From balance scores to fewer falls
What measured improvement means for falls prevention, exercise add-ons, spinal cord examination and medication review.
A better balance score is encouraging. Getting through another year without a fall is a different outcome. Two reports in today's reading put that distinction into focus: one pools exercise studies that measured balance and fall-risk tests; the other follows falls recorded in primary-care medical records. Reading them together shows why the name of the outcome matters as much as the size of the result.
That question runs through this edition. Does a cuff-assisted exercise programme improve lasting function? Do vitamins add anything to knee exercises? Can part of a spinal cord examination be dropped? Does reducing blood-pressure medication improve life in long-term care? Each answer depends on the comparison actually made, the people studied and the uncertainty left behind.
All 88 registered sources and 15 study domains were inspected. Discovery returned 807 journal-article metadata records plus one book identifier. Compared with October 16, 98 article records entered and 102 left the capped results; these are retrieval changes, not publication dates or retractions. All 12 unavailable sources were retried without recovery. Seven research reports and two review protocols receive focused appraisal below. Main text or selected main-text sections were reviewed for four reports; three reports use abstracts or subscription previews. A retrieved portal does not establish review of its linked chapters, courses or recordings.
Balance improved. Actual falls were a question for another study.
A review of 31 randomized trials involving 3,622 older adults examined traditional mind-body exercise, including Tai Chi and Baduanjin. Seven studies contributed fall-risk test scores; 24 contributed the Berg Balance Scale. The pooled results favored exercise: a standardized mean difference of −0.56, 95% confidence interval −0.84 to −0.28, on fall-risk tests, and a 2.60-point improvement on the Berg scale, 95% CI 1.47–3.72.
The first number does not mean 56% fewer falls. It expresses a difference in test scores using the studies' variability as the unit. The second describes performance on a balance scale. Neither is a count of people falling, fractures prevented or emergency visits avoided. The review explicitly says its findings do not establish a reduction in actual fall incidence.
For a patient, the useful message is that these activities may help measured balance, with a programme adapted to ability and circumstances. The evidence does not identify a single best exercise style or guarantee protection from falling. The authors rated certainty low for fall-risk test scores and very low for Berg scores in their evidence assessment.
Clinical notes: surrogate outcomes and heterogeneous exercise trials
Heterogeneity was I² 56.2% for fall-risk tests and 95.6% for Berg scores. Different populations, programmes, controls and measurement schedules limit how well the pooled average describes one clinical setting. Searches ended December 10, 2025; an October publication does not mean its evidence search includes October trials.
A test associated with falls is not automatically a validated surrogate for a treatment's effect on falls. Subgroup estimates for different exercise styles are also not head-to-head proof of superiority or an established dose-response relationship. The main methods, comparative results and limitations were read; all included trials, complete extraction tables, figure images and supplements were not independently appraised.
A modest reduction in medically recorded falls, with an analysis caveat
The US eSTEPS trial assigned 22 primary-care practices in matched pairs to usual care or electronic decision support for tailored exercise, optional physical therapy referral and patient materials. It included 11,945 adults aged 65 or older with a positive fall-risk screen, followed for 12–23 months in one integrated health system.
Its primary outcome was the rate of patients with at least one fall recorded through diagnostic codes. In the adjusted analysis, rates were 19.3 versus 21.2 per 100 patient-years: rate ratio 0.91, 95% CI 0.88–0.95. The adjusted absolute difference was −1.90 per 100 patient-years, 95% CI −3.03 to −0.77. This is a modest reduction in coded fall events affecting patients, rather than a count of every recurrent fall.
The distinction between analyses matters. The unadjusted coded-fall comparison did not show a statistically clear benefit. The favorable primary result emerged after the authors' prespecified adjustment for patient and practice characteristics. Injury-related fall rates also favored the intervention after adjustment, 15.7 versus 16.7 per 100 patient-years; rate ratio 0.94, 95% CI 0.90–0.97. Mortality did not differ clearly.
This supports testing practical ways to bring exercise into routine care. It does not show that an app alone prevents falls: the intervention combined decision support, clinicians, exercise materials and referral options. Exercise prescriptions reached 22.1% of intervention patients, and actual exercise adherence was unavailable.
