This Week in Physical Medicine & Rehabilitation — Oct 1, 2026

This AudioScholar briefing is generated by artificial intelligence for healthcare professionals and trainees. It is not medical advice.

Welcome to This Week in Physical Medicine and Rehabilitation. This week we're covering 10 notable papers spanning exercise dosing and its limits, stroke recovery from digital biomarkers to fatigue and falls, and the quieter but essential business of measurement — the scores and consensus statements that tell us whether rehabilitation is working. Let's dive in.

We start with supervised exercise, and the headline trial comes from Annals of Internal Medicine, where Culvenor and colleagues randomised 184 young adults aged 18 to 40 with persistent knee problems nine to 36 months after anterior cruciate ligament reconstruction [1]. One group received four months of twice-weekly physiotherapist-supervised exercise therapy with education; the other got a single education session plus a self-directed exercise resource. At four months the supervised group improved more on the composite Knee injury and Osteoarthritis Outcome Score, with an adjusted between-group difference of about five points, and the global rating of change data were more striking — roughly three quarters of the supervised group reported success for pain, against about 40 percent of controls. But by 12 months the gap had narrowed, because the control group kept improving while the supervised gains simply held. There were no differences in cartilage thickness or composition on magnetic resonance imaging. So this supports supervised exercise for faster symptom relief in this young post-reconstruction population, while tempering any claim that it alters structural disease or confers durable advantage over a well-designed self-directed program.

The dose question carries over into stroke. In Disability and Rehabilitation, Ball and colleagues report a nested multi-method study of the HiWalk program — up to three hours of walking training daily for three weeks in community-dwelling stroke survivors [9]. Attendance averaged about 91 percent and participants managed well over 500 repetitions per hour, and the interviews, framed by the Theoretical Framework of Acceptability, described strong motivation and the value of therapist and family support, with fatigue and competing life responsibilities as the main barriers. This is feasibility and acceptability evidence from a Phase II context, not efficacy — it tells us very high doses are tolerable in motivated survivors, nothing yet about whether walking outcomes improve.

And exercise has limits. A systematic review in Clinical Rehabilitation by Scrivener and colleagues looked at falls-prevention exercise for community-dwelling people after stroke and found only three randomised trials, together enrolling 677 participants [2]. The pooled estimate trended towards a lower rate of falls but remained compatible with no effect, and exercise clearly did not reduce the number of individuals who fell at least once. Balance showed only a trend, and the one firm finding was a small improvement in preferred walking speed, on the order of four centimetres per second. The authors conclude there is a suggestion exercise reduces fall rate without changing who falls — an honest negative-to-uncertain result, and a reminder that the stroke falls evidence base is thinner than our enthusiasm for it.

Falls also anchor the risk-prediction theme. In Archives of Physical Medicine and Rehabilitation, Bindawas and Vennu followed 4,783 adults aged 45 to 79 with or at risk for knee osteoarthritis in the Osteoarthritis Initiative cohort over eight years [6]. Baseline use of central nervous system-active analgesics — opioids or antidepressants — was associated with modestly higher odds of any fall and about a 30 percent increase in the odds of recurrent falls, and poorer physical performance on the repeated chair-stand and 400-metre walk tests predicted falls independently. Notably there was no statistical interaction between the two, meaning these look like separate, additive signals rather than one amplifying the other. This is observational and susceptible to confounding by indication, but it does strengthen the case that objective performance testing adds information beyond medication review in this population.

The stroke story continues on the measurement side, with the week's most technically ambitious paper in Science Translational Medicine. Wang and colleagues analysed roughly 23,000 hours of wrist-worn accelerometer data from 215 participants — healthy individuals plus subacute and chronic stroke survivors — to build what they call the digital arm performance scale [3]. By decomposing continuous accelerometry into lower-level movement segments and aggregating extracted features in a linear mixed-effects model, they produced an interpretable biomarker with strong reliability, concurrent and known-groups validity, and responsiveness. The eye-catching claim is from their power analysis: compared with traditional clinical assessments, this metric could cut the required sample size for an upper-limb recovery trial by more than 60 percent. If that holds in prospective use, it matters most for trial efficiency rather than bedside care, and it remains a development and validation study, not yet a tool with external replication.

Alongside that, PM and R brings us fatigue. Brycke and colleagues followed 268 people, 91 of them women, after a first-ever ischaemic stroke, with a median age of 73 [10]. Using the Swedish Fatigue Assessment Scale, significantly more women than men reported clinically relevant fatigue at both three months, where it was roughly 46 percent of women against 30 percent of men, and at 12 months, where the figures were 42 and 28 percent. Moderate fatigue predominated in both genders. Total scores fell modestly between three and 12 months in both men and women, but a substantial share of patients remained fatigued at a year, with the most common complaints being tiring very quickly and accomplishing little during the day. The authors argue for reassessing severity and consequences over time, not just presence or absence — an observational single-centre finding, but one consistent with the gender gap reported elsewhere.

