FluSurvey works by recruiting members of the public who agree to complete weekly symptom surveys throughout the respiratory illness season. Rather than waiting for people to seek care, the system asks them directly whether they have experienced fever, cough, and other symptoms consistent with influenza-like illness, or ILI. This approach, sometimes called participatory or crowd-sourced surveillance, has gained prominence since the COVID-19 pandemic demonstrated how quickly digital reporting tools can complement traditional clinical data streams. The new study set out to characterise the wider impact of ILI among participants and to assess how closely their reports correlate with established influenza surveillance systems, including GP consultation rates, hospital admissions, and laboratory-confirmed PCR test positivity.

The analysis included data from 3,057 participants across the two winter seasons. In 2023-24, 2,540 people took part, roughly 63 percent of them female, with a mean age of 60 years. The following season drew 2,273 participants, 64 percent female, with a mean age of 61. The researchers identified 1,868 ILI episodes over the two winters and examined what happened during each one: whether the person sought healthcare, took medication, missed work or school, used a test, or found their daily activities curtailed.

The headline finding is striking in its consistency: only 14 percent of ILI episodes resulted in contact with healthcare services, most frequently a visit to the GP. In other words, roughly six out of seven episodes of flu-like illness were managed entirely outside the formal healthcare system. This has important implications for how surveillance data are interpreted. Traditional systems that rely on GP consultations see only the tip of the iceberg—the small fraction of illness severe enough, or concerning enough, to prompt a clinical visit. FluSurvey, by contrast, captures the full spectrum of community illness, including the mild and moderate cases that never appear in clinical records.

Yet the fact that most episodes avoided the doctor’s office does not mean they were trivial. A large proportion of episodes—89 percent—involved medication use, suggesting that people were actively treating their symptoms with over-the-counter remedies or prescription drugs. Three-quarters of episodes, 75 percent, were reported to have an impact on daily life, and nearly half, 47 percent, resulted in missed school or work. Absenteeism on this scale carries substantial economic and social costs, from lost productivity to disrupted childcare, that are largely invisible to health systems focused on clinical encounters.

One notable shift between the two seasons concerned testing. The frequency with which participants used tests for respiratory illness declined notably in 2024-25 compared with 2023-24. This pattern likely reflects changing habits as the acute phase of the COVID-19 pandemic recedes further into the past, with fewer people reaching for lateral flow tests or seeking PCR confirmation when they fall ill. The decline complicates efforts to track infection rates through testing data alone and underscores the value of symptom-based surveillance that does not depend on test availability or willingness to test.

The second major aim of the study was to determine whether FluSurvey’s volunteer-reported ILI rates move in step with the official signals that inform public health action. The researchers applied established methodologies, including omitting each participant’s first report—a step designed to reduce bias from people joining the survey precisely because they are ill—and weighting the data to match the age and sex structure of England. They then computed cross-correlations between weekly FluSurvey ILI rates and three national surveillance indicators: GP ILI consultations, influenza hospital admissions, and influenza PCR test positivity, at time lags ranging from two weeks ahead to two weeks behind.

The correlations were strong across the board, and their timing tells a coherent story about the natural history of an influenza season. FluSurvey ILI rates predominantly led GP ILI consultations, with a maximum correlation coefficient of 0.73. This makes intuitive sense: people experience symptoms at home before some of them eventually decide to see a doctor, so community symptom reporting provides an early warning signal of rising clinical activity. FluSurvey rates coincided most closely with influenza hospital admissions, reaching a maximum correlation of 0.88, suggesting that peak community illness aligns tightly with the period of greatest pressure on hospitals. Meanwhile, FluSurvey ILI lagged behind influenza PCR test positivity, with a maximum correlation of 0.88, which fits the expectation that laboratory-confirmed cases among tested patients rise before broader community symptoms peak—or that testing is concentrated early in illness episodes among those who choose to test.