Clinical notes: clusters, ascertainment and adjusted estimates
Practices, rather than individual patients, were randomized: 11 matched pairs require analysis that accounts for clustering. The authors used Poisson generalized estimating equations with follow-up time and adjustment for patient/practice factors, including the number of completed fall-screen assessments. Raw proportions and covariate-adjusted rates are different estimates; the reported relative reduction is not an absolute nine-percentage-point reduction.
Medical-record codes miss falls that never reach care, and documentation or surveillance can differ between groups. Clinicians were unblinded, uptake was incomplete and applicability beyond this health system is uncertain. The favorable adjusted result should be read alongside the unadjusted comparison. No independent audit of raw data, code, supplements or figure images was performed. This edition omits the model-derived number needed to treat, which depends on additional rate-to-risk assumptions.
Date transparency: the actual JAMA Network Open page identifies October 16, 2026 as publication, while the paper's own PubMed publication and electronic-date fields say October 1. Both are preserved here. Discovery date is not used to resolve that discrepancy.
Primary eSTEPS record · Actual publisher report and tables · DOI: 10.1001/jamanetworkopen.2026.39240
Blood-flow restriction is an adjunct with uneven evidence
Blood-flow restriction training pairs lower-load exercise with a pressure cuff. A new upper-limb review included 20 reports: nine parallel-group randomized trials, one crossover trial, two case series and eight case-report publications. Although 340 people were enrolled or randomized, 326 contributed to analyzed or treated samples. Counting reports alone would hide how much of the evidence lacks a randomized comparator.
Some comparisons were promising. A 46-person sham-controlled elbow tendinopathy trial reported a 12-week pain difference of −1.54 points, 95% CI −2.89 to −0.18, and a Patient-Rated Tennis Elbow Evaluation (PRTEE) difference of −15.23 points, 95% CI −23.57 to −6.90. Another, pragmatic elbow trial found no clear 12-month PRTEE advantage: adjusted contrast +4.32, 95% CI −9.17 to +17.81.
These are individual trials reported within the review, not one pooled treatment effect or newly conducted experiments. Differences in active versus sham comparators, diagnosis, duration and accompanying therapy matter. The review supports considering the approach within professional rehabilitation; it does not establish that a cuff is routinely better than well-designed active exercise.
Clinical notes: count designs, then inspect the search and harms
The authors used narrative synthesis because prescriptions and conditions were too heterogeneous for a useful common estimate. Among the nine parallel trials, one was rated low risk of bias, seven had some concerns and one high risk. The database search ended before February 18, 2026; 15 of 20 reports came from reference/citation checking, and the authors acknowledge incomplete retained database search history. These limit reproducibility and confidence in coverage.
Cuff pressure, load and co-interventions varied. Incomplete adverse-event ascertainment means an absence of reported major events cannot establish general safety. This edition supplies no self-treatment pressure or exercise prescription. Detailed study tables, figure images, supplements and the underlying trial reports were not fully audited.
Primary upper-limb blood-flow restriction review · Publisher report
Adding a vitamin did not clearly improve the exercise result
A Japanese trial randomized 88 postmenopausal women with symptomatic knee osteoarthritis to home exercise alone, exercise plus the prescription active vitamin-D analogue eldecalcitol, or exercise plus vitamin E for 24 weeks. Eighty completed follow-up and entered the per-protocol analysis.
The primary measure was the Japanese Knee Osteoarthritis Measure (JKOM), where lower symptom/function scores are better. Mean total scores fell from 35.2 to 21.7 with active vitamin D, 34.9 to 23.8 with vitamin E and 33.6 to 17.0 with exercise alone. Improvement within every group does not demonstrate an advantage from either add-on.
The adjusted treatment coefficients, relative to exercise alone, were +6.67 JKOM points, 95% CI −1.99 to +15.33, for active vitamin D, and +8.67, 95% CI −0.28 to +17.62, for vitamin E. Neither showed a statistically clear added symptom benefit. Muscle-strength and walking comparisons likewise did not establish an additional benefit.
Clinical notes: the between-group question and the population studied
The adjusted model included age, body mass index and radiographic grade. Positive coefficients on this lower-is-better scale do not favor the add-on, but confidence intervals cross zero; this small study does not prove equivalence. Eight losses and per-protocol analysis can affect randomization's protection against bias.
