Verification and consensus mechanisms move practice, while advance care planning's documentation gains still lack an outcomes check

This week’s evidence points the same way at three different scales: formal verification and consensus mechanisms, not new technology or new guidance on its own, are what actually shift practice in frailty and end-of-life care. And last week’s optimism about advance care planning needs a caveat, because documentation gains do not automatically mean better outcomes for patients.

Verification, not guidance, moves assessment practice

A US multicentre retrospective cohort of 6,631 patients aged 75 and over across 20 hospitals found that formal implementation of the American College of Surgeons’ Geriatric Surgery Verification Program — an accreditation-style verification process, not a guidance document — was associated with a modest but measurable rise in geriatric vulnerability assessment before elective surgery, from 43.4% to 47.3%, largest in cardiac and vascular surgery (the verification programme is a directly relevant precedent for what actually moves assessment rates). The same week, a UK Delphi-consensus statement from the Association of Anaesthetists, the British Association for Day Surgery and the Association of Paediatric Anaesthetists embedded frailty explicitly into peri-operative patient selection for day surgery for the first time at this scale, naming frailty, obesity and obstructive sleep apnoea as comorbidities that must be actively assessed rather than treated as automatic exclusions. Read together with an appraised evidence summary for emergency department frailty management — thirteen high-quality sources synthesised into thirty-four evidence-based insights, including a stratified screening-assessment-intervention-discharge protocol that any emergency department could benchmark its own pathway against rather than sourcing each element separately — this week’s clearest lesson is that a formal verification or consensus mechanism, not guidance alone, is what actually shifts frailty-assessment practice at scale.

This extends a pattern this record has been tracking since a provisional register entry on structural and measurement fidelity at the point of handover: both this week’s US surgical-verification cohort and the UK’s day-surgery consensus statement show the same mechanism — a formal, monitored process rather than a document — operating in a different setting again, pre-operative geriatric assessment rather than ward-level handover. Neither study, on its own, is enough to promote that pattern further; each simply extends where the mechanism has now been observed.

Psychology embedded in primary care, and a quiet week for core rehabilitation

A UK scoping review of 72 studies found that co-located, integrated psychology professionals working in primary care were consistently described positively by both patients and staff — improved access, better patient knowledge and skills, and reduced primary care workload and stress — while psychology professionals themselves reported role-demand pressures, organisational challenges and unmet training needs. The review’s most striking finding is not the benefit itself but its absence at scale: a standing national recommendation for routine embedding of psychology in primary care remains unimplemented, which is precisely why a high-deprivation, high-demand primary care population would be a natural test case for closing that gap rather than relying solely on referral into secondary mental health services.

It was a genuinely quiet week for core post-acute rehabilitation, AHP-led discharge pathways, dysphagia and workforce evidence specifically. The only item to clear this week’s bar for the domain was a Danish prospective cohort of 60 patients — 100% women, mean age 58.2 — testing a three-month physiotherapist-led education and home-exercise programme for MRI-verified hip abductor tendon pathology, which produced a clinically relevant improvement in hip pain and in chair-stand performance with 85% adherence. It is MSK-adjacent rather than core post-acute rehabilitation, and single-site, so its value is as a transferable service-design template — structured education plus a modest home-exercise programme — rather than as evidence for this population’s frailty and rehabilitation caseload specifically. Reported plainly, rather than padded to look busier than the week was.

Advance care planning: documentation is rising, but the outcomes case is not yet made

A systematic narrative review of 41 clinical trials of interventions to improve goals-of-care discussions in serious illness — predominantly US, two-thirds randomised controlled trials — found that patient-focused interventions increased readiness and documentation but showed mixed clinical-outcome results, and that clinician- or system-level interventions improved documentation and conversation quality but had inconsistent effects on healthcare utilisation; combined interventions improved communication but not consistently quality of life or utilisation. A separate Swedish participatory action research project, developing “Death Literacy” among residential care staff to promote proactive end-of-life conversations, found the validated Death Literacy Index showed no measurable quantitative change — high baseline scores left little room to move — even though a qualitative read of the same intervention found genuine, action-oriented change in how staff actually practised. As a single-site, mixed-methods project without a control arm, it cannot establish that the training caused better outcomes; its real value is methodological, a caution that a validated index can miss the change a qualitative read captures, worth carrying into how any similar staff-training initiative is evaluated rather than treated only as a result in itself.