For public health officials, these timing relationships matter. A surveillance signal that leads GP consultations by even a week or two can buy valuable time for interventions such as vaccination campaigns, antiviral distribution, and hospital surge planning. The strong correlation with hospital admissions indicates that FluSurvey is not merely detecting trivial sniffles; the aggregate pattern of community-reported symptoms tracks the severe end of the epidemic curve remarkably well. And the fact that the signal holds up across two distinct winter seasons, with different circulating virus profiles and participant cohorts, lends confidence that the findings are robust rather than a one-off artefact.

The study also highlights the complementary nature of participatory surveillance. It does not replace laboratory testing, which provides virological specificity, or clinical reporting, which captures healthcare burden. Instead, it adds a community-level layer that contextualises the clinical data: how much illness is out there that never reaches a clinic, how it affects work and daily life, and how people manage it on their own. As the authors conclude, the majority of ILI reported to FluSurvey does not involve contact with healthcare services, but it carries wider impacts on daily life, and FluSurvey ILI corresponds well with other national influenza surveillance while providing broader context on community illness. In an era when respiratory virus threats demand rapid, granular situational awareness, systems that enlist the public as sensors may prove an increasingly essential part of the surveillance toolkit.

Subject of Research: Participatory surveillance of influenza-like illness in the UK using FluSurvey over the 2023-24 and 2024-25 winter seasons

Article Title: Monitoring influenza-like symptoms in the UK through participatory surveillance: Insights from FluSurvey over two winter seasons (2023-24 and 2024-25)

Article References: Green, R. E., Mellor, J., Rawlinson, C., Waller, E., Abdul Aziz, N., Watson, C. H., & Dabrera, G. (2026). Monitoring influenza-like symptoms in the UK through participatory surveillance: Insights from FluSurvey over two winter seasons (2023-24 and 2024-25). PLOS Digital Health, 5(10), e0001714. https://doi.org/10.1371/journal.pdig.0001714

Image Credits: AI Generated

DOI: 10.1371/journal.pdig.0001714

Keywords: influenza, FluSurvey, participatory surveillance, influenza-like illness, UK Health Security Agency, public health, epidemiology, respiratory viruses, GP consultations, hospital admissions, PCR test positivity, winter seasons

The research team, led by Xiaoxiao Song of the Second Affiliated Hospital of Guangxi University of Chinese Medicine together with colleagues from the Nanning Hospital of Traditional Chinese Medicine and Guangxi Medical University, analyzed 2,382 hospital admissions for acute ischemic stroke recorded between July 1, 2017, and June 30, 2020. The records came from a single tertiary hospital in Nanning, a city whose humid subtropical climate makes it a useful natural laboratory for studying how heat and cold shape cerebrovascular risk. On an average day, the hospital admitted just over two stroke patients, a modest daily count that nevertheless accumulates into a dataset rich enough to detect subtle weather-related patterns.

Methodologically, the study leans on two statistical workhorses of environmental epidemiology. The first is the quasi-Poisson generalized linear model, which handles daily count data such as hospital admissions and accommodates the overdispersion—variance exceeding the mean—that is typical of such series. The second is the distributed lag nonlinear model, or DLNM, a framework that allows researchers to estimate simultaneously how an exposure like temperature affects risk in a nonlinear way and how that effect is spread across time. Rather than asking only whether a hot day produces more strokes on that same day, the DLNM can trace the risk across a window of lag days, here spanning zero to seven days after exposure, and can compute cumulative effects over any sub-window within that range.

The team anchored its comparisons to the median daily mean temperature of 23.30 degrees Celsius, treating this as the reference point against which hotter and colder days were judged. When the mean temperature rose to 29.60 degrees Celsius—a level well within Nanning’s summer routine—the risk of an ischemic stroke admission increased substantially. The relative risk reached 1.347, with a 95 percent confidence interval of 1.058 to 1.714, when cumulative effects over lag days zero to one were considered, and climbed to 1.396 (95 percent CI: 1.085 to 1.797) over lag days zero to two. In practical terms, on such hot days the hospital could expect roughly 35 to 40 percent more ischemic stroke admissions than on a typical day at the median temperature.

