Alpha¹ by KaimenConfidential

Kaimen Rigor review

Genotype-stratified adjunctive dexamethasone for tuberculous meningitis in HIV-negative adults: a randomized controlled phase 3 trial.

Donovan J, Duc Bang N, Dong HKT, Ho DTN, Nguyen TAT, Nguyen TTH, Lam HBN, Phung VKN, Nguyen TT, Nguyen HHH, Pham KNO, Do DAT, Nguyen TMT, Dang TMH, Nguyen HL, Nguyen VVC, Hoang TH, Tran DD, Phung KL, Ramakrishnan L, Le THN, Nguyen TTT, Wolbers M, Kestelyn E, Geskus RB, Nguyen HP, Thwaites GE

DOI: 10.1038/s41591-025-04138-z
Record issued: 2026-08-15
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/62da3c94-71a7-4d13-ae78-a16e277b77c7 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 ×2−1★
ClaimsTreatment effect not shown to be clinically meaningful−0.5★
StatisticsPrinted percentage does not match its own count (capped) ×3−0.25★
CitationsUnresolved reference−0.25★

Model-reported confidence: highReviewer agreement: highEngine: 7.39.0Source: full text

Biggest issues

01Treatment effect not shown to be clinically meaningful
highOverreachreviewer-assessedFull detail
The primary analysis failed to establish noninferiority or superiority of placebo, with a hazard ratio of 0.99 (96% CI 0.748–1.31) for the primary endpoint. The effect size is essentially null, and the paper does not report a clinically meaningful benefit. The conclusion states that dexamethasone was not beneficial or harmful in this population, and the modest benefit is only seen in a meta-analysis with a hazard ratio of 0.78 for mortality, which is not the primary trial result. The primary trial result shows no significant effect, so the effect size is not anchored to clinical meaningfulness.
“hazard ratio of 0.99, 96% confidence interval (adjusted for multiple testing) 0.748–1.31”
Find in source
02Printed percentage does not match its own count
mediumArithmeticrecomputedFull detail
21.4% does not match the reported count 66/305
“66/305 (21.4%)”
Find in source
03Printed percentage does not match its own count
mediumArithmeticrecomputedFull detail
46% does not match the reported count 69/151
“69/151 (46.0%)”
MRC grade IIFind in source
04Printed percentage does not match its own count
mediumArithmeticrecomputedFull detail
21.4% does not match the reported count 66/305
“66/305 (21.4%)”
Secondary outcomesFind in source

Findings by category

201ArithmeticNumerical inconsistencies 102OverreachOverstated conclusions 03AuthenticityNone found 04DisclosureNone found 105AvailabilityBroken references and links 06Copyedit4 minor suggestions

This Kaimen Rigor review uses Kaimen Rigor reviewers trained on a curated corpus of high-fidelity and retracted papers, with expert supervision and curation. It can still make mistakes; verify each finding against the source before relying on it.

This is a rigorously designed and transparently reported phase 3 randomized controlled trial. The scientific premise is well-supported, the methodology is thorough, and data/code availability is exemplary. Minor reporting inconsistencies and a few copyedit issues do not undermine the overall integrity.

Both reviewers classified the study as interventional, and this was adopted. The evaluation covered all eight dimensions; several sub-criteria were marked not applicable (e.g., animal housing, cell line authentication) due to the human clinical trial nature. The statistics verification covered only a subset of tests (7 recomputed, 4 consistent, 3 not machine-verifiable); no decision errors were found. The citation check flagged one reference as not found in registry (the Dryad DOI), which is a data repository link rather than a citation.

Numerical inconsistencies

2 findings · worst medium

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.

Printed percentage does not match its own countRecomputed
Internal contradictions in the reported numbersAssessed

Recomputed 4 tests: 4 consistent, 0 inconsistent; 4 via agent-written checks. 3 printed percentages that do not match their own count.