Read together, these two studies sharpen a caution first raised in the previous week’s synthesis, which reported that documentation and completion rates for advance care planning were rising under deliberate implementation effort. This week’s evidence does not contradict that documentation rises with effort; it directly qualifies whether that rise reliably produces better use of services or better quality of life for patients — on this week’s evidence, it does not. That is worth stating plainly rather than letting last week’s optimism stand unqualified: a completed goals-of-care conversation is a process measure, not an outcome, and the two are not the same thing. If a service cannot say what happened to a patient’s own time and preferences after the conversation was documented — whether the plan travelled with them, whether it changed what happened at the point of crisis — counting conversations is measuring the trace of the encounter, not its benefit. The published pattern register currently carries no entry addressing advance care planning completion against outcome benefit specifically, despite two consecutive weeks of convergent, directly relevant evidence pointing the same way — a gap in how evidence of this kind gets promoted into the standing record, not a contradiction of anything currently published.

OPAT: two low-cost changes, one meta-analysis, and a safety question worth checking locally

A systematic review and meta-analysis of five comparative studies covering 3,497 adults — the first to compare outpatient parenteral antibiotic therapy against a genuine inpatient comparator for infective endocarditis, rather than pooling single-arm data — found outpatient antibiotic therapy associated with lower all-cause mortality and lower acute kidney injury than staying in hospital, with no significant difference in readmission, relapse or valve replacement. High statistical heterogeneity on the mortality outcome (I²=87%) means selection bias in the underlying comparative studies cannot be ruled out, and that caveat should travel with the headline figure rather than be dropped from it. Separately, a controlled UK comparison at one NHS foundation trust tested whether a closed-system transfer device could reduce contamination when nursing staff prepare home-antibiotic-pump devices outside a controlled pharmacy environment: across 46 device preparations by eight operators in three uncontrolled rooms, the closed-system device produced zero contaminated devices against 22% contamination with standard needle-and-syringe technique (p=0.049), with shorter and more consistent preparation times. As a single-site, non-randomised comparison it cannot be generalised as a national effect size, but the size of the gap it closes, and the low cost of the equipment change, make it worth any OPAT service checking against its own preparation practice.

A US single-centre retrospective cohort of 1,169 outpatient antibiotic therapy patients found bloodstream infection or venous thrombosis occurred markedly more often with midline catheters than with peripherally inserted central catheters (6.4 versus 1.1 complications per 1,000 line-days, adjusted hazard ratio 3.60). US line-selection practice may not map directly onto UK services, but the size of the safety differential is large enough that any OPAT service using midline catheters for antimicrobial therapy should check the finding against its own line-choice protocol and complication data, rather than assume the comparison does not apply.

International hospital-at-home programmes are setting a benchmark this system does not yet match

Two international hospital-at-home programmes reported hard outcomes at national or near-national scale this week — a materially stronger evidence class than the feasibility and satisfaction data that has dominated this benchmark strand since August. Israel’s Clalit Health Services, a named, integrated payer-provider system, matched 3,335 hospital-at-home patients with pneumonia, heart failure, urinary tract infection or cellulitis 1:1 against general medical ward admissions in a national retrospective cohort: 30-day all-cause mortality was 5.8% at home against 9.1% in hospital, and 30-day readmission 13% against 16%, with 84% of patients saying they would choose home care again. Taiwan’s National Health Insurance ran a prospective matched-controlled pilot across seven nursing homes, treating older residents with acute infections at home rather than admitting them: care episodes were shorter, costs fell by 45–65% depending on diagnosis, and emergency-department revisits, readmissions and mortality were similar between groups — with a subgroup of soft-tissue infection patients showing significantly fewer early emergency-department returns.

