This Week in Critical Care — Aug 14, 2026

Welcome to This Week in Critical Care. This week we're covering 10 notable papers spanning septic shock resuscitation and the timing of vasoactive support, kidney replacement therapy and fluid management, and the evolving evidence base in hypoxemic respiratory failure and ARDS — plus two smaller studies that will change how you interpret blood cultures and prescribe stress ulcer prophylaxis. Let's dive in.

We start with septic shock, where two papers pull in opposite directions on whether doing things sooner helps. In Intensive Care Medicine, a target trial emulation using the MIMIC-Four and eICU databases asked when adjunctive vasopressin should be started once norepinephrine has escalated [1]. Among nearly four thousand adults with septic shock whose norepinephrine reached a quarter of a microgram per kilogram per minute or higher, the investigators used a clone-censor-weight design to compare starting vasopressin within the first three hours against starting it between three and six hours. Twenty-eight-day mortality was about forty-eight percent with the ultra-early strategy versus fifty-three percent with the later one, a risk difference of five percentage points, with a gain of roughly one and a half days of survival time over the follow-up window and about an eighteen percent lower hazard of death. Ultra-early vasopressin was also associated with roughly a third less use of renal replacement therapy and slightly fewer arrhythmias requiring treatment, though not with less acute kidney injury. This is observational, built from electronic health records with all the residual confounding that implies, and adherence to the strategies was limited to a few hundred clones in each arm — but it is the best available signal that once you are escalating norepinephrine, the window for adding vasopressin is measured in hours, not half-days. Contrast that with a cluster-randomised trial in Critical Care Medicine testing whether moving the whole resuscitation bundle even earlier — into the ambulance — saves lives [2]. French mobile ICU teams delivered antibiotics, up to thirty-five millilitres per kilogram of saline, norepinephrine for persistent hypotension, and hydrocortisone for norepinephrine-dependent shock, all within the first hour of prehospital contact. Among 381 analysed patients, twenty-eight-day mortality was twenty-two percent with the one-hour bundle versus twenty-seven percent with usual care, and that difference was not statistically significant. Secondary outcomes were similar. The authors themselves point to enrolment imbalance across centres and, crucially, a small actual difference in care delivered between arms. The honest reading is that prehospital bundling did not reduce mortality here, and that speed alone, without a meaningful change in what is actually given, does not translate into survival.

Moving to the kidneys, two papers reframe questions we thought were settled. In Critical Care Medicine, a Japanese registry study of nearly nineteen thousand propensity-matched ICU patients found that, on average, receiving renal replacement therapy was associated with about a six percentage point higher hospital mortality [3]. But a causal forest analysis showed substantial heterogeneity: patients with markedly elevated creatinine, high blood urea nitrogen, and low urine output appeared to benefit, while those with relatively preserved renal function appeared to be harmed. This does not overturn the trials showing no benefit from early initiation; it suggests the average effect hides two populations, and that the discipline of waiting matters most in patients whose kidneys still have something left. The companion question — how aggressively to remove fluid once you are on continuous renal replacement therapy — is addressed in a two-centre European cohort of 973 adults across nearly seven thousand days of therapy, also in Critical Care Medicine [5]. After accounting for fluid balance, the net ultrafiltration rate itself had no independent association with twenty-eight-day mortality. Mediation analysis suggested the apparent survival edge among patients ultrafiltered faster was explained almost entirely by their achieving a more negative fluid balance, while more positive daily fluid balance tracked with worse outcomes. The practical message is that net ultrafiltration is a tool, not a target: set a fluid balance goal for the individual patient and let the rate follow, rather than fixating on millilitres per kilogram per hour.

