Alpha¹ by KaimenConfidential

Kaimen Rigor review

Effect of a clinical decision support system on stroke care quality and outcomes in patients with acute ischaemic stroke (GOLDEN BRIDGE II): cluster randomised clinical trial.

Zhang X, Ding L, Jing J, Wang C, Gu H, Jiang Y, Meng X, Liu T, Xie X, Xu M, Hu M, Zhang Y, Fu H, Liu P, Du C, Du K, Wang M, Li H, Gong X, Dong K, Xiong Y, Wang Y, Liu L, Zhang Z, Zang Y, Yang C, Xian Y, Peterson E, Fonarow GC, Schwamm LH, Zhao X, Wang Y, Li Z, GOLDEN BRIDGE II Investigators

DOI: 10.1136/bmj-2025-085810
Record issued: 2026-08-10
Engine: 7.29.0
Exported: 2026-09-21

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/7fc02005-81b1-4230-a5a7-bfecd85fff13 is authoritative.

Kaimen Rigor review

Kaimen Rigor reviewed

3 / 5

Adequate — real issues to address

Interventional Study

Critical

High-priority

How this rating was calculated

Started at 5★ — deductions

IntegrityIntegrity concern ×4−2★

Model-reported confidence: highReviewer agreement: moderateEngine: 7.29.0Source: full text

Findings by category

101ArithmeticNumerical inconsistencies 102OverreachOverstated conclusions 103AuthenticityData authenticity concerns 04DisclosureNone found 05AvailabilityNone found 106CopyeditCopyediting

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.

The GOLDEN BRIDGE II trial is a well-designed cluster RCT with strong methodological reporting, including pre-registration, CONSORT-AI adherence, and transparent outcome reporting. Minor issues are the absence of an explicit Declaration of Helsinki compliance statement, an undefined 'absolute difference' metric that is inconsistent with percentage-point differences, and analysis code only in the supplementary appendix rather than a public repository.

The three independent reviewers converged on most dimensions (scientific premise, study design, biological variables, statistical analysis, reporting transparency). There was disagreement on ethical approvals (Reviewer 2 flagged a missing compliance statement) and data code availability (Reviewer 2 flagged inadequate code sharing, but the scoring rule allows a pass with 3 of 4 criteria adequate). The key resources dimension was rated not applicable by Reviewer 2, but the bespoke software intervention qualifies as a resource, so the pass status from the other reviewers was adopted.

Numerical inconsistencies

1 finding · worst low

Values 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 25 tests: 25 consistent, 0 inconsistent; 11 recomputed directly from the reported test statistics, 14 via agent-written checks.

Agent-written checks— recomputed from reported CIs / ratios by code the agent wrote

