Home based, tailored intervention to reduce rate of falls after stroke (FAST): randomised trial.
Clemson L, Scrivener K, Lannin N, Ada L, Day S, Lin I, Isbel S, Cusick A, Gardner B, Preston E, Heller G, Dean CM, FAST Study Group
- DOI
- 10.1136/bmj-2025-085519
- Record issued
- 2026-08-15
- Engine
- 7.39.0
- Exported
- 2026-09-20
Prepared by Alpha1. This document is confidential: it is intended for the recipient it was shared with and must not be redistributed. The live record at alpha1science.com/verify/f4002020-46bf-4d04-a594-3d5faa38cd91 is authoritative.
How this rating was calculated
- IntegrityIntegrity concern ×2−1★
This Kaimen Rigor review uses Kaimen Rigor reviewers trained on a curated corpus of high-fidelity and retracted papers, with expert supervision and curation. It can still make mistakes; verify each finding against the source before relying on it.
This is a well-conducted and well-reported phase 3 randomized controlled trial with rigorous design, clear statistical methods, and transparent reporting. The main weakness is the lack of an explicit informed consent statement, which is a reporting gap rather than a substantive flaw. Minor copyedit issues and a few reporting details (e.g., outlier handling, assumptions verification) could be improved.
Both reviewers agreed on study type (interventional) and on most dimensions. Divergences were minor: outlier handling (R1 inadequate vs R2 adequate), assumptions verified (R1 inadequate vs R2 adequate), repository deposit (R1 inadequate vs R2 adequate), and limitations discussed (R1 adequate vs R2 inadequate). These were resolved conservatively where the paper lacked explicit detail. The statistics verification recomputed 4 tests consistently, but this is limited coverage; other statistics remain unverified.
Numerical inconsistencies
1 finding · worst lowValues that contradict each other or are impossible for the stated sample: recomputed p-values and test statistics, GRIM/GRIMMER checks on summary numbers, percentages against their own counts, totals against their parts, and estimates against their own confidence intervals.
- Internal contradictions in the reported numbersAssessed
Recomputed 4 tests: 4 consistent, 0 inconsistent; 2 recomputed directly from the reported test statistics, 2 via agent-written checks.
- CONSISTENTreported p = .020 · recomputed p = .020Recomputed incidence rate ratio 0.67 (95% CI 0.48–0.94), reported p=0.02
“incidence rate ratio 0.67, 95% confidence interval (CI) 0.48 to 0.94; P=0.02”
Taken as given: 0.48–0.94 is a two-sided 95% confidence interval for the incidence rate ratio of 0.67, not a range, an IQR, or a different interval level; the incidence rate ratio is a RATIO measure, so the interval is symmetric on the log scale; p=0.02 is the p for THIS estimate, not for another comparison in the same sentenceMethod: back the two-tailed p out of the log-scale CI width and compare it against the printed pHow we recomputed it: pCI(0.67, 0.48, 0.94, 1) - CONSISTENTreported p = .520 · recomputed p = .487Recomputed relative risk 0.94 (95% CI 0.79–1.12), reported p=0.52
“relative risk 0.94, 95% CI 0.79 to 1.12; P=0.52”
Taken as given: 0.79–1.12 is a two-sided 95% confidence interval for the relative risk of 0.94, not a range, an IQR, or a different interval level; the relative risk is a RATIO measure, so the interval is symmetric on the log scale; p=0.52 is the p for THIS estimate, not for another comparison in the same sentenceMethod: back the two-tailed p out of the log-scale CI width and compare it against the printed pHow we recomputed it: pCI(0.94, 0.79, 1.12, 1) - CONSISTENTreported p = .020 · recomputed p = .020Reviewers 1, 2Primary outcome: incidence rate ratio (IRR) for rate of falls
“incidence rate ratio 0.67, 95% confidence interval (CI) 0.48 to 0.94; P=0.02”
Taken as given: The IRR is 0.67 with 95% CI 0.48 to 0.94.; The CI is two-sided at 95%.; The p-value is two-tailed.Method: Recomputed p-value from the IRR and its 95% CI using the normal approximation for the log rate ratio.How we recomputed it: pCI(0.67, 0.48, 0.94, 1) - CONSISTENTreported p = .520 · recomputed p = .518Reviewers 1, 2Secondary outcome: proportion of fallers (relative risk)
“No (%) of fallers | 104 (56) | 109 (59) | RR 0.94 (0.79 to 1.12; 0.52); ARR 0.03 (−0.07 to 0.13; 0.52)”
Taken as given: The numbers 104 and 109 are the number of fallers in the experimental and control groups, respectively.; The group sizes are 186 and 184, so non-fallers are 82 and 75.; The p-value is from a chi-square test.Method: Recomputed p-value from the 2x2 table using Pearson's chi-square test.How we recomputed it: pChi2x2(104, 82, 109, 75)
- lowinternal contradictionThe abstract reports a 33% reduction in falls, but the annual fall rate in the experimental group is 1.8 vs 2.7 in control, which is a 33% reduction (1 - 1.8/2.7 = 0.333). This is consistent.
