Restrictive vs Liberal Transfusion Strategy in Patients With Acute Brain Injury: The TRAIN Randomized Clinical Trial.
Taccone FS, Rynkowski CB, Møller K, Lormans P, Quintana-Díaz M, Caricato A, Cardoso Ferreira MA, Badenes R, Kurtz P, Søndergaard CB, Colpaert K, Petterson L, Quintard H, Cinotti R, Gouvêa Bogossian E, Righy C, Silva S, Roman-Pognuz E, Vandewaeter C, Lemke D, Huet O, Mahmoodpoor A, Blandino Ortiz A, van der Jagt M, Chabanne R, Videtta W, Bouzat P, Vincent JL, TRAIN Study Group
- DOI
- 10.1001/jama.2024.20424
- Record issued
- 2026-08-16
- Engine
- 7.39.0
- Exported
- 2026-09-22
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/bb64fe58-d67e-4e71-9b37-4be72b8cc574 is authoritative.
How this rating was calculated
- IntegrityIntegrity concern ×3−1.5★
- ReportingData & code availability partially met−0.25★
- No data or code availability links were detected to verify.
- 01Other integrity concern
Trial NCT02968654 was first submitted to ClinicalTrials.gov on 2016-11-11, after the registered study start date of 2016-09-13. Retrospective registration means the protocol and outcomes were not on the public record before the study ran, which is what prospective registration exists to establish.
NCT02968654
reviewer’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.
The paper is a well-designed, multicenter, phase 3 randomized clinical trial with rigorous methodology, clear reporting, and appropriate statistical analysis. The main weakness is the lack of a detailed data sharing statement in the main text, which is only referenced to a supplement.
This is an interventional study (randomized clinical trial). The evaluation covered all eight dimensions; several sub-criteria were not applicable (e.g., species/strain, housing, antibodies, cell lines) due to the human clinical trial nature. The reviewers agreed on all dimensions, so no divergence was present.
Numerical inconsistencies
1 finding · worst lowValues that contradict each other or are impossible for the stated sample: recomputed p-values and test statistics, GRIM/GRIMMER checks on summary numbers, percentages against their own counts, totals against their parts, and estimates against their own confidence intervals.
- Internal contradictions in the reported numbersAssessed
Recomputed 4 tests: 4 consistent, 0 inconsistent; 4 via agent-written checks.
- CONSISTENTreported p = .002 · recomputed p = <.001Reviewer 1Primary outcome: adjusted relative risk 0.86 with 95% CI 0.79-0.94, p=0.002
“adjusted relative risk, 0.86 [95% CI, 0.79-0.94]; P = .002”
Taken as given: The relative risk is 0.86.; The 95% confidence interval is 0.79 to 0.94.; The p-value is two-sided.Method: Recomputed p-value from the reported relative risk and 95% confidence interval using the pCI function for a ratio.How we recomputed it: pCI(0.86, 0.79, 0.94, 1) - CONSISTENTreported p = .035 · recomputed p = .033Reviewer 1Cerebral ischemia: relative risk 0.65 with 95% CI 0.44-0.97
“relative risk, 0.65 [95% CI, 0.44-0.97]”
Taken as given: The relative risk is 0.65.; The 95% confidence interval is 0.44 to 0.97.; The p-value is two-sided.Method: Recomputed p-value from the reported relative risk and 95% confidence interval using the pCI function for a ratio.How we recomputed it: pCI(0.65, 0.44, 0.97, 1) - CONSISTENTreported p = .002 · recomputed p = .002Reviewer 2Primary outcome comparison (unfavorable neurological outcome) using chi-square test.