CONSISTENTreported p = .010 · recomputed p = .012
Recomputed adjusted hazard ratio 0.74 (95% CI 0.58–0.93), reported p=0.01
“adjusted hazard ratio 0.74, 95% confidence interval (CI) 0.58 to 0.93, P=0.01”
Find in source
Taken as given: 0.58–0.93 is a two-sided 95% confidence interval for the adjusted hazard ratio of 0.74, not a range, an IQR, or a different interval level; the adjusted hazard ratio is a RATIO measure, so the interval is symmetric on the log scale; p=0.01 is the p for THIS estimate, not for another comparison in the same sentence
Method: back the two-tailed p out of the log-scale CI width and compare it against the printed p
How we recomputed it: pCI(0.74, 0.58, 0.93, 1)
CONSISTENTreported p < .001 · recomputed p = <.001
Recomputed adjusted odds ratio 1.21 (95% CI 1.17–1.26), reported p<0.001
“adjusted odds ratio 1.21, 95% CI 1.17 to 1.26, P<0.001”
Find in source
Taken as given: 1.17–1.26 is a two-sided 95% confidence interval for the adjusted odds ratio of 1.21, not a range, an IQR, or a different interval level; the adjusted odds ratio is a RATIO measure, so the interval is symmetric on the log scale; p<0.001 is the p for THIS estimate, not for another comparison in the same sentence
Method: back the two-tailed p out of the log-scale CI width and compare it against the printed p
How we recomputed it: pCI(1.21, 1.17, 1.26, 1)
CONSISTENTreported p = .020 · recomputed p = .020
Recomputed adjusted hazard ratio 0.73 (95% CI 0.56–0.95), reported p=0.02
“adjusted hazard ratio 0.73, 95% CI 0.56 to 0.95, P=0.02”
Find in source
Taken as given: 0.56–0.95 is a two-sided 95% confidence interval for the adjusted hazard ratio of 0.73, not a range, an IQR, or a different interval level; the adjusted hazard 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 sentence
Method: back the two-tailed p out of the log-scale CI width and compare it against the printed p
How we recomputed it: pCI(0.73, 0.56, 0.95, 1)
CONSISTENTreported p = .006 · recomputed p = .006
Recomputed adjusted hazard ratio 0.70 (95% CI 0.54–0.90), reported p=0.006
“adjusted hazard ratio 0.70, 95% CI 0.54 to 0.90, P=0.006”
Find in source
Taken as given: 0.54–0.90 is a two-sided 95% confidence interval for the adjusted hazard ratio of 0.70, not a range, an IQR, or a different interval level; the adjusted hazard ratio is a RATIO measure, so the interval is symmetric on the log scale; p=0.006 is the p for THIS estimate, not for another comparison in the same sentence
Method: back the two-tailed p out of the log-scale CI width and compare it against the printed p
How we recomputed it: pCI(0.7, 0.54, 0.9, 1)
CONSISTENTreported p < .001 · recomputed p = <.001
Recomputed odds ratio 1.21 (95% CI 1.17–1.26), reported p<0.001
“odds ratio 1.21, 95% CI 1.17 to 1.26, P<0.001”
Find in source
Taken as given: 1.17–1.26 is a two-sided 95% confidence interval for the odds ratio of 1.21, 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<0.001 is the p for THIS estimate, not for another comparison in the same sentence
Method: back the two-tailed p out of the log-scale CI width and compare it against the printed p
How we recomputed it: pCI(1.21, 1.17, 1.26, 1)
CONSISTENTreported p = .290 · recomputed p = .292
Recomputed adjusted odds ratio 1.18 (95% CI 0.87–1.61), reported p=0.29
“adjusted odds ratio 1.18, 95% CI 0.87 to 1.61, P=0.29”
Find in source
Taken as given: 0.87–1.61 is a two-sided 95% confidence interval for the adjusted odds ratio of 1.18, not a range, an IQR, or a different interval level; the adjusted odds ratio is a RATIO measure, so the interval is symmetric on the log scale; p=0.29 is the p for THIS estimate, not for another comparison in the same sentence
Method: back the two-tailed p out of the log-scale CI width and compare it against the printed p
How we recomputed it: pCI(1.18, 0.87, 1.61, 1)
CONSISTENTreported p = .006 · recomputed p = .006
Recomputed adjusted hazard ratio 0.71 (95% CI 0.56–0.91), reported p=0.006
“adjusted hazard ratio 0.71, 95% CI 0.56 to 0.91, P=0.006”
Find in source
Taken as given: 0.56–0.91 is a two-sided 95% confidence interval for the adjusted hazard ratio of 0.71, not a range, an IQR, or a different interval level; the adjusted hazard ratio is a RATIO measure, so the interval is symmetric on the log scale; p=0.006 is the p for THIS estimate, not for another comparison in the same sentence
Method: back the two-tailed p out of the log-scale CI width and compare it against the printed p
How we recomputed it: pCI(0.71, 0.56, 0.91, 1)
CONSISTENTreported p = .290 · recomputed p = .274
Recomputed adjusted odds ratio 1.14 (95% CI 0.90–1.44), reported p=0.29
“adjusted odds ratio 1.14, 95% CI 0.90 to 1.44, P=0.29”
Find in source
Taken as given: 0.90–1.44 is a two-sided 95% confidence interval for the adjusted odds ratio of 1.14, not a range, an IQR, or a different interval level; the adjusted odds ratio is a RATIO measure, so the interval is symmetric on the log scale; p=0.29 is the p for THIS estimate, not for another comparison in the same sentence
Method: back the two-tailed p out of the log-scale CI width and compare it against the printed p
How we recomputed it: pCI(1.14, 0.9, 1.44, 1)
CONSISTENTreported p = .680 · recomputed p = .663
Recomputed adjusted hazard ratio 0.91 (95% CI 0.60–1.40), reported p=0.68
“adjusted hazard ratio 0.91, 95% CI 0.60 to 1.40, P=0.68”
Find in source
Taken as given: 0.60–1.40 is a two-sided 95% confidence interval for the adjusted hazard ratio of 0.91, not a range, an IQR, or a different interval level; the adjusted hazard ratio is a RATIO measure, so the interval is symmetric on the log scale; p=0.68 is the p for THIS estimate, not for another comparison in the same sentence
Method: back the two-tailed p out of the log-scale CI width and compare it against the printed p
How we recomputed it: pCI(0.91, 0.6, 1.4, 1)
CONSISTENTreported p = .740 · recomputed p = .745
Recomputed adjusted hazard ratio 0.89 (95% CI 0.44–1.79), reported p=0.74
“adjusted hazard ratio 0.89, 95% CI 0.44 to 1.79, P=0.74”
Find in source
Taken as given: 0.44–1.79 is a two-sided 95% confidence interval for the adjusted hazard ratio of 0.89, not a range, an IQR, or a different interval level; the adjusted hazard ratio is a RATIO measure, so the interval is symmetric on the log scale; p=0.74 is the p for THIS estimate, not for another comparison in the same sentence
Method: back the two-tailed p out of the log-scale CI width and compare it against the printed p
How we recomputed it: pCI(0.89, 0.44, 1.79, 1)
CONSISTENTreported p = .130 · recomputed p = .132
Recomputed adjusted hazard ratio 0.68 (95% CI 0.41–1.12), reported p=0.13
“adjusted hazard ratio 0.68, 95% CI 0.41 to 1.12, P=0.13”
Find in source
Taken as given: 0.41–1.12 is a two-sided 95% confidence interval for the adjusted hazard ratio of 0.68, not a range, an IQR, or a different interval level; the adjusted hazard ratio is a RATIO measure, so the interval is symmetric on the log scale; p=0.13 is the p for THIS estimate, not for another comparison in the same sentence
Method: back the two-tailed p out of the log-scale CI width and compare it against the printed p
How we recomputed it: pCI(0.68, 0.41, 1.12, 1)
CONSISTENTreported p = .010 · recomputed p = .012Reviewer 1
Primary outcome: New vascular events at 3 months, adjusted hazard ratio and p-value.
“New vascular events at three months occurred in 2.9% (320/11 054) in the intervention group compared with 3.9% (416/10 549) in the control group (adjusted hazard ratio 0.74, 95% confidence interval (CI) 0.58 to 0.93, P=0.01).”
Find in source
Taken as given: The hazard ratio is 0.74.; The 95% confidence interval is 0.58 to 0.93.; The confidence interval is two-sided.
Method: P-value derived from a reported hazard ratio and its 95% confidence interval.
How we recomputed it: pCI(0.74, 0.58, 0.93, 1)
CONSISTENTreported p < .001 · recomputed p = <.001Reviewer 1
Secondary outcome: Composite measure, adjusted odds ratio and p-value.
“Patients in the intervention group were more likely to have a higher composite measure (91.4% (77 049/84 276) v 89.8% (70 794/78 834), adjusted odds ratio 1.21, 95% CI 1.17 to 1.26, P<0.001).”
Find in source
Taken as given: The odds ratio is 1.21.; The 95% confidence interval is 1.17 to 1.26.; The confidence interval is two-sided.
Method: P-value derived from a reported odds ratio and its 95% confidence interval.
How we recomputed it: pCI(1.21, 1.17, 1.26, 1)
CONSISTENTreported p = .020 · recomputed p = .020Reviewer 1
Secondary outcome: New vascular events at 12 months, adjusted hazard ratio and p-value.
“New vascular events were significantly lower in the intervention group at 12 months (4.0% (440/11 054) v 5.5% (576/10 549), adjusted hazard ratio 0.73, 95% CI 0.56 to 0.95, P=0.02).”
Find in source
Taken as given: The hazard ratio is 0.73.; The 95% confidence interval is 0.56 to 0.95.; The confidence interval is two-sided.
Method: P-value derived from a reported hazard ratio and its 95% confidence interval.
How we recomputed it: pCI(0.73, 0.56, 0.95, 1)
CONSISTENTreported p = .010 · recomputed p = .014Reviewer 1
Table 2: Primary outcome, New vascular events at 3 months, unadjusted hazard ratio and p-value.
“New vascular events‡ at 3 months | 320/11 054 (2.9) | 416/10 549 (3.9) | 0.75 (0.60 to 0.95) | 0.74 (0.58 to 0.93) | — | 0.01”
Find in source
Taken as given: The unadjusted hazard ratio is 0.75.; The 95% confidence interval is 0.60 to 0.95.; The confidence interval is two-sided.
Method: P-value derived from a reported hazard ratio and its 95% confidence interval.
How we recomputed it: pCI(0.75, 0.60, 0.95, 1)
CONSISTENTreported p < .001 · recomputed p = <.001Reviewer 1
Table 2: Secondary outcome, Composite measure, unadjusted odds ratio and p-value.
“Composite measure | 77 049/84 276 (91.4) | 70 794/78 834 (89.8) | 1.21 (1.16 to 1.26) | 1.21 (1.17 to 1.26) | — | <0.001”
Find in source
Taken as given: The unadjusted odds ratio is 1.21.; The 95% confidence interval is 1.16 to 1.26.; The confidence interval is two-sided.
Method: P-value derived from a reported odds ratio and its 95% confidence interval.
How we recomputed it: pCI(1.21, 1.16, 1.26, 1)
CONSISTENTreported p = .290 · recomputed p = .348Reviewer 1
Table 2: Secondary outcome, All-or-none measure, unadjusted odds ratio and p-value.
“All-or-none measure | 6504/11 054 (58.8) | 5619/10 549 (53.3) | 1.16 (0.85 to 1.58) | 1.18 (0.87 to 1.61) | — | 0.29”
Find in source
Taken as given: The unadjusted odds ratio is 1.16.; The 95% confidence interval is 0.85 to 1.58.; The confidence interval is two-sided.
Method: P-value derived from a reported odds ratio and its 95% confidence interval.
How we recomputed it: pCI(1.16, 0.85, 1.58, 1)
CONSISTENTreported p = .006 · recomputed p = .009Reviewer 1
Table 2: Secondary outcome, New vascular events at 6 months, unadjusted hazard ratio and p-value.
“6 months | 379/11 054 (3.4) | 503/10 549 (4.8) | 0.73 (0.58 to 0.93) | 0.71 (0.56 to 0.91) | — | 0.006”
Find in source
Taken as given: The unadjusted hazard ratio is 0.73.; The 95% confidence interval is 0.58 to 0.93.; The confidence interval is two-sided.
Method: P-value derived from a reported hazard ratio and its 95% confidence interval.
How we recomputed it: pCI(0.73, 0.58, 0.93, 1)
CONSISTENTreported p = .020 · recomputed p = .037Reviewer 1
Table 2: Secondary outcome, New vascular events at 12 months, unadjusted hazard ratio and p-value.
“12 months | 440/11 054 (4.0) | 576/10 549 (5.5) | 0.75 (0.57 to 0.98) | 0.73 (0.56 to 0.95) | — | 0.02”
Find in source
Taken as given: The unadjusted hazard ratio is 0.75.; The 95% confidence interval is 0.57 to 0.98.; The confidence interval is two-sided.
Method: P-value derived from a reported hazard ratio and its 95% confidence interval.
How we recomputed it: pCI(0.75, 0.57, 0.98, 1)
CONSISTENTreported p = .010 · recomputed p = .012Reviewer 2
Primary outcome: three-month new vascular events, adjusted hazard ratio
“adjusted hazard ratio 0.74 (95% confidence interval (CI) 0.58 to 0.93, P=0.01)”
Find in source
Taken as given: 0.74 is the adjusted hazard ratio estimate; the 95% CI 0.58-0.93 is two-sided; the HR is analysed on the log scale (log=1) under the normal approximation
Method: Two-tailed p back-calculated from the estimate and its two-sided 95% CI using the normal approximation on the log scale.
How we recomputed it: pCI(0.74, 0.58, 0.93, 1)
CONSISTENTreported p = .006 · recomputed p = .006Reviewer 2
Secondary outcome: six-month new vascular events, adjusted hazard ratio
“significantly lower rates in the intervention group at six months (3.4% v 4.8%, adjusted hazard ratio 0.71, 95% CI 0.56 to 0.91, P=0.006)”
Find in source
Taken as given: 0.71 is the adjusted hazard ratio estimate; the 95% CI 0.56-0.91 is two-sided; the HR is analysed on the log scale (log=1)
Method: Two-tailed p back-calculated from the estimate and its two-sided 95% CI using the normal approximation on the log scale.
How we recomputed it: pCI(0.71, 0.56, 0.91, 1)
CONSISTENTreported p = .020 · recomputed p = .020Reviewer 2
Secondary outcome: twelve-month new vascular events, adjusted hazard ratio
“12 months (4.0% v 5.5%, 0.73, 0.56 to 0.95, P=0.02)”
Find in source
Taken as given: 0.73 is the adjusted hazard ratio estimate; the 95% CI 0.56-0.95 is two-sided; the HR is analysed on the log scale (log=1)
Method: Two-tailed p back-calculated from the estimate and its two-sided 95% CI using the normal approximation on the log scale.
How we recomputed it: pCI(0.73, 0.56, 0.95, 1)
CONSISTENTreported p < .001 · recomputed p = <.001Reviewer 2
Composite performance-measure outcome, adjusted odds ratio
“adjusted odds ratio 1.21, 95% CI 1.17 to 1.26, P<0.001”
Find in source
Taken as given: 1.21 is the adjusted odds ratio estimate; the 95% CI 1.17-1.26 is two-sided; the OR is analysed on the log scale (log=1)
Method: Two-tailed p back-calculated from the estimate and its two-sided 95% CI using the normal approximation on the log scale; reported as P<0.001 so the check verifies the computed p is below 0.001.
How we recomputed it: pCI(1.21, 1.17, 1.26, 1)
CONSISTENTreported p = .290 · recomputed p = .292Reviewer 2
All-or-none performance-measure outcome, adjusted odds ratio
“adjusted odds ratio 1.18, 95% CI 0.87 to 1.61, P=0.29”
Find in source
Taken as given: 1.18 is the adjusted odds ratio estimate; the 95% CI 0.87-1.61 is two-sided; the OR is analysed on the log scale (log=1)
Method: Two-tailed p back-calculated from the estimate and its two-sided 95% CI using the normal approximation on the log scale.
How we recomputed it: pCI(1.18, 0.87, 1.61, 1)
CONSISTENTreported p = .290 · recomputed p = .274Reviewer 2
Disability (mRS>=3) at three months, adjusted odds ratio
“adjusted odds ratio 1.14, 95% CI 0.90 to 1.44, P=0.29”
Find in source
Taken as given: 1.14 is the adjusted odds ratio estimate; the 95% CI 0.90-1.44 is two-sided; the OR is analysed on the log scale (log=1)
Method: Two-tailed p back-calculated from the estimate and its two-sided 95% CI using the normal approximation on the log scale.
How we recomputed it: pCI(1.14, 0.90, 1.44, 1)