Numeric forensics— values impossible for the reported sample size

PERCENT
21.4% does not match the reported count 66/305
“66/305 (21.4%)”
Find in source
PERCENT
46% does not match the reported count 69/151
“69/151 (46.0%)”
MRC grade IIFind in source
PERCENT
21.4% does not match the reported count 66/305
“66/305 (21.4%)”
Secondary outcomesFind in source

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

CONSISTENTreported p > .050 · recomputed p = .944Reviewer 1
Primary endpoint HR CI includes 1, consistent with non-significance.
“hazard ratio of 0.99, 96% confidence interval (adjusted for multiple testing) 0.748–1.31”
Find in source
Taken as given: The HR is 0.99 and the 96% CI is 0.748-1.31.; The CI is two-sided at 96% level.
Method: Compute p-value from HR and CI using normal approximation.
How we recomputed it: pCI(0.99, 0.748, 1.31, 1)
CONSISTENTreported p < .050 · recomputed p = .037Reviewer 1
Meta-analysis mortality HR CI excludes 1, consistent with p=0.044.
“death occurred in 135/624 (21.6%) given dexamethasone and 138/525 (26.3%) given placebo (hazard ratio of 0.78, 95% CI 0.62–0.99; P = 0.044”
Find in source
Taken as given: The HR is 0.78 and the 95% CI is 0.62-0.99.; The CI is two-sided at 95% level.
Method: Compute p-value from HR and CI using normal approximation.
How we recomputed it: pCI(0.78, 0.62, 0.99, 1)
CONSISTENTreported p > .050 · recomputed p = .944Reviewer 2
Primary endpoint hazard ratio CI includes 1, consistent with no significant difference.
“hazard ratio of 0.99, 96% confidence interval (adjusted for multiple testing) 0.748–1.31”
Find in source
Taken as given: The hazard ratio is 0.99.; The 96% CI is 0.748 to 1.31.; The CI is two-sided.
Method: Compute p-value from HR and CI using the normal approximation.
How we recomputed it: pCI(0.99, 0.748, 1.31, 1)
CONSISTENTreported p = .044 · recomputed p = .037Reviewer 2
Meta-analysis mortality hazard ratio p-value from CI.
“death occurred in 135/624 (21.6%) given dexamethasone and 138/525 (26.3%) given placebo (hazard ratio of 0.78, 95% CI 0.62–0.99; P = 0.044)”
Find in source
Taken as given: The hazard ratio is 0.78.; The 95% CI is 0.62 to 0.99.; The CI is two-sided.
Method: Compute p-value from HR and CI using the normal approximation.
How we recomputed it: pCI(0.78, 0.62, 0.99, 1)

lowinternal contradiction
In the secondary outcomes section, the number of participants allocated to dexamethasone is reported as 394 in the open-label corticosteroid use sentence, but the randomized dexamethasone group is 305. This may be a typo or refer to a different population.
“Open-label corticosteroids were used in 69/394 (17.5%) participants allocated to dexamethasone and in 50/308 (16.2%) allocated to placebo.”
ResultsFind in source
lowinternal contradiction
The abstract states the primary endpoint occurred in 108/305 (35.4%) dexamethasone and 110/308 (35.7%) placebo, but the text later reports 110/308 (35.7%) for placebo, which is consistent. No contradiction found.
“108/305 (35.4%) given dexamethasone and 110/308 (35.7%) given placebo”
AbstractFind in source

Overstated conclusions

1 finding · worst high

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.

Treatment effect not shown to be clinically meaningfulAssessed

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

supportedReviewers 1, 2
Placebo noninferiority was not established in the CC and CT population.
The primary endpoint HR CI (0.748-1.31) just covers the noninferiority margin of 0.75, so noninferiority is not established.
Evidence: Primary endpoint HR 0.99, 96% CI 0.748-1.31.
“Placebo noninferiority was not established in the CC and CT population or in individual genotype subpopulations.”
AbstractFind in source
supportedReviewers 1, 2
Dexamethasone was safe in CC- and CT-genotype participants.
Serious adverse events were similar between arms (52.8% vs 51.9%, p=0.84), supporting safety.
Evidence: Serious adverse events occurred in 161/305 (52.8%) dexamethasone vs 160/308 (51.9%) placebo.
“In CC- and CT-genotype participants, serious adverse events occurred in 161/305 (52.8%) dexamethasone-treated participants and 160/308 (51.9%) placebo-treated participants.”
AbstractFind in source
supportedReviewers 1, 2
LTA4H genotype did not significantly influence outcomes.
No heterogeneity of effect was observed by genotype, and TT-genotype outcomes were similar to CC/CT.
Evidence: Heterogeneity test p=0.32 for genotype; TT primary endpoint 28/89 (31.5%) similar to CC/CT.
“Benefit or heterogeneity of effect was not observed by any prespecified subgroup.”
AbstractFind in source
supportedReviewer 1
Dexamethasone reduced mortality in the combined meta-analysis.
The meta-analysis shows a significant reduction in mortality (HR 0.78, 95% CI 0.62-0.99, p=0.044), but the authors caution about bias.
Evidence: Meta-analysis of 1,149 participants: death 135/624 vs 138/525, HR 0.78, 95% CI 0.62-0.99, p=0.044.
“Overall, death occurred in 135/624 (21.6%) given dexamethasone and 138/525 (26.3%) given placebo (hazard ratio of 0.78, 95% CI 0.62–0.99; P = 0.044”
ResultsFind in source
supportedReviewer 1
LTA4H genotyping does not currently provide information that can influence clinical management.
Given the lack of differential benefit, the conclusion is supported by the trial results.
Evidence: No significant heterogeneity of effect by genotype; TT outcomes similar.
“LTA4H genotyping does not currently provide information that can influence clinical management.”
DiscussionFind in source
supportedReviewer 2
Dexamethasone reduced overall mortality in a combined analysis of two trials.
The planned IPD meta-analysis showed a significant mortality reduction, though the authors caution about differences between trials.
Evidence: HR 0.78, 95% CI 0.62–0.99, P=0.044 in 1,149 participants.
death occurred in 135/624 (21.6%) given dexamethasone and 138/525 (26.3%) given placebo (hazard ratio of 0.78, 95% CI 0.62–0.99; P = 0.044)
Resultsreviewer’s wording