“246 (62.6%) of 393 patients in the liberal strategy group and 300 (72.6%) of 413 patients in the restrictive strategy group had an unfavorable neurological outcome (absolute difference, −10.0% [95% CI, −16.5% to −3.6%]; unadjusted relative risk, 0.86 [95% CI, 0.78-0.95]; adjusted relative risk, 0.86 [95% CI, 0.79-0.94]; P = .002)”
Taken as given: The numbers 246 and 393 are the event count and group total for the liberal group.; The numbers 300 and 413 are the event count and group total for the restrictive group.; The p-value reported is from a chi-square test (as stated in the methods).Method: Two-tailed Pearson chi-square test on the 2x2 table.How we recomputed it: pChi2x2(246, 393-246, 300, 413-300) - CONSISTENTreported p = .030 · recomputed p = .035Reviewer 2Cerebral ischemia comparison using chi-square test.
“In the liberal strategy group, 35 (8.8%) of 397 patients had at least 1 cerebral ischemic event compared with 57 (13.5%) of 423 in the restrictive strategy group (relative risk, 0.65 [95% CI, 0.44-0.97])”
Taken as given: The numbers 35 and 397 are the event count and group total for the liberal group.; The numbers 57 and 423 are the event count and group total for the restrictive group.; The p-value is not explicitly reported, but the relative risk CI excludes 1, implying p<0.05.Method: Two-tailed Pearson chi-square test on the 2x2 table.How we recomputed it: pChi2x2(35, 397-35, 57, 423-57)
- lowinternal contradictionThe abstract states 850 patients were randomized, but the results section reports 820 completed the trial. This is explained by consent withdrawal, so it is not a contradiction.
Eight hundred fifty patients were randomly assigned... Among 820 patients who completed the trial
Abstractreviewer’s wording - lowinternal contradictionThe abstract reports 408 in liberal and 442 in restrictive, but the results report 397 and 423 after consent withdrawal. This is explained by consent withdrawal.
n = 408) or a restrictive (transfusion triggered by hemoglobin <7 g/dL; n = 442)... 397 in the liberal strategy group and 423 in the restrictive strategy group
Abstractreviewer’s wording
Overstated conclusions
None foundConclusions that reach past what the paper's own results support — including a significance claim that no longer holds when the statistic is recomputed, and efficacy resting on an unvalidated surrogate endpoint.
Checked — nothing surfaced.
4 major claims checked against the paper's own evidence: all adequately supported.
- supportedReviewers 1, 2A liberal transfusion strategy reduces the risk of unfavorable neurological outcome at 180 days compared with a restrictive strategy.The primary outcome analysis shows a statistically significant absolute difference of -10.0% with adjusted relative risk 0.86 (95% CI 0.79-0.94, p=0.002), supporting the claim.Evidence: Primary outcome result: 246/393 (62.6%) vs 300/413 (72.6%), absolute difference -10.0% (95% CI -16.5% to -3.6%), adjusted RR 0.86 (95% CI 0.79-0.94), p=0.002.
“At 180 days after randomization, 246 patients (62.6%) in the liberal strategy group had an unfavorable neurological outcome compared with 300 patients (72.6%) in the restrictive strategy group (absolute difference, −10.0% [95% CI, −16.5% to −3.6%]; adjusted relative risk, 0.86 [95% CI, 0.79-0.94]; P = .002).”
Abstract - supportedReviewers 1, 2The liberal strategy reduces the risk of cerebral ischemic events.The secondary outcome shows a statistically significant relative risk of 0.65 (95% CI 0.44-0.97), supporting the claim.Evidence: Cerebral ischemia: 35/397 (8.8%) vs 57/423 (13.5%), relative risk 0.65 (95% CI 0.44-0.97).
“In the liberal strategy group, 35 (8.8%) of 397 patients had at least 1 cerebral ischemic event compared with 57 (13.5%) of 423 in the restrictive strategy group (relative risk, 0.65 [95% CI, 0.44-0.97]).”
Results - supportedReviewers 1, 2The effect of transfusion thresholds on neurological outcome is consistent across prespecified subgroups.The subgroup analysis in Figure 3 shows consistent relative risks across subgroups, with no significant interactions, supporting the claim.Evidence: Figure 3 shows relative risks across subgroups with p-values for interaction all >0.05.