lowinternal contradiction
In Table 3, the 'Difference (95% CI)' for early antithrombotics is reported as positive (0.1, CI -0.3 to 0.5) even though the intervention rate (98.2%) is lower than the control rate (98.5%); the sign is inconsistent with the raw percentages unless the value is a model-derived adjusted estimate, which is not stated.
“Early antithrombotics* | 8582/8735 (98.2) | 8234/8361 (98.5) | 0.1 (−0.3 to 0.5) | 0.9”
Table 3Find in source
lowinternal contradiction
The reported absolute differences for several performance measures in the 'Adherence to evidence based performance measures' section do not match the direct calculation from the provided percentages.
“dual antiplatelet treatment in patients with non-disabling ischaemic cerebrovascular disease within 24 hours of disease onset (76.2% v 69.6%, absolute difference 14.8%)”
ResultsFind in source
lowinternal contradiction
The reported absolute difference for anticoagulation in patients with atrial fibrillation at discharge does not match the direct calculation from the provided percentages.
“anticoagulation in patients with atrial fibrillation at discharge (77.3% v 67.5%, 22.2%)”
ResultsFind in source

Overstated conclusions

1 finding · worst low

Conclusions 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 only partially backed by the presented evidenceAssessed

10 major claims checked against the paper's own evidence: all adequately supported.

