Fatigue after stroke, shoulder-injection evidence and the science of measuring progress—with a ventilator safety check.
Radu Physiatry editorial desk · About 14 minutes · AI-assisted, source-checked; not physician-reviewed
A faster walk, a lower pain score, a stronger push against a testing device: each can mark progress. The harder question is what that progress means. Has the change exceeded ordinary measurement variation? Is it large enough to matter? Can the person do something important that was difficult before?
Today’s reading room starts with three papers published online October 2: a large analysis of fatigue after stroke, a pilot shoulder-injection trial and a review of injections for chronic shoulder disorders. An earlier strength-measurement review and a 2023 walking study supply the measurement lesson. The dates and reading boundaries stay attached: the four recent research summaries are abstract-only; the older walking paper was read in full.
Today’s reading lens
Ask what changed, how confidently, and for whom.
A stable population percentage can conceal substantial movement between categories. A small P value can accompany a modest benefit. A reproducible test can still have limits as a substitute for another instrument.
Same test conditionsChange measuredUncertainty checkedPatient goal revisited
Stroke / October 2, 2026 / individual participant data analysis / abstract only
Fatigue’s headline barely moved. The people did.
A pooled analysis used data from three large trials of fluoxetine for stroke recovery. Among 5,907 participants, 5,025 had fatigue measurements at both six and twelve months. Researchers defined fatigue using a vitality questionnaire score of 45 or lower. The proportion meeting that definition was almost unchanged: 29.6% at six months and 29.3% at twelve months. That apparent stability hides two different trajectories. Read the original analysis.
Of 1,489 participants classified as fatigued at six months, 568 moved out of that category by twelve months: 38.1%, 95% confidence interval 35.7–40.7%. Meanwhile, 553 participants who had not met the fatigue definition at six months did so at twelve months—15.6% of that starting group, interval 14.5–16.9%. Improvement and new fatigue largely offset each other in the overall percentage.
For a patient, this supports revisiting energy and participation even after the early recovery period. A group average cannot describe a personal course. It also gives no reason to dismiss persistent exhaustion because strength or walking has improved.
Clinical notes: a trajectory analysis is not a treatment effect
The parent trials randomized fluoxetine 20 mg daily versus placebo for six months; this report studies subsequent fatigue trajectories and associations. The abstract does not report a randomized fatigue treatment contrast. Questionnaire data were missing for about 15% of participants, and trial participation limits generalization. The authors found no consistent pattern across predictor analyses. Category crossing is also distinct from the separately defined 10-point vitality improvement; do not assume that all 568 participants achieved that change. The questionnaire threshold is a study definition, not a diagnostic test for every cause of exhaustion.
Established guidance / Canadian Stroke Best Practices / 2025 update
An energy complaint deserves its own assessment.
The Canadian recommendations call for ongoing fatigue monitoring and assessment of treatable contributors, including sleep apnea, pain, mood conditions, sedating medicines, anemia and thyroid disease. They also support education about balancing activity and rest, with a graded return to activity. These are established recommendations, not treatments tested by the new trajectory analysis. Read the actual fatigue and sleep recommendations.
Clinical notes: recommendation strength and evidence quality differ
Ongoing monitoring is a strong recommendation with moderate-quality evidence. Several management recommendations, including progressive activity and energy conservation, have low-quality supporting evidence. Antidepressants are not recommended for fatigue alone without another indication such as depression or anxiety; that recommendation is strong, with moderate-quality evidence. Medication choices require individual clinical assessment. This edition reviewed the recommendations and rationale, rather than independently reassessing all contributing trials.
Musculoskeletal / October 2, 2026 / pilot randomized trial / abstract only
A shoulder result can be statistically clear and clinically modest.
A double-blind pilot randomized 46 adults with ultrasound-confirmed subacromial-subdeltoid bursitis and shoulder pain lasting three months or less. Both groups received a twelve-week exercise program. One also received a single ultrasound-guided injection of 10 mL 20% dextrose; the other received saline. Forty-two completed thirteen-week follow-up. The primary outcome was the Shoulder Pain and Disability Index, or SPADI. Read the trial abstract.
The adjusted score trajectory favored dextrose plus exercise (group-by-time P = 0.004). At thirteen weeks, the unadjusted difference was 6.40 points in its favor, 95% interval 4.08–8.72. The point estimate fell below the trial’s prespecified eight-point clinically important difference, although the interval included effects above that threshold. DASH arm-function and quality-of-life outcomes did not show differential trajectories.