All groups received exercise, so this trial cannot by itself measure exercise efficacy against no exercise. Within-group biomarker changes are not proof of superior clinical outcomes. The active prescription analogue is not interchangeable with ordinary over-the-counter vitamin D. Results in older Japanese women over six months do not settle deficiency treatment, fracture prevention, long-term disease modification or other populations. This edition gives no dosing recommendation; complete supplements and all secondary tables were not independently audited.
Primary active-vitamin/exercise pilot trial · Publisher report, Tables 2 and 4
Two sensory tests contribute different information
In a secondary analysis of 7,449 people with traumatic spinal cord injury, investigators used complete standardized examinations recorded from 2011 through 2023 to ask what changes when light touch or pinprick information is omitted. The neurological level of injury (NLI) remained unchanged in 88.6% of cases without light touch and 76.8% without pinprick. Removing either could also move up to 2.3% of cases to a more severe impairment grade.
These numbers do not license a shorter examination. They show that the components sometimes supply information the other cannot replace. For a patient, an apparently repetitive sensory examination can therefore have a classification purpose beyond asking whether sensation is simply present or absent.
Clinical notes: classification agreement is not diagnostic accuracy
The reference here is the classification obtained from a complete ISNCSCI examination. The proportions are agreement after information is removed, not sensitivity or specificity for discovering an injury. NLI and ASIA Impairment Scale grade are distinct classification outputs. The analysis found contributions from both sensory modalities, particularly where motor testing cannot contribute.
This is an observational analysis of existing examinations, not a randomized test of an abbreviated protocol or a newly issued classification standard. Complete methods, missing-examination handling and detailed subgroup tables remained unavailable. Do not substitute its abstract for the official examination manual.
Primary sensory/motor contribution study · DOI: 10.1016/j.apmr.2026.10.007
Fewer blood-pressure medicines is a process, not a guaranteed benefit
A review found three randomized long-term-care studies involving 1,548 residents that tested planned antihypertensive deprescribing. Populations, reduction strategies, monitoring, restart rules and follow-up differed enough that the authors declined to pool the results. That is an informative finding: a single reassuring average would conceal different clinical questions.
For all-cause mortality, RETREAT-FRAIL reported an adjusted hazard ratio of 1.02, 95% CI 0.86–1.21, over median potential follow-up of 38.4 months. DANTON reported 1.65, 95% CI 0.95–2.85, to 32 weeks. Those intervals do not establish a common survival benefit or uniform safety. In DANTON, systolic pressure rose by 11.0 mmHg, 95% CI 5.6–16.4, at 16 weeks while medication exposure fell.
The relevant discussion is a supervised medication review matched to goals, symptoms and monitoring. The findings are not a reason for a resident or caregiver to stop blood-pressure treatment independently.
Clinical notes: why not to pool, and what a null result leaves open
Searches were updated August 20, 2026. The review required compatible estimates of a common interpretable effect before meta-analysis. The third study, Gulla/COSMOS, had no cluster-adjusted comparative mortality estimate. Evidence for falls, hospitalization, cardiovascular events, function and quality of life was sparse or limited by bias.
A mortality interval crossing one does not demonstrate equivalence; DANTON's interval includes substantial possible harm. Follow-up horizons and trial protocols differ, so the two hazard ratios cannot be averaged casually. This edition reviewed the primary abstract and publisher subscription preview, not the complete review, extraction tables, deprescribing protocols or original trials. It supplies no medication-withdrawal algorithm.
Primary long-term-care deprescribing review · Publisher abstract and access boundary
Support needs remain, even when a programme misses its target
The BRITE caregiver report concerns secondary outcomes from a pragmatic traumatic-brain-injury trial at six rehabilitation centres. Of 925 randomized patients, 591 had enrolled caregivers: 295 in transition support and 296 in standard discharge planning. Up to 12 remote follow-up contacts over six months did not produce statistically clear differences in caregiver burden or quality of life at six or twelve months.
The abstract provides no effect confidence intervals. Full methods, received contact dose and missing-data details were unavailable. This result tests that programme and those measures; it does not establish that caregiver support is unnecessary or ineffective in every form. Exploratory associations with greater patient impairment or assistance needs are not randomized treatment effects. Primary BRITE caregiver report.