partialReviewer 2
The positive impact resulted from the combined effects of various system components (imaging analysis, cause classification, evidence-based recommendations).
The paper shows improved process measures (anticoagulation, dual antiplatelet, dysphagia screening, DVT prophylaxis) that plausibly mediate the outcome, but the causal pathway from specific components to the vascular-event reduction is inferred, not directly demonstrated.
Evidence: Higher adherence rates in the CDSS group for several performance measures (e.g., dual antiplatelet 76.2% v 69.6%, anticoagulation for AF 77.0% v 69.3%).
“We considered the positive impact resulted from the combined effects of various system components.”
DiscussionFind in source
partialReviewer 3
The CDSS intervention effect was driven by improved guideline adherence (dual antiplatelet, anticoagulation, dysphagia screening, DVT prophylaxis).
Higher adherence in the CDSS group is demonstrated, but the causal attribution linking these performance measures to the reduced vascular events is reasonable inference, not directly proven.
Evidence: Higher rates of dual antiplatelet (76.2% vs 69.6%), anticoagulation for AF (77.0% vs 69.3%), dysphagia screening (98.5% vs 91.2%), and DVT prophylaxis (37.1% vs 30.0%) in the intervention group.
“We considered the positive impact resulted from the combined effects of various system components.”
DiscussionFind in source
supportedReviewers 1, 2, 3
Use of the stroke CDSS in patients with acute ischaemic stroke in China led to a significant decrease in new vascular events at three months.
The primary outcome results directly support this claim with a statistically significant adjusted hazard ratio.
Evidence: New vascular events at three months occurred in 2.9% (320/11 054) in the intervention group compared with 3.9% (416/10 549) in the control group (adjusted hazard ratio 0.74, 95% confidence interval (CI) 0.58 to 0.93, P=0.01).
“New vascular events at three months occurred in 2.9% (320/11 054) in the intervention group compared with 3.9% (416/10 549) in the control group (adjusted hazard ratio 0.74, 95% confidence interval (CI) 0.58 to 0.93, P=0.01).”
AbstractFind in source
supportedReviewers 1, 3
The stroke CDSS intervention was also effective in improving stroke care quality.
The study reports a significantly higher composite measure of evidence-based performance measures in the intervention group.
Evidence: Patients in the intervention group were more likely to have a higher composite measure (91.4% (77 049/84 276) v 89.8% (70 794/78 834), adjusted odds ratio 1.21, 95% CI 1.17 to 1.26, P<0.001).
“Patients in the intervention group were more likely to have a higher composite measure (91.4% (77 049/84 276) v 89.8% (70 794/78 834), adjusted odds ratio 1.21, 95% CI 1.17 to 1.26, P<0.001).”
AbstractFind in source
supportedReviewer 1
The stroke CDSS intervention was also effective in decreasing long term vascular events.
The results show significantly lower new vascular events at 12 months in the intervention group.
Evidence: New vascular events were significantly lower in the intervention group at 12 months (4.0% (440/11 054) v 5.5% (576/10 549), adjusted hazard ratio 0.73, 95% CI 0.56 to 0.95, P=0.02).
“New vascular events were significantly lower in the intervention group at 12 months (4.0% (440/11 054) v 5.5% (576/10 549), adjusted hazard ratio 0.73, 95% CI 0.56 to 0.95, P=0.02).”
AbstractFind in source
supportedReviewer 1
The stroke CDSS offers a promising approach to providing high quality care for patients with acute ischaemic stroke admitted to hospital, particularly for resource constrained regions with a heavy burden of cerebrovascular diseases like China.
The study's findings on reduced vascular events and improved care quality in a multicenter Chinese setting support the promise of the CDSS in such regions.
Evidence: Use of the stroke CDSS in patients with acute ischaemic stroke in China led to a significant decrease in new vascular events at three months. The system was also effective in improving stroke care quality and decreasing long term vascular events.
“Use of the stroke CDSS in patients with acute ischaemic stroke in China led to a significant decrease in new vascular events at three months. The system was also effective in improving stroke care quality and decreasing long term vascular events.”
ConclusionFind in source
supportedReviewer 2
The stroke CDSS was effective in improving stroke care quality.
The composite performance-measure result supports the claim: 91.4% vs 89.8%, adjusted OR 1.21 (95% CI 1.17-1.26), P<0.001.
Evidence: Composite measure result in Table 2: 77 049/84 276 (91.4%) vs 70 794/78 834 (89.8%), adjusted OR 1.21 (1.17-1.26), P<0.001.
“Patients in the CDSS intervention group were more likely to have a higher composite measure of evidence based performance measures (91.4% v 89.8%, odds ratio 1.21, 95% CI 1.17 to 1.26, P<0.001).”
Table 2Find in source
supportedReviewers 2, 3
The stroke CDSS decreased long term (six and 12 month) vascular events.
Both the six-month (HR 0.71, P=0.006) and 12-month (HR 0.73, P=0.02) analyses are statistically significant and support the claim.
Evidence: Six-month HR 0.71 (0.56-0.91), P=0.006; 12-month HR 0.73 (0.56-0.95), P=0.02.
“The reduction in new vascular events persisted through longer term follow-up, with significantly lower rates in the intervention group at six months (3.4% v 4.8%, adjusted hazard ratio 0.71, 95% CI 0.56 to 0.91, P=0.006) and 12 months (4.0% v 5.5%, 0.73, 0.56 to 0.95, P=0.02).”
ResultsFind in source
supportedReviewer 2
Application of the stroke CDSS led to a 25.6% decrease in new vascular events within three months.
The relative reduction is arithmetically consistent with the reported 3-month event rates (1.0 pp / 3.9% = 25.6%).
Evidence: 3-month rates 2.9% vs 3.9%: absolute reduction 1.0 percentage point over control 3.9% = 25.6% relative reduction.
“application of the stroke CDSS led to a 25.6% decrease in new vascular events within three months in patients with acute ischaemic stroke.”
DiscussionFind in source
supportedReviewer 3
The findings show the safety of the stroke CDSS.
No significant differences in bleeding or mortality outcomes support the safety claim.
Evidence: No significant differences in moderate/severe bleeding or all bleeding at 3, 6, 12 months; no significant differences in all-cause mortality.
“We observed a significant reduction in ischaemic events, but no differences in all cause mortality and moderate or severe bleeding events. These findings also showed the safety of the stroke CDSS.”
DiscussionFind in source

Efficacy claim is anchored to an adequate endpoint and a meaningful effect.

ADEQUATESurrogate endpoint
The primary outcome is a composite of hard clinical events (ischaemic stroke, haemorrhagic stroke, myocardial infarction, vascular death) at 3 months, not a surrogate biomarker. The efficacy claim is based on this clinical endpoint.
“The primary outcome was a new vascular event (composite of ischaemic stroke, haemorrhagic stroke, myocardial infarction, or vascular death) within three months after stroke onset.”
Find in source
ADEQUATEEffect size
Primary effect: new vascular events at 3 months reduced from 3.9% to 2.9% (adjusted HR 0.74, 95% CI 0.58-0.93), a 25.6% relative reduction. The paper explicitly anchors this as clinically significant, noting the 1% absolute decrease could carry significance in practice and aligning with the pre-specified 26% relative reduction hypothesis.
“the 1% decrease could carry significance in clinical practice”
Find in source

Data authenticity concerns

1 finding · worst low

An 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

4 integrity concerns flagged (0 high).

lowother
The 'absolute difference (%)' metric used in Tables 1 and 3 and the Results text does not equal the percentage-point difference between the groups (e.g., dual antiplatelet 76.2% v 69.6% is labelled 'absolute difference 14.8%' though the pp difference is 6.6; prestroke mRS 6.5% v 3.8% is labelled 12.2% though the pp difference is 2.7). Because this undefined metric is also the basis for the '>10% absolute difference' covariate-adjustment rule, the adjustment rationale is not verifiable from the printed numbers.
“Patients in the CDSS intervention group were more likely to have a higher rate of some performance measures, including dual antiplatelet treatment in patients with non-disabling ischaemic cerebrovascular disease within 24 hours of disease onset (76.2% v 69.6%, absolute difference 14.8%)”
Table 3Find in source

Reporting gaps

None found

Required 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.

Eight Dimensions8 pass

The paper cites previous studies like GOLDEN BRIDGE-AIS and GWTG-Stroke to support the idea that quality improvement interventions improve stroke outcomes. It also highlights the current limitations of AI applications in stroke care, specifically the lack of rigorous evaluation through randomized controlled trials, which the current study aims to address.

Sub-criteria

prior work citedADEQUATE

premise rationaleADEQUATE

limitations addressedADEQUATE

Per reviewer (3) — the votes behind the verdict

PASSReviewer 1· 90% conf

The study clearly establishes its premise by citing prior work on AI in healthcare and stroke quality improvement, and provides a logical rationale for the CDSS intervention.

