This Week in Critical Care — Sep 1, 2026

Welcome to This Week in Critical Care. This week we're covering 10 notable papers spanning mechanical circulatory support and resuscitation, hemodynamic assessment and vasoactive management, and the metabolic and microbial side of critical illness — from protein delivery to the glycocalyx. Let's dive in.

We start with extracorporeal support, where two papers from different journals converge on an uncomfortable message about expectations. In Critical Care, Dettling and colleagues used a target trial emulation across two high-volume centres to ask whether mechanically unloading the left ventricle changes outcomes in refractory cardiogenic shock on peripheral veno-arterial ECMO. Among 264 patients supported for at least 48 hours, roughly half received early unloading within the first 24 hours, split between intra-aortic balloon pump and Impella. After propensity-score overlap weighting, there was no difference in 60-day mortality — the weighted risk difference was about one percent, which is to say nothing — and no difference in weaning rates, while device-related complications were more frequent in the unloading arm. This is retrospective and hypothesis-generating, and the authors are careful to frame it as an argument for phenotype-driven selection rather than a blanket verdict. But if your unit has drifted toward routine early unloading on the assumption of survival benefit, this analysis does not support that assumption, and it does remind you that each additional device carries a real complication cost. Sitting alongside that, Critical Care Medicine published a systematic review and meta-analysis by Stebbins and colleagues covering 163 studies and more than 78,000 adults, looking at outcomes at six months and beyond. One-year mortality was about 37 percent after veno-venous ECMO, around 55 percent after veno-arterial, and just under three quarters after extracorporeal CPR. Those are numbers worth having at hand for family conversations. The more sobering finding is that fewer than a quarter of the studies reported any functional outcome at all, and those that did used a scattergun of instruments, most often Cerebral Performance Category. So we can tell families how likely survival is; we still cannot tell them reliably what survival looks like.

Staying with resuscitation, Resuscitation carried a study from Kano and colleagues on cerebral tissue oxygen saturation measured by near-infrared spectroscopy during ongoing CPR after out-of-hospital cardiac arrest. Looking at the first monitored in-hospital defibrillation in patients with a shockable rhythm, median cerebral saturation was just under 49 percent in those who achieved return of spontaneous circulation versus about 38 percent in those who did not, and each one-percent increment in saturation was associated with roughly a fifty percent increase in the odds of ROSC after that shock. Discrimination was very high, and a cutoff around 44 percent was both sensitive and specific. Two important caveats: only 19 patients achieved ROSC after that first monitored shock, and the authors explicitly warn that this exploratory threshold needs external validation and must never delay guideline-directed defibrillation. Read this as a signal that cerebral oximetry may eventually help gauge perfusion quality during resuscitation — not as permission to shock selectively.

Our second theme is hemodynamic assessment and vasoactive strategy, where the week offers one solid validation study and one provocative critique. In Chest, D'Alto and colleagues performed a multicentre invasive validation of the Venous Excess Ultrasound score across seven Italian pulmonary hypertension centres, comparing bedside ultrasound with right heart catheterisation performed within an hour. Across 187 patients, mean right atrial pressure rose stepwise with each VExUS grade — roughly 4 millimetres of mercury at grade zero climbing to about 16 at grade three — and the score discriminated a right atrial pressure above 12 millimetres of mercury almost perfectly, outperforming isolated echocardiographic markers and performing comparably to full echocardiographic estimation of right atrial pressure. Consistency held across pre-capillary and post-capillary phenotypes. This is the invasive confirmation VExUS has been missing, and it supports using the score as a credible non-invasive readout of venous congestion. Then, in the Journal of Critical Care, Zaidi and colleagues offer a critical appraisal arguing that the vasopressor weaning literature — thirteen comparative studies, over 2,500 patients — has been asking the wrong question. They highlight a genuine paradox: both randomised trials suggest weaning norepinephrine first produces more hypotension, while observational data consistently point the other way. Their argument is that the contradiction is methodological, not biological, driven by hypotension definitions that conflate clinician dose-escalation behaviour with real hemodynamic events, by failure to separate weaning sequence from whether vasopressin is stopped abruptly or titrated, and by uncontrolled corticosteroid use. They flag a possible signal favouring vasopressin-first weaning in patients with reduced left ventricular function and propose a two-by-two factorial trial addressing sequence and method together. For now, the practical message is that the method of vasopressin withdrawal may matter as much as the order.

Our third theme covers metabolism, host biology, and populations we under-study. In the Journal of Critical Care, Chaba and colleagues report a nested cohort within the TARGET Protein trial, and the results should temper enthusiasm for augmented protein. Comparing usual intake of about 0.84 grams per kilogram per day against augmented delivery of about 1.35, the augmented group achieved a less negative nitrogen balance — about five grams per day better — but roughly sixty percent of that additional nitrogen was accounted for by urea generation, with higher plasma urea concentrations. More concerning, among patients without kidney injury at admission, augmented protein was independently associated with incident acute kidney injury, with roughly a fivefold increase in odds, though the confidence around that estimate was wide. The implication is that much of the extra protein we push is going to nitrogen waste rather than anabolism. Two papers examine specific populations. In Chest, Subileau and colleagues describe septic shock after allogeneic stem-cell transplantation across a multicentre ICU cohort — 444 patients, with 90-day mortality of 63 percent rising to 81 percent at three years. Fungal documentation, higher SOFA score, and neutropenia were each independently associated with worse survival, and early aminoglycoside use in a landmark analysis was associated with better 90-day survival, though the authors explicitly caution that the association is time-varying and does not support a causal reading. And in the Journal of Critical Care, Azevedo and colleagues report a decade of dengue ICU admissions across Brazil, more than 32,000 episodes in the national registry. Organ support in the first 24 hours was relatively uncommon — mechanical ventilation in about six percent, vasopressors in about seven percent — yet in-hospital mortality was about eight percent in the registry and nearly nineteen percent in the public-system dataset, with markedly higher mortality among older patients and those in public hospitals. That disparity is the finding to sit with. Rounding out the week, Critical Care Medicine published a narrative review by Klingensmith and colleagues on the microbiome in critical illness, arguing that both disease and our own therapeutics reshape the gut microbiome in ways that plausibly influence recovery, and the Journal of Critical Care carried a small prospective substudy by Heldeweg and colleagues linking glycocalyx-derived heparan sulfate, but not syndecan-1, to a persistent strong ion gap after cardiopulmonary bypass — an early mechanistic hint at what those unmeasured anions actually are.

If you only have time for one paper this week, make it the Critical Care target trial emulation on left ventricular unloading during veno-arterial ECMO. It directly challenges a practice that has spread faster than its evidence, and it quantifies the complication cost of adding a second device.

Here are the key takeaways from this week in Critical Care. Routine early left ventricular unloading on veno-arterial ECMO was not associated with better 60-day survival and came with more device-related complications — select patients, don't default. Long-term ECMO mortality varies enormously by modality, from roughly a third at one year with veno-venous support to about three quarters after extracorporeal CPR, and functional outcomes remain poorly characterised. VExUS now has invasive validation in pulmonary hypertension and tracks right atrial pressure closely enough to trust at the bedside. Augmented enteral protein improves nitrogen balance mostly on paper — most of the gain appears as urea, with a signal toward new acute kidney injury. And in the vasopressor weaning debate, the method of vasopressin withdrawal may be the variable we have been ignoring.

That's your roundup for This Week in Critical Care. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And if it saves you time, a quick rating in your podcast app helps other physicians find the show.

This is an automated summary generated by artificial intelligence, which can make mistakes. Always review the original source materials.