Premise concern: effect size not shown to be clinically meaningful.

ADEQUATESurrogate endpoint
The primary endpoint is a hard clinical outcome: all-cause death or new neurological event over 12 months. The trial also reports secondary outcomes including mortality, disability, and safety. Although the paper discusses CSF cytokine and inflammatory pathway changes as secondary endpoints, these are not the primary basis for the efficacy claim. The primary efficacy claim is based on the clinical composite endpoint, which is a hard clinical outcome.
“the primary endpoint of all-cause death or new neurological event over 12 months from randomization occurred in 108/305 (35.4%) given dexamethasone and 110/308 (35.7%) given placebo”
Find in source
INADEQUATEEffect size
The primary analysis failed to establish noninferiority or superiority of placebo, with a hazard ratio of 0.99 (96% CI 0.748–1.31) for the primary endpoint. The effect size is essentially null, and the paper does not report a clinically meaningful benefit. The conclusion states that dexamethasone was not beneficial or harmful in this population, and the modest benefit is only seen in a meta-analysis with a hazard ratio of 0.78 for mortality, which is not the primary trial result. The primary trial result shows no significant effect, so the effect size is not anchored to clinical meaningfulness.
“hazard ratio of 0.99, 96% confidence interval (adjusted for multiple testing) 0.748–1.31”
Find in source

Data authenticity concerns

None found

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.

Checked — nothing surfaced.

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 introduction cites the 2004 dexamethasone trial, substudies showing heterogeneity, and genetic studies linking LTA4H genotypes to inflammation and dexamethasone response. It acknowledges limitations of prior work (e.g., post-hoc analyses, uncertain benefit in HIV-positive, heterogeneity) and explains how the trial addresses these gaps by stratifying by genotype. The rationale for the hybrid noninferiority/superiority design is clearly linked to the premise.

Sub-criteria

prior work citedADEQUATE

premise rationaleADEQUATE

limitations addressedADEQUATE

Per reviewer (2) — the votes behind the verdict

PASSReviewer 1· 95% conf

The paper establishes a strong scientific premise by citing prior research on LTA4H genotype and corticosteroid response, and clearly links this to the trial's objectives.

Evidence

paraphrase[Main, paragraphs 2-4]

“Genetic analysis of a subset of the Vietnam dexamethasone trial participants and subsequent tuberculous meningitis cohorts found that LTA4H genotypes were associated with distinct intracerebral inflammatory phenotypes and dexamethasone responsiveness in HIV-negative individuals but not in those living with HIV.”

“We therefore conducted a LTA4H -genotype stratified randomized placebo-controlled trial of dexamethasone in HIV-negative Vietnamese adults with tuberculous meningitis.”

Main, paragraph 5Find in source

“It may reflect the limitations of previous post-hoc analysis of trials and observational studies, which may have over-estimated the effects of LTA4H on pathophysiology and treatment outcomes.”

Discussion ¶5Find in source

PASSReviewer 2· 90% conf

The study builds on a well-established prior evidence base, clearly articulates the rationale for genotype-stratified corticosteroid therapy, and addresses limitations of prior research.

