This Week in Physical Medicine & Rehabilitation — Sep 3, 2026

Welcome to This Week in Physical Medicine and Rehabilitation. This week we're covering 10 notable papers spanning procedural safety and musculoskeletal pain, technology-assisted rehabilitation from wearable robots to wearable sensors, and the long-term medical and functional consequences of neurological and spinal injury — plus one paper about our own certification pipeline. Let's dive in.

We start with musculoskeletal practice, and specifically with a question many of us get asked in clinic: is platelet-rich plasma riskier than a steroid shot? In PM&R, Ford and colleagues did the arithmetic across two large academic medical centres, reviewing more than a million peripheral joint and bursa injections in roughly 366,000 patients — just over a million corticosteroid injections and about 13,500 platelet-rich plasma injections. They then searched for any joint infection encounter between three and sixty days afterwards, with physician chart review to confirm the link. There were 69 confirmed cases of septic arthritis after corticosteroid injection and exactly one after platelet-rich plasma. That works out to roughly one infection per fifteen thousand steroid injections and one per thirteen and a half thousand platelet-rich plasma injections — which is to say, both are rare, and this is the first dedicated incidence estimate for platelet-rich plasma that we have. The obvious caveat is that with a single event, the estimate for platelet-rich plasma is imprecise, and this is retrospective code-and-chart work rather than surveillance. But the theoretical concern that injecting an autologous blood product raises infection risk above that of steroid is not supported here, and that is a useful number to have at your fingertips during consent.

Staying with musculoskeletal care, two papers this week look at who ends up in pain and why. Also in PM&R, Pier and colleagues surveyed 100 people with knee osteoarthritis in a large urban centre about diet and exercise — our first-line treatments and the ones patients most often don't do. The encouraging finding is that motivation is not the problem: over eight in ten believed exercise matters and wanted to exercise more, and around three quarters said the same about diet. The barriers were elsewhere. Knee pain itself was the leading obstacle to exercise, reported by close to half OF RESPONDENTS, while the leading dietary barriers were lack of guidance from their health care team, uncertainty about which foods to eat, and time. Critically, patients of non-White racial identity were more likely to cite pain and absent clinical guidance as barriers to exercise, and those living in lower-income zip codes were more likely to report inadequate exercise counselling and less confidence about food choices. The message for practice is uncomfortable but actionable: telling a patient with knee osteoarthritis to lose weight and move more is not counselling, and the patients least likely to have received real guidance are the ones already carrying the most social disadvantage. Treat the pain enough to make movement possible, and be specific about the diet. Alongside that, the American Journal of Physical Medicine and Rehabilitation published a cross-sectional survey by Kalbian and colleagues of 428 former NCAA Division One female athletes. Just over seven in ten reported current musculoskeletal pain, concentrated in the knee, spine, and hip and pelvis. Injury before college independently predicted current pain, nearly tripling the odds, and a career-ending injury raised them roughly fivefold. Those with pain were also more likely to have trained through injury and to have used anti-inflammatories during their careers. Notably, despite persistent symptoms, 94 percent of these women remained physically active. This is self-reported and cross-sectional, so recall bias is real, but it reframes the college athlete as a patient with a cumulative injury ledger that we will be reading for decades.

The second theme this week is technology in rehabilitation, and the papers sit at very different points on the evidence ladder. In Archives of Physical Medicine and Rehabilitation, Soma and colleagues report ten years of experience with the Hybrid Assistive Limb, a wearable exoskeletal robot that uses the patient's own voluntary muscle signals to drive assisted movement. Across 229 patients with a wide range of central nervous system disorders and 238 intervention episodes, in both inpatient and outpatient settings, there were no serious adverse events. Adverse events of any kind occurred in about eight percent of episodes, all mild, all resolving with conservative management, and none forcing permanent discontinuation. Event rates varied by diagnosis and by which device configuration was used, but not by care setting. This is a single-arm feasibility and safety study, so it says nothing about efficacy — what it does establish is that this technology can be delivered safely to a heterogeneous neurological population outside a tightly controlled trial. Efficacy evidence, meanwhile, comes from Topics in Stroke Rehabilitation, where Khalid and colleagues randomised 42 subacute stroke patients in Pakistan to task-oriented upper limb training alone or the same training plus Xbox Kinect exergaming, five sessions a week for six weeks. The exergaming group did better across the board — Fugl-Meyer upper extremity scores, the Chedoke Arm and Hand Inventory, fatigue, pain, and social participation — and the embedded qualitative interviews described the sessions as enjoyable and appropriately challenging. It's a small single-centre trial without blinding and with more total therapist-guided engagement in the intervention arm, so treat the effect size cautiously, but the adherence signal from a low-resource setting is worth noting.

