Identification is the easy part: frailty scores predict mortality, and social disadvantage marks who is held up

A quiet week nationally and a moderate one for academic evidence. No new national clinical or policy development was identified this week. Twelve items across the five domains cleared review (nine primary studies, two syntheses and one preprint), most of them single-centre or from outside the UK, with one large UK dataset in outpatient antibiotic therapy. Alongside them, three international comparators and one Welsh guidance document were read for a single question: how patients’ time can be measured across a whole system. Two threads run through all of it. Frailty identification is now cheap and predictive, yet nothing this week shows it changes what happens to the patient next. And the two best-measured effects of the week, social need on recovery and homelessness on delayed discharge, are about who is held up, and both come from outside the UK.

Frailty scores identify who is at risk of dying; no study this week shows they change what happens next

A prospective observational cohort at the emergency department of Aarhus University Hospital in Denmark followed 402 patients aged 65 and over, recruited by convenience sampling between April 2023 and January 2025 Primary Study (Gregersen et al.). The Clinical Frailty Scale classed 40% as frail and the bedside Brief Multidimensional Prognostic Index classed 50% as moderately or severely frail. Both predicted 30- and 90-day mortality (5.7% and 12.9% in the cohort) with no significant difference in accuracy, but neither effectively predicted unplanned readmission. The authors caution that the absence of a significant difference between the tools is not evidence that they perform equally. It is a single-centre study and did not compare either tool against usual identification practice.

In a single-centre retrospective cohort at an urban tertiary hospital in Japan, covering 3,127 emergency visits by patients aged 65 and over between November 2025 and March 2026, 1,526 ended in hospitalisation Primary Study (Hata et al.). Adjusted odds of hospitalisation with a Clinical Frailty Scale score of 5 or more were 1.00 (95% confidence interval 0.65–1.55) in Japan Triage and Acuity Scale (JTAS) category 2, 1.54 (1.19–1.99) in category 3 and 2.01 (1.13–3.58) in categories 4 and 5 combined; the most urgent category could not be estimated. The global test of whether the association differed across categories was not significant (p=0.091), so the study provides insufficient evidence that it does. The larger estimates in lower-acuity categories are suggestive, not established, and a Japanese triage scale in one hospital limits transfer.

A multicentre prospective cohort across ten Spanish hospitals reached a parallel conclusion from liaison psychiatry: in 465 patients aged 65 and over, frailty, not psychiatric variables, was independently associated with early mortality, psychotropic polypharmacy was associated with falls, and older age was associated with a lower likelihood of specialist mental-health follow-up Primary Study (Narvaiza-Grau et al.). These are associations in an observational cohort; the authors’ recommendation to build geriatric assessment and medication review into liaison work has not been tested.

Set against the record, the three studies bound what the published identification patterns can claim. The pattern that structured identification outperforms usual practice is supported for identification and mortality risk but not for downstream outcomes: neither Danish tool predicted readmission, and the study did not test against usual practice. The provisional pattern that identification tooling can substitute for capacity is less well supported than when it was written, and nothing this week adds to it. Three studies show scoring predicts risk and none shows it releases capacity, although none set out to test capacity, so this is a weakening rather than a refutation. The Japanese result also qualifies the wider reading that frailty status organises outcomes in the acute setting: that holds for mortality, and for hospitalisation within triage categories it does not rest on a significant interaction. The Spanish cohort concerns identification and medication review rather than handover, so it does not extend the published pattern on fidelity at the point of handover.

The same gap between a process and an outcome appears in advance care planning. A scoping review of 16 studies drawn from 1,479 records, searched in December 2025, mapped how vignettes (text, video or virtual-reality scenarios) are used in advance care planning Synthesis (Wang and Ho). They improved knowledge, clarified preferences and increased planning behaviour and communication, working as cognitive anchors, low-stakes simulations and a depersonalised route into difficult conversations. The review maps the field and does not establish comparative effectiveness; the authors call for longitudinal work on whether hypothetical preferences become real-world behaviour, and the record identifies no UK-specific evidence within it. This confirms the caution in the week ending 18 September synthesis that documentation gains in advance care planning have not yet been checked against outcomes.

Social need and homelessness are measurably linked to lower recovery and longer delay, but only outside the UK so far

A prospective cohort of 164 adults receiving in-person outpatient orthopaedic physical therapy within a large urban public hospital system in Houston, Texas, between September 2021 and May 2022, found that after adjusting for baseline function, greater social-need burden was independently associated with lower function at discharge (B = −2.35, p<.001) Primary Study (Butcher et al.). Better self-rated health was associated with higher discharge function, and so was lower community-level social vulnerability. The primary model explained 61.7% of the variance in discharge function. Only 112 of the 164 completed treatment and discharge assessment, so attrition limits interpretation, and it is a single-system, orthopaedic cohort in a different health system from the NHS.

