This Week in Critical Care — Aug 21, 2026

Welcome to This Week in Critical Care. This week we're covering 10 notable papers spanning resuscitation decisions in septic shock, the mechanics and pharmacology of the airway, and a set of studies on diagnostics, glucose timing and system-level safety. Let's dive in.

We start with septic shock, where two studies asked whether doing something earlier or more aggressively actually helps. In Intensive Care Medicine, Pène and colleagues report TRANSPORT, a multicentre randomised trial across eighteen French centres testing liberal versus restrictive red cell transfusion in patients with cancer and septic shock. Adults with haematological or solid malignancies, a lactate above 2 and a haemoglobin under 9 were assigned to a transfusion threshold of either 9 or 7 grams per decilitre during the first 48 hours of resuscitation. The trial was designed for 260 patients but was stopped early for safety, with 187 enrolled. Lactate clearance at 12 hours was essentially identical, achieved in about half of patients in each arm, and 28-day mortality did not differ, at roughly 46 percent with the liberal strategy and 53 percent with the restrictive one. What stopped the trial was thrombosis: arterial or venous thrombotic events occurred in about 13 percent of the liberal group versus one percent of the restrictive group. The authors are appropriately cautious, noting that the absence of any mortality benefit alongside an unexplained imbalance in harm does not support routinely preferring either strategy. But for the practising intensivist, this is another data point telling you that a haemoglobin of 7 remains a defensible floor even in the anaemic patient with cancer, and that transfusing to a target of 9 buys you no faster lactate clearance.

Alongside that, in the Journal of Critical Care, Lee and colleagues looked at the timing of vasopressin in a single-centre cohort of 674 patients with septic shock, comparing initiation within four hours of shock onset against four to twenty-four hours. The hypothesis was renal protection, and it was not borne out: the risk of needing renal replacement therapy over the following week was not significantly different between early and late starters. What early initiation was associated with was more shock reversal within 24 hours, roughly 42 percent versus 30 percent, and lower fluid volumes, with no difference in mortality or ventilation. So if you start vasopressin early, do it for haemodynamics and fluid sparing, not on the promise of saving the kidneys. Setting both of these in context, Critical Care Medicine published a nationwide South Korean cohort from Lee and colleagues covering nearly 1.15 million sepsis hospitalisations between 2011 and 2022. Recorded incidence more than doubled, in-hospital mortality fell from about 27 percent to 17 percent, and yet the absolute number of sepsis deaths rose from roughly 15,600 to over 25,000 a year. Socioeconomic gradients persisted throughout, with higher mortality among medical aid beneficiaries and rural residents, and better outcomes at tertiary hospitals. The authors flag that coding practices may have shifted over time, but the message about volume outrunning case fatality is hard to escape.

Turning to the airway and the ventilator, Chest published a systematic review and network meta-analysis from McDougall and colleagues on induction agents for tracheal intubation in the critically ill, pooling 21 randomised trials and just over 6,000 patients across the intensive care unit, emergency department, operating room and prehospital settings. Compared with etomidate, ketamine probably causes about 30 percent more peri-intubation haemodynamic instability, a moderate-certainty finding that will surprise clinicians who reach for ketamine precisely to protect the blood pressure. The picture then complicates: etomidate may increase the need to start post-intubation vasopressor infusions relative to ketamine, and it clearly causes more adrenal suppression. Ketamine combined with propofol looked best for haemodynamic stability against both single agents, but on low-certainty evidence from limited data. The honest reading is that the competing effects roughly offset, the certainty is modest, and this analysis does not license you to abandon your usual agent — but it should temper the confidence with which ketamine is described as the haemodynamically safe choice.

