Home exercise rehabilitation for frailty after hospital discharge: the HERO trial

The HERO trial (Health Technology Assessment 30(4):1-40; PMID 41557460) is the largest NIHR-funded randomised controlled trial of community rehabilitation for frailty to date. Across 15 NHS sites, 740 older adults aged 65 and over with mild, moderate or severe frailty (Clinical Frailty Scale 5–7) following hospital discharge were randomised to a 24-week therapist-delivered home exercise programme or usual care. The primary outcome was physical health-related quality of life measured by the SF-36 Physical Component Summary at 12 months. The trial found no statistically significant improvement: adjusted mean difference −0.22 (95% CI −1.47 to 1.03; p=0.73). The economic analysis returned an incremental cost-effectiveness ratio of £58,375 per quality-adjusted life year, well above the conventional NICE threshold of £20,000–£30,000.

A secondary hospitalisation signal runs in the opposite direction to the one a reader might expect: the trial reports some evidence of a higher rate of all-cause hospitalisation in the CONTROL arm, not the intervention arm (incidence rate ratio 1.12, 95% confidence interval 1.00 to 1.25; p=0.05), with no evidence of differences in any other outcome. The process evaluation found the intervention was largely delivered as intended and acceptable to most participants, and over half of participants completed more than 75% of prescribed exercises — though fewer than half (45.9%) completed the full 24 weeks, and 64.7% completed 12-month follow-up. The authors do not recommend routine commissioning of extended rehabilitation on this model, and direct future work at how core intermediate care and community rehabilitation services are organised rather than at extending standalone exercise programmes beyond them. That a well-delivered programme did not move its primary outcome points to the model rather than its delivery — a reading this publication draws, and not a conclusion the trial states.

This finding sits alongside a concurrent Japanese multicentre retrospective cohort study (Umegaki et al., PMID 42105048) validating a four-item bedside risk score for hospital-associated disability in older adults. The score — based on routine admission data — achieved an AUC of 0.796 across the validation cohort. Taken together, these two studies illustrate a complementary direction: where uniform post-discharge exercise programmes do not demonstrate benefit at the population level, validated admission-based stratification tools may enable more precise targeting of rehabilitation resource to those most likely to benefit.

The clinical practice implication is specific. The HERO trial does not argue against rehabilitation for frail older adults; it argues against undifferentiated delivery. Objective functional measurement at admission — to identify patients at highest risk of hospital-associated disability and to define rehabilitation goals — is the capability this result points toward, though the trial tested no such stratified model and so cannot establish that it would succeed where undifferentiated delivery did not. Services that cannot yet stratify by functional trajectory should treat the HERO result as an active prompt to build that capability, not as evidence against rehabilitation itself.