“representing a 33% reduction in falls (incidence rate ratio 0.67, 95% confidence interval (CI) 0.48 to 0.94; P=0.02)”
AbstractFind in source
Overstated conclusions
None foundConclusions that reach past what the paper's own results support — including a significance claim that no longer holds when the statistic is recomputed, and efficacy resting on an unvalidated surrogate endpoint.
Checked — nothing surfaced.
7 major claims checked against the paper's own evidence: all adequately supported.
- supportedReviewer 1The intervention reduced the rate of falls over 12 months.The primary outcome analysis shows a significant IRR of 0.67 (95% CI 0.48-0.94, P=0.02), supporting the claim.Evidence: Primary outcome: IRR 0.67 (95% CI 0.48-0.94; P=0.02)
“a significant between group difference was seen in the rate of falls in favour of the experimental group, representing a 33% reduction in falls (incidence rate ratio 0.67, 95% confidence interval (CI) 0.48 to 0.94; P=0.02)”
AbstractFind in source - supportedReviewer 1The intervention improved community participation, self-efficacy, mobility, and balance.Secondary outcomes show significant between-group differences in these domains at 12 months, supporting the claim.Evidence: Secondary outcomes: LLFDI limitation mean diff 3% (1-6; P=0.02), self-efficacy 0.6 (0.2-1.0; P=0.004), fast walking speed 0.13 (0.06-0.19; P<0.001), Step Test 0.06 (0.01-0.12; P=0.03)
“The main between group differences in favour of the experimental group were in community participation (Late Life Function and Disability Instrument disability limitation: mean difference 3% (95% CI 1% to 6%); P=0.02), self-efficacy (mean difference 0.6 (0.2 to 1.0); P=0.004), mobility (fast walking speed: mean difference 0.13 (0.06 to 0.19) m/s (P<0.001); preferred walking speed: 0.06 (0.02 to 0.10) m/s (P=0.02)), and balance (Step Test: mean difference 0.06 (0.01 to 0.12) steps/s; P=0.03)”
AbstractFind in source - supportedReviewer 1The intervention did not significantly reduce the proportion of participants having a fall.The secondary analysis shows no significant difference in fallers (RR 0.94, 95% CI 0.79-1.12, P=0.52), supporting the claim.Evidence: Secondary outcome: RR 0.94 (0.79-1.12; P=0.52)
“No significant between group difference was seen in the number of participants having a fall (absolute risk reduction 0.03, 95% CI −0.07 to 0.13; P=0.52)”
AbstractFind in source - supportedReviewer 1This is the first effectiveness trial to show prevention of falls after stroke.The paper states that only three previous trials found no effect, and this trial found a significant reduction, supporting the claim.Evidence: Comparison with other studies: 'Only three previous randomised trials have investigated the effectiveness of a non-drug intervention to reduce falls after stroke and reported the annual rate of falls. All found no effect, making ours the first effectiveness trial to find a reduction in the rate of falls relative to the control group after stroke.'
“Only three previous randomised trials have investigated the effectiveness of a non-drug intervention to reduce falls after stroke and reported the annual rate of falls. All found no effect, making ours the first effectiveness trial to find a reduction in the rate of falls relative to the control group after stroke.”