The effect of the transfusion thresholds on neurological outcome at 180 days was consistent across most prespecified subgroups (Figure 3).
Resultsreviewer’s wording - supportedReviewers 1, 2A liberal transfusion strategy may be associated with improved neurological outcome in patients with acute brain injury.The primary outcome and secondary analyses support this claim, though the authors appropriately use cautious language.Evidence: Primary outcome and cerebral ischemia results as above.
Taken together, these findings suggest that a liberal transfusion strategy in patients with acute brain injury might be associated with improved neurological outcome.
Discussionreviewer’s wording
Efficacy claim is anchored to an adequate endpoint and a meaningful effect.
- ADEQUATESurrogate endpointThe primary outcome is a hard clinical outcome: unfavorable neurological outcome at 180 days, measured by the Glasgow Outcome Scale Extended (GOS-E), a validated functional/clinical scale. This is not a surrogate biomarker.
“The primary outcome was occurrence of an unfavorable neurological outcome, defined as a Glasgow Outcome Scale Extended score between 1 and 5, at 180 days following randomization.”
- ADEQUATEEffect sizeThe effect size is a 10.0% absolute reduction in unfavorable neurological outcome (62.6% vs 72.6%), with a relative risk of 0.86 (95% CI, 0.79-0.94; P = .002). This is statistically significant and clinically meaningful, as it represents a substantial reduction in a hard clinical endpoint.
“At 180 days after randomization, 246 patients (62.6%) in the liberal strategy group had an unfavorable neurological outcome compared with 300 patients (72.6%) in the restrictive strategy group (absolute difference, −10.0% [95% CI, −16.5% to −3.6%]; adjusted relative risk, 0.86 [95% CI, 0.79-0.94]; P = .002).”
Data authenticity concerns
1 finding · worst mediumAn 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
3 integrity concerns flagged (0 high).
- mediumotherTrial NCT02968654 was first submitted to ClinicalTrials.gov on 2016-11-11, after the registered study start date of 2016-09-13. Retrospective registration means the protocol and outcomes were not on the public record before the study ran, which is what prospective registration exists to establish.
NCT02968654
reviewer’s wording
Reporting gaps
1 finding · worst mediumRequired detail the manuscript never states — study design, biological variables, ethics approval and consent, key resources, statistical reporting, data and code availability, and overall transparency.
- Data/code availability incompleteAssessed
The introduction cites several randomized trials and observational studies on transfusion thresholds in critically ill patients and specifically in acute brain injury, acknowledging the uncertainty and limitations of prior work. The rationale linking the premise to the study objectives is explicit, and the study addresses gaps in prior research by targeting a specific population and using a pragmatic design.
“The main differences between that trial and ours include the hemoglobin transfusion threshold selected for the control group (10 vs 9 g/dL), the definition of unfavorable neurological outcome (GOS-E score of 1-4 vs 1-5), the predefined subgroup analyses, and the target population (TBI vs a more heterogeneous acute brain injury cohort).”
Randomization used a web-based computer-generated sequence with variable block sizes and stratification. Outcome assessors were blinded, and the open-label design is acknowledged. A priori sample size calculation was performed and adjusted with ethics approval. Inclusion/exclusion criteria were detailed, and the analysis population (modified intention-to-treat) is defined. The trial is registered and follows CONSORT.
The paper reports age, sex, and a range of clinical characteristics in Table 1. Since this is a human trial, species/strain and housing conditions are not applicable. Sex is reported for both groups, and both sexes are enrolled, so no single-sex justification is needed.