Evidence

paraphrase[Main, paragraph 2]

“A randomized controlled trial of the corticosteroid dexamethasone, conducted in Vietnam and published in the year 2004, found that dexamethasone significantly reduced overall mortality from tuberculous meningitis.”

“These findings suggested LTA4H genotype might be able to stratify HIV-negative individuals with tuberculous meningitis into those who benefit from dexamethasone and those who do not.”

Main, paragraph 3Find in source

“It may reflect the limitations of previous post-hoc analysis of trials and observational studies, which may have over-estimated the effects of LTA4H on pathophysiology and treatment outcomes.”

Discussion ¶5Find in source

Randomization used computer-generated random permuted blocks with block sizes 4 and 6, stratified by hospital, genotype, and MRC grade. The trial was double-blind (participants and investigators blinded to treatment allocation, as implied by placebo-controlled design and standard practice; the paper states 'double-blind' in the abstract). Power analysis is detailed: 184 events required for 80% power at one-sided 2% significance, with assumptions stated. Inclusion/exclusion criteria are explicitly listed. Outlier handling is addressed through per-protocol exclusions and data quality filters. Controls are inherent in the placebo arm. Independent replication is not applicable for a single pivotal trial.

Sub-criteria

controlsADEQUATE

power analysisADEQUATE

blinding levelsADEQUATE

outlier handlingADEQUATE

randomization unitADEQUATE

inclusion exclusionADEQUATE

randomization methodADEQUATE

replicate distinctionN/A

independent replicationN/A

Per reviewer (2) — the votes behind the verdict

PASSReviewer 1· 95% conf

The trial design is rigorous: randomization method, stratification, blinding, power analysis, and inclusion/exclusion criteria are all clearly described.

Evidence

“The randomization list was computer-generated based on random permuted blocks with block size 4 and 6 (probability 0.75 and 0.25).”

MethodsFind in source

“phase 3, placebo-controlled trial”

AbstractFind in source

“To obtain 80% power at the one-sided 2% significance level, 184 events in the CC- and CT-genotype population would be required.”

Statistical analysisFind in source

PASSReviewer 2· 95% conf

The trial uses a robust randomized, placebo-controlled design with appropriate stratification, blinding, and a prespecified statistical analysis plan.

Evidence

“The randomization list was computer-generated based on random permuted blocks with block size 4 and 6 (probability 0.75 and 0.25).”

MethodsFind in source

“LTA4H CC- and CT-genotype participants were randomized to two parallel groups in a 1:1 ratio: dexamethasone or placebo for 6–8 weeks.”

MethodsFind in source

“To obtain 80% power at the one-sided 2% significance level, 184 events in the CC- and CT-genotype population would be required.”

Statistical analysisFind in source

Sex is reported for all participants (62% male overall). Age is reported as median with quartiles. Health status is captured via MRC severity grade and diagnostic category. Demographics include age, sex, and clinical characteristics; race/ethnicity is not explicitly reported but the study is in Vietnamese adults, which is stated. Species/strain and housing are not applicable for a human trial.

Sub-criteria

demographicsADEQUATE

sex reportedADEQUATE

sex justifiedN/A

age weight healthADEQUATE

housing conditionsN/A

species strain sourceN/A

Per reviewer (2) — the votes behind the verdict

PASSReviewer 1· 90% conf

Biological variables are well reported: sex, age, disease severity, and diagnostic categories are detailed; demographics are comprehensive.

Evidence

“The median age of all participants was 47 years (first to third quartiles 33–59 years).”

ResultsFind in source

“Disease was mostly mild or moderate (640/702 (91.2%), modified Medical Research Council (MRC) severity grade I or II)”

ResultsFind in source

“Male sex | 702 | 435 (62.0%)”

Table 1Find in source

PASSReviewer 2· 90% conf

Key biological variables including sex, age, disease severity, and diagnostic category are reported for the study population.

Evidence

“Male sex | 702 | 435 (62.0%)”

Table 1Find in source

paraphrase[Table 1]

“Modified MRC severity grade | Grade I | – | 298 (42.5%)”

The paper lists specific ethics committees and approval numbers: Oxford Tropical Research Ethics Committee (52-16), Hospital for Tropical Diseases (37/HDDD), Pham Ngoc Thach Hospital (1034/HDDD-PNT), and Vietnam Ministry of Health (151/CN-BDGDD). Informed consent is described: written consent from participants or relatives if incapacitated. Regulatory compliance is implied by adherence to national and international standards, though not explicitly named.