Two new Cochrane records are review protocols: exercise for temporomandibular disorders and statins after a first episode of venous thromboembolic disease. Their abstracts describe planned evidence assessments. They contain no completed pooled treatment effects and support no new exercise or statin recommendation today.
Bring the evidence back to an individual plan
Established guidance: NICE's 2025 falls guideline, NG249, advises against using fall-risk prediction tools to predict an individual's fall risk. Its recommendations instead link fall history, gait/balance assessment and comprehensive assessment where indicated to individualized action. For community-dwelling people eligible for prevention interventions, it describes progressive, tailored exercise addressing function, balance, coordination, strength and power, with professional delivery and review. It also addresses medication review and home hazards. This is established guidance, not a new October 17 guideline or an independent appraisal of all its underlying evidence. Actual NICE recommendations.
Safety carry-forward: FDA's October 15 warning about 10 Day Forecast was reread today. The sexual-enhancement product contained undeclared sildenafil and acetaminophen. FDA advises against buying or using it: sildenafil can interact dangerously with nitrates, while hidden acetaminophen can add unrecognized exposure and liver-injury risk. This is a continuing product-specific warning, not a newly dated alert or a survey proving other supplements safe. The notice text was reviewed; the product photograph was not. Actual FDA notification.
Education calendar: AAPM&R lists its October journal club for October 19, 7:30–8:30 p.m. Central, discussing autonomic/endothelial dysfunction after spinal cord injury and a review of adverse events with knee platelet-rich-plasma injections. The actual course description says members are eligible and “No Credit Offered.” These linked papers were not appraised for clinical conclusions in this edition; the listing is not attendance, lecture review or CME credit. Official journal-club description.
Test the interpretation
A review reports a standardized mean difference of −0.56 in fall-risk test scores after exercise. It does not measure actual fall incidence. Which conclusion is justified?
- Exercise prevented 56% of falls.
- The result favors exercise on the measured tests, but does not quantify a reduction in actual falls.
- Every individual improved by 0.56 points on the same scale.
- The intervention has been shown to prevent fractures.
Reveal answer and reasoning
B. A standardized mean difference expresses a between-group score difference relative to variability. It is neither a percentage fall reduction nor a raw point change on one shared scale. Without actual fall or fracture outcomes, those clinical effects cannot be calculated from this estimate. Low certainty and heterogeneity further limit interpretation.
Words worth keeping
- Surrogate outcome
- A measurement used in place of a clinical outcome; improvement may not reliably predict the treatment's effect on what patients experience.
- Standardized mean difference
- A difference between group averages expressed in units of variability, allowing different scales to be combined.
- Mean difference
- A difference expressed in the original measurement units, such as points on the Berg Balance Scale.
- Patient-years
- Combined follow-up time across patients. A rate per patient-year is not the same as the percentage of people affected.
- Cluster randomization
- Assignment of groups, such as practices, rather than individual people; analysis must account for similarities within groups.
- Adjusted estimate
- A comparison accounting statistically for specified characteristics. Its interpretation depends on the model and variables used.
- Confidence interval
- A range describing uncertainty around an estimate under the statistical assumptions.
- Per-protocol analysis
- An analysis restricted to people meeting specified treatment or follow-up conditions; exclusions can introduce bias.
- Neurological level of injury
- A standardized classification output based on sensory and motor examination after spinal cord injury.
- Deprescribing
- Planned, supervised reduction or stopping of medication when a clinical review supports it, with appropriate monitoring.
- Review protocol
- A plan for how a review will be conducted, rather than its completed findings.
