Adjunctive Intravenous Argatroban or Eptifibatide for Ischemic Stroke.
Adeoye O, Broderick J, Derdeyn CP, Grotta JC, Barsan W, Bentho O, Berry S, Concha M, Davis I, Demel S, Elm J, Gentile N, Graves T, Hoffman M, Huang J, Ingles J, Janis S, Jasne AS, Khatri P, Levine SR, Majjhoo A, Panagos P, Pancioli A, Pizzella S, Ranasinghe T, Sabagha N, Sivakumar S, Streib C, Vagal A, Wilson A, Wintermark M, Yoo AJ, Barreto AD
- DOI
- 10.1056/NEJMoa2314779
- 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/4c665d3a-e57a-40e7-919f-e4ec38c53764 is authoritative.
How this rating was calculated
- IntegrityIntegrity concern−0.5★
- ReportingData & code availability partially met−0.25★
- The numeric-impossibility checks (GRIM/GRIMMER/DEBIT/SPRITE) covered 2 of 6 reported means. The other 4 do not state their group size where the value is printed, and these checks need the count the mean was averaged over, so they were not checked.
This Kaimen Rigor review uses Kaimen Rigor reviewers trained on a curated corpus of high-fidelity and retracted papers, with expert supervision and curation. It can still make mistakes; verify each finding against the source before relying on it.
This is a well-conducted phase 3 randomized trial with rigorous design, clear ethical approvals, and appropriate statistical methods. The main weaknesses are minor reporting gaps: the data sharing statement is not detailed, statistical software is not named, and a few copyedit issues exist.
Both reviewers independently scored all eight dimensions and agreed on all statuses. The study type is interventional. Non-applicable sub-criteria (e.g., animal housing, cell line authentication) were excluded. The statistics verification checked only 1 test; other statistics were not machine-verified.
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 1 test: 1 consistent, 0 inconsistent; 1 via agent-written checks.
- CONSISTENTreported p = .002 · recomputed p = .004Reviewer 1Posterior probability for argatroban vs placebo
“The posterior probability that argatroban was better than placebo was 0.002 (posterior mean difference in utility-weighted modified Rankin scale score, −1.51±0.51)”
Taken as given: The posterior mean difference is -1.51 and the standard deviation is 0.51, so the 95% credible interval is approximately -1.51 ± 1.96*0.51 = (-2.51, -0.51).; The posterior probability is approximated by a two-sided p-value from the normal distribution.Method: Approximated the posterior probability using the normal distribution with the given mean and SD.How we recomputed it: pCI(-1.51, -2.53, -0.49, 0)
- lowinternal contradictionThe abstract states 514 patients were assigned, but the safety analysis uses N=217 for placebo, N=54 for argatroban, N=212 for eptifibatide, which sum to 483, not 514. This is explained by the safety population excluding patients who did not receive the investigational product.
A total of 514 patients were assigned to receive argatroban (59 patients), eptifibatide (227 patients), or placebo (228 patients). ... Only the patients in the safety sample (i.e., those who had received any amount of an investigational product) were included in the analysis of safety outcomes.
Table 3reviewer’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.
3 major claims checked against the paper's own evidence: all adequately supported.
- supportedReviewers 1, 2Adjunctive argatroban or eptifibatide did not reduce poststroke disability.The primary outcome analysis shows posterior probabilities of benefit of 0.002 and 0.041, both below the threshold, supporting the claim.Evidence: Primary efficacy analysis: posterior probability of argatroban better than placebo = 0.002, eptifibatide = 0.041.
“adjunctive treatment with intravenous argatroban or eptifibatide did not reduce poststroke disability”
ConclusionFind in source - supportedReviewers 1, 2Adjunctive argatroban or eptifibatide was associated with increased mortality.Mortality was higher in the active groups (24% and 12%) compared to placebo (8%), and the paper discusses this association.Evidence: Mortality at 90 days: 24% argatroban, 12% eptifibatide, 8% placebo.
“Mortality at 90 days was higher in the argatroban group (24%) and the eptifibatide group (12%) than in the placebo group (8%).”
AbstractFind in source - supportedReviewers 1, 2The trial was stopped early due to futility.The interim analysis met futility criteria, and enrollment was discontinued.Evidence: Interim analysis showed predictive probabilities of success <0.001 for both active groups.
“Both active-treatment groups had predictive probabilities of success of less than 0.20, which indicated that the trial should be stopped.”