Evidence

“However, the majority of AI applications for stroke healthcare have not been rigorously evaluated through randomised controlled trials.”

Introduction ¶2Find in source

“CDSS may provide a new way to facilitate clinical decision making and consequently improve care quality and clinical outcomes in patients with ischaemic cerebrovascular disease.”

Introduction ¶4Find in source

“The clustered randomised controlled GOLDEN BRIDGE II trial was designed to determine whether the stroke CDSS could improve care quality and clinical outcomes of patients with acute ischaemic stoke in China.”

Introduction ¶5Find in source

PASSReviewer 2· 90% conf

The rationale is well grounded in prior stroke quality-improvement and AI literature, with clear acknowledgement of the evidence gap and a hypothesis that follows logically.

Evidence

“However, the majority of AI applications for stroke healthcare have not been rigorously evaluated through randomised controlled trials.”

Introduction ¶1Find in source

“CDSS may provide a new way to facilitate clinical decision making and consequently improve care quality and clinical outcomes in patients with ischaemic cerebrovascular disease.”

Introduction ¶2Find in source

PASSReviewer 3· 90% conf

The paper clearly establishes its scientific foundation by citing prior quality-improvement trials and RCTs, articulating a logical rationale for the CDSS, and identifying the lack of rigorous RCT evaluation of AI stroke applications as the gap addressed.

Evidence

“According to GOLDEN BRIDGE-AIS (Intervention to Bridge the Evidence-based Gap in Stroke Care Quality) and GWTG-Stroke (Get With The Guidelines-Stroke), implementing multifaceted quality improvement interventions improves acute ischaemic stroke care quality and outcomes.”

Introduction ¶2Find in source

“However, the majority of AI applications for stroke healthcare have not been rigorously evaluated through randomised controlled trials.”

Introduction ¶1Find in source

Introduction ¶3Find in source

Hospitals were randomized using a random number generator, stratified by location and hospital grade. Blinding was maintained for follow-up interviewers and statisticians. A power analysis was performed to determine sample size, and clear inclusion/exclusion criteria for both hospitals and patients were defined. The control group received usual care.

Sub-criteria

controlsADEQUATE

power analysisADEQUATE

blinding levelsADEQUATE

outlier handlingADEQUATE

randomization unitADEQUATE

inclusion exclusionADEQUATE

randomization methodADEQUATE

replicate distinctionN/A

independent replicationN/A

Per reviewer (3) — the votes behind the verdict

PASSReviewer 1· 90% conf

The study design is robust for a cluster randomized trial, detailing randomization, blinding, and inclusion/exclusion criteria. A power analysis was conducted, and appropriate controls (usual care) were used.

Evidence

“Hospitals were randomly assigned in a 1:1 ratio to the CDSS intervention group or the usual care group through a random number generator.”

MethodsFind in source

“Hospitals were randomly assigned in a 1:1 ratio to the CDSS intervention group or the usual care group through a random number generator.”

MethodsFind in source

“Therefore, we estimated that 21 689 participants from 80 hospitals would provide 80% power to detect a 26% relative reduction in the primary outcome, with a two sided significance level of 5%, an intraclass correlation coefficient of 0.01, and a 10% loss to follow-up.”

MethodsFind in source

PASSReviewer 2· 90% conf

Methodologically strong cluster RCT: randomisation method and unit stated, assessor/statistician blinding, a priori power calculation with ICC, detailed inclusion/exclusion criteria, and ITT analysis population.

Evidence

“Hospitals were randomly assigned in a 1:1 ratio to the CDSS intervention group or the usual care group through a random number generator. Randomisation was stratified by the hospital location (eastern, central, or western region) and hospital grade (secondary or tertiary).”

MethodsFind in source

“we estimated that 21 689 participants from 80 hospitals would provide 80% power to detect a 26% relative reduction in the primary outcome, with a two sided significance level of 5%, an intraclass correlation coefficient of 0.01, and a 10% loss to follow-up.”

MethodsFind in source

“All primary and secondary analyses were based on the intention-to-treat principle.”

Statistical analysisFind in source

PASSReviewer 3· 90% conf

The cluster-randomised design is rigorously executed: randomisation method and unit stated, blinding of outcome assessors and statisticians described for an open-label trial, a priori power calculation reported, and detailed inclusion/exclusion criteria plus ITT analysis population defined.

Evidence

“Hospitals were randomly assigned in a 1:1 ratio to the CDSS intervention group or the usual care group through a random number generator.”

MethodsFind in source

“To ensure blinding to cluster assignments, follow-up data were collected by interviewers who were masked to patients’ cluster assignments. Statisticians were masked to the cluster allocation.”

MethodsFind in source

The median age of patients and the percentage of men are reported for both groups. The paper notes that patient characteristics were balanced between the two groups except for prestroke mRS score ≥3, which was adjusted for in multivariable models.

Sub-criteria

demographicsADEQUATE

sex reportedADEQUATE

sex justifiedN/A

age weight healthADEQUATE

housing conditionsN/A

species strain sourceN/A

Per reviewer (3) — the votes behind the verdict

PASSReviewer 1· 90% conf

Demographic information for patients, including age and sex, is adequately reported and balanced between groups, with a note on one imbalanced characteristic.

Evidence

“Men | 7129 (64.5) | 6796 (64.4) | 0.1”

Table 1Find in source

“Age (years), median (IQR) | 67 (58-74) | 66 (57-74) | 3.2”

Table 1Find in source

“The median age of patients was 67 (57-74) years and 7678 (35.5%) were female.”

Results ¶2Find in source

PASSReviewer 2· 85% conf

Sex, age, BMI, and extensive comorbidities are reported in Table 1; the study enrolled both sexes so sex justification is not applicable.

Evidence

“The median age of patients was 67 (57-74) years and 7678 (35.5%) were female.”

ResultsFind in source

“Hypertension | 6782 (61.4) | 6344 (60.1) | 2.5”

Table 1Find in source

“Men | 7129 (64.5) | 6796 (64.4) | 0.1”

Table 1Find in source

PASSReviewer 3· 90% conf

Patient demographics and biological variables are thoroughly reported for a human trial, including sex, age, BMI, medical history, and NIHSS; both sexes are enrolled so sex justification is not applicable.

Evidence

“The median age of patients was 67 (57-74) years and 7678 (35.5%) were female.”

ResultsFind in source

“Age (years), median (IQR) | 67 (58-74) | 66 (57-74) | 3.2”

Table 1Find in source

The central institutional review board at Beijing Tiantan Hospital (KY 2020-016-02) and each participating site approved the trial protocol. Written informed consent was obtained from all patients or their legal guardians.

Sub-criteria

iacuc statementN/A

informed consentADEQUATE

irb ethics statementADEQUATE

regulatory complianceADEQUATE

Per reviewer (3) — the votes behind the verdict

PASSReviewer 1· 90% conf

The study reports explicit ethical approval from a named IRB and states that informed consent was obtained from all participants.

Evidence

“The central institutional review board at Beijing Tiantan Hospital (KY 2020-016-02) and each participating site approved the trial protocol.”

MethodsFind in source

“Written informed consent was obtained.”

MethodsFind in source

“The study complied with the CONSORT-AI (consolidated standards of reporting trials—artificial intelligence) extension guidelines.”

