Edaravone dexborneol versus placebo on functional outcomes in patients with acute ischaemic stroke undergoing endovascular thrombectomy (TASTE-2): randomised controlled trial.
Wang C, Gu H, Huo X, Yuan B, Li S, Xu J, Jiang Y, Jing J, Yao X, Li Z, Long F, Ma Z, Zhuang X, Xu L, Jin Y, Huang W, Zhang Y, Wen J, Wang A, Pan Y, Ye W, Yu W, Cheng A, Wang M, Dong Q, Xu A, Wang N, Yang Y, Meng X, Liu L, Zhao X, Li H, Miao Z, Li Z, Wang Y, TASTE-2 investigators
- DOI
- 10.1136/bmj-2025-086850
- 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/37254882-36ab-470d-bdc5-bf32ca5ba3a3 is authoritative.
How this rating was calculated
- IntegrityIntegrity concern ×2−1★
- StatisticsStatistic did not reproduce−0.5★
- ClaimsUnsupported claim (uncorroborated)−0.5★
- CitationsUnresolved reference−0.25★
- The numeric-impossibility checks (GRIM/GRIMMER/DEBIT/SPRITE) did not run: 2 reported means were read, and their group size is not stated where the values are printed (this source has no machine-readable table structure). These checks need the count the mean was averaged over, so none was performed.
- 01Printed percentage does not match its own count
33% does not match the reported count 354/690
“354/690 (33.0)”
Table 2Find in source - 02Conclusion not supported by the paper’s own evidence
Edaravone dexborneol may exert a 'penumbra freezing' effect by mitigating ischaemia-reperfusion injury.
This suggests that the drug may exert a 'penumbra freezing' effect by mitigating ischaemia-reperfusion injury and secondary inflammation when administered before and after endovascular therapy.
Discussionreviewer’s wording
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 rigorously designed and reported phase 3 RCT of edaravone dexborneol in acute ischaemic stroke patients undergoing thrombectomy. The paper demonstrates strong methodology across all eight dimensions, with minor reporting issues (a percentage typo in Table 2, missing explicit CONSORT statement, and a speculative mechanistic claim) that do not undermine the core findings.
Both reviewers independently scored all eight dimensions as pass, with high agreement. The study type is interventional (RCT). The statistics component recomputed 8 tests, finding 7 consistent and 1 inconsistent (the Table 2 percentage typo). The citation check found 1 reference not found in registry. The claim audit flagged one unsupported mechanistic claim. These deterministic findings are folded into the relevant dimensions and action items.
Numerical inconsistencies
2 findings · worst highValues 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.
- Summary statistic impossible for the stated N (GRIM/GRIMMER)Recomputed
- Internal contradictions in the reported numbersAssessed
Recomputed 7 tests: 7 consistent, 0 inconsistent; 2 recomputed directly from the reported test statistics, 5 via agent-written checks. 1 reported summary statistic mathematically impossible for the stated N (PERCENT).