DiscussionFind in source - supportedReviewer 2The intervention reduces the rate of falls after stroke.The primary outcome shows a significant reduction in fall rate with IRR 0.67 (95% CI 0.48-0.94, P=0.02).Evidence: Primary outcome analysis: IRR 0.67 (95% CI 0.48 to 0.94; P=0.02).
“a significant between group difference was seen in the rate of falls in favour of the experimental group, representing a 33% reduction in falls (incidence rate ratio 0.67, 95% confidence interval (CI) 0.48 to 0.94; P=0.02)”
AbstractFind in source - supportedReviewer 2The intervention improves community participation, self-efficacy, mobility, and balance.Secondary outcomes show significant improvements in these domains at 12 months.Evidence: Table 3 shows significant between-group differences for LLFDI disability limitation, self-efficacy, fast walking speed, preferred walking speed, SPPB, and Step Test.
“At month 12, we found a statistically significant between group difference in community participation (LLFDI disability limitation: mean difference 3% (95% CI 1% to 6%); P=0.02), self-efficacy (mean difference 0.6 (0.2 to 1.0); P=0.004), mobility (fast walking speed: mean difference 0.13 (0.06 to 0.19) m/s (P<0.001); preferred walking speed: mean difference 0.06 (0.02 to 0.10) m/s (P=0.02); self-reported walking distance: mean difference 0.3 (0 to 0.5) out of 4 points (P=0.03)), and balance (SPPB: mean difference 0.3 (0.0 to 0.6) out of 4 points (P=0.04); Step Test: mean difference 0.06 (0.01 to 0.12) steps/s (P=0.03))”
ResultsFind in source - supportedReviewer 2The intervention is the first effective intervention to prevent falls after stroke.The paper states that previous trials found no effect, and this trial found a significant reduction.Evidence: Discussion states 'Only three previous randomised trials have investigated the effectiveness of a non-drug intervention to reduce falls after stroke and reported the annual rate of falls. All found no effect, making ours the first effectiveness trial to find a reduction in the rate of falls relative to the control group after stroke.'
“All found no effect, making ours the first effectiveness trial to find a reduction in the rate of falls relative to the control group after stroke.”
DiscussionFind in source
Efficacy claim is anchored to an adequate endpoint and a meaningful effect.
- ADEQUATESurrogate endpointThe primary outcome is rate of falls, which is a hard clinical outcome (an event-based measure), not a surrogate biomarker. Therefore, the surrogate criterion is not applicable.
“The primary outcome was rate of falls over 12 months.”
- ADEQUATEEffect sizeThe primary effect is a 33% reduction in falls (incidence rate ratio 0.67, 95% CI 0.48 to 0.94; P=0.02). This is statistically significant and the paper anchors it as clinically worthwhile by noting that no prior trial had shown such a reduction, and by linking it to improvements in self-efficacy, mobility, and balance. The effect size is not a small fraction of normal but a meaningful reduction in a clinically important outcome.
“a 33% reduction in falls (incidence rate ratio 0.67, 95% confidence interval (CI) 0.48 to 0.94; P=0.02)”
Data authenticity concerns
1 finding · worst lowAn adversarial read for patterns associated with data that may not be genuine: results that look too clean, implausibly large effects, duplicated data or images, and methods that do not match the results reported.
- Other integrity concernAssessed
2 integrity concerns flagged (0 high).
- lowotherThe paper reports that 370 participants were enrolled, but the CONSORT flow diagram (Fig 1) is not shown in the text; the numbers in Table 1 sum to 370 (186+184).
“769 people with stroke were screened and 370 enrolled.”
Results ¶1Find in source
Reporting gaps
None foundRequired detail the manuscript never states — study design, biological variables, ethics approval and consent, key resources, statistical reporting, data and code availability, and overall transparency.
Checked — nothing surfaced.
The introduction cites three prior randomized trials that found no effective non-drug interventions to reduce falls after stroke, and explains how the current intervention differs by integrating exercise into daily activities. The rationale is logically linked to the study objectives and hypotheses. Limitations of prior work are implicitly addressed by the design, though not explicitly detailed.
“In three randomised trials, the non-drug interventions investigated to reduce falls after stroke were not effective and involved either home modifications or exercise that was segregated from daily activities.”
“The intervention combined exercise, home safety, and a community mobility goal, but most important is that the exercise is habit forming because it is integrated into daily activities.”