Turning to the lungs, three papers together map how much of our ARDS playbook rests on solid ground. A registry cohort from nine Toronto-affiliated ICUs in Critical Care Medicine screened almost twenty-two thousand patients and found that half OF THE COHORT met criteria for acute hypoxemic respiratory failure within the first twenty-four hours, defined by a P-to-F ratio at or below three hundred [7]. Of those patients, seventy-six percent required invasive ventilation, most received lung-protective settings, and ICU mortality was twenty-three percent overall, rising to forty-seven percent among ventilated patients with severe hypoxemia. The point is that we are already applying ARDS-derived strategies to a much broader population than ARDS trials enrolled, which argues for a pragmatic, standardised definition of hypoxemic respiratory failure to anchor future research. On what actually changes mortality, a systematic review in the Journal of Critical Care identified thirty-three randomised trials and twenty-three interventions with a statistically significant mortality difference in ARDS, then compared them against the European Society of Intensive Care Medicine and American Thoracic Society guidelines [10]. Roughly three quarters OF THOSE TRIALS showed a mortality reduction and about a quarter showed an increase. Guideline recommendations backed by mortality-changing trials include low tidal volume ventilation, prone positioning, venovenous ECMO, steroids, and avoiding high-frequency oscillatory ventilation — but eight interventions with a mortality benefit were not mentioned in the guidelines at all, and five further interventions were associated with increased mortality. That is a gap worth knowing about, both for research planning and for humility at the bedside. Steroids get a dedicated narrative review in Intensive Care Medicine [9], which argues that corticosteroids in ARDS are best understood as context-dependent therapy. The evidence supports use in moderate-to-severe ARDS, particularly COVID-19 ARDS and severe community-acquired pneumonia, but the effect is heterogeneous: hyperinflammatory and septic phenotypes appear more likely to benefit, while data in influenza-associated and non-septic ARDS remain limited or conflicting. Long-term harms, including metabolic complications and ICU-acquired weakness, are still poorly characterised. And on getting patients off the ventilator, a meta-analysis in Critical Care pooled twenty-three studies and two thousand patients with acute exacerbations of chronic obstructive pulmonary disease to test diaphragmatic ultrasound for predicting weaning outcome [8]. Diaphragmatic excursion performed best, with pooled sensitivity and specificity both around eighty-five percent and an area under the curve of zero point nine two; the diaphragmatic rapid shallow breathing index was highly specific but less sensitive. The authors are appropriately cautious — mostly small single-centre studies, post hoc thresholds, substantial heterogeneity, and geographically concentrated evidence — so this remains an adjunct to clinical judgement, not a substitute for it.

Two shorter papers round out the week. In the Journal of Critical Care, a meta-analysis of seven randomised trials and roughly nine thousand patients found that pantoprazole for stress ulcer prophylaxis cut clinically important gastrointestinal bleeding by about half, with no effect on overall mortality and no increase in ventilator-associated pneumonia or Clostridioides difficile infection [4]. That is reassurance on the infection question that has dogged acid suppression for years. And a single-centre self-controlled study, also in the Journal of Critical Care, compared blood cultures drawn from the central line hub during insertion against same-day cultures from other sites in the same patient [6]. Cultures taken from the hub at insertion had roughly two and a half times the odds of contamination, a finding robust across different matching windows. If you draw from a fresh central line, treat a positive result with real scepticism, and prefer another site when you can.

If you only have time for one paper this week, make it the vasopressin timing emulation in Intensive Care Medicine [1]. It speaks directly to a decision most of us make several times a week, and it shifts the default from watchful escalation toward adding vasopressin within hours of hitting a moderate norepinephrine dose — with the caveat that this is emulation, not randomisation.

Here are the key takeaways from this week in Critical Care. First, once norepinephrine has escalated in septic shock, earlier adjunctive vasopressin — within three hours — was associated with lower mortality and less dialysis, while pushing the entire bundle into the prehospital phase did not reduce mortality. Second, renal replacement therapy is not neutral: the patients most likely to benefit are those with genuinely failing kidneys, and once on continuous therapy, fluid balance, not ultrafiltration rate, is what tracks with survival. Third, acute hypoxemic respiratory failure affects half of ventilated ICU admissions and is routinely managed with ARDS-derived strategies, several of which lack guideline endorsement while others carry mortality risk. Fourth, corticosteroids in ARDS should be a phenotype- and aetiology-guided decision rather than a reflex. And fifth, pantoprazole reduces clinically important gastrointestinal bleeding without an infection penalty, while blood cultures drawn from a central line hub at insertion should be interpreted with real caution.

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 you have a paper you have been meaning to read, upload the PDF, or paste any link, at audioscholar dot C C. We will turn it into audio like this one, in any of thirty-one languages.

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