This Week in Critical Care — Sep 22, 2026

Welcome to This Week in Critical Care. This week we're covering 10 notable papers spanning fluid and kidney management in the sickest patients, cardiac arrest and ventilator liberation, and the softer but no less consequential science of how we make and communicate high-stakes decisions. Let's dive in.

We start with resuscitation fluids and the kidney, where one large cohort pushes an old question further upstream and two other papers deliver sobering messages. In Chest, Kim and colleagues asked whether the crystalloid a septic patient receives before they ever reach the intensive care unit matters. This was a nationwide multicentre retrospective cohort of nearly 3,700 adults with sepsis in Korea, of whom about 37 percent received a balanced crystalloid and the remainder saline during initial resuscitation in the emergency department, the ward, or the operating room. After propensity score matching left just over 1,300 patients in each arm, balanced crystalloids were associated with roughly a 30 percent lower odds of death at 30 days, a signal that held up under inverse probability weighting and across age, sex, illness severity, and whether or not patients got more than 30 millilitres per kilogram of fluid or early vasopressors [1]. The secondary findings are coherent with the mechanism you'd expect: less continuous renal replacement therapy during the stay, less dialysis dependence at intensive care unit discharge, and more ventilator-free days. The effect was concentrated in patients without chronic kidney disease, those resuscitated in the emergency department or on the ward, and those whose baseline chloride was at or below 110 and whose baseline pH was under 7.35. This is observational and unmeasured confounding by institution and era is a real concern, particularly since balanced fluid use rose over the study period. But it is a well-aligned addition to the randomised literature, and it argues that the fluid decision worth auditing is the one made in the emergency department, not the one made on arrival to your unit.

On the pharmacological side of kidney protection, Intensive Care Medicine published a phase two trial that came up empty. Mourisse and colleagues randomised 244 patients with pre-existing estimated glomerular filtration rates between 25 and 65 undergoing complex on-pump cardiac surgery to two perioperative doses of ilofotase alfa, a recombinant alkaline phosphatase, or placebo, with 204 patients receiving both doses and entering the analysis. The primary endpoint, the ratio of peak postoperative creatinine to the preoperative value over five days, was essentially identical between groups, and major adverse kidney events by day 60 occurred in about 16 percent of patients in each arm [4]. There were no safety concerns, but there is also no efficacy signal here. Alkaline phosphatase joins the long list of agents that attenuate renal injury in models and not in patients, and for now prevention of cardiac surgery associated acute kidney injury remains a matter of haemodynamics, nephrotoxin avoidance, and perfusion strategy.

And when the kidneys do fail in the presence of a failing liver, Critical Care Medicine offers an uncomfortable epidemiological snapshot. Karvellas and colleagues analysed two contemporaneous North American registries, HARMONY and CRRTnet, covering roughly 450 critically ill patients with cirrhosis or acute-on-chronic liver failure started on continuous renal replacement therapy. Survival to hospital discharge was only around a third in both cohorts in the absence of transplantation, with acute tubular necrosis rather than hepatorenal syndrome driving most of the need for dialysis. The striking process finding is that these patients were commonly prescribed effluent doses above 30 millilitres per kilogram per hour, and that higher dose was independently associated with nearly threefold higher odds of in-hospital death [5]. That is almost certainly confounding by indication — sicker patients get more aggressive prescriptions — but it is a clear reminder that dose escalation beyond guideline targets has no demonstrated benefit and may mark a group for whom we should be having goals-of-care conversations. Reassuringly, regional citrate anticoagulation was used in well over half of patients despite theoretical toxicity concerns, and among survivors, about two thirds of patients transitioned to intermittent haemodialysis before discharge.

Turning to cardiac arrest, two papers address very different links in the chain of survival. In Chest, Niu and colleagues used the nationwide, prospective Chinese BASIC registry of in-hospital cardiac arrest, screening more than 32,000 events to identify 612 adults with refractory ventricular fibrillation or pulseless ventricular tachycardia who received either lidocaine or amiodarone, with every case adjudicated against guideline indications. Survival to discharge or 30 days was around 20 percent with lidocaine and about 24 percent with amiodarone, sustained return of spontaneous circulation was around half in both groups, and after multilevel adjustment there was no significant difference in survival, return of circulation, or favourable neurological outcome [3]. Subgroups were consistent. Since nearly all of our antiarrhythmic evidence comes from out-of-hospital arrest, this fills a genuine gap and supports current guidance that either drug is acceptable — so use the one your team can draw up fastest.

The more provocative arrest paper is in the American Journal of Respiratory and Critical Care Medicine, where Sutton and colleagues applied the airway opening index to paediatric resuscitation. In a prospective multicentre cohort of 99 arrest events in children with invasive airways and capnography, the index was computed from the change in carbon dioxide with each chest compression during ventilation, scaled from zero for a closed airway to one for a fully patent one, with a median value of about 0.38. Events with an average index at or above 0.35 had roughly 50 percent higher likelihood of both survival to discharge and favourable neurological outcome, and after return of circulation their peak arterial lactate was about half that of the low-index group despite similar compression durations [6]. This is association, not causation, and the cohort is small, but it is a mechanistically plausible, real-time, non-invasive physiological target derivable from the capnograph you already have taped to the tube — and it suggests that during paediatric resuscitation, alveolar ventilation efficiency deserves the same attention we give to compression depth.