“No effective intervention that prevents falls after stroke is available.”
“The intervention combined exercise, home safety, and a community mobility goal, but most important is that the exercise is habit forming because it is integrated into daily activities.”
“In three randomised trials, the non-drug interventions investigated to reduce falls after stroke were not effective and involved either home modifications or exercise that was segregated from daily activities.”
Randomization used an offsite automated service with stratification and block randomization, and allocation was concealed. Blinding of outcome assessors and statistician was described, with a rationale for not blinding participants/therapists. A priori sample size calculation was provided. Inclusion/exclusion criteria were pre-specified. Outlier handling is not explicitly described, but the analysis population (ITT) and missing data approach are defined. Controls are appropriate (usual care). Independent replication is not applicable for a single pivotal trial.
“We used an offsite, independent, and automated service (Clinical Trials Centre, University of Sydney) to generate the sequence with stratification and block randomisation, to conceal the schedule from the recruiter, and to reveal allocation to the trial coordinators.”
“Researchers masked to group allocation measured outcomes at baseline, at six months (end of intervention), and at 12 months in the home. Masking participants and therapists to group allocation was not possible. A statistician masked to group allocation did the statistical analysis.”
“Assuming a 15% loss to follow-up, with a two sided level of significance of 5% and power of 80%, we needed a sample size of 185 per group to detect a 30% lower rate of falls in the experimental group than in the control group.”
“We used an offsite, independent, and automated service (Clinical Trials Centre, University of Sydney) to generate the sequence with stratification and block randomisation, to conceal the schedule from the recruiter, and to reveal allocation to the trial coordinators.”
“Researchers masked to group allocation measured outcomes at baseline, at six months (end of intervention), and at 12 months in the home. Masking participants and therapists to group allocation was not possible. A statistician masked to group allocation did the statistical analysis.”
“Assuming a 15% loss to follow-up, with a two sided level of significance of 5% and power of 80%, we needed a sample size of 185 per group to detect a 30% lower rate of falls in the experimental group than in the control group.”
The paper reports sex, age, time since stroke, comorbidities, and other relevant characteristics in Table 1. Both sexes are included, so sex justification is not applicable. Age and health status are reported. Species/strain and housing conditions are not applicable for a human trial.
“Mean (SD) age , years | 75 (10) | 76 (9) | 76 (11) | 79 (8) | | Female sex | 87 (47) | 82 (45) | 12 (50) | 7 (37)”
“Mean (SD) comorbidities (Groll scale 0-18) | 4.7 (2.5) | 4.4 (2.0) | 5.1 (2.8) | 4.2 (2.4)”
“Female sex | 87 (47) | 82 (45)”
“Mean (SD) age , years | 75 (10) | 76 (9)”
The study was approved by a named ethics committee (Macquarie University Human Research Ethics Committee) with protocol numbers. Informed consent is implied by the inclusion criteria 'capable of providing consent' and the ethics approval, though a specific consent statement is not explicitly quoted. Regulatory compliance is stated via adherence to CONSORT and the Declaration of Helsinki is not explicitly mentioned, but the ethics approval suffices.
“The study was approved by the Macquarie University Human Research Ethics Committee Human Research Ethics Committee (5201500456 and 520231547551507) and respective hospital institutional committees”
“capable of providing consent (fewer than five errors on the Short Portable Mental Status Questionnaire)”
“The study was approved by the Macquarie University Human Research Ethics Committee Human Research Ethics Committee (5201500456 and 520231547551507) and respective hospital institutional committees”
The intervention components are described in detail, including the LiFE program and the Westmead Home Safety Assessment. The statistical software (R and SPSS) is identified with versions. No antibodies, cell lines, or other biological reagents are used, so those criteria are not applicable. The intervention is a behavioral intervention, not a drug or device, so reagents_identified is not applicable.
“We used R version 4.4.2 to analyse falls data and IBM SPSS version 30 to analyse other secondary data.”
“The experimental group received the FAST intervention, a home based intervention consisting of three components—a habit forming functional exercise, home fall hazard reduction, and goal directed community mobility coaching—which were tailored to their level of disability.”
“We used R version 4.4.2 to analyse falls data and IBM SPSS version 30 to analyse other secondary data.”
