Rodríguez-Lima DR, Rodríguez-Villegas A, Toro Egas JD, Campo Bautista EM. The Effectiveness of Methylene Blue in Adult Shock: A Systematic Review, Meta-Analysis, and Trial Sequential Analysis of Randomized Controlled Trials.
Methylene blue blocks the nitric oxide–guanylate cyclase pathway that drives vasoplegia. Whether that hemodynamic effect converts to a survival benefit has been asked, and answered differently, several times now.
This is at least the fourth systematic review/meta-analysis on methylene blue (MB) in adult shock published since 2023, each with a different scope: some restricted to septic shock specifically, others pooling septic shock with post-cardiac-surgery vasoplegia. Rodríguez-Lima et al. is the newest and, at 9 RCTs (535 patients), the largest septic-shock-focused pool to date, searched through February 2026 and registered on PROSPERO.
| Design | Systematic review, meta-analysis, and trial sequential analysis (TSA) of RCTs, PRISMA-reported |
| Pooled for mortality | 9 RCTs, 535 patients (8 septic-shock trials used for TSA; the Levin 2004 post-cardiac-surgery trial excluded from TSA for its distinct population and outsized graphical influence) |
| Primary mortality result | OR 0.73 (95% CI 0.40–1.36), 28–30 day mortality — directionally favorable, not statistically significant |
| TSA required information size (RIS) | 1,773 patients, calculated from an assumed 25% relative risk reduction, α=0.05, 80% power |
| TSA result | Cumulative Z-curve never crossed benefit, harm, or futility boundaries — the trial's own conclusion is "not enough data," not "no effect" |
| Secondary findings | No significant difference in renal replacement therapy or length of stay; individual trials showed faster vasopressor discontinuation and improved MAP |
This is the part the single-post summary can't hold: at least two prior meta-analyses on overlapping trial literature found a statistically significant mortality benefit.
| Analysis | Population | N (trials) | Mortality result |
|---|---|---|---|
| Ng et al., 2024 (Braz J Anesthesiol) | Septic shock only | 257 (5 RCTs) | OR 0.49 (95% CI 0.27–0.88), p=.02, low certainty |
| Pisano/Wieruszewski et al., 2023 (J Cardiothorac Vasc Anesth) | Critically ill + perioperative (incl. cardiac surgery) | — (11 RCTs) | RR 0.60 (95% CI 0.43–0.84), p=.003 |
| Chinese group, 2024 (BMJ Open, distributive shock) | Distributive shock, broader | 265 (6 RCTs) | No significant mortality reduction; TSA also inconclusive |
| Rodríguez-Lima et al., 2026 (this paper) | Septic shock, largest pool to date | 535 (9 RCTs) | OR 0.73 (95% CI 0.40–1.36), not significant |
The differences trace to scope, not to contradictory data. Ng et al. and Rodríguez-Lima et al. both restrict to septic shock, but Ng's pool is smaller (5 vs. 9–8 trials) and happened to catch a more favorable point estimate before later, larger trials diluted it. Pisano/Wieruszewski's larger, significant result comes from pooling septic shock together with post-cardiac-surgery vasoplegia — a different, generally lower-risk population responding to the same drug for a related but distinct physiologic problem. Rodríguez-Lima's team explicitly excluded the cardiac-surgery outlier trial from their own TSA for exactly this reason. None of these papers are wrong; they're answering slightly different questions with overlapping but non-identical trial sets.
"Trial sequential analysis says inconclusive" sounds like an objective fact. It's actually downstream of a chosen number that reasonable people have publicly disagreed about.
TSA requires the analyst to pre-specify an assumed relative risk reduction (RRR) to calculate how many patients would be needed to detect it (the required information size). Rodríguez-Lima et al. used an assumed 25% RRR, yielding the 1,773-patient RIS the source post cites. But a published letter responding to the Pisano/Wieruszewski meta-analysis raised exactly this point about a similar TSA: that meta-analysis's original TSA used an assumed 20% RRR and found the Z-curve inconclusive — but the actual, observed mortality reduction in that pooled data was 40.7% (15% MB vs. 25.3% control). When the letter's authors recalculated TSA using the observed 40.7% RRR instead of the conservative 20% assumption, the Z-curve did cross the significance boundary.
This isn't a criticism specific to Rodríguez-Lima et al. — their 25% assumption is a defensible, moderately conservative choice, and their own point estimate (OR 0.73, a smaller effect than the 40.7% RRR debated above) makes their inconclusive verdict the more internally consistent one. The broader lesson, useful well beyond this drug: a TSA's "we need more data" conclusion is only as trustworthy as the effect-size assumption baked into its required-information-size calculation. Two honest analysts can look at similar data and land on different verdicts depending on how optimistic that upfront assumption is.
A dedicated, adequately powered trial answering the septic-shock-specific mortality question is already underway: EMPRESS (Early Methylene Blue for Mortality Reduction in High-Dose VasoPREssor-Dependent Septic Shock), a multicenter RCT targeting exactly the vasoplegic, high-vasopressor-dose subpopulation where MB's mechanism should matter most.
The source post is accurate and its framing is right: this is evidence of not-enough-data, not evidence of no effect. What's worth adding is that this newest, largest, most conservative analysis sits at one end of a spread of overlapping meta-analyses on the same drug — two smaller ones found statistically significant mortality benefit, and the TSA method itself is sensitive to an upfront assumption that a published methodological letter has directly disputed in a closely related analysis. The hemodynamic benefit (MAP, vasopressor-sparing) is the most reproducible finding across all four papers; the mortality question remains genuinely open, and EMPRESS is the trial to watch for a real answer.