The third cluster is about recovery and prognostication. The single most eye-catching result of the week is a randomised trial in Chest from Liu and colleagues, conducted at a single centre in Beijing, enrolling 150 patients aged 80 or older on prolonged mechanical ventilation and randomising them to a multicomponent pulmonary rehabilitation programme or standard rehabilitation with assessor blinding. Successful weaning was achieved in about 45 percent of patients in the intervention group versus 12 percent with usual care, a roughly fivefold higher rate of liberation that persisted in per-protocol and competing-risk analyses. Diaphragm thickening fraction, cough strength, oxygen saturation off the ventilator, and muscle strength all improved, ventilator-associated pneumonia was less frequent at 60 and 90 days, and deep vein thrombosis rates did not differ [2]. Ninety-day mortality was low in both arms and not significantly different, which tells you this was a selected, relatively stable long-stay population rather than an acutely dying one. The effect size is large enough to demand external replication, but the direction is consistent with everything we know about diaphragm atrophy and immobility, and it should make you ask whether your own long-stay octogenarians are getting a structured programme or just passive physiotherapy.

Two prognostic papers round this out, both essentially cautionary. In the Journal of Critical Care, Cao and colleagues systematically reviewed lymphopenia as a mortality marker in adult sepsis; of 29 studies, only four were poolable, and adverse lymphopenia phenotypes were associated with roughly double the risk of death, with moderate to substantial heterogeneity and sensitivity to individual studies [7]. The biology is compelling, but the authors are explicit that definitions and timing are too inconsistent to use the lymphocyte count for risk stratification or immunotherapy selection at the bedside. And in Critical Care Medicine, Ferraris and colleagues used population-based Ontario data on 1,204 older survivors of traumatic brain injury who filled a new antipsychotic prescription within 30 days of discharge, comparing haloperidol and risperidone against quetiapine for the development of epilepsy over five years. Neither comparison showed a higher epilepsy risk [8]. The caveat is severe: five-year mortality reached nearly 90 percent in the haloperidol group, so competing risk and confounding by frailty dominate, and the estimates are imprecise. Read it as an absence of a clear seizure signal rather than reassurance about antipsychotic safety in this group.

Finally, two papers in Chest examine the human systems around critical care. Rubin and colleagues performed a secondary qualitative analysis of interviews with 24 extracorporeal membrane oxygenation clinicians at international centres, applying regret theory, and found that although clinicians prefer utility-based reasoning about veno-venous candidacy, it is often unavailable — so both regret of omission and regret of commission shape who gets cannulated, with clinicians actively deploying strategies to minimise anticipated regret [9]. That is a useful explanation for why candidacy decisions are inconsistent even within a single centre, and an argument for structured, multidisciplinary candidacy review rather than relying on the memory of the last bad outcome. Alongside it, the Task Force for Mass Critical Care published a modified Delphi consensus report distilling 147 evidence statements into 22 suggestions on communication and coordination during pandemics and disasters, emphasising proactive, transparent engagement including social media to counter misinformation, medical operations coordination centres and interoperable information systems for real-time situational awareness, integration of emergency department expertise into incident command, attention to the last mile of the supply chain, and avoiding restrictive visitation policies that obstruct surrogate decision-making [10].

If you only have time for one paper this week, make it the balanced crystalloid cohort in Chest [1]. It shifts the fluid-choice question into the pre-intensive-care window where most septic resuscitation volume is actually given, and it is the rare observational study whose secondary outcomes all point the same way.

Here are the key takeaways from this week in Critical Care. First, the crystalloid chosen in the emergency department for sepsis may matter as much as the one chosen in the unit, with balanced solutions associated with lower 30-day mortality and less dialysis. Second, ilofotase alfa did not protect the kidneys after on-pump cardiac surgery, and in cirrhosis with acute kidney injury on continuous renal replacement therapy, effluent doses above 30 millilitres per kilogram per hour were associated with worse, not better, survival. Third, for refractory ventricular fibrillation during in-hospital arrest, lidocaine and amiodarone performed equivalently, while in children, a capnography-derived index of airway patency during compressions tracked both survival and post-arrest lactate. Fourth, a structured multicomponent rehabilitation programme markedly increased successful weaning in patients over 80 on prolonged ventilation, though in a single centre and awaiting replication. And fifth, lymphopenia is not yet ready as a sepsis risk-stratification tool, and clinician regret is a measurable, addressable driver of inconsistent extracorporeal membrane oxygenation candidacy decisions.

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 think of one colleague — in any specialty — who never has time to keep up with the literature. Tell them about AudioScholar: a free ten-minute weekly for every specialty, to listen to in any podcast app, or to read at audioscholar dot C C.

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