Just as striking as the size of the effect is its timing. The association between high temperature and stroke admissions attenuated as the lag window lengthened, fading over longer cumulative periods. This pattern suggests that heat acts as a near-immediate trigger rather than a slow-burning risk factor: the physiological damage it inflicts on vulnerable patients appears to translate into arterial blockages within hours to a couple of days. That short latency has real operational implications, because it means emergency services and stroke units can anticipate surges in demand almost in real time as heat waves roll through a city, rather than bracing for a delayed wave of cases.

The biological plausibility of a rapid heat effect is well supported by prior research. Heat stress promotes dehydration, which hemoconcentrates the blood and increases viscosity, tilting the hemostatic balance toward clot formation. Sweating-driven fluid loss also reduces plasma volume, and studies of heat-stressed humans have documented measurable changes in coagulation responses. Heat further strains the cardiovascular system by increasing cardiac output and cutaneous blood flow to shed excess warmth, while aging blood vessels lose some of their thermoregulatory reflex capacity. Endothelial function, the ability of blood vessel linings to dilate and maintain smooth flow, is itself temperature-sensitive. Inflammatory and coagulation markers rise in hot conditions, and blood pressure—normally lower in warm weather—can fluctuate in ways that destabilize existing atherosclerotic plaques. Any of these pathways could, in a patient with narrowed cerebral arteries, tip the balance toward an occlusive event within a single hot day.

Cold told a different and less conclusive story. When the mean temperature dropped to 10.50 degrees Celsius, the same-day risk estimate was lower than for heat, and the cumulative associations over longer lag windows were inconsistent. The authors are careful here: they note that the low-temperature findings require cautious interpretation. This asymmetry is not unusual in subtropical settings, where winters are mild and cold extremes are relatively rare, limiting the statistical power to detect cold effects. It also contrasts with studies from temperate and northern Chinese cities, such as Beijing and Guangzhou, where distributed lag analyses have often found robust cold-related increases in stroke admissions with longer lag times. The divergence underscores a central theme in climate-health research: temperature effects are regionally heterogeneous, shaped by local climate norms, housing, air conditioning prevalence, and the physiological adaptation of the population.

Recognizing how easily time-series findings can be artifacts of modeling choices, the researchers ran an extensive battery of sensitivity analyses. They varied how the long-term trend and seasonality were adjusted in the models, added air pollutant concentrations as covariates to rule out confounding by poor air quality, changed the maximum lag period, excluded the year 2020—a year distorted by the COVID-19 pandemic’s disruption of hospital care—and adjusted for the Spring Festival window, during which hospital utilization patterns in China shift dramatically. The main findings held up across these checks, lending confidence that the heat-stroke link is not a statistical mirage. The robustness of the hot-temperature effect, contrasted with the fragility of the cold-temperature signal, reinforces the study’s central conclusion.

The stakes of this line of research are rising with the thermometer. Ischemic stroke imposes an enormous global burden, and analyses of the Global Burden of Disease data show it remains a leading cause of death and long-term disability worldwide, with China bearing a particularly heavy share. The World Stroke Organization has issued a scientific statement on stroke and climate change, warning that warming temperatures will translate into additional cerebrovascular events. Meanwhile, studies using hourly heat exposure data have begun to show that even short bursts of high temperature can precipitate ischemic stroke, and occupational health research documents widespread heat stress in working populations. Against that backdrop, a study pinpointing a one-to-two-day window of elevated risk gives public health authorities something actionable: heat-health warning systems can be tuned not just to warn the general population but to alert hospitals, ambulance dispatch, and thrombolysis-capable stroke centers to prepare for demand spikes within 48 hours of extreme heat.

The study’s limitations are those inherent to its design. It draws on admissions from a single tertiary hospital in one city over three years, so the results may not generalize to regions with different climates or to populations with different demographics and healthcare access. Hospital admissions capture only patients who reach care, and prehospital delay—known to be common in stroke—could interact with weather in ways the data cannot reveal. The authors also emphasize that the cold-temperature association, being less consistent, should not be overinterpreted. Still, the core message is clear and increasingly hard to ignore: in a warming world, the days immediately following a heat spike are precisely when vulnerable brains are most at risk, and health systems that plan for that window may save not just comfort but lives.

