Whole-person care is being counted before it is shown to matter, and the national audit shows required follow-up not happening

A modest week for new clinical evidence, and a coherent one. Three of the five clinical domains produced items, nothing cleared the bar in Clinical Psychology or OPAT, and the most useful material came from a strand of work on how time across the whole system is counted. What joined the items was a mechanism, not a topic: delivery of a process is now being counted, and counted accurately, before anyone has shown that the process changes what happens to the patient. Where delivery is counted, the whole-person approach turns out to be rarely delivered in full. Where delivery is raised, function has not moved. And where the effect on hospital time is measured, the figures cannot separate better care from simpler patients.

Full delivery of whole-person assessment is rare, and its link to shorter stays cannot yet be separated from case mix

Three papers from US health systems measured how often the Age-Friendly Health Systems “4Ms” approach (what matters, medication, mentation, mobility) is delivered to older inpatients, and what followed. None is a UK study, and none has an NHS comparator; there is no 4Ms outcome evaluation from the UK, Canada or Australia in this week’s evidence.

The largest is a retrospective analysis of 40,411 inpatient encounters of adults aged 70 and over in an eight-hospital US health system between July 2024 and June 2025 Primary Study (Pan et al., DOI 10.1056/CAT.26.0364). Adherence was derived from the electronic record across 11 process measures, with inverse probability weighting. On average 75.3% of shift opportunities met compliance, yet 67% of encounters did not meet the first threshold, which required full compliance with each opportunity to assess, and only 3% reached the higher level of documentation. At the highest adherence tier, length of stay was 41% lower (adjusted incidence rate ratio 0.59, 95% confidence interval 0.57–0.60) and 30-day unplanned readmission was lower (adjusted odds ratio 0.60, 0.45–0.79). Higher adherence was also associated with more discharge to post-acute facilities, which the authors tie to patient complexity. The measure is documented adherence, not observed care, and the design is retrospective.

A second retrospective cross-sectional study, of 12,905 patients and 20,512 inpatient encounters (aged 65 and over, stays over 24 hours) at one urban US academic medical centre between December 2022 and May 2025, found the same shape Primary Study (Gleason et al., DOI 10.1056/CAT.26.0363). Of the encounters, 36.6% met full adherence to the assessment components; 8.4% also met complete adherence to acting on them for a quarter of the stay, 2.8% for half the stay and 0.4% for three-quarters. Greater adherence was associated with shorter stays, with reductions from 1.4 days (1.2–1.6) to 2.9 days (2.5–3.2), and with fewer discharges to post-acute facilities across the first three thresholds. Because full delivery is so rare, the benefit signal rests on a small minority of encounters.

A third item describes scaling rather than measuring. Across 26 hospitals in a US health system spanning Wisconsin and Illinois, executive governance, a system lead, site champions, electronic-record integration with dashboards and phased spread by cohort were reported to have improved the reliability of “what matters” documentation and cognition screening significantly, with delirium assessment and medication review reaching high reliability and mobility approaching its threshold Audit/Data (Ryer et al., DOI 10.1056/CAT.26.0337). The authors link the gains to lower delirium rates and shorter stays in pre-post analyses; the abstract does not give the magnitudes, and there is no comparator.

None of these designs can say whether fuller documentation marks better care or marks patients who are less complex and stay a shorter time. That is carried below as an open question. Against the record, these are the first data that point, in direction, against the published pattern that process measures move while patient-outcome measures stay flat: higher documented adherence tracked shorter stays and, in one system, fewer readmissions. They do not overturn it. They are retrospective, they measure documentation, and last week’s synthesis had already bounded the “flat outcomes” reading using the Cochrane review of comprehensive geriatric assessment. The same datasets confirm the published pattern on measurement and coordination fidelity, because full delivery is rare, and the scale-up report extends it weakly. The pattern to watch is the first: counter-evidence in direction for two weeks running, against a supporting set that is the weakest in the record.

Raising how often a process happens has not been shown to change function

Three further studies this week sit on the other side of the argument, and all are of low or moderate weight.

An Australian-team repeated cross-sectional implementation study of self-directed therapy in inpatient rehabilitation audited 278 patients and 61 staff at baseline, after implementation and at six months across physiotherapy, occupational therapy, speech pathology and music therapy Audit/Data (Peiris et al., PMID 42904788). Independent practice programmes rose significantly for occupational therapy and music therapy, rose early for speech pathology and stayed stable for physiotherapy; 88% of staff adopted the approach and fidelity to its four pillars improved. Functional independence at discharge did not change (p = 0.681), and no related adverse events were reported. Self-directed practice can multiply therapy time without more staff, but this uncontrolled evaluation found no functional gain, so a benefit to the patient is not shown.