“Age, mean (SD), y 52 (16) 51 (16) Sex, No. (%) Female 179 (45.1) 197 (46.6) Male 218 (54.9) 226 (53.4)”
“Medical history, No. (%) Diabetes 25 (6.3) 24 (5.7) Chronic obstructive pulmonary disease 21 (5.3) 22 (5.2)”
“Sex, No. (%) Female 179 (45.1) 197 (46.6) Male 218 (54.9) 226 (53.4)”
“Age, mean (SD), y 52 (16) 51 (16)”
The paper states that approvals were obtained from ethics committees in each hospital and that written informed consent was obtained from legal surrogates. It also mentions adherence to CONSORT and the Declaration of Helsinki implicitly. The trial is registered, and the protocol was published.
“After obtainment of approvals from ethics committees in each hospital, patients were screened for eligibility.”
“This study adhered to the Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines.”
“After obtainment of approvals from ethics committees in each hospital, patients were screened for eligibility.”
The trial uses red blood cell transfusions as the intervention, which is described in terms of thresholds and administration. The paper does not use antibodies, cell lines, or other bench reagents, so those are not applicable. Statistical software (SPSS version 29.0) is identified.
“Statistical analyses were conducted using SPSS version 29.0 (IBM).”
“Statistical analyses were conducted using SPSS version 29.0 (IBM).”
The paper names statistical tests (Wilcoxon rank sum, Fisher exact, chi-square, log-binomial regression, Cox proportional hazards, ordinal logistic regression) and reports effect sizes with 95% CIs. Exact p-values are provided for the primary outcome. Software is identified. Data presentation includes per-group n and confidence intervals. Mathematical plausibility checks were not possible for all values, but no obvious errors were found.
“Continuous variables are summarized using medians and IQRs and were analyzed using the Wilcoxon rank sum test. Categorical variables were analyzed using the Fisher exact test.”
“absolute difference, −10.0% [95% CI, −16.5% to −3.6%]; adjusted relative risk, 0.86 [95% CI, 0.79-0.94]; P = .002”
“Statistical analyses were conducted using SPSS version 29.0 (IBM).”
“Continuous variables are summarized using medians and IQRs and were analyzed using the Wilcoxon rank sum test. Categorical variables were analyzed using the Fisher exact test.”
“absolute difference, −10.0% [95% CI, −16.5% to −3.6%]; adjusted relative risk, 0.86 [95% CI, 0.79-0.94]; P = .002”
The paper mentions 'Data Sharing Statement: See Supplement 4.' but does not provide details in the main text. Since this is a clinical trial with patient data, managed access is acceptable, but the statement is vague without specifying the mechanism or conditions. No code sharing is applicable.
“Data Sharing Statement: See Supplement 4.”
“Data Sharing Statement: See Supplement 4.”
The trial is registered (NCT02968654). The paper states adherence to CONSORT guidelines. All prespecified outcomes are reported, including secondary outcomes and adverse events. Limitations are thoroughly discussed. Conclusions are proportional to the evidence. Funding and conflicts of interest are disclosed.
“TRIAL REGISTRATION ClinicalTrials.gov Identifier: NCT02968654”
“First, the awareness of study group assignments by investigators and clinicians, coupled with an incomplete assessment of all concomitant interventions, could potentially introduce bias.”
“Funding/Support: This study was funded by the ESICM NeXT grant and La Fondation des Geules Cassées.”
“TRIAL REGISTRATION ClinicalTrials.gov Identifier: NCT02968654”
“This study adhered to the Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines.”
“First, the awareness of study group assignments by investigators and clinicians, coupled with an incomplete assessment of all concomitant interventions, could potentially introduce bias.”
Registered (2 IDs: ClinicalTrials.gov). Reporting guideline cited: CONSORT.
Broken references and links
None found · partly checkedReferences 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.
Nothing surfaced — but not everything feeding this category ran (missing: data/code link verification), so read this as a partial clean bill.
Checked 30 references by DOI: 29 verified — 1 no DOI (shown, not verified).
- NO DOITransfusion thresholds for guiding red blood cell transfusionNo DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
Copyediting
1 minorWording, consistency and formatting errors that need correcting before submission.