- PERCENT33% does not match the reported count 354/690
“354/690 (33.0)”
Table 2Find in source
- CONSISTENTreported p = .050 · recomputed p = .048Recomputed risk ratio 1.11 (95% CI 1.00–1.23), reported p=0.05
“risk ratio 1.11, 95% confidence interval (CI) 1.00 to 1.23; P=0.05”
Taken as given: 1.00–1.23 is a two-sided 95% confidence interval for the risk ratio of 1.11, not a range, an IQR, or a different interval level; the risk ratio is a RATIO measure, so the interval is symmetric on the log scale; p=0.05 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(1.11, 1, 1.23, 1) - CONSISTENTreported p = .050 · recomputed p = .051Recomputed odds ratio 1.24 (95% CI 1.00–1.54), reported p=0.05
“odds ratio 1.24, 95% CI 1.00 to 1.54; P=0.05”
Taken as given: 1.00–1.54 is a two-sided 95% confidence interval for the odds ratio of 1.24, not a range, an IQR, or a different interval level; the odds ratio is a RATIO measure, so the interval is symmetric on the log scale; p=0.05 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(1.24, 1, 1.54, 1) - CONSISTENTreported p = .050 · recomputed p = .048Reviewers 1, 2Primary outcome risk ratio p-value from reported RR and 95% CI
“risk ratio 1.11, 95% confidence interval (CI) 1.00 to 1.23; risk difference 5.4%, 95% CI 0.1% to 10.7%; P=0.05”
Taken as given: The risk ratio is 1.11 with 95% CI 1.00 to 1.23.; The CI is two-sided at 95%.; The p-value is for the risk ratio from a logistic regression model.Method: Recomputed two-tailed p from the reported RR and CI using the normal approximation for the log-RR.How we recomputed it: pCI(1.11, 1.00, 1.23, 1) - CONSISTENTreported p = .530 · recomputed p = .511Reviewer 1Serious adverse events risk ratio p-value from reported RR and 95% CI
“risk ratio 1.06, 0.89 to 1.26; risk difference 1.5%, −3.2% to 6.2%: P=0.53”
Taken as given: The risk ratio is 1.06 with 95% CI 0.89 to 1.26.; The CI is two-sided at 95%.; The p-value is for the risk ratio from a logistic regression model.Method: Recomputed two-tailed p from the reported RR and CI using the normal approximation for the log-RR.How we recomputed it: pCI(1.06, 0.89, 1.26, 1) - CONSISTENTreported p = .050 · recomputed p = .046Reviewer 1Primary outcome risk difference p-value from reported RD and 95% CI
“risk difference 5.4%, 95% CI 0.1% to 10.7%; P=0.05”
Taken as given: The risk difference is 5.4% with 95% CI 0.1% to 10.7%.; The CI is two-sided at 95%.; The p-value is for the risk difference from a logistic regression model.Method: Recomputed two-tailed p from the reported RD and CI using the normal approximation.How we recomputed it: pCI(5.4, 0.1, 10.7, 0) - CONSISTENTreported p = .530 · recomputed p = .511Reviewer 2Primary safety outcome: risk ratio 1.06, 95% CI 0.89 to 1.26, P=0.53
“risk ratio 1.06, 0.89 to 1.26; P=0.53”
Taken as given: The risk ratio is 1.06; The 95% CI is 0.89 to 1.26; The CI is for a ratio (log scale); The test is two-sidedMethod: p-value derived from the reported risk ratio and its 95% confidence interval using the pCI function for a ratio on the log scale.How we recomputed it: pCI(1.06, 0.89, 1.26, 1) - CONSISTENTreported p > .990 · recomputed p = 1.000Reviewer 2All-cause mortality: risk ratio 1.00, 95% CI 0.79 to 1.27, P>0.99
“All cause mortality within 90 days: 114/690 (16.5) vs 111/672 (16.5); risk ratio 1.00, 0.79 to 1.27; P>0.99”
Taken as given: The risk ratio is 1.00; The 95% CI is 0.79 to 1.27; The CI is for a ratio (log scale); The test is two-sidedMethod: p-value derived from the reported risk ratio and its 95% confidence interval using the pCI function for a ratio on the log scale. The reported P>0.99 is consistent with a p-value very close to 1.How we recomputed it: pCI(1.00, 0.79, 1.27, 1)
- lowinternal contradictionIn Table 2, the adverse events row for the edaravone dexborneol group shows 354/690 (33.0%), but 354/690 = 51.3%, not 33.0%. This appears to be a typographical error in the percentage.
“Adverse events | 354/690 (33.0) | 332/672 (49.4) | 1.04 (0.93 to 1.15) | 1.9 (−3.4 to 7.2) | 0.48”
Table 2Find in source - lowinternal contradictionThe recurrence of stroke outcome uses denominators 686 and 667, which are smaller than the group totals (690 and 672), without explanation in the table.