MethodsFind in source

WARNReviewer 2· 65% conf

Named IRB approval with protocol number and written informed consent are both reported and adequate, but no explicit regulatory-compliance framework (e.g., Declaration of Helsinki) is stated.

Evidence

“The central institutional review board at Beijing Tiantan Hospital (KY 2020-016-02) and each participating site approved the trial protocol.”

MethodsFind in source

“Written informed consent was obtained.”

MethodsFind in source

absence[Methods, Study design; Ethics statements]

No Declaration of Helsinki / ICH-GCP compliance statement is present; only CONSORT-AI (a reporting guideline) is cited

PASSReviewer 3· 90% conf

Ethical compliance is fully documented: a named institutional review board with protocol number approved the trial, written informed consent was obtained from all participants, and adherence to a recognised reporting/ethics framework is stated.

Evidence

“The central institutional review board at Beijing Tiantan Hospital (KY 2020-016-02) and each participating site approved the trial protocol.”

MethodsFind in source

“All patients or their legal guardians provided written informed consent before enrolment.”

MethodsFind in source

“The study complied with the CONSORT-AI (consolidated standards of reporting trials—artificial intelligence) extension guidelines.”

MethodsFind in source

The statistical software (SAS version 9.4) is identified. The AI-assisted imaging analysis uses deep learning algorithms derived from a high-quality dataset (CNSR-III), and the CDSS knowledge base is rooted in clinical guidelines. No other biological or chemical resources are used.

Sub-criteria

mycoplasma testingN/A

reagents identifiedN/A

organisms identifiedN/A

antibodies identifiedN/A

cell line authenticationN/A

software tools identifiedADEQUATE

Per reviewer (3) — the votes behind the verdict

PASSReviewer 1· 80% conf

The study adequately identifies the software used for statistical analysis and describes the AI algorithms, but other wet-lab specific resources are not applicable.

Evidence

“All the analyses were performed using SAS software, version 9.4 (SAS Institute).”

Statistical analysisFind in source

paraphrase[Methods, Interventions]

“The AI assisted imaging analysis used deep learning algorithms derived from the Third China National Stroke Registry (CNSR-III).”

N/AReviewer 2· 60% conf

The study is a software-based procedural intervention with no drug, biologic, device, or bench reagent; no key biological or chemical resource is used, and the analysis software (SAS 9.4) is identified.

Evidence

“All the analyses were performed using SAS software, version 9.4 (SAS Institute).”

Statistical analysisFind in source

paraphrase[Methods, Interventions]

“The intervention is a clinical decision support system (software) integrated into hospital information systems; no pharmaceutical or biological product is involved.”

PASSReviewer 3· 80% conf

The trial's relevant software resources are identified: the CDSS intervention is described in detail in an appendix, and the analysis software (SAS 9.4) is named with version; no bench reagents/cell lines are used.

Evidence

“Supplementary appendix 2 gives a detailed description of the stroke CDSS.”

MethodsFind in source

“All the analyses were performed using SAS software, version 9.4 (SAS Institute).”

Statistical analysisFind in source

The paper names specific statistical tests (t-tests, Wilcoxon rank sum tests, χ² tests, Fisher’s exact tests, mixed effects Cox regression, mixed effects linear regression). Exact p-values and 95% confidence intervals are consistently reported for primary and secondary outcomes. Statistical software (SAS 9.4) is identified. Data presentation includes counts and percentages for categorical variables, and medians with IQRs for continuous variables, with per-group Ns clearly stated. Mathematical plausibility checks for reported percentages and counts did not reveal any inconsistencies.

Sub-criteria

tests namedADEQUATE

exact p valuesADEQUATE

effect sizes ciADEQUATE

data presentationADEQUATE

software identifiedADEQUATE

assumptions verifiedADEQUATE

mathematical plausibilityPASS

Per reviewer (3) — the votes behind the verdict

PASSReviewer 1· 90% conf

The statistical analysis is well-described, naming tests, reporting exact p-values and confidence intervals, and identifying software. Data presentation includes per-group Ns and percentages, and the mathematical plausibility checks did not reveal any inconsistencies.

Evidence

“T tests or Wilcoxon rank sum tests were used for continuous variables, while χ 2 tests or Fisher’s exact tests were performed for categorical variables.”

Statistical analysisFind in source

“adjusted hazard ratio 0.74, 95% confidence interval (CI) 0.58 to 0.93, P=0.01”

AbstractFind in source

“adjusted hazard ratio 0.74, 95% confidence interval (CI) 0.58 to 0.93, P=0.01”

AbstractFind in source

PASSReviewer 2· 85% conf

Statistical methods are clearly named, effect estimates reported with 95% CIs and exact p-values, SAS 9.4 identified, and the clinical-trial data presentation idiom (flowchart, Kaplan-Meier curves, per-group Ns) is followed; all reported percentages checked against counts are consistent.

Evidence

paraphrase[Methods, Statistical analysis]

“T tests or Wilcoxon rank sum tests were used for continuous variables, while χ 2 tests or Fisher's exact tests were performed for categorical variables.”

“adjusted hazard ratio 0.74 (95% confidence interval (CI) 0.58 to 0.93, P=0.01)”

ResultsFind in source

“All the analyses were performed using SAS software, version 9.4 (SAS Institute).”

Statistical analysisFind in source

PASSReviewer 3· 85% conf

Statistical methods are named, effect sizes are reported with 95% CIs alongside exact p-values, software is identified, and the clinical reporting idiom (per-group n, percentages, Kaplan-Meier curves) is followed; mathematical plausibility is not applicable for large-N model-based statistics.

Evidence

“T tests or Wilcoxon rank sum tests were used for continuous variables, while χ 2 tests or Fisher’s exact tests were performed for categorical variables.”

Statistical analysisFind in source

“with adjusted hazard ratio 0.74 (95% confidence interval (CI) 0.58 to 0.93, P=0.01).”

ResultsFind in source

“All primary and secondary analyses were based on the intention-to-treat principle.”

Statistical analysisFind in source

A data availability statement is present, directing readers to a public repository (NCMI) with a persistent identifier. The paper also states that the code used for analysis can be found in the supplementary appendix.

Sub-criteria

code sharingADEQUATE

accession numbersADEQUATE

repository depositADEQUATE

data availability statementADEQUATE

Per reviewer (3) — the votes behind the verdict

PASSReviewer 1· 90% conf

The study provides a data availability statement with a public repository link and accession number for the underlying data, and states that code is available in the supplementary appendix.

Evidence

“The data underlying the study findings are openly and publicly available ( https://www.ncmi.cn/phda/dataDetails.do?id=CSTR:17970.11.A004X.202603.36.V2.0 ).”

Data availabilityFind in source

“The data underlying the study findings are openly and publicly available ( https://www.ncmi.cn/phda/dataDetails.do?id=CSTR:17970.11.A004X.202603.36.V2.0 ).”

Data availabilityFind in source

“The code used to analyse the data in the paper can be found in the supplementary appendix.”

Data availabilityFind in source

WARNReviewer 2· 70% conf

The data availability statement is concrete and strong (open public repository with CSTR identifier), but the analysis code is only placed in the supplementary appendix with no version-controlled public repository or DOI.

Evidence

“The data underlying the study findings are openly and publicly available ( https://www.ncmi.cn/phda/dataDetails.do?id=CSTR:17970.11.A004X.202603.36.V2.0 ).”

Data availabilityFind in source

“The code used to analyse the data in the paper can be found in the supplementary appendix.”