Subject of Research: The short-term association between ambient mean temperature and hospital admissions for acute ischemic stroke in a subtropical Chinese city.

Article Title: Association between mean temperature and hospital admissions for acute ischemic stroke: a time-series study

Article References: Association between mean temperature and hospital admissions for acute ischemic stroke: a time-series study. (n.d.). https://doi.org/10.1007/s11869-026-02102-5

Image Credits: AI Generated

DOI: 10.1007/s11869-026-02102-5

Keywords: acute ischemic stroke, ambient temperature, heat, hospital admissions, distributed lag nonlinear model, time-series analysis, Nanning, China, climate change, cerebrovascular disease, public health, epidemiology

The research, published in the International Journal of Disaster Risk Science, was led by Congkai Hong of Tsinghua University together with colleagues from the Tianjin Institute of Meteorological Science, the China Meteorological Administration, the Beijing Institute of Technology, and Beihang University. The team examined the period from January 2021 to September 2022, during which four cold-spell-related stroke risk warnings were issued in Tianjin. Their analysis found that during warning periods, stroke hospital admissions fell by an average of 6.92 percent. Scaled to the entire city, that translates into roughly 1,593 avoided stroke hospitalizations over the study window, a figure the authors link to more than USD 2.66 million in savings for government medical insurance funds and USD 849.9 thousand in reduced out-of-pocket costs for residents.

What distinguishes this evaluation from earlier attempts to judge the value of weather-based health alerts is its statistical architecture. Previous studies often relied on simple comparisons of disease rates before and after warnings, or on regression models that captured only the average effect of temperature. Both approaches stumble over the fact that the relationship between meteorological conditions and health outcomes is neither linear nor immediate. Cold can elevate stroke risk for days after the temperature drop, and the dose-response curve bends in complicated ways across the thermal spectrum. To handle this, the team employed a distributed lag nonlinear model, or DLNM, a framework originally developed by epidemiologists to disentangle exactly these exposure-lag-response relationships. The model builds a cross-basis function that simultaneously characterizes how temperature affects admissions and how that effect is distributed over a 21-day lag window, using natural cubic splines with knots placed at the 75th and 90th percentiles of the temperature distribution.

Into this temperature-lag structure, the researchers inserted a binary warning variable, allowing the regression to separate the effect of the alert itself from the underlying weather. Controls included daily average relative humidity and air pressure, adjusted with splines, along with terms for the sequential day number within the year, interannual variation, the weekday, public holidays, and a dummy variable capturing the peaks of COVID-19 control measures, which visibly depressed hospital admissions during the study period. The estimated warning coefficient was negative and statistically significant at the 1 percent level, and the corresponding 6.92 percent reduction in admissions survived an unusually thorough battery of robustness checks.

Those checks matter because the study had only four warning events to work with, a small sample by any standard. The team re-ran the model replacing average temperature with maximum and minimum values, altering lag windows of 7, 14, and 30 days, adjusting spline degrees of freedom, and swapping in alternative knot placements. They also iteratively excluded each single warning event, and every combination of two and three events, to confirm the result was not driven by one outlier alert. The estimated effect varied by an average of only 11.2 percent across these specifications and retained significance throughout the single-event exclusions. Power analysis returned a value of 0.9, above the conventional 0.8 threshold, and 1,000 bootstrap resamples produced a 95 percent confidence interval for the warning coefficient that excluded zero. The researchers also tested for a so-called hospital admissions displacement effect, in which warnings merely postpone admissions rather than preventing them, and found no evidence of such harvesting among stroke patients, plausibly because stroke is acute enough that patients cannot delay care for long.