“Recurrence of stroke | 10/686 (1.5) | 8/667 (1.2)”
Table 2Find in source
Overstated conclusions
2 findings · worst highConclusions 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.
- Conclusions not supported by the paper’s own evidenceAssessed
- Conclusions only partially backed by the presented evidenceAssessed
8 major claims checked against the paper's own evidence: 1 not fully backed by the presented evidence (unsupported or overstated), 3 only partially supported (evidence backs part of the claim; gaps or caveats remain).
- unsupportedReviewer 2Edaravone dexborneol may exert a 'penumbra freezing' effect by mitigating ischaemia-reperfusion injury.This mechanistic claim is speculative and not directly tested in this trial. The paper presents no biomarker or imaging data to support this mechanism; it is inferred from subgroup analyses. The claim is unsupported by the evidence presented.Evidence: No direct evidence; the claim is based on subgroup analysis and prior preclinical studies.
This suggests that the drug may exert a 'penumbra freezing' effect by mitigating ischaemia-reperfusion injury and secondary inflammation when administered before and after endovascular therapy.
Discussionreviewer’s wording - partialReviewers 1, 2Edaravone dexborneol improves functional independence at 90 days compared with placebo in patients with acute ischaemic stroke undergoing endovascular thrombectomy.The primary outcome reached borderline significance (P=0.05) with a risk ratio of 1.11 (95% CI 1.00 to 1.23), which is not a definitive improvement.Evidence: Primary outcome: 379/689 (55.0%) vs 333/671 (49.6%), RR 1.11 (1.00-1.23), P=0.05.
“379 (55.0%) of 689 patients in the edaravone dexborneol group and 333 (49.6%) of 671 patients in the placebo group achieved functional independence on day 90 (risk ratio 1.11, 95% confidence interval (CI) 1.00 to 1.23; P=0.05”
AbstractFind in source - partialReviewer 1The treatment effect is clinically meaningful despite borderline significance.The paper argues that a 5.4% absolute increase is clinically meaningful based on external references, but the primary outcome is not statistically significant at the conventional level.Evidence: Discussion cites STAIR recommendations that absolute effect sizes of 2-8% are acceptable.
“The Stroke Treatment Academic Industry Round table has suggested that for cytoprotective therapies, absolute effect sizes of 2-8% (dichotomous) would be acceptable.”
DiscussionFind in source - partialReviewer 2The study population is representative of the general population undergoing endovascular thrombectomy.The authors compare baseline characteristics to HERMES and AURORA studies and find alignment, but they also note limitations: only Chinese patients, high eTICI 2B-3 rates (95%) due to centre selection, and only 9.6% of screened patients enrolled. The claim is partially supported with caveats.Evidence: Discussion: 'the median age, baseline NIHSS score, baseline ASPECTS, and distribution of arterial occlusion sites at admission closely aligned with those reported in the HERMES and AURORA studies'.
“Additionally, the median age, baseline NIHSS score, baseline ASPECTS, and distribution of arterial occlusion sites at admission closely aligned with those reported in the HERMES and AURORA studies, indicating that the study participants were representative of the general population undergoing endovascular thrombectomy.”
DiscussionFind in source - supportedReviewer 1The effect is primarily driven by the subgroup with mismatch present at admission.The subgroup analysis shows a significant interaction (P=0.003) with a larger effect in the mismatch subgroup.Evidence: Subgroup analysis: 55.5% (178/321) vs 42.9% (134/312), RR 1.29 (1.10-1.52), P for interaction=0.003.
“Patients with mismatch at admission (defined as NIHSS score ≥10 and ASPECTS ≥9 or NIHSS score ≥20 and ≥7) were more likely to achieve functional independence in the subgroup analysis (55.5% (178/321) versus 42.9% (134/312); risk ratio 1.29, 1.10 to 1.52; risk difference 13.0%, 5.6% to 20.3%; P for interaction=0.003).”