Data availabilityFind in source

PASSReviewer 3· 85% conf

The data availability statement names a concrete public repository with a persistent identifier, and the analysis code location is stated in the supplementary appendix; three of four criteria are adequate.

Evidence

“The data underlying the study findings are openly and publicly available ( https://www.ncmi.cn/phda/dataDetails.do?id=CSTR:17970.11.A004X.202603.36.V2.0 ).”

Data availabilityFind in source

“The code used to analyse the data in the paper can be found in the supplementary appendix.”

Data availabilityFind in source

Trial is registered (ClinicalTrials.gov NCT04524624); CONSORT-AI compliance is stated; non-significant outcomes (disability, mortality, bleeding) are transparently reported; limitations are explicitly discussed (cluster randomisation, mild-stroke skew, no thrombectomy coverage, no cost analysis); conclusions are proportional to the RCT evidence; and funding grants and an ICMJE-style COI statement are provided. All seven criteria are adequate.

Sub-criteria

funding coiADEQUATE

trial registrationADEQUATE

reporting guidelineADEQUATE

methods completenessADEQUATE

all outcomes reportedADEQUATE

limitations discussedADEQUATE

conclusions proportionalADEQUATE

Per reviewer (3) — the votes behind the verdict

PASSReviewer 1· 90% conf

The study reports methods in sufficient detail for replication, is registered as a clinical trial, references a reporting guideline, discusses limitations, and provides funding and COI statements.

Evidence

paraphrase[Methods, Interventions]

“The intervention section describes the functions of the CDSS, its integration, AI-assisted imaging analysis, classification of stroke causes, and treatment recommendations.”

“ClinicalTrials.gov NCT04524624 (https://clinicaltrials.gov/ct2/show/NCT04524624)”

AbstractFind in source

“The study complied with the CONSORT-AI (consolidated standards of reporting trials—artificial intelligence) extension guidelines.”

MethodsFind in source

PASSReviewer 2· 95% conf

Reporting is exemplary for a clinical trial: trial registration, CONSORT-AI adherence, all outcomes (including null results) reported, a dedicated limitations section, proportional conclusions, and full funding/COI disclosures.

Evidence

“Trial registration ClinicalTrials.gov NCT04524624 (https://clinicaltrials.gov/ct2/show/NCT04524624)”

AbstractFind in source

“The study complied with the CONSORT-AI (consolidated standards of reporting trials—artificial intelligence) extension guidelines.”

MethodsFind in source

“The trial randomised hospitals rather than individual patients.”

LimitationsFind in source

PASSReviewer 3· 90% conf

Reporting is complete: trial registration number, CONSORT-AI adherence, all prespecified outcomes reported including null results, limitations discussed, proportional conclusions, and funding/COI statements are all present.

Evidence

“Trial registration ClinicalTrials.gov NCT04524624 (https://clinicaltrials.gov/ct2/show/NCT04524624)”

AbstractFind in source

“This study has several limitations. The trial randomised hospitals rather than individual patients.”

LimitationsFind in source

paraphrase[Funding / Competing interests]

“Funding: This study was supported by grants from the National Key Research and Development Program of China (2022YFC2504902)...”

Registered (3 IDs: ClinicalTrials.gov). Reporting guideline cited: CONSORT.

ClinicalTrials.gov

NCT04524624

ClinicalTrials.gov

NCT03785678

ClinicalTrials.gov

NCT03342664

Reporting guidelines: CONSORT

Broken references and links

None found

References 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 48 references by DOI: 45 verified — 3 no DOI (shown, not verified).

NO DOI
Acts for Hospital Classification
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Brief introduction: 2013 China statistical yearbook
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Medical quality control indicators for cerebral infarction (2020 edition)
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).

2 data/code links checked; 1 live.

datahttps://clinicaltrials.gov/ct2/show/NCT04524624LIVEHTTP 200
Resolves, but the content could not be matched to the paper.
datahttps://www.ncmi.cn/phda/dataDetails.do?id=CSTR:17970.11.A004X.202603.36.V2.0UNVERIFIEDHTTP 403
Liveness indeterminate — content not checked.

Copyediting

1 finding · worst low

Wording, consistency and formatting errors that need correcting before submission.

Wording or formatting errors that need correctingAssessed

22 copyedit issues flagged (1 major): mostly consistency, typo, clarity.

MAJORconsistencyTable 3 and Results (performance measures)
“dual antiplatelet treatment in patients with non-disabling ischaemic cerebrovascular disease within 24 hours of disease onset (76.2% v 69.6%, absolute difference 14.8%)”
→ Define the 'absolute difference' metric; the percentage-point difference between 76.2% and 69.6% is 6.6, not 14.8.
The 'absolute difference' column is undefined and does not equal the percentage-point difference; it is used as the >10% covariate-adjustment threshold, so the formula should be stated.
MINORconsistencyAbstract, Results
“2.9% (320/11 054)”
→ Ensure consistent formatting for percentages and counts throughout the abstract and results sections.
Some percentages are followed by counts in parentheses, others are not, and some counts are missing for percentages.
MINORclarityAbstract, Results
“The CDSS intervention effect remained significant in the cluster level analysis (−0.01, −0.02 to −0.004, P=0.003).”
→ Clarify the unit or meaning of the effect size (-0.01) in the abstract for the cluster level analysis.
The effect size for the cluster level analysis is presented as a raw difference without clear units in the abstract, which could be confusing.
MINORgrammarIntroduction, paragraph 5
“acute ischaemic stoke”
→ acute ischaemic stroke
Typo in 'stroke'.
MINORconsistencyMethods, Participants
“taking into account their geographical region and hospital grade.”
→ Consider using 'geographical region' or 'geographic region' consistently.
Minor inconsistency in spelling 'geographical'.
MINORconsistencyMethods, Interventions
“The stroke CDSS intervention and usual care were provided in the intervention group and control group, respectively. shows the intervention workflow in the CDSS group during hospital admission.”
→ Ensure figures are properly referenced in the text, e.g., 'Figure 1 shows the intervention workflow...'
Missing figure reference before 'shows the intervention workflow'.
MINORconsistencyResults, Primary outcome
“The stroke CDSS intervention effect remained significant in the cluster level linear regression analysis (−0.01, 95% CI −0.02 to −0.004, P=0.003; ).”
→ Ensure figures are properly referenced in the text, e.g., '...P=0.003; (Table 2).'
Missing table reference after the cluster level analysis result.
MINORconsistencyResults, Secondary outcomes
“but did not reach statistical significance (adjusted odds ratio 1.18, 95% CI 0.87 to 1.61, P=0.29; ).”
→ Ensure figures are properly referenced in the text, e.g., '...P=0.29; (Table 2).'
Missing table reference after the all-or-none measure result.
MINORconsistencyResults, Secondary outcomes
“and 12 months (3.0% v 3.5%, 0.77, 0.52 to 1.13, P=0.18; ).”
→ Ensure figures are properly referenced in the text, e.g., '...P=0.18; (Table 2).'
Missing table reference after the all cause mortality result.
MINORconsistencyResults, Safety outcomes
“and 12 months (1.0% v 1.4%, 0.66, 0.41 to 1.07, P=0.09; )”
→ Ensure figures are properly referenced in the text, e.g., '...P=0.09; (Table 2).'
Missing table reference after the all bleeding events result.
MINORconsistencyResults, Adherence to evidence based performance measures
“dual antiplatelet treatment in patients with non-disabling ischaemic cerebrovascular disease within 24 hours of disease onset (76.2% v 69.6%, absolute difference 14.8%)”
→ Ensure consistent reporting of absolute differences, as 14.8% is not the direct difference between 76.2% and 69.6%.
The reported absolute difference of 14.8% is inconsistent with the percentages 76.2% and 69.6% (difference is 6.6%). This occurs multiple times in this section.
MINORconsistencyResults, Adherence to evidence based performance measures
“anticoagulation for atrial fibrillation in hospital admission (77.0% v 69.3%, 17.4%)”
→ Ensure consistent reporting of absolute differences, as 17.4% is not the direct difference between 77.0% and 69.3%.
The reported absolute difference of 17.4% is inconsistent with the percentages 77.0% and 69.3% (difference is 7.7%).
MINORconsistencyResults, Adherence to evidence based performance measures
“dysphagia screening (98.5% v 91.2%, 33.6%)”
→ Ensure consistent reporting of absolute differences, as 33.6% is not the direct difference between 98.5% and 91.2%.
The reported absolute difference of 33.6% is inconsistent with the percentages 98.5% and 91.2% (difference is 7.3%).
MINORconsistencyResults, Adherence to evidence based performance measures
“DVT prophylaxis (37.1% v 30.0%, 15.2%)”
→ Ensure consistent reporting of absolute differences, as 15.2% is not the direct difference between 37.1% and 30.0%.
The reported absolute difference of 15.2% is inconsistent with the percentages 37.1% and 30.0% (difference is 7.1%).
MINORtypoIntroduction, last sentence
“patients with acute ischaemic stoke in China”
→ patients with acute ischaemic stroke in China
Misspelling of 'stroke'.
MINORconsistencyTable 3, Early antithrombotics row
“8582/8735 (98.2) | 8234/8361 (98.5) | 0.1 (−0.3 to 0.5)”
→ Clarify whether the 'Difference (95% CI)' is model-derived; the sign is positive though the intervention rate (98.2%) is lower than control (98.5%).
Sign appears inconsistent with the raw percentages; likely an adjusted estimate but not stated.
MINORconsistencyTable 2, All cause mortality 6 months
“243/10 549(2.3)”
→ 243/10 549 (2.3)
Missing space before the parenthetical percentage, unlike all other cells.
MINORtypoIntroduction, final sentence
“patients with acute ischaemic stoke in China”
→ patients with acute ischaemic stroke in China
'stoke' misspelled for 'stroke'.
MINORtypoCompeting interests
“Boehinger Ingelheim”
→ Boehringer Ingelheim
Misspelling of the company name.
MINORpunctuationResults, Safety outcomes
“0.66 (0.41 to 1.07), P=0.09; )”
→ 0.66 (0.41 to 1.07), P=0.09.
Stray closing parenthesis after the semicolon.
MINORclarityMethods, Interventions; Results
“shows the intervention workflow in the CDSS group during hospital admission.”
→ Fig 1 shows the intervention workflow in the CDSS group during hospital admission.
Figure callout missing the figure number in the text.
MINORconsistencyThroughout
“all cause mortality”
→ all-cause mortality
Inconsistent hyphenation of 'all-cause'.