Sub-criteria

iacuc statementN/A

informed consentADEQUATE

irb ethics statementADEQUATE

regulatory complianceADEQUATE

Per reviewer (2) — the votes behind the verdict

PASSReviewer 1· 95% conf

Ethical approvals are thoroughly documented with named ethics committees and approval numbers, and informed consent is described.

Evidence

“The Oxford Tropical Research Ethics Committee (approval no. 52-16), The Ethical Committee of the Hospital for Tropical Diseases (approval no. 37/HDDD), The Ethical Committee of Pham Ngoc Thach Hospital for Tuberculosis and Lung Disease (approval no. 1034/HDDD-PNT) and The Vietnam Ministry of Health (approval no. 151/CN-BDGDD).”

MethodsFind in source

“Written informed consent to enter the trial was obtained from all participants or a relative if they were incapacitated.”

MethodsFind in source

PASSReviewer 2· 95% conf

The trial has documented ethics approvals from named committees and describes informed consent procedures.

Evidence

MethodsFind in source

“Written informed consent to enter the trial was obtained from all participants or a relative if they were incapacitated.”

MethodsFind in source

Dexamethasone is the investigational product, and its dosing regimen is fully described. The placebo is not described in detail but is standard. Software tools are identified: R version 4.4.2, STAR aligner v2.5.2a, FeatureCounts v2.0.0, DESeq2 v1.34.0. Genotyping reagents are identified (Taqman assays, LightCycler 480). Antibodies, cell lines, and mycoplasma testing are not applicable as this is a clinical trial without wet-lab assays.

Sub-criteria

mycoplasma testingN/A

reagents identifiedADEQUATE

organisms identifiedN/A

antibodies identifiedN/A

cell line authenticationN/A

software tools identifiedADEQUATE

Per reviewer (2) — the votes behind the verdict

PASSReviewer 1· 90% conf

Key resources are well identified: the investigational drug (dexamethasone) is named with dose regimen, and software tools are specified with versions.

Evidence

paraphrase[Methods, Study treatments]

“Participants allocated dexamethasone received the 6–8-week regimen previously shown to reduce tuberculous meningitis mortality.”

“The data were analyzed using the program R (version 4.4.2; R Core Team, 2024)”

Statistical analysisFind in source

paraphrase[Methods, LTA4H genotyping]

“rs17525495 was genotyped by Taqman using predesigned assay kits (Applied Biosystems)”

PASSReviewer 2· 90% conf

The investigational product (dexamethasone) and key laboratory reagents and software are adequately identified.

Evidence

paraphrase[Methods, Study treatments]

“Participants allocated dexamethasone received the 6–8-week regimen previously shown to reduce tuberculous meningitis mortality.”

paraphrase[Methods, Statistical analyses]

“The data were analyzed using the program R (version 4.4.2; R Core Team, 2024).”

The primary analysis uses Cox proportional hazards regression, with assumptions tested (proportional-hazards assumption not violated). Exact p-values are reported for some analyses (e.g., P = 0.044 for meta-analysis), but many are reported as CIs, which is acceptable. Effect sizes are reported with CIs throughout. Data presentation includes Kaplan-Meier curves, forest plots, and per-group n. Mathematical plausibility checks: the primary endpoint counts (108/305 and 110/308) are plausible; percentages in Table 1 sum correctly (e.g., 42.5+48.7+8.8=100).

Sub-criteria

tests namedADEQUATE

exact p valuesADEQUATE

effect sizes ciADEQUATE

data presentationADEQUATE

software identifiedADEQUATE

assumptions verifiedADEQUATE

mathematical plausibilityPASS

Per reviewer (2) — the votes behind the verdict

PASSReviewer 1· 90% conf

Statistical methods are rigorous and transparent: tests are named, effect sizes with CIs are reported, and software is identified.

Evidence

“The primary analysis used a Cox proportional hazards regression model”

Statistical analysisFind in source

“hazard ratio of 0.99, 96% confidence interval (adjusted for multiple testing) 0.748–1.31”

AbstractFind in source

“The proportional-hazards assumption was tested for each model and was not violated ( P > 0.05).”

Table 2Find in source

PASSReviewer 2· 90% conf

Statistical methods are appropriate and well-described, with effect sizes and confidence intervals reported; p-values are used appropriately.