AbstractFind in source - supportedReviewer 1Edaravone dexborneol does not increase safety concerns.Serious adverse events were similar between groups (27.2% vs 25.7%, P=0.53), and no significant differences in secondary safety outcomes.Evidence: Serious adverse events: 188/690 (27.2%) vs 173/672 (25.7%), RR 1.06 (0.89-1.26), P=0.53.
“The rates of serious adverse events were similar in the two groups (27.2% (188/690) versus 25.7% (173/672); risk ratio 1.06, 0.89 to 1.26; risk difference 1.5%, −3.2% to 6.2%: P=0.53).”
AbstractFind in source - supportedReviewer 2The effect is primarily driven by the subgroup with mismatch at admission.The subgroup analysis shows a significant interaction (P for interaction=0.003) with a larger effect in the mismatch subgroup (risk ratio 1.29, 95% CI 1.10 to 1.52). This is appropriately described as exploratory.Evidence: Subgroup analysis: mismatch subgroup (55.5% vs 42.9%, risk ratio 1.29, 95% CI 1.10 to 1.52, P for interaction=0.003).
Patients with mismatch at admission... were more likely to achieve functional independence in the subgroup analysis (55.5% (178/321) versus 42.9% (134/312); risk ratio 1.29, 1.10 to 1.52; risk difference 13.0%, 5.6% to 20.3%; P for interaction=0.003)
Abstractreviewer’s wording - supportedReviewer 2Edaravone dexborneol is safe with no increased safety concerns.The primary safety outcome (serious adverse events) showed no significant difference (27.2% vs 25.7%, P=0.53), and secondary safety outcomes were similar. The claim is supported.Evidence: Primary safety outcome: serious adverse events 27.2% vs 25.7%, risk ratio 1.06 (95% CI 0.89 to 1.26), P=0.53.
“The rates of serious adverse events were similar in the two groups (27.2% (188/690) versus 25.7% (173/672); risk ratio 1.06, 0.89 to 1.26; risk difference 1.5%, −3.2% to 6.2%: P=0.53)”
AbstractFind in source
Efficacy claim is anchored to an adequate endpoint and a meaningful effect.
- ADEQUATESurrogate endpointThe primary endpoint is functional independence at 90 days measured by the modified Rankin Scale (mRS 0-2), which is a validated clinical outcome scale for stroke recovery. This is a hard clinical outcome, not a surrogate biomarker. The trial also reports target engagement indirectly through the mechanism of action and prior trials, but the primary endpoint itself is clinically meaningful.
“The primary efficacy outcome was the proportion of patients achieving functional independence, characterised by a modified Rankin Scale score of 0-2 at 90 days.”
- ADEQUATEEffect sizeThe primary effect size is a 5.4% absolute increase in functional independence (55.0% vs 49.6%), with a risk ratio of 1.11 (95% CI 1.00-1.23) and P=0.05. The authors anchor this as clinically meaningful by citing Stroke Treatment Academic Industry Roundtable recommendations that absolute effect sizes of 2-8% are acceptable for cytoprotective therapies, and they note that even 2% improvements can have significant public health impact. The effect is statistically borderline but explicitly anchored to clinical meaningfulness.
“The Stroke Treatment Academic Industry Round table has suggested that for cytoprotective therapies, absolute effect sizes of 2-8% (dichotomous) would be acceptable. Although the primary efficacy outcome reached only borderline statistical significance, this magnitude of difference may still represent a clinically meaningful benefit...”
Data authenticity concerns
None foundAn 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.
Checked — nothing surfaced.
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 the TASTE trial and preclinical studies, provides a logical rationale linking the drug's cytoprotective mechanism to the need for adjunctive therapy in endovascular thrombectomy, and explicitly states the gap: 'whether edaravone dexborneol confers additional cytoprotective benefits in patients undergoing reperfusion therapy... remains uncertain.' Limitations of prior work (e.g., TASTE trial excluded reperfusion patients) are addressed.