Kaimen Rigor Action ItemsFinal pass before journal submission

The published work is robust overall, but informed readers should note the undefined 'absolute difference' metric used for covariate adjustment, the missing Declaration of Helsinki statement, and the code location (supplementary appendix only). These issues do not invalidate the core findings but could warrant a correction or clarification from the authors.

1.
HIGHstatistics
Define the 'absolute difference (%)' metric used in Tables 1 and 3 and throughout the Results section: state the formula explicitly, as it does not equal the simple percentage-point difference (e.g., 76.2% vs 69.6% yields 14.8%, not 6.6).
This metric is the basis for the '>10% absolute difference' covariate-adjustment rule; without a clear definition, the adjustment rationale is not verifiable.
2.
HIGHethics
Add an explicit statement of compliance with the Declaration of Helsinki (or ICH-GCP) to the Ethics statements section.
The paper currently cites only a reporting guideline (CONSORT-AI), not a recognized ethics framework, which is a reporting gap for a human trial.
3.
HIGHdata code
Deposit the custom analysis code (currently in the supplementary appendix) in a version-controlled public repository (e.g., GitHub with a Zenodo DOI) and cite the DOI in the Data availability statement.
Code in a supplementary appendix is not version-controlled and lacks a persistent identifier, reducing reproducibility and long-term access.
4.
HIGHstatistics
Clarify the sign of the 'Difference (95% CI)' for early antithrombotics in Table 3: the reported positive difference (0.1) is inconsistent with the raw rates (98.2% vs 98.5%); state whether this is a model-derived adjusted estimate.
An apparent sign error or missing explanation could confuse readers and undermine confidence in the results.
5.
HIGHreporting
Fix the typo 'acute ischaemic stoke' to 'acute ischaemic stroke' in the final sentence of the Introduction.
A typo in a key term is easily corrected and improves professionalism.
6.
MEDIUMreporting
Add missing figure and table references in the Results text (e.g., 'Figure 1 shows the intervention workflow' and 'Table 2' after relevant results).
Missing callouts reduce clarity and make it harder for readers to locate supporting data.
7.
MEDIUMreporting
Fix the stray closing parenthesis in the Results, Safety outcomes section: '0.66 (0.41 to 1.07), P=0.09; )' should be '0.66 (0.41 to 1.07), P=0.09.'
A punctuation error is minor but suggests lack of attention to detail.
8.
MEDIUMreporting
Correct the misspelling 'Boehinger Ingelheim' to 'Boehringer Ingelheim' in the Competing interests section.
A misspelled company name is a simple error that should be fixed.
9.
MEDIUMreporting
Ensure consistent hyphenation of 'all-cause mortality' throughout the paper (currently appears as 'all cause mortality' in some places).
Inconsistent hyphenation is a minor style issue but can be corrected for clarity.
10.
LOWreporting
Add a space before the parenthetical percentage in Table 2, All cause mortality 6 months: '243/10 549(2.3)' should be '243/10 549 (2.3)'.
Consistent formatting improves readability.

Full text (49,230 chars)Analyzed: Aug 10, 2026

RigorReviewer 1

RigorReviewer 2

RigorReviewer 3

RigorPre-Submission Reviewer

CopyeditAgent Scribe

CitationsCitation Integrity

StatisticsStatistics Consistency

ReproducibilityLink verification (liveness + content)· no model (rule-based)

PreregistrationDeterministic check· no model (rule-based)

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.”

How stars are deducted

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.

The eight reporting dimensions

Scientific PremisePrior work, rationale, and how it addresses gaps.

Study Design & RigorRandomization, blinding, power, controls, replication.

Biological VariablesSex, age, species/strain, housing, demographics.

Ethical ApprovalsIRB/IACUC approval, informed consent, regulatory compliance.

Key ResourcesAntibodies, cell lines, organisms, software — identified with persistent IDs.

Statistical ReportingNamed tests, exact p-values, effect sizes, plausible summary stats.

Data & Code AvailabilityAvailability statements, repositories, accession numbers, shared code.

Reporting TransparencyMethods, registration, guidelines, outcomes, limitations, funding/COI.

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.

Effect of a clinical decision support system on stroke care quality and outcomes in patients with acute ischaemic stroke (GOLDEN BRIDGE II): cluster randomised clinical trial.

DOI

10.1136/bmj-2025-085810

Record issued

2026-08-10

Engine

7.29.0

Exported

2026-09-21