A quality-improvement project across eight internal medicine wards in Singapore, over April to October 2025 and three improvement cycles, identified patients by diagnosis code, engaged doctors and nurses, sent weekly digital reminders and built advance care planning into multidisciplinary meetings Audit/Data (Tang et al., PMID 42901706). In life-limiting illness, completed plans rose by 285% (0.7 to 2.0 a week), with conversion from 24.7% to 36.0%; across all medical inpatients they rose by 440% (1.4 to 6.2 a week). The comparison is with a prior-year baseline, with no control. It counts plans completed, not their quality, their concordance with what the person wanted, or where they died, and the authors themselves flag that the manual process is hard to sustain.

A single-blind randomised trial by an Australian team of a multimodal exercise programme for 126 adults with mild cognitive impairment (median age 76) compared six months of walking, Balance Yourself and Clock Yourself, plus six months of maintenance, with control Primary Study (Burton et al., PMID 42904808). It did not reach the sample size needed for either primary outcome, falls or balance, so the lack of a significant between-group difference at 6 and 12 months is inconclusive and not negative. The 59% of intervention participants who adhered had significantly better balance than low adherers and controls, and fell less, although not significantly; the authors describe the adherer analyses as exploratory. It is a caution for unsupervised community falls programmes, and it is relevant to the published pattern on remote care in dispersed populations only by analogy.

Taken together, My Therapy and the advance care planning project confirm the shape of the process-versus-outcome pattern: process measures rose and outcomes were not shown to change. They confirm its shape; they do not test it, because both are uncontrolled.

Identification and exercise both carry signal, with uncertainty that limits what they can justify

Two meta-analyses on frailty completed the domain. A pooled analysis of the electronic frailty index (12 studies, 13 cohorts) found a graded mortality association (hazard ratios of 1.81, 2.95 and 4.41 for mild, moderate and severe frailty against non-frail) with substantial heterogeneity Meta-analysis (Du et al., PMID 42890831). Its authors say this is not direct evidence of predictive performance. It qualifies the published pattern on structured frailty identification: the identification side holds, but the controlled evaluation showing that identification changes readmission, time at home or place of death has still not appeared. The abstract does not say how many included studies were UK-based.

A meta-analysis of exercise modalities of ten randomised trials and 1,361 participants favoured exercise on frailty scores (standardised mean difference −0.57), with I² of 88% and a prediction interval of −1.73 to 0.57 that crosses the null Meta-analysis (Li et al., PMID 42891303). Resistance exercise had the clearest pooled signal (−0.78), but superiority over walking or Tai Chi is not established because no study compared them directly. This qualifies the published pattern on resistance exercise and remote delivery without meeting its condition for lifting the qualification, which is a patient-centred trial of remote delivery.

Neither paper shows that identifying frailty, or exercising, changes what happens to the patient next. The next section is what the national audit says happens after identification.

A national audit finds the follow-up required after identification is not happening

The National Audit Office’s report on primary and community support for people living with frailty in England, published in December 2025, is carried here because its figures set the baseline for any whole-system measure Audit/Data (National Audit Office, HC 1518, 5 December 2025). It is covered in full in its own signal. Of 226,000 people diagnosed with severe frailty in 2024-25, 16% had the medication review the GP contract requires, 18% a falls risk assessment and 29% had consented to an enriched summary care record, and the audit found no single joined-up national strategy. It corroborates the published pattern on community settings being structurally under-equipped, now verified against the report. Its dataset has no deprivation or rurality breakdown, which is itself the point: the gap cannot be located.

Two further national and international documents add context at lower weight. Healthcare Improvement Scotland’s October 2025 report on hospital and community services for older people with frailty lists, among its priorities, dedicated frailty teams in every hospital with an emergency department, and states that its Focus on Frailty programme was in 2025-2026 supporting 15 of 28 hospitals that accept emergency admissions Audit/Data (Healthcare Improvement Scotland, October 2025). It reports no outcome data. An Australian consultancy series argues that hospital delays for older patients trace back to aged care capacity and calls for coordinated reform across health and aged care Synthesis (FTI Consulting, three-part series, 24 February 2026). It is advisory, not an evaluation, and it complements the interface-delay measure proposed in last week’s evidence.

Where access and continuity fail, the studies place the cause in service design

Four primary studies, none a UK frailty study, agree on where the failure sits.