“In the TASTE trial, a phase 3 clinical study, treatment with edaravone dexborneol (37.5 mg twice daily) showed superior efficacy and a favourable safety profile compared with edaravone alone in improving functional outcomes in patients with acute ischaemic stroke who did not receive reperfusion therapy.”
“Nevertheless, whether edaravone dexborneol confers additional cytoprotective benefits in patients undergoing reperfusion therapy—a population that may derive the greatest potential benefit from cytoprotective agents—remains uncertain.”
“However, translating basic research into effective stroke treatments has been a challenge in the development of cytoprotective agents.”
“In the TASTE trial, a phase 3 clinical study, treatment with edaravone dexborneol (37.5 mg twice daily) showed superior efficacy and a favourable safety profile compared with edaravone alone in improving functional outcomes in patients with acute ischaemic stroke who did not receive reperfusion therapy.”
“Nevertheless, whether edaravone dexborneol confers additional cytoprotective benefits in patients undergoing reperfusion therapy—a population that may derive the greatest potential benefit from cytoprotective agents—remains uncertain.”
Randomisation used random permuted fixed-size blocks generated centrally with SAS 9.4; allocation concealment and blinding of participants and investigators were maintained. A sample size calculation with 90% power and 5% dropout was provided. Inclusion/exclusion criteria were detailed, and the modified intention-to-treat population was defined. Outlier handling is addressed through the pre-specified analysis population and sensitivity analyses including multiple imputation.
“The randomisation sequence numbers were generated centrally by using random permuted, fixed size blocks methods with SAS 9.4 software.”
“Complete blinding of participants and site investigators, as well as strict allocation concealment, was maintained throughout the entire study.”
“a total estimated sample size of 1362 patients (with a predicted 5% dropout rate) with an equal distribution of 681 patients per group would yield a statistical power of 90% to detect the superiority of edaravone dexborneol over placebo in improving neurological function at a two sided α significance level of P=0.05.”
“The randomisation sequence numbers were generated centrally by using random permuted, fixed size blocks methods with SAS 9.4 software.”
“Edaravone dexborneol and placebo were prepared as colourless solutions in vials that were visually identical.”
“On the basis of these assumptions, a total estimated sample size of 1362 patients (with a predicted 5% dropout rate) with an equal distribution of 681 patients per group would yield a statistical power of 90% to detect the superiority of edaravone dexborneol over placebo in improving neurological function at a two sided α significance level of P=0.05.”
Table 1 reports sex (64.0% male), age (median 67 years), body mass index, medical history, and clinical characteristics. Demographics include ethnicity (Han vs. others). Sex is reported for both groups, and since both sexes are enrolled, sex_justified is not applicable. Age, weight (via BMI), and health status are reported. Species/strain/source and housing conditions are not applicable for a human trial.
The paper states approval by the ethics committee at Beijing Tiantan Hospital with a protocol number (KY2021-177-01) and endorsement by each site. Written informed consent was obtained from patients or legal representatives. Regulatory compliance is implied through adherence to ethical principles, though not explicitly naming a framework; however, the named approval and consent are sufficient.
“The study was approved by the ethics committee at Beijing Tiantan Hospital (No KY2021-177-01) and each participating site.”
“Written informed consent was obtained from the patients or their legal representatives before the study drugs were used”
“The study was approved by the ethics committee at Beijing Tiantan Hospital (No KY2021-177-01) and each participating site.”
“Written informed consent was obtained from the patients or their legal representatives before the study drugs were used”
“The trial was approved by a central institutional review board and endorsed by the ethics committee of each research site; written informed consent was obtained from the patients or their legal representatives before the drugs were used.”
The investigational product is clearly described: edaravone dexborneol 37.5 mg (edaravone 30 mg, (+)-dexborneol 7.5 mg), twice daily for 10-14 days, with manufacturer (Simcere Pharmaceutical Group) stated. The placebo is described as containing excipients only. Software tools are identified: iStroke version 3.13, SAS version 9.4. Antibodies, cell lines, mycoplasma testing, and organisms are not applicable for this clinical trial. Reagents are covered by the investigational product description.