That brings us to the outcome-measurement papers, which are unglamorous and quietly consequential. In Spinal Cord, Palermo and colleagues tested three seated trunk control assessments in 30 adults during early inpatient spinal cord injury rehabilitation, with injury levels from C4 to L2 and 80 percent motor-incomplete [5]. The Trunk Control Test and the Function in Sitting Test for spinal cord injury both showed excellent inter- and intra-rater reliability, the modified version slightly lower but still good, all three distinguished patients independent in transfers from those needing assistance, and all correlated with the Berg Balance Scale. Responsiveness, however, was limited by attrition, and motor-complete injury remains untested — so these are validated for use, with the change-detection question still open.

Prediction of discharge function comes from the American Journal of Physical Medicine and Rehabilitation, where Matsuzaki and colleagues retrospectively studied 158 patients aged 65 and over in a Japanese subacute orthopaedic rehabilitation ward [8]. Admission comfortable walking speed, Timed Up and Go, and Berg Balance Scale each improved prediction of the discharge Functional Independence Measure motor score beyond a base clinical model, and a simple equal-weight composite of the three performed best on repeated cross-validation, remaining independently associated with discharge score. It is single-centre and retrospective with only internal validation, so it is a promising, interpretable tool awaiting external testing.

Two more papers round out the week. In Archives of Physical Medicine and Rehabilitation, Earley and colleagues ran a three-round international Delphi study with up to ten clinicians experienced in bone-anchored limb care and reached consensus on 28 statements about prosthetic component prescription for transtibial bone-anchored limbs [4]. Delegates unanimously endorsed an appropriate safety connector, a torque absorber, and a high-energy-return or compliant carbon fibre or fibreglass foot with a split toe, and prioritised torsion absorption over shock absorption when a choice is forced. They also agreed that sex, age, and body mass index should not drive component choice, while physical ability and functional goals may. This is expert opinion from a very small pool of delegates — a starting point for guidelines, not evidence of outcome benefit. And in the same journal family, the American Journal of Physical Medicine and Rehabilitation published a small randomised trial from Türköz and colleagues comparing extracorporeal shock wave therapy, complex decongestive therapy, and the two combined in 33 patients with lipoedema, 11 per arm [7]. Fat mass fell significantly only in the arms containing shock wave therapy, but the magnitude was small and not superior to decongestive therapy alone, and all three groups improved comparably in limb volume, pain, pressure pain threshold, and quality of life, with no between-group differences. With 11 patients per arm this is underpowered to detect anything but a large effect, so the authors' suggestion that shock wave therapy may be preferred on cost and accessibility grounds should be read as a hypothesis.

If you only have time for one paper this week, make it the Annals of Internal Medicine anterior cruciate ligament trial [1]. It is the largest, cleanest test yet of whether supervised exercise therapy beats good self-directed care in symptomatic young adults after reconstruction, and its split verdict — a real early symptom benefit that converges by a year, with no structural signal — reframes the question from whether to supervise to for how long and for whom.

Here is what this week's evidence adds up to in Physical Medicine and Rehabilitation. First, supervised exercise after anterior cruciate ligament reconstruction produces faster symptom improvement than a single education session, but a well-supported self-directed group caught up by 12 months, and neither approach altered cartilage on imaging. Second, high-dose walking training after stroke is acceptable and achievable at remarkably high adherence, though efficacy data are still pending, while the pooled randomised evidence that exercise prevents falls after stroke remains inconclusive. Third, risk prediction is maturing: objective physical performance and central nervous system-active analgesic use independently predicted falls over eight years in knee osteoarthritis, and a simple composite of mobility and balance measures improved prediction of discharge function in subacute orthopaedic rehabilitation, both still needing external validation. Fourth, measurement itself is advancing on two fronts — a wrist-worn accelerometer biomarker that could substantially shrink upper-limb trial sample sizes, and seated trunk control tests now shown reliable and valid early after spinal cord injury, with responsiveness unresolved. And finally, post-stroke fatigue remains common at a year and consistently more prevalent in women, which the authors frame as an argument for repeated assessment of severity rather than a one-time screen.

That's your roundup for This Week in Physical Medicine and Rehabilitation. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And if it saves you time, a quick rating in your podcast app helps other physicians find the show.

This is an automated summary generated by artificial intelligence, which can make mistakes. Always review the original source materials.