In Ontario, a population-based matched cohort of 51,377 hospital patients discharged between April 2022 and March 2024 found that patients with recent homelessness were designated as awaiting alternative care in 9.1% of cases against 2.0% of matched patients from the most materially deprived neighbourhoods (rate ratio 4.51, 95% confidence interval 4.11–4.95), and spent 4.7 days longer in that status among those designated Primary Study (Richard et al.). The rate ratio ranged from 2.39 to 7.88 across health regions, which the authors attribute to local capacity and supports. The cohort was younger than a typical frailty population (mean age about 49), and it is observational administrative data.

Two settings, two related exposures: social need linked to lower function at discharge in one study, and recent homelessness linked to longer delay in the other. The Ontario comparison group was drawn from the most materially deprived neighbourhoods, so that finding describes a gap beyond area deprivation rather than a measure of it, and the two studies do not measure the same thing. Neither is a UK study, and neither tested an intervention. The published patterns do not currently include one on social disadvantage as a modifier of recovery or delay, which is a gap in the standing record for a population with a steep deprivation gradient. The practical reading is that a whole-system measure of patients’ time should be reported by social disadvantage rather than as an average, because an average would conceal the people with the least of it. A UK anchor is what both studies lack.

Community-delivered care performs comparably in selected patients; carer readiness, rural access and selection set the limits

A cohort of 1,064 outpatient parenteral antimicrobial therapy episodes at a large UK teaching hospital between January 2023 and January 2026 found treatment failure in 9.1% of self-administered episodes, not significantly different from home-nurse or clinic-based delivery (adjusted relative risk 0.72 and 0.94 respectively) Primary Study (Durojaiye et al.). Within self-administration, older age, comorbidity, diabetes, peripheral vascular disease and central venous access were associated with failure, and the authors suggest enhanced monitoring for those groups. It partly tempers a US retrospective cohort of 202 people with osteomyelitis of the foot, 77.7% of whom had diabetes, in which self-administration carried higher rehospitalisation (odds ratio 2.10) and lower 90-day wound healing (odds ratio 0.49) Primary Study (Tarricone et al.; week ending 17 July synthesis). The populations differ, so neither overturns the other. The abstract reports no rurality analysis and nothing on digital exclusion or carer capacity, so the cohort qualifies rather than confirms the published pattern that these bind who benefits from home-based care.

A prospective non-randomised controlled pilot in Taiwan enrolled 144 homebound older adults with acute infections requiring hospital-level treatment (83 treated at home, 61 as conventional inpatients) across five hospitals and 13 community clinics between October 2022 and July 2023 Primary Study (Wang et al.). Remission or treatment completion was 86.7% against 95.1% (p=0.153), episode mortality 2.4% against 3.3%, and 10.8% of home-treated patients needed escalation to hospital. Length of stay was shorter at home (9.72 against 11.85 days, p=0.035), and home-treated patients reported significantly higher satisfaction and convenience. The model depends on a home-medical-care infrastructure with no direct NHS equivalent, and the trial was non-randomised. It extends the week ending 18 September synthesis, which found hospital-at-home evidence at scales the NHS virtual ward evidence has not matched, to a community-initiated admission route for frail patients who would otherwise reach hospital through the emergency department.

The palliative and rehabilitation items stress what limits such models. Semi-structured interviews with 29 multidisciplinary staff from 22 US nursing homes, 14 of them rural, between June 2021 and July 2022 Primary Study (Alexander et al.) found that telehealth widened access to specialist input, reduced transport burden and brought geographically distant family into goals-of-care discussions. Barriers were technical problems, workflow strain and digital-literacy gaps, together with limits for assessment, rapport and sensitive end-of-life conversations, and staff saw it as a bridge to, not a replacement for, in-person palliative care. This is a pandemic-era qualitative study in a different payment system, so its findings on access carry further than any on funding. The barriers it names sit with staff and care-home workflow, which qualifies the published pattern that digital exclusion and carer capacity are the binding constraints, since that pattern concerns patients and carers.

A scoping review of 15 studies of family caregivers in home-based cancer palliative care Synthesis (Utama et al.) found needs centred on emotional support, knowledge for physical care, psychological resilience and navigating the system. Psychoeducation, mindfulness, communication tools and digital services showed benefits for symptom-management knowledge, anxiety, coping and quality of life, with mixed or non-significant results depending on delivery method and context. It confirms the carer-capacity thread of the published pattern, within a cancer-only, non-UK boundary, and it states a cost plainly: where care moves home, preparing and supporting unpaid carers is part of the model, not an implementation detail.