Having established effectiveness, the study moved into economic territory where earlier evaluations had gone shallow. Most prior work on heat and cold health warning systems counted only direct healthcare savings, such as avoided emergency visits and hospitalizations. That accounting ignores the enormous downstream costs of stroke, particularly the labor lost when survivors live with disability or die prematurely. The Tianjin team combined the human capital method with the disability-adjusted life year, or DALY, framework, dividing the population into eight age groups and calculating how many productive working years the avoided admissions preserved. Each avoided stroke hospitalization represents not only a medical expense dodged but a potential loss of earnings, workplace continuity, and household income averted. By this broader measure, the four warnings generated an additional USD 1.61 million in economic benefit through avoided labor losses alone.

The authors then pushed the analysis one step further, asking what happens when those saved expenses ripple through the wider economy. Using Tianjin’s 2017 input-output table covering 42 industrial sectors, they modeled how government insurance savings and resident out-of-pocket savings, if redirected to other forms of consumption, would propagate through supply chains via the Leontief model, while avoided labor losses entered the Ghosh model as restored employee compensation. Under a baseline redistribution scenario, the combined effects suggested a potential increase in economic gross output of USD 5.76 million across the warning periods. An optimized scenario, in which saved funds are steered toward sectors with stronger backward linkage coefficients, meaning industries whose demand stimulates production more powerfully throughout the economy, raised the plausible output gain to as much as USD 1.32 million from out-of-pocket savings alone, a sum comparable to the annual profits of Tianjin’s entire culture and arts industry.

The financial case for the warning system itself is unusually favorable. The initial investment by the Tianjin Environmental Meteorological Center in developing the system was approximately USD 607.1 thousand. Against the combined medical insurance and out-of-pocket savings, the benefit-cost ratio reached 4.39, a figure the authors note is comparable to annual benefit-cost ratios reported for tobacco cessation programs targeting cardiovascular disease in Massachusetts. Counting avoided labor losses separately yielded a benefit-cost ratio of 2.66, in the range reported for flood early warning systems in Europe. The high returns reflect a basic asymmetry: information is cheap to produce and distribute, while stroke treatment and long-term care are extraordinarily expensive, so even a modest reduction in admissions translates into substantial savings.

The mechanisms behind the reduction are straightforward to describe even if they are hard to observe directly. Cold-weather stroke warnings likely work through two channels. First, alerts prompt high-risk individuals, particularly the elderly and those with hypertension or other chronic conditions, to limit outdoor exposure, bundle up, and adhere strictly to their medication regimens during dangerous conditions. Second, warnings give healthcare providers a window to act proactively, intensifying patient education and preparing clinical resources before the surge in cases that typically follows a cold snap. Similar logic underpins heat-health warning systems now operating in many countries, but stroke adds a distinctive challenge because its onset is rapid and its outcomes, from paralysis to death, generate some of the largest disability burdens of any disease worldwide, ranking fourth among global risk factors for disability-adjusted life years in 2021.

The study’s authors are candid about its limits. The admission data covered seven major hospitals, which together handle roughly half of Tianjin’s stroke cases but cannot resolve differences between districts and communities. Only four warning events fell within the study window, constrained by the system’s November 2021 launch and by pandemic disruptions, and the analysis relied on aggregate daily counts rather than individual patient records, which may understate protective effects for the most vulnerable groups. The input-output modeling rests on static assumptions and hypothetical redistribution, so the macroeconomic figures should be read as theoretical potential rather than realized gains. Even so, the framework is designed to travel. The authors argue it offers a replicable blueprint for evaluating meteorological health warning systems across regions and diseases, from asthma to heat-related illness, and they urge governments to count labor losses and cross-sector effects, not just hospital bills, when deciding whether to invest. As climate change intensifies extreme weather and the United Nations pushes its Early Warnings for All initiative toward full global coverage by 2027, the Tianjin experience suggests that a modest investment in telling people when the weather threatens their health can return its cost several times over, in hospitals, households, and the broader economy alike.