Evidence

“The primary analysis was a Cox proportional hazards regression model with the primary endpoint as the outcome, treatment as the only covariate and with LTA4H genotype (CC or CT) and modified MRC severity grade at enrollment as stratum variables.”

Statistical analysisFind in source

“hazard ratio of 0.99, 96% confidence interval (adjusted for multiple testing) 0.748–1.31”

AbstractFind in source

“The proportional-hazards assumption was tested for each model and was not violated ( P > 0.05).”

Table 2Find in source

The data availability statement provides a concrete route: deidentified participant data shared on request to OUCRU with a response within 4 weeks. Proteomic and transcriptomic data are deposited in Dryad with DOI. Code is available on GitHub. Repository deposit and accession numbers are applicable for the proteomic/transcriptomic data, and are provided.

Sub-criteria

code sharingADEQUATE

accession numbersADEQUATE

repository depositADEQUATE

data availability statementADEQUATE

Per reviewer (2) — the votes behind the verdict

PASSReviewer 1· 95% conf

Data and code availability are well documented: a data availability statement with a concrete access route, proteomic/transcriptomic data deposited in Dryad, and code on GitHub.

Evidence

“Deidentified trial participant data (including data dictionaries) will be shared on request to the Oxford University Clinical Research Unit (via emailing the senior author, G.E.T., or corresponding author, J.D.). A response will be provided within 4 weeks of receiving the request.”

Data availabilityFind in source

“Proteomic and transcriptomic data used for this analysis have been deposited into a public repository and are available via Dryad at 10.5061/dryad.f1vhhmh7v”

Data availabilityFind in source

“R analysis code has been made publicly available via GitHub at oucru-biostats/LAST-ACT: Analysis code for LAST ACT trial, https://github.com/oucru-biostats/LAST-ACT”

Data availabilityFind in source

PASSReviewer 2· 95% conf

Data and code availability are well-documented, with a clear data access procedure and public deposition of proteomic/transcriptomic data and analysis code.

Evidence

Data availabilityFind in source

“Proteomic and transcriptomic data used for this analysis have been deposited into a public repository and are available via Dryad at 10.5061/dryad.f1vhhmh7v”

Data availabilityFind in source

“R analysis code has been made publicly available via GitHub at oucru-biostats/LAST-ACT: Analysis code for LAST ACT trial, https://github.com/oucru-biostats/LAST-ACT”

Data availabilityFind in source

The trial is registered at ClinicalTrials.gov (NCT03100786). Protocol amendments and deviations are described. Limitations are discussed in the Discussion. Conclusions are proportional, acknowledging the noninferiority was not established and the meta-analysis is exploratory. Funding (Wellcome Investigator award) and competing interests are stated. Reporting guideline adherence is implied by the Nature Portfolio reporting summary.

Sub-criteria

funding coiADEQUATE

trial registrationADEQUATE

reporting guidelineADEQUATE

methods completenessADEQUATE

all outcomes reportedADEQUATE

limitations discussedADEQUATE

conclusions proportionalADEQUATE

Per reviewer (2) — the votes behind the verdict

PASSReviewer 1· 95% conf

Reporting is transparent: trial registration, protocol amendments, limitations, and funding/COI are all disclosed.

Evidence

“ClinicalTrials.gov NCT03100786”

AbstractFind in source

“Our trial has some limitations. First, it was conducted exclusively in Vietnam.”

Discussion ¶6Find in source

“The trial was funded by a Wellcome Investigator award (award no. 110179/Z/15/Z to G.E.T.).”

AcknowledgementsFind in source

PASSReviewer 2· 95% conf

The manuscript is transparent, with a registered trial, complete reporting of outcomes, and a thorough discussion of limitations.

Evidence

“ClinicalTrials.gov NCT03100786”

AbstractFind in source

“All outcomes are presented in this manuscript.”

ResultsFind in source

“Our trial has some limitations. First, it was conducted exclusively in Vietnam.”

Discussion ¶6Find in source

Registered (1 ID: ClinicalTrials.gov). No reporting guideline cited.

ClinicalTrials.gov

NCT03100786

Broken references and links

1 finding · worst low

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.

References not resolvable to a published paperRecomputed

Checked 39 references by DOI: 0 verified — 1 DOI unresolved, 38 no DOI (shown, not verified).