The primary analysis used negative binomial regression, and secondary analyses used ANOVA and Mann-Whitney U tests, all named. Exact p-values are reported for primary and secondary outcomes. Effect sizes are reported with 95% confidence intervals. Software is identified. Data presentation includes per-group n and dispersion measures. Mathematical plausibility checks were not performed due to continuous data and large N, but no obvious errors were noted.
“We analysed the between group difference in annual rate of falls by using negative binomial regression and presented it as an incidence rate ratio. We analysed the between group difference in proportion of fallers by using the χ 2 test and presented it as relative risk and risk difference.”
“incidence rate ratio 0.67, 95% confidence interval (CI) 0.48 to 0.94; P=0.02”
“community participation (Late Life Function and Disability Instrument disability limitation: mean difference 3% (95% CI 1% to 6%); P=0.02)”
“the between group difference at 12 months was an incidence rate ratio of 0.67 (95% confidence interval (CI) 0.48 to 0.94; P=0.02)”
“We analysed interval secondary outcomes by using analysis of variance and presented them as mean between group differences with 95% confidence intervals. We analysed ordinal secondary outcomes by using the non-parametric Mann-Whitney U statistic but presented them as mean between group differences with 95% confidence intervals.”
The data availability statement indicates that the statistical analysis plan and data and analysis code files are lodged on the Sydney University eScholarship Repository, providing a concrete access route. Repository deposit is mentioned, though no DOI or accession number is given. Code sharing is implied by the statement. For patient-level data, repository deposit and accession numbers are not applicable.
The trial is registered with ACTRN12619001114134. The paper states it is reported according to the updated CONSORT statement 2025. All pre-specified outcomes are reported, including non-significant ones. Limitations are discussed. Conclusions are proportional to the evidence. Funding and competing interests are disclosed.
“Australian New Zealand Clinical Trials Registry ACTRN12619001114134”
“The study is reported in accordance with the updated CONSORT statement 2025.”
“Trial registration Australian New Zealand Clinical Trials Registry ACTRN12619001114134.”
“The study is reported in accordance with the updated CONSORT statement 2025.”
“The main limitation of the study is the large number of therapists needed to provide intervention across multiple sites and regions.”
Registered (1 ID: ANZCTR). Reporting guideline cited: CONSORT.
Broken references and links
None foundReferences checked against Crossref, OpenAlex and Retraction Watch for retractions and resolvability, plus declared data and code links probed for whether they resolve to content matching the paper.
Checked — nothing surfaced.
Checked 32 references by DOI: 31 verified — 1 no DOI (shown, not verified).
- NO DOINational Stroke Audit: Rehabilitation Services Report 2018No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
1 data/code link found; not probed for liveness in this run.
- datahttps://ses.library.usyd.edu.au/handle/2123/33822.3UNVERIFIEDHTTP 403Liveness indeterminate — content not checked.
Copyediting
5 minorWording, consistency and formatting errors that need correcting before submission.
No major wording or formatting errors. 5 minor suggestions below.
5 copyedit issues flagged: mostly consistency, clarity, typo.
- MINORconsistencyEthics statements“Macquarie University Human Research Ethics Committee Human Research Ethics Committee”→ Remove duplicate 'Human Research Ethics Committee'.Duplicate phrase.
- MINORclarityAbstract, Results“The main between group differences in favour of the experimental group were in community participation (Late Life Function and Disability Instrument disability limitation: mean difference 3% (95% CI 1% to 6%); P=0.02)”→ Consider clarifying that the mean difference is in percentage points.Potential ambiguity.
- MINORtypoEthics statements“Macquarie University Human Research Ethics Committee Human Research Ethics Committee”→ Remove duplicate 'Human Research Ethics Committee'Duplicate phrase in ethics approval statement.
- MINORconsistencyResults, primary outcome“annual fall rate for the experimental group was 1.8 (SD 3.0; range 0-19; total 327) per participant”→ Clarify that 'total 327' is the total number of falls, not the number of participants.Ambiguous phrasing.
- MINORclarityMethods, Statistical analysis“We analysed ordinal secondary outcomes by using the non-parametric Mann-Whitney U statistic but presented them as mean between group differences with 95% confidence intervals.”→ Clarify how mean differences were derived from Mann-Whitney U tests.Potential methodological clarity issue.