Neither is UK evidence, and absolute figures will not transfer directly. But both are a materially stronger evidence class — large, matched, national or near-national cohorts with hard mortality, readmission and cost endpoints — than most of what this benchmark strand has surfaced since August, and neither has an equivalent in the evidence this record has gathered from the NHS’s own virtual ward programme. The honest comparison is no longer whether hospital-at-home care works; increasingly, elsewhere, at scale, the answer is yes. The sharper question is why an equivalent national outcomes dataset for this country’s own virtual ward programme does not yet exist, several years after these and other health systems began treating hospital-at-home as a funded, protocolised alternative to admission rather than a bolt-on pilot.

Two further pieces of evidence widen that same lever rather than the technology behind it. A national qualitative study of Canadian community workshops on hospital-at-home, spanning healthcare providers, patients, carers and administrators, named technology support for staff, managing provider and carer workload, and ensuring access for rural and remote communities as the standing unresolved issues, even amid genuinely high reported satisfaction — independently confirming a Pattern already in the published register: Pattern digital exclusion and carer capacity, not clinical suitability, are what actually bind who benefits from remote and home-based care. A cross-border scoping review of rural telehealth in Western Canada and the Western United States reached a comparable conclusion from a different literature: moderate confidence that telehealth reduces travel and supports specialist consultation, but only low confidence that it durably reduces emergency department use, and an explicit recommendation that telehealth works best as a hybrid, locally anchored access model rather than a stand-alone substitute for in-person care — a second, independent source now extending the same published pattern on remote monitoring redistributing clinical workload without a matching outcome gain, moving that finding beyond the single comparison it previously rested on.

Two smaller international items sit alongside these without individually clearing the same bar. A single-site pilot in rural eastern Taiwan linked on-site physicians to remote specialists in real time using point-of-care ultrasound and teleconsultation across 676 examinations and 26 consultation sessions over two years, with financial modelling suggesting break-even at under three cases a month — a small, single-service pilot, but a concrete design example of extending specialist input, not just remote monitoring, into a geographically dispersed population through a modest equipment investment. And a rapid policy review comparing home health care and telehealth policy across eight jurisdictions, including the UK, mapped service delivery, workforce training, financing and governance arrangements; because its own purpose was adapting policy recommendations for the Philippines, the UK entry is described rather than evaluated in depth, but it is still useful raw material for understanding how the UK’s policy settings compare with those of other health systems.

Where this leaves the week

Nothing this week contradicts a published register pattern outright. What is more striking is how much of it is new ground: most of what cleared this week does not confirm or contradict a standing claim so much as add territory the published record has not yet mapped, which is a legitimate feature of an evidence base that is still being built rather than a sign anything has gone wrong. Two gaps are worth naming plainly rather than letting them pass unremarked. OPAT and clinical psychology evidence has never yet produced an entry in the published pattern register despite recurring most weeks, and advance care planning evidence has now produced two consecutive weeks of caution-worthy findings with no formal place to land. Neither is a contradiction of anything currently published; both are questions about how quickly a standing record keeps pace with the evidence it is already collecting.

Underneath the different subjects, the same question about a patient’s own time keeps recurring. A frailty assessment completed at the point of surgical decision-making, rather than after admission, is time returned to a patient who is not later found to have deteriorated unassessed. A home antibiotic infusion prepared without contamination is a course of treatment that does not turn into an unplanned hospital admission. A documented goals-of-care conversation is only worth the time it took to hold if it travels with the patient and changes what happens to them later — and this week’s evidence says that link cannot yet be assumed. And a hospital-at-home model that reaches a rural or digitally excluded patient only by quietly shifting monitoring work onto an unpaid carer has not reduced the burden on that person’s time; it has moved it. Where this week’s evidence is silent on any of these questions, that silence is itself worth recording rather than filled in.

Open question

Two consecutive weeks of advance care planning evidence show documentation and staff-competence gains without a demonstrated link to utilisation or quality-of-life benefit, and the published pattern register has no entry addressing that gap specifically — nor, more broadly, any entry at all for OPAT or clinical psychology despite both domains recurring in most weeks to date. Is that a deliberate scope boundary on what counts as a registrable pattern, or a gap in how evidence gets promoted into the standing record?