ResultsFind in source
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
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 prior trials with conflicting findings for argatroban and equivocal phase 2 results for eptifibatide, and explains the biological rationale for each drug. The premise logically leads to the trial objective of comparing these agents against placebo. Limitations of prior research are implicitly addressed by the design of a larger, phase 3 trial, though not explicitly framed as addressing prior limitations.
“Two recent trials yielded conflicting findings regarding the benefit of argatroban in ischemic stroke.”
“We previously conducted three phase 2 trials that showed equivocal effects of the combination of eptifibatide with thrombolysis.”
“On the basis of these findings regarding argatroban and eptifibatide, we conducted the Multiarm Optimization of Stroke Thrombolysis (MOST) trial.”
“Two recent trials yielded conflicting findings regarding the benefit of argatroban in ischemic stroke.”
“On the basis of these findings regarding argatroban and eptifibatide, we conducted the Multiarm Optimization of Stroke Thrombolysis (MOST) trial.”
The trial is described as randomized, single-blind, with response-adaptive randomization. The randomization method is described in terms of ratios and adaptation rules, but the specific algorithm (e.g., biased coin) is mentioned without full detail. Blinding is described: investigators were aware, but patients and outcome assessors were blinded. A power analysis is not explicitly stated, but the maximum planned sample size of 1200 is given, and the adaptive design includes futility and efficacy stopping rules. Inclusion/exclusion criteria are described. Outlier handling is not explicitly addressed, but missing data imputation is described. Controls are the placebo group. Independent replication is not applicable for a single pivotal trial.
“The randomization algorithm controlled for imbalances in baseline NIHSS score and age with the use of a biased coin if the maximum tolerated imbalance was exceeded.”
“Investigators were aware of the trial-group assignments, but patients and legally authorized representatives were unaware of the trial-group assignments throughout the trial.”
“The maximum planned sample size was 1200.”
“The first 150 patients were randomly assigned in a 1:1:1 ratio to receive argatroban, eptifibatide, or placebo.”
“Investigators were aware of the trial-group assignments, but patients and legally authorized representatives were unaware of the trial-group assignments throughout the trial.”
“The maximum planned sample size was 1200.”
The paper reports sex, age, race, ethnicity, and various comorbidities in Table 1. Since both sexes are enrolled, sex_justified is not applicable. Age and health status are reported. Demographics are adequately reported. Species/strain and housing conditions are not applicable for a human trial.
“Male sex — no. (%) | 118 (52) | 31 (53) | 109 (48)”
“Median age (IQR) — yr | 66 (58–78) | 68 (60–79) | 68 (58–79)”
“Male sex — no. (%) | 118 (52) | 31 (53) | 109 (48)”
“Median age (IQR) — yr | 66 (58–78) | 68 (60–79) | 68 (58–79)”
The trial was approved by the University of Cincinnati central institutional review board and the FDA, and written informed consent was obtained from all patients or their representatives. Regulatory compliance is implied through FDA oversight and the trial being conducted under an IND.
“All the patients or their legally authorized representatives provided written informed consent.”
“All the patients or their legally authorized representatives provided written informed consent.”
The drugs are named with doses and regimens. The statistical software is not explicitly named, but the analysis methods are described. Since this is a drug trial, the investigational products are the key resources and are adequately identified. Antibodies, cell lines, mycoplasma, and organisms are not applicable.
“Patients in the argatroban group received a bolus (100 μ g per kilogram) followed by a 12-hour infusion (3 μ g per kilogram per minute).”
“Patients in the argatroban group received a bolus (100 μ g per kilogram) followed by a 12-hour infusion (3 μ g per kilogram per minute).”
The primary analysis uses a Bayesian normal dynamic linear model, and posterior probabilities are reported. Secondary outcomes use odds ratios with 95% CIs. The paper reports exact posterior probabilities (e.g., 0.002, 0.041) and CIs. Statistical software is not identified. Data presentation includes means, SDs, and percentages. Mathematical plausibility checks are not applicable due to continuous outcomes and large N.
“The primary analysis compared each active-treatment group with the placebo group regarding 90-day utility-weighted modified Rankin scale scores, with adjustment for baseline NIHSS score in a Bayesian normal dynamic linear model.”
“The primary analysis compared each active-treatment group with the placebo group regarding 90-day utility-weighted modified Rankin scale scores, with adjustment for baseline NIHSS score in a Bayesian normal dynamic linear model.”
“90-Day modified Rankin scale score of 0 to 2 or return to prestroke score — no. (%) | 139 (61) | 26 (44) | 0.50 (0.28 to 0.90) | 126 (56) | 0.80 (0.55 to 1.16)”
The paper mentions a data sharing statement is available with the full text, but the specific mechanism is not described in the text. No repository deposit or accession numbers are provided. Code sharing is not applicable as no custom code is mentioned.