Subject of Research: Evaluation of the effectiveness and economic benefits of a meteorological stroke risk warning system in Tianjin, China

Article Title: Evaluating Effectiveness and Economic Benefits of Meteorological Health Warning Systems: A Case Study of Stroke Warnings in Tianjin, China

Article References: Hong, C., Han, S., Shang, J., Zhang, S., Zhao, M., Xia, C., Hao, T., Yao, Q., & Cai, W. (2026). Evaluating Effectiveness and Economic Benefits of Meteorological Health Warning Systems: A Case Study of Stroke Warnings in Tianjin, China. International Journal of Disaster Risk Science. https://doi.org/10.1007/s13753-026-00768-0

Image Credits: AI Generated

DOI: 10.1007/s13753-026-00768-0

Keywords: meteorological health warning system, stroke, early warning, cold spell, hospital admissions, distributed lag nonlinear model, economic benefits, DALY, input-output analysis, climate change adaptation, Tianjin, public health

The research, published in European Geriatric Medicine by a team led by Sarah J. Richardson of Newcastle University and the NIHR Newcastle Biomedical Research Centre, drew on routinely collected electronic health record data from Newcastle upon Tyne Hospitals NHS Foundation Trust, one of the largest hospital trusts in the United Kingdom, with roughly 1,700 inpatient beds across two sites. The analysis covered all index admissions of people aged 75 or older between 1 April 2021 and 31 October 2021, a total of 6,068 admissions. The work forms part of a wider programme by the ADMISSION Research Collaborative, an initiative funded to tackle multimorbidity at scale using large-scale clinical data.

Delirium itself is an acute, fluctuating disturbance of attention, arousal and cognition that predominantly strikes older people in hospital. It is distressing for patients and families, strongly associated with poor outcomes including higher mortality, and extraordinarily costly: a recent policy analysis estimated that delirium cost the NHS £10.8 billion in 2022 alone. National guidance recommends that everyone aged 65 and over be screened for delirium on admission to hospital, and higher screening rates are strongly linked to higher rates of diagnosis. Yet the largest point prevalence study of delirium detection in the UK found that only 27.3 percent of older inpatients were screened. Patients in whom delirium is missed fare measurably worse, experiencing higher mortality than those in whom the condition is recognised.

The Newcastle team set out to ask whether a patient’s burden of multiple long-term conditions, often abbreviated MLTC and also known as multimorbidity, influenced whether delirium screening was documented. Multimorbidity, defined as the coexistence of two or more chronic conditions in one individual, is rising in prevalence worldwide and has been identified as a global research priority. It is a recognised risk factor for developing delirium and worsens outcomes after delirium occurs, but no previous study had examined whether it shapes the screening process itself. To operationalise the concept, the researchers used the ADMISSION collaborative’s standardised framework of 60 long-term conditions, each defined by lists of ICD-10 diagnostic codes drawn from the index admission and up to ten years of prior records. This approach deliberately treats multimorbidity as a continuum, analysing the number of conditions rather than a simple present-or-absent cut-off, in line with earlier work showing that outcomes scale with condition count.

Delirium screening at the trust is implemented through a non-mandatory, clinician-initiated electronic form embedded in the electronic health record. The form asks whether the patient has a clinical diagnosis of delirium and instructs clinicians to use the 4AT assessment tool alongside clinical judgement. Local policy mandates screening for everyone aged 75 and over, whether admitted electively or as an emergency. For the purposes of the study, screening documentation was deemed complete if the form existed at all; the researchers did not assess how thoroughly the form was filled in, the outcome of screening, or whether the screening itself was accurate.

The headline numbers were sobering. Delirium screening documentation was completed in only 42.7 percent of admissions, 2,594 of 6,068. The gap between admission types was stark: 48.6 percent of emergency admissions had documented screening, compared with just 28.2 percent of elective admissions. The cohort had a median age of 82 years, and 52 percent were women. Multimorbidity was highly prevalent, as expected in a hospitalised older population, and patients admitted electively carried fewer long-term conditions on average than those admitted as emergencies. Older age, emergency admission, female sex, living in the most deprived neighbourhoods as measured by the Index of Multiple Deprivation, and increasing numbers of long-term conditions were all associated with higher odds of screening documentation, and similar patterns held when elective and emergency admissions were analysed separately.