“Patients allocated to the intervention group were given intravenous edaravone dexborneol injection 37.5 mg, comprising 30 mg of edaravone and 7.5 mg of (+)−dexborneol, twice a day over a consecutive period of 10-14 days lasting for 30 minutes each time”
“We used SAS version 9.4 for statistical analyses”
“Patients allocated to the intervention group were given intravenous edaravone dexborneol injection 37.5 mg, comprising 30 mg of edaravone and 7.5 mg of (+)−dexborneol, twice a day over a consecutive period of 10-14 days”
“We used iStroke software (version 3.13) to calculate the infarct core volume, mismatch volume, and mismatch ratio.”
“We used SAS version 9.4 for statistical analyses”
The primary analysis used logistic regression with risk ratios and risk differences. Sensitivity analyses included multiple imputation and adjusted models. Exact p-values are reported (e.g., P=0.05, P=0.53). Effect sizes with 95% CIs are reported throughout. Software (SAS 9.4) is identified. Data presentation includes per-group n and CIs. Mathematical plausibility checks: the primary outcome percentages (55.0% of 689 = 378.95, rounds to 379; 49.6% of 671 = 332.8, rounds to 333) are consistent. The adverse events percentage in Table 2 (33.0% of 690 = 227.7, but reported 354/690 = 51.3%) appears inconsistent, but this may be a typographical error in the table; however, the text reports 354/690 (51.3%)? Actually the table shows 354/690 (33.0) which is arithmetically impossible (354/690 = 51.3%). This is a potential reporting error, but it does not affect the primary outcome. Given the overall rigor, this is a minor issue.
“We assessed differences in the proportion of modified Rankin Scale score 0-2 at 90 days between study groups with a logistic regression model.”
“risk ratio 1.11, 95% confidence interval (CI) 1.00 to 1.23; risk difference 5.4%, 95% CI 0.1% to 10.7%; P=0.05”
“We assessed differences in the proportion of modified Rankin Scale score 0-2 at 90 days between study groups with a logistic regression model. We presented the results as risk ratios and risk differences, both with 95% confidence intervals.”
“risk ratio 1.11, 95% confidence interval (CI) 1.00 to 1.23; risk difference 5.4%, 95% CI 0.1% to 10.7%; P=0.05”
“We used SAS version 9.4 for statistical analyses”
A data availability statement is present, stating that data are openly available at the National Clinical Medical Sciences Data Center (https://www.ncmi.cn//phda/dataDetails.do?id=CSTR:17970.11.A004X.202509.2.V1.0). This is a concrete route with a persistent identifier. Code is provided in appendix S15. Repository deposit is satisfied by the public data repository. Accession numbers are not applicable for this type of clinical data (the repository URL serves as the identifier).
“The data underlying the findings in this paper are openly and publicly available and can be found at https://www.ncmi.cn//phda/dataDetails.do?id=CSTR:17970.11.A004X.202509.2.V1.0”
“The code used to analyse the data in the paper can be found in appendix S15.”
“The data underlying the findings in this paper are openly and publicly available and can be found at https://www.ncmi.cn//phda/dataDetails.do?id=CSTR:17970.11.A004X.202509.2.V1.0”
“The code used to analyse the data in the paper can be found in appendix S15.”
The trial is registered at ClinicalTrials.gov (NCT05249920). Methods are detailed enough for replication. Limitations are explicitly discussed (e.g., overestimated effect, selection bias, single ethnicity, centre selection bias). Conclusions are proportional: the primary outcome is described as 'borderline significance' and the need for further validation is stated. Funding sources and conflicts of interest are disclosed. A reporting guideline (CONSORT) is implied but not explicitly named; however, the paper follows CONSORT-like structure. All pre-specified outcomes are reported, including negative secondary outcomes.
