical depression. It does not reduce chronic work-related stress. It does not address the presenteeism problem, where employees come to work whilst mentally unwell and operate at a fraction of their capacity. According to the CIPD’s 2024 Health and Wellbeing at Work report, presenteeism is still far more common in UK workplaces than absenteeism, yet it receives comparatively little attention in wellbeing strategies.

    This connects to a broader pattern: organisations tend to adopt interventions that are visible, easy to communicate in a press release, and unlikely to require meaningful structural change. Yoga classes are easier to organise than a proper job-demand audit. A mental health day makes for good social media content. Reviewing line manager behaviour does not.

    What actually shows a reasonable evidence base

    I do not want to leave this as a pure critique without acknowledging what does appear to work. Several interventions have a more credible evidence base, even if the effect sizes remain modest.

    Cognitive behavioural therapy-based programmes delivered by qualified practitioners, either face-to-face or through accredited digital platforms, consistently outperform generic mental health apps in reducing anxiety and depression symptoms. The key word is qualified: the evidence does not transfer to self-guided mindfulness apps, which have proliferated in UK workplace wellbeing packages despite very limited trial evidence for population-level benefit.

    Occupational health referrals, when they happen early rather than as a last resort, are associated with better return-to-work outcomes for employees with musculoskeletal problems and common mental health conditions. The problem is that most UK employers still do not have direct access to occupational health services, a gap the government’s Occupational Health Task Force acknowledged in its 2023 report.

    Physical activity, when integrated into the working day through flexible scheduling or active travel incentives, does show some meaningful benefit for cardiovascular health and mood, especially in sedentary desk workers. I’ve written elsewhere about the specific cardiovascular risks of sedentary work and how little exercise it actually takes to begin reversing them. The point is not that movement is pointless; it is that a lunchtime yoga session offered as an optional extra, with no structural support for attendance, does not qualify as a meaningful physical activity intervention.

    What employers should actually be asking

    If you are involved in designing or commissioning workplace health programmes, a few questions are worth putting plainly. What is the uptake rate, and who is using this? Is the intervention targeting individuals or the work environment? Is there a qualified clinical professional involved, or is this essentially a branded app and a poster campaign? And perhaps most importantly: what does the data actually show, not what do employees say they like about it?

    Likeability and efficacy are not the same thing. A fruit basket scores well on both employee satisfaction surveys and Instagram. It does not prevent burnout.

    The wellbeing industry in the UK is substantial and growing, and much of it is selling products and programmes on the basis of intent rather than outcome. Organisations genuinely committed to employee health need to apply the same rigour they would to any other business decision. That means reading the NICE guidance, looking at the Cochrane evidence, and being honest about whether what is currently on offer is improving anything that actually matters.

    There is real hope here too. Organisations that shift their focus toward management training, workload control, and proper clinical referral pathways do see improvements. The evidence for those approaches is stronger. They are just harder to put in a wellbeing brochure, and that is probably why most employers still have not done it.

    Frequently Asked Questions

    Do workplace wellbeing programmes actually work in the UK?

    The evidence is mixed, and often weak. Cochrane reviews and NICE guidance suggest that popular interventions like yoga sessions, mental health days, and EAP hotlines show limited or inconsistent effects on measurable health outcomes. Structural changes to workload and management practice tend to show stronger results.

    What does NICE recommend for employee mental health at work?

    NICE guidance (NG212) prioritises reducing excessive job demands, improving employee autonomy, and addressing management behaviour over individual-focused perks. It emphasises organisational and environmental changes rather than optional wellness activities.

    Are employee assistance programme (EAP) hotlines effective?

    EAPs are widely available in UK workplaces but uptake is typically below 5% of the workforce. Evidence for their ability to reduce clinical anxiety or depression at a population level is limited, though individual users often report satisfaction with the service.

    Which workplace health interventions have the best evidence behind them?

    CBT-based programmes delivered by qualified practitioners, early occupational health referrals, and genuine integration of physical activity into the working day have stronger evidence bases than most standard wellbeing packages. The quality of delivery matters significantly.

    Why do employers keep offering wellbeing programmes if the evidence is weak?

    CIPD surveys suggest many employers choose interventions that are easy to implement and communicate, rather than those with the strongest clinical evidence. Visible, low-friction activities like yoga classes or app subscriptions are simpler to deploy than meaningful structural workplace changes.

The Latest