The finding that a greater burden of chronic disease predicts a greater likelihood of screening has not been reported before, but it aligns with earlier evidence that multimorbidity increases the chance of delirium symptoms being noted in nursing documentation. The authors suggest a plausible mechanism: multimorbidity correlates with greater delirium severity, which may make symptoms more obvious to clinical teams and prompt the completion of screening. By contrast, patients with fewer long-term conditions, and those arriving for planned procedures, may be implicitly judged by staff to be fitter and at lower risk of delirium. This informal risk stratification, the researchers argue, may be quietly depressing screening rates in precisely the groups where it should not. That matters because post-operative delirium rates after elective surgery are high, and screening in elective admissions is vital even when patients appear robust on arrival.

How do these figures compare with the wider landscape? The emergency admission screening rate of 49 percent matches rates reported from a large teaching hospital in Salford, England, and sits well above the 27.3 percent recorded in the UK-wide point prevalence study of 45 hospitals. But it falls far short of the 77 percent achieved in emergency admissions in Edinburgh, Scotland, demonstrating that high screening rates are achievable and that wide regional variation persists. The authors note that consistently high screening is the exception rather than the rule, which underscores the importance of understanding the factors that hold screening back at lower-performing sites, including the subtle patient-level biases documented in this study.

The study has notable strengths, including the large cohort size and the inclusion of both elective and emergency admissions, a group frequently under-represented in analyses of routine data. The use of agreed, standardised definitions of multimorbidity designed specifically for hospital data is another methodological advance. Limitations remain, however. The team could only examine the completion of screening documentation, not screening itself or its accuracy, because of constraints in the available data. Local policy screening only those aged 75 and over, rather than 65 as national guidance recommends, may limit generalisability, as may the fact that the data window fell during the COVID-19 pandemic. The analysis was deliberately descriptive, focused on clinician behaviour, and future work with multivariable modelling will need to address confounding, including measures of frailty, though the authors caution that frailty definitions built for routine data overlap heavily with multimorbidity counts and require careful handling.

The practical message is clear and potentially actionable for hospitals everywhere. Delirium screening needs to improve for all older inpatients, because missing the diagnosis carries serious consequences, but improvement efforts should explicitly target the groups identified here: people with fewer long-term conditions and those admitted electively. Embedding screening as a default, mandatory step in the electronic record, decoupled from staff impressions of who looks like a delirium risk, could help close the gap. As populations age and multimorbidity becomes the norm rather than the exception, ensuring that every older patient, whatever their apparent fitness, receives a simple, validated cognitive check on admission may be one of the highest-value interventions available to hospital systems, with the potential to avert deaths, shorten stays and spare families the distress of an overlooked, treatable brain failure.

Subject of Research: The influence of multiple long-term conditions on delirium screening documentation in hospitalised older adults, analysed using routinely collected electronic health record data.

Article Title: Understanding the impact of Multiple Long-Term Conditions (MLTC) on the completion of delirium screening documentation: analysis using routinely collected clinical data

Article References: Richardson, S. J., Bunn, J. G., Evison, F., Gallier, S., Harris, S., le Roux, P., Plummer, C., Prendergast, E., Sapey, E., Sayer, A. A., Singer, M., Shahmandi, M., Witham, M. D., & on behalf of the ADMISSION Collaborative (2026). Understanding the impact of Multiple Long-Term Conditions (MLTC) on the completion of delirium screening documentation: analysis using routinely collected clinical data. European Geriatric Medicine. https://doi.org/10.1007/s41999-026-01594-9

Image Credits: AI Generated

DOI: 10.1007/s41999-026-01594-9

Keywords: delirium, screening, multimorbidity, multiple long-term conditions, older people, electronic health records, hospital admissions, 4AT, geriatrics, elective admission, emergency admission, routinely collected data