In US Medicare claims data for 2022, enrolees with Alzheimer’s disease and related dementias in Medicare Advantage (47,830) received 25% fewer home-based medical care visits than those in Traditional Medicare (49,194) (incidence rate ratio 0.75, 95% confidence interval 0.69–0.81), arising from lower user incidence and lower visit intensity Primary Study (Yang et al., PMID 42894207). The gap was larger for Black (47%), Hispanic (67%) and dual Medicaid-eligible (42%) enrolees, and telemedicine use was similar. It is a US payer finding and does not transfer as such. The transferable point is that home-based care reaches fewer people when access is shaped by plan design, which extends last week’s reading on access depending on design, and the published pattern that disadvantage lengthens time in the system, towards access.

A national qualitative study led by UK-based authors, drawing on experience-based co-design, interviewed and held focus groups with 121 bereaved people from 13 ethnically diverse backgrounds, interviewed 23 providers across 12 services, and held co-design workshops Primary Study (Selman et al., PMID 42902353). Barriers arose across the access pathway but were concentrated mainly within bereavement support organisations and the wider system: poor signposting and referral, long waiting lists, limited staff diversity, language barriers, weak organisational commitment to equity, precarious funding and white-centred, individualised models of grief support. It concerns bereavement, not preferred place of death, and bereavement is the least-measured end of the palliative pathway. The topic is new ground here; the access-by-design finding extends an existing theme.

A French qualitative study shadowed 15 hospital nurses for about 92 hours in four units of a university hospital and held one focus group of seven self-employed community nurses Primary Study (Montaigne et al., PMID 42905017). Discharge coordination was fragmented, nurse-to-nurse communication rare and wound-specific handover thin; hospital and community workflows diverged, with fee-for-service time pressure in the community; patient involvement was limited; and continuity was kept going by compensatory work. The authors conclude that continuity depends less on nursing expertise than on the organisational conditions that let it be used. It is small, single-country, nursing rather than allied health, and set in a fee-for-service system; it confirms the published pattern on measurement and coordination fidelity at the point of handover, and fits the community under-equipment pattern.

A very small interview study of six people with chronic post-stroke aphasia and five significant others found that accessing community healthcare was complex and distressing, that a family support person played an exceptional and underacknowledged role, and that co-designed technology could help only if providers were supported at system level to adapt how they communicate Primary Study (Shorland et al., PMID 42904560). It is Australian and low weight, but it extends the published pattern on digital exclusion and carer capacity to community speech and language need: here the family is the service.

What the week says about whether patients’ time is being well spent: mostly that nobody is counting it

The one direct link to time in hospital is the 4Ms association with shorter stays, and its authors cannot separate care from case mix. The other items measure throughput, plans completed and therapy practice rates, not the time a person spends on something that does them good. The acute end of the pathway is being scaled and audited at encounter level, while the community end is described only qualitatively: the handover that depended on individual effort, the bereaved people who met long waits and poor signposting, the family who carry community speech and language need, the home-based care that reached fewer people under one plan design. Those are costs in the person’s time, and in carers’ time, that nobody counted.

The patient’s own account is thin. The bereavement and aphasia studies are the only ones this week that report experience directly, and both found effort and distress at the point of access. The advance care planning project did not report whether the plans matched what people wanted, and the trial of the exercise programme found balance better only among those who kept going with it, a warning for dispersed populations in which unsupervised programmes are hardest to sustain. No system anywhere in this week’s evidence counts patient time in value-added terms, which is now the fourth week running with no comparator found. A whole-system time measure for frailty should record whether identification, assessment and handover happened, not only how long a stay lasted, and should be stratified by deprivation and rurality, because the national dataset the audit relies on cannot do so.

The national frameworks that would set direction remain overdue

The government committed in May 2026 to publish an interim product for the dementia and frailty framework in September, with the full framework by the end of 2026 Guidance (Department of Health and Social Care, 22 May 2026). A targeted search this week found no publication of it. That is secondary-source level only and not confirmation that it has not appeared, but it follows the forecast in the October monthly review that it was overdue, and it is now about three weeks past its stated month, undelivered and undated. The consultation on the allied health professions framework for 2027 to 2032, noted in the September monthly review as closing on 18 September, has published no outcome that was found five weeks on, and a professional body’s notice says the final framework is due in 2027 Guidance (Society of Radiographers, 11 August 2026). National direction is for now waiting, neither aligned nor misaligned with the evidence, so the useful work is to count delivery of identification, assessment and handover so the frameworks can be read against it when they land.

Open questions

• Does documented adherence to the 4Ms mark better whole-person care, or does it track patients who are less complex and stay a shorter time?
• Does NHS home-based medical care reach housebound and rural people with dementia in proportion to need, and is that measured?
• Does anything count patient time in value-added terms? The fourth week running with no comparator found.
• When will the dementia and frailty interim product and the outcome of the allied health professions framework consultation appear, and will they rest on evidence from settings such as areas of high deprivation and rural or dispersed populations?