On rehabilitation, an Indian single-centre randomised trial compared a 12-week home-based programme (exercise, education, nutritional and psychological support) with drug therapy alone in people with interstitial lung disease Primary Study (Joseph et al.). Six-minute walk distance and the K-BILD health-status score improved significantly more in the intervention arm but did not reach the minimum clinically important difference on average; more intervention participants reached it (relative risk 2.32, 95% confidence interval 1.47–3.68 for walk distance; 1.94, 1.28–2.96 for K-BILD), with no significant change in spirometry or anxiety and depression scores. A small trial at one non-UK centre, it shows a modest functional gain from a model that avoids travel to a facility. A [Preprint] from rural Appalachian Ohio (Jesus et al.) reports a community-based participatory codesign with seven steering-committee members and five county codesigners across two workshops, which produced a six-component concept: community-based occupational therapy, a trusted community hub, multiple delivery modes, task sharing, local health-system and community partnerships, and community-academic partnerships. It is a concept framework with no outcome data and has not been peer reviewed, so it should be treated as provisional; it extends the published pattern of converging mandates for community allied-health investment into rural delivery design but carries little weight.

The week’s comparators describe policy and bed-days; none measures patients’ time

A comparative analysis by experts of the European Observatory on Health Systems and Policies identified 17 policies or initiatives for timely discharge in Denmark, France, Germany, the Netherlands, Norway and Sweden, up to November 2023 Synthesis (Eriksen et al.). Its common principles are clarity of responsibility, effective planning and communication, resourcing of community-based capacity, the possible unintended consequences of financial penalties, and a systemic approach. It is an expert-network comparison reported here from the abstract, so it cannot be said which initiative has the strongest evaluation, and it describes policy rather than measuring effect on patients.

At four hospitals of the University of Rochester Medical Center, an analysis of 112,930 patient visits between January 2021 and July 2022 (with a detailed analysis of 33,771 Medicine and Paediatric patients) examined a programme asking clinicians to estimate and record a Medically-Ready-for-Discharge Date and a new measure of delay, “unnecessary bed days” Audit/Data (Dwyer-Matzky et al.). Later entry of the date after admission was associated with more unnecessary bed days, and a longer interval between first and last entry with fewer. These are observational associations, hospital-only, with no measure of the community capacity waiting to receive the patient.

An NHS Wales framework updated in October 2025 brings discharge to recover then assess, structured board rounds and the Red2Green approach into one whole-system hospital-flow approach, adds a Deconditioning Early Warning Indicator and criteria-led discharge, and sets out mandated and suggested flow measures Guidance (NHS Wales Performance and Improvement). Its foreword frames the aim as making time in the last 1,000 days of life as valuable as possible and minimising time away from home. It is guidance, not an evaluation: the sections read contain no outcome data, and Wales is a devolved system with its own targets.

Between them, these describe what a whole-system approach to patients’ time might look like, but none supplies an outcome expressed as patient time in value-added terms. No international comparator was found that measures it that way rather than as bed-days.

Patients’ time is lost after the score, not before it

On this week’s evidence, the loss is not at identification. A frailty score that predicts death but not readmission, and that no study shows triggering a different offer, gives the patient a label rather than time well spent. The Ontario study counts delay in days and finds it concentrated on people with recent homelessness; the Houston study finds the same disadvantage leaving people with lower function at discharge, so the time spent in rehabilitation returned less. In Taiwan, home-treated patients stayed fewer days and reported higher satisfaction and convenience, while 10.8% needed escalation to hospital, and the UK antibiotic cohort found self-administration comparable overall but not equally safe for older, multimorbid patients. The work that moves with the patient lands on carers: the caregiver review names needs for knowledge, emotional support and navigation, and the home-delivery studies as summarised here report no outcome for the carer’s time. Where the evidence says nothing about a patient’s own time, as with the Danish, Japanese and Spanish cohorts and the Welsh framework, that silence is itself the finding.

One early signal extends outpatient antibiotics beyond antibacterials

A case series of eight patients at a Dutch teaching hospital stepped down from inpatient treatment to continuous-infusion intravenous acyclovir through an outpatient programme for herpes simplex and varicella zoster neurological infections Primary Study (Manders et al.). Plasma concentrations stayed above the therapeutic threshold in all seven patients measured, no persistent complications of the infusion were observed, hospital admissions were shortened with no readmissions, and the authors estimate a saving of EUR 3,486 per patient from daily inpatient and outpatient costs. With eight patients, a non-UK cost model and no comparison group, it is an early signal only, and the authors call for larger studies.

Open questions

• Does a direct-from-community admission route shorten length of stay for frail older people in a dispersed rural population, where visit travel time is the binding constraint?
• Would systematic social-need screening at the first allied health assessment change rehabilitation intensity or completion in the most deprived areas, and with what effect on discharge function?
• For dispersed rural patients, does the access benefit of self-administered outpatient antibiotics outweigh the monitoring burden in older, comorbid patients? The question was raised in the week ending 17 July synthesis, and the new UK cohort has no rurality analysis to answer it.
• Do the published patterns need a place for social disadvantage as a modifier of recovery and delay, and should the identification patterns be re-bounded in light of three studies that predict mortality but not readmission or capacity release?