References & the coverage record
- Effectiveness of traditional mind-body exercises on fall risk test scores and balance in older adults: A systematic review and meta-analysis. Medicine · PMID 42888580 · DOI 10.1097/MD.0000000000051150
- Clinical Decision Support for Fall Prevention in Primary Care: The eSTEPS Cluster Randomized Clinical Trial. JAMA network open · PMID 42886484 · DOI 10.1001/jamanetworkopen.2026.39240
- Effects of blood flow restriction training in adult upper-limb musculoskeletal rehabilitation: A systematic review. Medicine · PMID 42888637 · DOI 10.1097/MD.0000000000051205
- Additive effects of active vitamin D3 or vitamin E administration on therapeutic lower-limb home exercise for knee osteoarthritis: A pilot prospective randomized controlled trial. Medicine · PMID 42888606 · DOI 10.1097/MD.0000000000051197
- How Sensory and Motor Components Shape the Neurological Level of Injury. Archives of physical medicine and rehabilitation · PMID 42887630 · DOI 10.1016/j.apmr.2026.10.007
- Planned Antihypertensive Deprescribing in Long-Term Care Residents: A Systematic Review of Randomized Evidence. Drugs & aging · PMID 42888381 · DOI 10.1007/s40266-026-01342-5
- Caregiver Outcomes from the BRITE Study: Secondary Outcomes of a Multisite Pragmatic Randomized Controlled Trial of Post-Rehabilitation Discharge Transitional Support after Traumatic Brain Injury. Journal of neurotrauma · PMID 42885621 · DOI 10.1177/08977151261495028
- Exercise therapy for temporomandibular disorders. The Cochrane database of systematic reviews · PMID 42886005 · DOI 10.1002/14651858.CD016402
- Statins for adults with a first episode of venous thromboembolic disease. The Cochrane database of systematic reviews · PMID 42886002 · DOI 10.1002/14651858.CD016403
Seven research reports and two review protocols were appraised: a mind-body exercise meta-analysis, the eSTEPS cluster randomized trial, a mixed-design upper-limb blood-flow restriction review, an active vitamin/exercise knee osteoarthritis pilot trial, a secondary spinal-cord classification analysis, a long-term-care antihypertensive deprescribing review, BRITE caregiver secondary outcomes, and two planned Cochrane reviews. Main methods, results and limitations were read for the mind-body review, with complete study tables/figures and supplements not audited. The eSTEPS main publisher text, methods, displayed tables, discussion, limitations and disclosures were read through verified publisher web access after direct DOI access failed; supplements, raw data, code and figure images were not audited. The publisher says October 16 while the own PubMed publication and electronic fields say October 1; both are disclosed. Adjusted and unadjusted results, coded first-event outcome, limited uptake, absent patient adherence and model-dependent NNT were checked. Main methods, selected comparative results and limitations of the BFR review were read; detailed extraction tables and underlying trials were not fully appraised. Main knee-trial methods, selected Tables 2 and 4, results and limitations were read; prescription active analogue is distinguished from ordinary vitamin D, and uncertain dose conversions are not reproduced. SCI, BRITE and deprescribing summaries use primary abstracts; the deprescribing publisher page is explicitly a subscription preview, not full text. The two Cochrane primary abstracts identify protocols, not completed evidence. All 88 source statuses, 15 capped/truncated domain results, 98 newly retrieved article titles/dates and 38 portal comparisons were inspected. Twelve unavailable sources were retried without recovery. Added/removed records are capped retrieval changes, not publication dates or retractions. The 16 focused-search titles were screened and selected abstracts appraised; 16 selected primary abstracts were read across discovery and focused results. FDA device/drug update lists were inspected beyond registry landing pages; no exhaustive absence-of-updates claim is made. The actual October 15 hidden-ingredient notification was reread as safety carry-forward, not a new alert. Actual NICE NG249 recommendation sections 1.1 and 1.3 were read as established 2025 guidance, not underlying evidence-review appraisal. The actual AAPMR October 19 journal-club description, time, member eligibility and no-credit label were checked; linked papers, lectures and recordings were not appraised for clinical conclusions. Portal changes included food recalls and ATS career navigation; dynamic hashes are not evidence of chapter updates. Discovery contains 807 article metadata records and one book identifier, not 808 appraised publications. Licensed textbooks, paid courses, podcasts, recordings and unindexed sources remain unavailable or unreviewed. No physician review, CME, official-board-item status or exhaustive global source coverage is claimed.
The discovery window was 2026/09/17–2026/10/17. 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.
See every registered source and its status → · Download this edition’s machine-readable coverage record
Educational, not medical advice. Discuss treatment choices with your clinician. Sources and evidence evolve; each dated edition preserves what was checked for that issue.