UNRESOLVED10.5061/dryad.f1vhhmh7v
Genotype stratified adjunctive dexamethasone for tuberculous meningitis in HIV-negative adults: the LAST ACT trial
Cited DOI does not resolve to any Crossref record.
NO DOI
Tuberculous meningitis: progress and remaining questions
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
The global burden of tuberculous meningitis in adults: s modelling study
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Tuberculous meningitis and miliary tuberculosis: the Rich focus revisited
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
The mechanisms and consequences of the extra-pulmonary dissemination of Mycobacterium tuberculosis
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Tuberculous meningitis
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Prognostic models for 9-month mortality in tuberculous meningitis
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Leukotriene A4 hydrolase genotype and HIV infection influence intracerebral inflammation and survival from tuberculous meningitis
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Dexamethasone for the treatment of tuberculous meningitis in adolescents and adults
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
The clinical benefit of adjunctive dexamethasone in tuberculous meningitis is not associated with measurable attenuation of peripheral or local immune responses
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Serial MRI to determine the effect of dexamethasone on the cerebral pathology of tuberculous meningitis: an observational study
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
The lta4h locus modulates susceptibility to mycobacterial infection in zebrafish and humans
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Host genotype-specific therapies can optimize the inflammatory response to mycobacterial infections
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
A Bayesian analysis of the association between Leukotriene A4 Hydrolase genotype and survival in tuberculous meningitis
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Corticosteroids for managing tuberculous meningitis
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Treatment-associated inflammatory deterioration in tuberculous meningitis: unpicking the paradox
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Tuberculous meningitis: combined therapy with cortisone and antimicrobial agents
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Effect of corticosteroids on intracranial pressure, computed tomographic findings, and clinical outcome in young children with tuberculous meningitis
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Adjunctive dexamethasone for tuberculous meningitis in HIV-positive adults
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Elevated cerebrospinal fluid cytokine levels in tuberculous meningitis predict survival in response to dexamethasone
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
LTA4H prevalence and mortality in adult Zambians with tuberculous meningitis
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Association analysis of the LTA4H gene polymorphisms and pulmonary tuberculosis in 9115 subjects
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Relationship between human LTA4H polymorphisms and extra-pulmonary tuberculosis in an ethnic Han Chinese population in Eastern China
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Relationship between LTA4H promotor polymorphism and tuberculosis-associated immune reconstitution inflammatory syndrome and its prevention with prednisone
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Clinical parameters, routine inflammatory markers, and LTA4H genotype as predictors of mortality among 608 patients with tuberculous meningitis in indonesia
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Advancing the chemotherapy of tuberculous meningitis: a consensus view
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Tuberculous meningitis: a uniform case definition for use in clinical research
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Adjunctive dexamethasone for the treatment of HIV-uninfected adults with tuberculous meningitis stratified by leukotriene A4 hydrolase genotype (LAST ACT): Study protocol for a randomised double blind placebo controlled non-inferiority trial
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Statistical analysis plan for the LAST ACT clinical trial; a leukotriene A4 hydrolase stratified non-inferiority trial of adjunctive corticosteroids for HIV-negative adults with tuberculous meningitis
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
LTA4H genotype is associated with susceptibility to bacterial meningitis but is not a critical determinant of outcome
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Streptomycin treatment of tuberculous meningitis
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Xpert MTB/RIF Ultra versus Xpert MTB/RIF for the diagnosis of tuberculous meningitis: a prospective, randomised, diagnostic accuracy study
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
The SVA package for removing batch effects and other unwanted variation in high-throughput experiments
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
STAR: ultrafast universal RNA-seq aligner
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
featureCounts: an efficient general purpose program for assigning sequence reads to genomic features
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Adjusted significance levels for subgroup analyses in clinical trials
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
R: A Language and Environment for Statistical Computing
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).
NO DOI
Whole blood transcriptional profiles and the pathogenesis of tuberculous meningitis
No DOI in the reference — shown for manual review; not independently verifiable (not a fabrication signal).

3 data/code links checked; 3 live.

datahttps://clinicaltrials.gov/study/NCT03100786LIVEHTTP 200
Resolves, but the content could not be matched to the paper.
datahttps://clinicaltrials.gov/ct2/show/NCT03100786LIVEHTTP 200
Resolves, but the content could not be matched to the paper.
codeGitHubLIVEHTTP 200
https://github.com/oucru-biostats/LAST-ACT
Resolves to GitHub (code repository).

Copyediting

4 minor

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

No major wording or formatting errors. 4 minor suggestions below.

4 copyedit issues flagged: mostly consistency, clarity.