The trial is registered (NCT03735979). Methods are detailed enough for replication. No reporting guideline is explicitly mentioned, but the paper follows CONSORT-like structure. All outcomes are reported, including negative results. Limitations are discussed. Funding and COI are disclosed.
“One limitation of the trial is the single-blind design.”
“One limitation of the trial is the single-blind design.”
“Supported by a grant (1U01NS100699-01A1) from the National Institute of Neurological Disorders and Stroke.”
Registered (1 ID: ClinicalTrials.gov). No reporting guideline cited.
Broken references and links
None foundReferences checked against Crossref, OpenAlex and Retraction Watch for retractions and resolvability, plus declared data and code links probed for whether they resolve to content matching the paper.
Checked — nothing surfaced.
Checked 31 references by DOI: 31 verified.
Every extracted reference resolved against Crossref/OpenAlex with no retraction flags.
1 data/code link checked; 1 live.
- datahttps://clinicaltrials.gov/ct2/show/NCT03735979LIVEHTTP 200Resolves, but the content could not be matched to the paper.
Copyediting
3 minorWording, consistency and formatting errors that need correcting before submission.
No major wording or formatting errors. 3 minor suggestions below.
3 copyedit issues flagged: mostly consistency, typo.
- MINORtypoResults, Secondary Efficacy Analyses“odd ratio”→ odds ratioTypographical error in the text.
- MINORconsistencyTable 3“Placebo (N = 217)”→ Ensure N matches the safety population definition.The safety population N differs from the randomized N; this is expected but should be clearly labeled.
- MINORconsistencyAbstract vs. Results“514 patients were assigned to receive argatroban (59 patients), eptifibatide (227 patients), or placebo (228 patients)”→ Ensure consistent use of 'assigned' vs 'randomized' throughout.Abstract says 'assigned', Results says 'randomization'.
The published work is robust and methodologically sound. An informed reader should weigh the minor reporting gaps (data sharing details, software identification) and the copyedit issues, but none undermine the validity of the findings. No erratum is warranted for the identified issues.
- 1.HIGHdata codeIn the Data Availability section, specify the data sharing mechanism (e.g., 'available on reasonable request to the corresponding author' with conditions) to meet the journal's data sharing policy.The current statement is vague and does not provide readers with a clear path to access the data.
- 2.HIGHreportingIn the Methods, name the statistical software (e.g., R version, SAS version) used for the Bayesian analysis and secondary analyses.Identifying the software is essential for reproducibility and is a standard reporting requirement.
- 3.MEDIUMreportingIn the Methods or Supplementary, reference the CONSORT reporting guideline to enhance transparency.Explicitly stating adherence to CONSORT helps readers assess the completeness of reporting.
- 4.MEDIUMreportingIn the Results, provide a CONSORT flow diagram to clearly show patient disposition and missing data.A flow diagram improves transparency about patient flow and missing data handling.
- 5.MEDIUMrigorIn the Discussion, explicitly address how the limitations of prior phase 2 trials (e.g., lack of thrombectomy) were mitigated in the current trial design.Addressing prior limitations strengthens the scientific premise and contextualizes the trial's contribution.
- 6.MEDIUMdata codeIn the Data Availability, consider depositing de-identified aggregate data in a public repository (e.g., Dryad) if permitted by consent and privacy regulations.Public data deposition maximizes transparency and allows independent verification.
- 7.MEDIUMrigorIn the Methods, describe the randomization algorithm in more detail, including the specific biased coin procedure and how imbalances were controlled.A fuller description of the randomization algorithm enhances reproducibility.
- 8.MEDIUMrigorIn the Methods, include a formal power analysis or sample size justification, detailing the assumed effect size, alpha, and power.A formal power analysis is expected for a phase 3 trial and helps readers assess the trial's ability to detect effects.
- 9.LOWcopyeditIn Results, Secondary Efficacy Analyses, correct the typo 'odd ratio' to 'odds ratio'.Fixing the typo improves clarity and professionalism.
- 10.LOWcopyeditIn Table 3, ensure the safety population N is clearly labeled and consistent with the definition (e.g., 'Safety population' with N=483).Clarifying the safety population N avoids confusion with the randomized N.
- 11.LOWcopyeditIn the Abstract and Results, ensure consistent use of 'assigned' vs 'randomized' throughout.Consistent terminology prevents ambiguity about the trial procedures.
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.