Clinical notes: keep the comparison and analysis attached
This tests adding dextrose to exercise against adding saline to the same exercise program. It does not compare an injection with doing nothing. The adjusted longitudinal test and unadjusted thirteen-week difference are different analyses. A clinically important threshold helps interpretation; an average group effect is not an individual responder rate. Four dextrose recipients reported temporary soreness; no serious events were reported. A small, short pilot cannot establish uncommon harms or durable benefit.
Patient implication: Discuss the likely size of added benefit, the exercise plan and uncertainty before treating a positive P value as a compelling reason for a procedure.
Pain and sports / October 2, 2026 / systematic review / abstract only
The chronic-shoulder evidence remains uncertain.
A separate review searched through March 2026 and included fourteen studies of dextrose prolotherapy for shoulder pain lasting more than three months: thirteen randomized trials and one observational study. Ten studies had high risk of bias; most were small and short. In mixed rotator-cuff tendinopathy or bursitis, the medium-term functioning comparison against saline was standardized mean difference −0.3, interval −1.0 to 0.4, from three studies and 149 participants. Certainty was very low. The interval crosses zero and permits quite different effects. Read the original review.
This review and the pilot do not study identical patients: chronic mixed disorders differ from relatively recent, imaging-confirmed bursitis. The review’s search also predates the new pilot’s publication. Their combined message is to inspect diagnosis, symptom duration and comparator before generalizing a result.
Clinical notes: a pooled standardized effect is not a SPADI point change
A standardized mean difference expresses an effect in standard-deviation units so different scales can be combined. It cannot be read as −0.3 SPADI points. Concentrations, injection schedules and comparators varied; harms evidence was sparse. Full study tables, risk-of-bias judgments and complete disclosures were not appraised here. These abstracts support careful discussion, rather than a universal injection recommendation.
Principles and outcomes / July 7 online; October 2026 issue / abstract only
A portable strength test needs a consistent setup.
A review of portable dynamometers—devices that measure force—included 123 studies and 5,042 participants; 113 studies entered meta-analyses. Across most knee and ankle muscle groups and testing setups, reliability met the authors’ threshold of intraclass correlation coefficient at least 0.70, with moderate- to high-certainty evidence. External fixation had high-certainty support for testing knee extensors, knee flexors and ankle plantarflexors. Read the review record.
Evidence of agreement with a reference method was much less extensive. Seated, externally fixed knee testing had correlations of 0.82–0.90, but the authors caution against treating portable devices as direct replacements for isokinetic dynamometry. This is earlier research, not an October 3 discovery claim.
Clinical notes: reliability, validity and functional recovery
Reliability addresses consistency. Criterion validity asks how a test relates to a reference measure. A high correlation alone does not establish interchangeable numerical results. Keep position and fixation consistent when comparing visits, and interpret force alongside the activity the person wants to regain. The review included uninjured people and people with lower-limb musculoskeletal conditions; it does not independently validate every neurological population or every testing protocol.
Established teaching / July 19, 2023 / measurement study / full text reviewed
How much faster is more than ordinary variation?
The minimal detectable change (MDC) estimates how large a change must be to exceed expected measurement error under specified conditions. It answers a different question from how much change is meaningful to the patient.
A single-hospital study tested 84 people after stroke, averaging about 76 days from onset. They walked twice at comfortable speed and twice at maximum safe speed, timing the middle ten meters with three-meter acceleration and deceleration areas. Walking aids and ankle-foot orthoses were permitted; therapist physical assistance was not. At comfortable speed, the estimated MDC was 0.05 m/s in the slow group, 0.11 m/s in the middle group and 0.21 m/s in the fast group. Original record · Full paper.
The lesson is to use measurement-error estimates from a similar population and protocol. These are not universal recovery cutoffs. The small groups, variable time after stroke, aid use and same-session repeated testing limit transfer to longer-term follow-up.
Clinical notes: detectable change, important change and participation
MDC concerns measurement error; a minimal clinically important difference concerns meaningful change. Neither automatically establishes that a treatment caused the change. Speed is distance divided by time, so a given time difference does not produce the same speed difference at every baseline. This study’s comfortable-speed groups were below 0.4 m/s, 0.4–0.8 m/s and the fastest stratum around or above 0.8 m/s. A walking score should be paired with the person’s experience of activities and participation.
For established administration guidance, the RehabMeasures entry emphasizes documenting the test version, preferred versus fast speed, assistive device and assistance level. Its page is dated January 22, 2014; it is not a new guideline.