On the measurement side, also in Archives, Dhamrongsirivadh and colleagues asked whether we are putting our accelerometers in the wrong place. In 20 chronic stroke survivors monitored for about six days in free-living conditions, sensors worn on the finger — capturing fine hand movement — correlated better with Fugl-Meyer, the Wolf Motor Function Test, and the Motor Activity Log than wrist-worn devices, discriminated impairment levels better, and had higher test-retest reliability. They also found that reliability improved with longer monitoring, with about four days needed for a stable estimate. Small sample, chronic and mildly-to-moderately impaired patients only, but if you are designing outcome measurement for upper limb trials or remote monitoring, wrist placement and single-day snapshots may both be underselling what your patients actually do with their hands. Rounding out the digital theme, the Journal of Rehabilitation Medicine published a propensity-matched retrospective cohort by Xu and colleagues comparing twelve weeks of home-based digital exercise against hospital-supervised physiotherapy for plantar fasciitis in nearly 600 matched Chinese adults. Home-based care was non-inferior for first-step morning pain at three months, slightly better on average, and less costly from both payer and societal perspectives. The authors are refreshingly candid that only about six percent of screened patients were eligible and that most baseline covariates remained imbalanced after matching, so they explicitly label the findings provisional and not ready to inform reimbursement.

Our third theme is long-term risk and patient heterogeneity. In Spinal Cord, Bae and colleagues used Korean national insurance and health screening data to follow 1,412 adults with newly diagnosed spinal cord injury against twenty propensity-matched controls each. Spinal cord injury was associated with roughly a 75 percent higher adjusted risk of incident heart failure, and for cervical injuries the risk was about doubled, with thoracic and lumbar injuries showing only a non-significant trend. Risk was concentrated in men with cervical injury and in those over 60. Heart failure–free survival was lower across all injury groups. Coding-based outcomes are a limitation, but the practical implication is straightforward: cardiac surveillance belongs in the long-term care plan after cervical spinal cord injury, not just bladder, bowel, skin, and spasticity. And in the Journal of Rehabilitation Medicine, Koivunen and colleagues applied latent class analysis to Neck Disability Index item responses from 962 patients with persistent neck pain, finding not one continuum but three distinct profiles: close to half OF PATIENTS with no or mild disability, a small middle group with selective limitation in reading, driving, and lifting but preserved personal care and concentration, and just over a third with severe global disability. Lower education, obesity, physical inactivity, worse pain, and lower occupational status all predicted membership in the severe class. It's cross-sectional and needs replication, but it argues against treating a single total score as the whole clinical picture.

Finally, a workforce note. In the American Journal of Physical Medicine and Rehabilitation, Driscoll and colleagues reviewed 26 years of American Board of Physical Medicine and Rehabilitation data and identified 161 examinees — about one and a half percent — who scored in the top quartile on Part One and then failed Part Two. Higher risk was seen among physicians over 40, non-white physicians, and graduates of newer and smaller residency programmes, while women and those in fellowship training were at lower risk. The practical point for programme directors is that a strong written score is not reassurance about the oral examination, and structured Part Two preparation should be universal.

If you only have time for one paper this week, make it the septic arthritis study in PM&R. Over a million injections give us the first real denominator for infection risk after platelet-rich plasma, and it is a number you can use in a consent conversation tomorrow.

Here are the key takeaways from this week in Physical Medicine and Rehabilitation. Septic arthritis after peripheral joint injection is rare, and platelet-rich plasma does not appear riskier than corticosteroid. Patients with knee osteoarthritis want to exercise and eat better; pain and absent clinical guidance are what stop them, and that gap falls hardest on non-White and lower-income patients. Wearable robotics can be delivered safely across diverse neurological diagnoses, exergaming added to task-oriented training improved upper limb outcomes in a small subacute stroke trial, and finger-worn sensors monitored over several days may measure real-world hand use better than the wrist. After cervical spinal cord injury, expect and screen for elevated heart failure risk, particularly in older men. And persistent neck pain resolves into distinct disability profiles rather than one uniform pattern.

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.