The published work is robust and generally trustworthy, but an informed reader should note the missing explicit informed consent statement and the lack of explicit outlier handling and assumption checks. These are reporting gaps that could warrant a correction or clarification, but they do not undermine the main conclusions. The copyedit issues are minor and do not affect scientific integrity.
- 1.HIGHethicsAdd an explicit statement in the Ethics statements section describing how informed consent was obtained (e.g., written informed consent from all participants).Informed consent is a required reporting element for human research; its absence is a reporting gap that should be corrected.
- 2.HIGHreportingIn the Methods, Statistical analysis section, explicitly state how outliers were handled (e.g., none excluded) and how assumptions for statistical tests (e.g., normality) were verified.Both reviewers flagged these as inadequately reported; explicit statements would strengthen the statistical reporting.
- 3.MEDIUMcopyeditIn the Ethics statements, remove the duplicate phrase 'Human Research Ethics Committee' so it reads 'Macquarie University Human Research Ethics Committee'.The duplicate phrase is a copyedit error that should be fixed for clarity.
- 4.MEDIUMcopyeditIn the Abstract, Results, clarify that the mean difference of 3% is in percentage points (e.g., 'mean difference 3 percentage points').The current phrasing is ambiguous and could be misinterpreted.
- 5.MEDIUMcopyeditIn the Results, primary outcome, clarify that 'total 327' refers to the total number of falls, not participants.The phrasing is ambiguous and could confuse readers.
- 6.MEDIUMreportingIn the Methods, Statistical analysis, clarify how mean differences were derived from Mann-Whitney U tests for ordinal outcomes.The current description is methodologically unclear and could be questioned by reviewers.
- 7.MEDIUMreportingExpand the limitations section to include potential biases such as lack of blinding of participants/therapists and the impact of missing data due to COVID-19.A more comprehensive limitations discussion would strengthen transparency.
- 8.MEDIUMdata codeConsider providing a DOI or persistent identifier for the repository deposit in the Data Availability statement.A DOI would enhance accessibility and citability of the data/code.
- 9.LOWreportingAdd a note on how missing data were handled (e.g., complete-case analysis) in the statistical methods.Clarifying missing data handling is important for reproducibility.
- 10.LOWreportingProvide more detail on the randomization sequence generation (e.g., block size) in the Methods.Additional detail would improve transparency of the randomization process.
- 11.LOWreportingInclude a statement on patient and public involvement, even if it was not done, to explain its absence.This is a common reporting expectation and would address a potential reviewer comment.
- 12.LOWreportingConsider providing a link to the full statistical analysis plan in the data availability statement.This would enhance transparency and reproducibility.
The star rating is the report’s one-glance summary. Every paper starts at 5★ and loses stars for the concrete problems the review finds — so a rating is never a vague average, it’s a running total you can read line by line under “How this rating was calculated.”
- Reporting — 8 dimensionseach dimension that fully fails−½★
- each dimension partially met−¼★
- Statistics · Integrity · Claimseach serious problem−1★
- each medium problem−½★
- Citationseach retracted or unverifiable reference−¼★
- Copyeditonly when the manuscript needs a full edit−½★
The rating never drops below 1★, and a demonstrable critical failure (an impossible statistic, a proven ethics violation) caps it at 1★ on its own — so the stars can never look healthy when the verdict is CRITICAL.
The rating draws on a panel of agents. Three independent Kaimen Rigor reviewers grade the eight dimensions below across several independent passes (the shown verdict is their majority vote — steadier than any single run), isolate the paper’s major claims and check its own evidence backs them, and flag integrity concerns. Alongside them, a citation agent resolves every reference against Crossref, OpenAlex, and Retraction Watch; a statistics agent recomputes reported tests; and rule-based checks verify that declared data/code links actually resolve. Full text is required — an abstract-only submission is not analyzed.
Graded against NIH, MDAR, ARRIVE 2.0, CONSORT, EQUATOR, and RRID guidelines. A dimension that doesn’t apply to the study type is skipped, never penalized.
This Kaimen Rigor review is model-assisted and is not a substitute for formal expert review. It complements human evaluation by surfacing potential methodological concerns — verify each finding against the source.