Safety desk / FDA early alert posted September 23, 2026 / reviewed October 3
For ventilator users, readiness includes the backup.
An FDA early alert concerns the listed VOCSN ventilator configurations with serial numbers 112786 and higher. A cable connector can loosen, producing reboots and recurring System Fault 10 alarms. The notice reports no reboots during active patient ventilation in available reports and no serious injuries or deaths reported by the company as of August 12. An interruption nevertheless poses particular danger to someone unable to breathe adequately without support.
The instructions include keeping functional alternative ventilation immediately available, caregiver readiness, careful device handling and the specified checkout procedure after a drop or impact. Recurring faults require backup ventilation as needed and contacting React Health for service. Do not stop prescribed ventilation without an appropriate alternative and instructions from the treating clinician. Check the model list and full actions in the actual FDA notice. This is an earlier, device-specific early alert, not a newly classified recall or a notice covering every ventilator.
The learning calendar and the coverage boundary.
AAPM&R currently lists an October 13, 7 pm National Grand Rounds session on establishing a PM&R pathway for brain-computer interfaces. The official series page describes a free webinar series and links registration. Check the organizer’s current details and time zone before attending. We reviewed the listing, not the future lecture; this magazine confers no CME credit.
All 88 registered sources and 15 study domains were attempted: 38 journal indexes checked, 38 landing pages retrieved and twelve sources unavailable after retry. The scan retrieved 774 unique candidate records, including 75 PMIDs absent from yesterday’s set. That comparison measures retrieval, not seventy-five new advances. A focused recent-entry trial/guideline search supplied additional leads. Larger registry searches stop at twenty candidates and can miss relevant work.
The complete source desk preserves statuses and search boundaries. Licensed chapters, paid courses and linked recordings were not appraised. Canadian guidance and the older full-text measurement paper supplement the registered discovery sources. Reading selected evidence does not establish that every linked publication was reviewed or that no other safety update exists.
One question to study.
Original teaching question: A hypothetical stroke survivor’s comfortable ten-meter walking speed rises from 0.90 to 1.02 m/s under the same test conditions. Assume the older study’s fast-stratum MDC of 0.21 m/s applies. Which interpretation is best?
The change proves there was no improvement.
The 0.12 m/s gain is below that MDC; this score alone does not establish change beyond expected error.
The patient has met a universal clinically important threshold.
The treatment caused a 12% improvement in community participation.
Reveal the answer and reasoning
B. 1.02 − 0.90 = 0.12 m/s, less than 0.21 m/s. Falling below an applicable MDC does not prove no improvement. It limits what that measurement alone establishes. Meaningfulness, causation and community participation require other evidence. The assumed applicability is part of this teaching example, not an individual clinical judgment. Source: the gait-speed measurement study.
A small glossary for a deeper read.
Vitality
A questionnaire measure of perceived energy and fatigue.
Individual participant data analysis
An analysis using participants’ records across studies rather than only published group summaries.
Confidence interval
A range expressing statistical uncertainty under a model; it does not account for every possible bias.
Bursitis
Inflammation or irritation of a fluid-filled sac that helps tissues move against one another.
SPADI
Shoulder Pain and Disability Index, a questionnaire about shoulder symptoms and limitations.
Standardized mean difference
A group difference expressed in standard-deviation units to combine outcomes measured on different scales.
Reliability
The consistency of a measurement.
Criterion validity
How a measurement relates to an appropriate reference measure.
Minimal detectable change
A threshold for change exceeding expected measurement error under defined conditions.
Participation
Involvement in life situations, such as work, family roles or community activities.
Four research summaries were checked against primary PubMed abstracts: a stroke-fatigue trajectory analysis, a pilot randomized shoulder trial and two systematic reviews. These remain abstract-only; the portable-dynamometry review was published online in July. The established 2023 gait-measurement paper was read in full, including methods, results and limitations. The actual Canadian 2025 fatigue recommendations and rationale, FDA device-specific early alert, RehabMeasures administration entry and AAPM&R education listing were separately consulted. This is independent education, not accredited CME or official examination material.
The discovery window was 2026/09/03–2026/10/03. A journal index check can miss work not yet indexed. A retrieved landing page does not verify every linked publication, paid course, podcast or textbook chapter. Candidate counts are discovery records, not validated practice changes.
Educational, not medical advice. Discuss treatment choices with your clinician. Sources and evidence evolve; each dated edition preserves what was checked for that issue.