On the ventilator itself, two papers push individualisation. Also in Intensive Care Medicine, Grieco and colleagues performed a post hoc analysis of the VENTIBRAIN study in 1,158 patients with acute brain injury, asking whether respiratory system elastance modifies the relationship between ventilator settings and outcome. Tidal volumes were similar across elastance tertiles at roughly 7 to 8 millilitres per kilogram, which produced a wide gradient in driving pressure, from a median of 5 up to 12 centimetres of water. In patients with low elastance — compliant lungs — higher tidal volumes were associated with about half the odds of intensive care unit mortality, whereas in stiff lungs there was no such association. Their Bayesian modelling suggests the optimal tidal volume falls continuously as elastance rises, from over 11 millilitres per kilogram in the most compliant lungs down to around 4 in the stiffest, with the optimal respiratory rate rising in parallel from about 15 to 19 breaths per minute. They quantify the trade-off as roughly one centimetre of water of driving pressure being prognostically equivalent to about three breaths per minute, and the findings held regardless of whether the patient had acute respiratory distress syndrome. This is observational and hypothesis-generating, but it argues against a reflexive 6 millilitres per kilogram in every brain-injured patient with normal lungs. That dovetails with a narrative review in the same journal from Wongtirawit and colleagues, which walks through the practical bedside questions guidelines do not answer in acute respiratory distress syndrome — how to set initial ventilation, what to do when the driving pressure target is unreachable, how to assess recruitability and set positive end-expiratory pressure, and how to titrate sedation to respiratory drive during the transition to assisted breathing. Worth reading in full if you teach ventilation.

Still on the respiratory theme, Critical Care published a randomised crossover physiological study from Xu and colleagues in 20 patients, comparing a T-piece spontaneous breathing trial with high-flow oxygen delivered through the endotracheal tube at two flow rates and two expiratory port diameters. The narrower port at 60 litres per minute generated substantial positive airway pressure — a median positive end-expiratory pressure around 6 centimetres of water — attenuated end-expiratory lung volume loss by around 300 millilitres, lowered dynamic transpulmonary driving pressure, improved oxygenation and slowed the respiratory rate. Importantly, inspiratory effort measured by oesophageal pressure swings did not differ significantly across any of the trial conditions or after extubation. So tracheal high-flow behaves as a physiologically distinct option that preserves realistic effort while limiting derecruitment. Whether that translates into better extubation outcomes is entirely untested.

Three remaining papers address diagnostics, glucose and system safety. In Intensive Care Medicine, Millot and colleagues randomised 156 patients with suspected ventilator-associated or ventilated hospital-acquired pneumonia to rapid multiplex polymerase chain reaction testing added to conventional microbiology, or conventional microbiology alone. The primary outcome — targeted antimicrobial therapy at 24 hours — was not significantly increased overall, at about 35 percent versus 24 percent. Only in the subgroup with culture-confirmed pneumonia, fewer than half of those enrolled, did the difference reach significance. This is a negative trial for the primary endpoint, and it suggests the bottleneck is clinician behaviour and diagnostic uncertainty rather than turnaround time. In Chest, Gantzel and colleagues used target trial emulation in nearly 5,800 Swedish intensive care patients with moderate hyperglycaemia, between 10 and 13.9 millimoles per litre. Insulin started within six hours was associated with about a 2 percentage point lower adjusted 30-day mortality, and starting within one to two hours was associated with a reduction closer to 7 percentage points. That is a large effect for an observational emulation, and residual confounding by indication is the obvious concern, so treat this as a signal justifying a trial rather than a mandate. Finally, in the Journal of Critical Care, Mendes and colleagues used a large language model pipeline to screen more than 8,500 Australian coronial reports, identifying 563 potentially preventable intensive care-related deaths since 1997. Communication, handover and documentation failures dominated, coded in nearly two thirds of cases, ahead of treatment or device problems and diagnostic process failure, with perioperative deaths carrying the heaviest multiple-failure burden. Agreement between the model and clinician reviewers was substantial. The actionable lesson is structural: strengthen handover interfaces and postoperative ward rescue after major surgery.

If you only have time for one paper this week, make it the TRANSPORT trial in Intensive Care Medicine [1]. It is the first randomised evidence on transfusion thresholds specifically in patients with cancer and septic shock, and it removes the intuitive argument for transfusing to a higher haemoglobin in this anaemic, high-mortality group.

Here are the key takeaways from this week in Critical Care. A liberal transfusion target in cancer patients with septic shock did not improve lactate clearance or survival and was stopped early over thrombotic events, so a restrictive threshold remains reasonable [1]. Early vasopressin in septic shock reversed shock faster but did not reduce renal replacement therapy [8]. Ketamine is not the haemodynamically neutral induction agent it is often assumed to be, though etomidate's adrenal suppression and vasopressor requirement complicate the comparison [3]. In brain-injured patients with compliant lungs, a fixed low tidal volume may not be optimal — think in terms of driving pressure traded against respiratory rate [5]. And rapid molecular pneumonia panels did not significantly increase targeted therapy at 24 hours overall, so faster results alone will not change prescribing [2].

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 a quick rating on Apple Podcasts or Spotify helps other physicians discover the show.

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