“Trial registration ClinicalTrials.gov NCT05249920 (https://clinicaltrials.gov/ct2/show/NCT05249920)”
“This trial had several limitations. Firstly, the effect of edaravone dexborneol compared with placebo was initially overestimated in the endovascular thrombectomy population; the anticipated 20% relative increase in neurological improvement actually amounted to about 11%.”
“Funding: This trial was supported by Simcere Pharmaceutical Group, the Ministry of Science and Technology of the People’s Republic of China (2022YFC2504904, 2022YFC2504902, the National Natural Science Foundation of China (81801152, 92046016), CAMS Innovation Fund for Medical Sciences (2019-I2M-5-029), and Beijing Municipal Committee of Science and Technology (Z201100005620010, Z211100002921064).”
“Trial registration ClinicalTrials.gov NCT05249920 (https://clinicaltrials.gov/ct2/show/NCT05249920)”
Registered (1 ID: ClinicalTrials.gov). No reporting guideline cited.
Broken references and links
1 finding · worst lowReferences 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.
- References not resolvable to a published paperRecomputed
Checked 32 references by DOI: 31 verified — 1 DOI unresolved.
- UNRESOLVED10.1016/s0140-6736(20Announcement No. 46 of 2022 from the Center for Drug Evaluation of the National Medical Products Administration on Issuing the "Guiding Principles for General Considerations in Organizing Patient Participation in Drug Development (Trial)"Cited DOI does not resolve to any Crossref record.
3 of 4 data/code links checked; 3 live; 1 not probed.
- datahttps://www.ncmi.cn//phda/dataDetails.do?id=CSTR:17970.11.A004X.202509.2.V1.0UNVERIFIEDHTTP 403Liveness indeterminate — content not checked.
- datahttps://public.flourish.studio/visualisation/26783154LIVEHTTP 200Resolves, but the content could not be matched to the paper.
- datahttps://public.flourish.studio/visualisation/26783332/LIVEHTTP 200Resolves, but the content could not be matched to the paper.
- datahttps://public.flourish.studio/visualisation/26824711/LIVEHTTP 200Resolves, but the content could not be matched to the paper.
Copyediting
1 finding · worst lowWording, consistency and formatting errors that need correcting before submission.
- Wording or formatting errors that need correctingAssessed
7 copyedit issues flagged (1 major): mostly consistency, typo, clarity.
- MAJORconsistencyTable 2, Adverse events row“354/690 (33.0)”→ Correct the percentage to 51.3% or verify the numerator/denominator.354/690 equals 51.3%, not 33.0%.
- MINORtypoMethods, Participants“lager values”→ Change to 'larger values'.Typographical error.
- MINORconsistencyTable 2, Recurrence of stroke“10/686 (1.5) | 8/667 (1.2)”→ Clarify why denominators differ from the group totals (686 vs 690, 667 vs 672).Denominators are not the full group sizes; likely due to missing data, but should be explained.
- MINORclarityDiscussion, Strengths and limitations“owing to convergence issues, the random centre effect was not included in the analysis”→ Provide more detail on the convergence issues and potential impact.This is a methodological limitation that could affect results.
- MINORtypoAbstract, Results“risk difference 5.4%, 95% CI 0.1% to 10.7%: P=0.05”→ Change colon to semicolon for consistency: 'risk difference 5.4%, 95% CI 0.1% to 10.7%; P=0.05'Inconsistent punctuation in the same sentence.
- MINORconsistencyTable 2, Adverse events row“354/690 (33.0)”→ Check calculation: 354/690 = 51.3%, not 33.0%. The percentage appears to be a typo; should be 51.3%.The percentage for adverse events in the edaravone dexborneol group is inconsistent with the numerator/denominator.