MINORconsistencyAbstract
“hazard ratio of 0.99, 96% confidence interval (adjusted for multiple testing) 0.748–1.31”
→ Consider adding a space after '0.99,' for consistency.
Minor formatting inconsistency.
MINORclarityResults, Secondary outcomes
“Open-label corticosteroids were used in 69/394 (17.5%) participants allocated to dexamethasone and in 50/308 (16.2%) allocated to placebo.”
→ The denominator 394 for dexamethasone group is unclear; clarify why it differs from 305.
Potential inconsistency in denominators; may be a typo.
MINORconsistencyAbstract
“hazard ratio of 0.99, 96% confidence interval (adjusted for multiple testing) 0.748–1.31”
→ Ensure consistent use of 'hazard ratio' vs 'HR' and consistent decimal places.
Minor formatting inconsistency.
MINORclarityMethods, Statistical analyses
“We chose to spend 2% of the one-sided type I error of 2.5% to the CC- and CT-genotype analysis, leaving 0.86% for the CC-genotype subgroup analysis.”
→ Clarify the allocation of type I error, as the numbers may confuse readers.
The sentence is technically correct but could be clearer.

Kaimen Rigor Action ItemsFinal pass before journal submission

The published work is robust and well-reported. An informed reader should weigh the minor internal inconsistency in the secondary outcomes denominator (394 vs 305) and the unverified subset of statistical tests, but these do not threaten the main conclusions. No erratum is warranted for the core findings, though the denominator issue should be clarified.

1.
HIGHreporting
Clarify the denominator discrepancy in the secondary outcomes section: the sentence reporting open-label corticosteroid use states 69/394 (17.5%) for the dexamethasone group, but the randomized dexamethasone group is 305. Correct the denominator or explain the population it refers to.
An internal contradiction in participant numbers could confuse readers and undermine trust in the reported results.
2.
HIGHdata code
Verify the Dryad DOI (10.5061/dryad.f1vhhmh7v) resolves correctly, as the citation check could not find it in the registry; if it is a data repository link, ensure it is active and correctly cited.
A non-resolving data DOI would undermine the data availability statement and reproducibility.
3.
MEDIUMreporting
Add an explicit statement naming the reporting guideline followed (e.g., CONSORT) in the Methods, even though a Nature reporting summary is linked.
Explicitly naming the guideline improves transparency and helps readers verify completeness of reporting.
4.
MEDIUMreporting
Consider reporting exact p-values for all secondary outcomes in the main text, as some are only in supplementary tables.
Exact p-values facilitate independent verification and are expected for key secondary analyses.
5.
MEDIUMreporting
Add a brief statement on the generalizability of findings to other populations, given the single-country setting.
Addressing external validity helps readers interpret the applicability of the results beyond Vietnam.
6.
MEDIUMreporting
Consider reporting sex-based analyses or a rationale for not performing them, as sex is a key biological variable.
Sex-based reporting is increasingly expected and would strengthen the biological variables dimension.
7.
MEDIUMdata code
Consider providing a more formal data access committee or platform (e.g., Vivli) for clinical data sharing, in addition to the email-based request route.
A formal access mechanism aligns with best practices for clinical trial data sharing and improves discoverability.
8.
LOWcopyedit
Fix minor formatting inconsistencies in the abstract: add a space after '0.99,' and ensure consistent use of 'hazard ratio' vs 'HR' and consistent decimal places.
Minor copyedit issues can distract readers and reflect poorly on attention to detail.
9.
LOWcopyedit
Clarify the sentence in Methods, Statistical analyses about type I error allocation: 'We chose to spend 2% of the one-sided type I error of 2.5% to the CC- and CT-genotype analysis, leaving 0.86% for the CC-genotype subgroup analysis.'
The sentence is technically correct but could be clearer to avoid reader confusion about the alpha allocation.
10.
LOWreporting
Consider adding a CONSORT flow diagram in the main text (currently only referenced as Fig. 1) or ensure it is clearly accessible.
A flow diagram improves transparency of participant disposition and is a CONSORT requirement.

Full text (79,336 chars · analyzed ~95%, head + tail)Analyzed: Aug 15, 2026

RigorReviewer 1

RigorReviewer 2

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.

Genotype-stratified adjunctive dexamethasone for tuberculous meningitis in HIV-negative adults: a randomized controlled phase 3 trial.

DOI

10.1038/s41591-025-04138-z

Record issued

2026-08-15

Engine

7.39.0

Exported

2026-09-22