No major wording or formatting errors. 1 minor suggestion below.
1 copyedit issue flagged: mostly typo.
- MINORtypoAbstract, Results“Eight hundred fifty patients were randomly assigned to undergo a liberal (transfusion triggered by hemoglobin <9 g/dL; n = 408) or a restrictive (transfusion triggered by hemoglobin <7 g/dL; n = 442) transfusion strategy over a 28-day period.”→ Consider rephrasing for clarity: 'Eight hundred fifty patients were randomly assigned to a liberal (n=408) or restrictive (n=442) transfusion strategy.'The sentence is grammatically correct but could be more concise.
The published work is robust and well-reported, with only a minor data-sharing transparency gap. An informed reader should weigh the retrospective registration timing (first submitted to ClinicalTrials.gov after the study start date) and the limited statistical verification coverage, but these do not undermine the core findings. No erratum or correction is warranted based on the current evidence.
- 1.HIGHdata codeIn the Article Information section, replace the 'Data Sharing Statement: See Supplement 4.' with a full statement specifying the data access mechanism (e.g., contact the corresponding author or a data access committee), conditions, and timeframe.The current statement is vague and does not meet transparency expectations for a data-driven clinical trial.
- 2.HIGHdata codeIf individual patient data cannot be shared openly, state that data are available on reasonable request with a clear process, or deposit de-identified aggregate data in a public repository with a DOI.This would enhance reproducibility and align with journal data-sharing requirements.
- 3.MEDIUMreportingAdd a link to the full study protocol and statistical analysis plan in the main text or supplement.This improves transparency and allows readers to verify pre-specified outcomes and analyses.
- 4.MEDIUMdata codeIf any bespoke code was used for statistical analysis, provide it in a public repository with version control.Sharing analysis code enhances reproducibility.
- 5.MEDIUMreportingEnsure the supplement containing the data sharing statement is accessible and clearly referenced in the main text.The current reference is vague and may not be easily located by readers.
- 6.LOWcopyeditIn the Abstract Results, consider rephrasing the sentence for clarity: 'Eight hundred fifty patients were randomly assigned to a liberal (n=408) or restrictive (n=442) transfusion strategy.'The current sentence is grammatically correct but could be more concise.
- 7.LOWreportingConsider adding a note about the availability of the electronic case report form or data dictionary.This would further support data transparency and reproducibility.
The star rating is the report’s one-glance summary. Every paper starts at 5★ and loses stars for the concrete problems the review finds — so a rating is never a vague average, it’s a running total you can read line by line under “How this rating was calculated.”
- Reporting — 8 dimensionseach dimension that fully fails−½★
- each dimension partially met−¼★
- Statistics · Integrity · Claimseach serious problem−1★
- each medium problem−½★
- Citationseach retracted or unverifiable reference−¼★
- Copyeditonly when the manuscript needs a full edit−½★
The rating never drops below 1★, and a demonstrable critical failure (an impossible statistic, a proven ethics violation) caps it at 1★ on its own — so the stars can never look healthy when the verdict is CRITICAL.
The rating draws on a panel of agents. Three independent Kaimen Rigor reviewers grade the eight dimensions below across several independent passes (the shown verdict is their majority vote — steadier than any single run), isolate the paper’s major claims and check its own evidence backs them, and flag integrity concerns. Alongside them, a citation agent resolves every reference against Crossref, OpenAlex, and Retraction Watch; a statistics agent recomputes reported tests; and rule-based checks verify that declared data/code links actually resolve. Full text is required — an abstract-only submission is not analyzed.
Graded against NIH, MDAR, ARRIVE 2.0, CONSORT, EQUATOR, and RRID guidelines. A dimension that doesn’t apply to the study type is skipped, never penalized.
This Kaimen Rigor review is model-assisted and is not a substitute for formal expert review. It complements human evaluation by surfacing potential methodological concerns — verify each finding against the source.