- MINORclarityMethods, 'Statistical analysis'“We did the primary efficacy evaluation in the modified intention-to-treat population, which included all randomly assigned patients except those lost to follow-up at 90 days.”→ Clarify that the modified ITT population excludes only those lost to follow-up, not other exclusions.The definition is clear but could be more precise.
The published work is robust and well-reported, but an informed reader should weigh the minor arithmetic error in Table 2 (354/690 reported as 33.0% instead of 51.3%), the lack of an explicit CONSORT statement, and the speculative 'penumbra freezing' mechanism that is not directly supported by the data. These issues warrant a correction or clarification from the authors, but they do not invalidate the primary outcome.
- 1.HIGHcopyeditCorrect the percentage in Table 2, Adverse events row: change '354/690 (33.0)' to '354/690 (51.3)' or verify the numerator/denominator.The current percentage is arithmetically impossible (354/690 = 51.3%), which is a demonstrable error that could undermine trust in the safety data.
- 2.HIGHrigorAdd direct evidence for the 'penumbra freezing' mechanism claim (e.g., imaging or biomarker data) or temper the claim to state it is speculative and not directly tested.The claim audit found this mechanistic claim unsupported by the trial's data; an unsupported headline conclusion can mislead readers.
- 3.HIGHreportingExplicitly name the reporting guideline followed (e.g., CONSORT) in the Methods section and consider submitting a CONSORT checklist as supplementary material.Both reviewers noted the guideline is implied but not explicitly stated, which is a transparency gap for a clinical trial.
- 4.HIGHstatisticsAdd a statement in the Statistical analysis section on how statistical test assumptions were verified (e.g., logistic regression robustness) and how outliers were handled.Both reviewers flagged assumptions verification and outlier handling as not explicitly reported; this is a reproducibility concern.
- 5.HIGHreportingClarify in the Statistical analysis section how the random centre effect convergence issue was handled (e.g., sensitivity analysis without random effects) and its potential impact.The copyedit pass and reviewers noted this methodological limitation is only mentioned in the limitations section; readers need to know how it was addressed.
- 6.HIGHotherVerify or correct the reference 'Announcement No. 46 of 2022 from the Center for Drug Evaluation of the National Medical Products Administration...' (DOI 10.1016/s0140-6736(20) which could not be found in any registry.A reference not found in any registry may be fabricated or contain an incorrect DOI; this is a potential integrity issue.
- 7.MEDIUMcopyeditFix the typo in Methods, Participants: change 'lager values' to 'larger values'.Typographical error that should be corrected for professionalism.
- 8.MEDIUMcopyeditClarify in Table 2 why the recurrence of stroke outcome uses denominators 686 and 667 instead of the full group totals (690 and 672).The denominators are inconsistent with group sizes and need explanation (e.g., missing data) to avoid confusion.
- 9.MEDIUMcopyeditChange the colon to a semicolon in the Abstract Results: 'risk difference 5.4%, 95% CI 0.1% to 10.7%: P=0.05' to '...10.7%; P=0.05'.Inconsistent punctuation in the same sentence is a minor copyedit issue.
- 10.MEDIUMdata codeSpecify the license under which the data are shared in the Data availability statement.Both reviewers suggested specifying the license to clarify reuse conditions.
- 11.MEDIUMdata codeProvide the analysis code in a version-controlled public repository (e.g., GitHub) with a DOI, rather than only in an appendix.Appendix code is less reproducible and not version-controlled; a repository would enhance transparency.
- 12.LOWreportingAdd a note in the Discussion that subgroup findings are exploratory and not confirmatory.To avoid overinterpretation of subgroup analyses, as suggested by one reviewer.
- 13.LOWreportingConsider reporting the exact p-value for the primary outcome (e.g., 0.049 or 0.051) instead of P=0.05 to avoid ambiguity.P=0.05 is borderline and exact value would be more informative.
- 14.LOWreportingAdd a statement on whether any post-hoc analyses were performed beyond the pre-specified subgroups.Enhances transparency about the analysis plan.
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.