This Week in Emergency Medicine — Jun 18, 2026

Welcome to This Week in Emergency Medicine. This week we're covering 10 notable papers spanning resuscitation controversies in sepsis and cardiac arrest, new data on optimizing systems of care for trauma and stroke, and the growing role of artificial intelligence in our daily workflow. Let's dive in.

We begin this week in the resuscitation bay, with major papers challenging long-held practices in critical care. First, in The New England Journal of Medicine, the ARISE FLUIDS trial tackled the perennial question of vasopressors versus fluids in early septic shock [9]. This randomized trial assigned over 960 adults in the emergency department with septic shock to either a fluid-restricted strategy with early vasopressor use, or a strategy with higher fluid volumes and later vasopressor initiation. The primary outcome was days alive and out of the hospital at day 90. The results showed no difference between the groups, with a median of 76 days in both arms. However, a key secondary finding was that pulmonary edema was significantly less common in the fluid-restricted, early vasopressor group, occurring in just 0.6% of patients compared to 5.0% in the higher-volume fluids group. This suggests that while a restrictive strategy may not change 90-day survival, it may reduce iatrogenic harm from fluid overload.

Staying with resuscitation, a large randomized trial in JAMA investigated another common but unproven intervention: sodium bicarbonate for in-hospital cardiac arrest [10]. This double-blind, placebo-controlled trial in Denmark randomized nearly 800 adults to receive either sodium bicarbonate or placebo after at least one dose of epinephrine. The primary outcome was sustained return of spontaneous circulation, or ROSC. The study found no significant difference, with ROSC occurring in 39% of the bicarbonate group versus 37% of the placebo group. There were also no significant differences in the key secondary outcomes of survival at 30 days or survival with a favorable neurologic outcome, although there were non-significant trends favoring bicarbonate. The authors conclude that these findings do not support the routine administration of sodium bicarbonate during in-hospital cardiac arrest.

Finally in our critical care section, another paper from The New England Journal of Medicine provides a clear answer on antibiotic choice for methicillin-susceptible bacteremia [1]. In an international trial, nearly 1300 patients were randomized to receive either cefazolin or an antistaphylococcal penicillin like flucloxacillin. The study found that cefazolin was noninferior with respect to 90-day mortality, which was 15% in the cefazolin group and 17% in the penicillin group. Importantly, cefazolin was associated with a significantly lower incidence of acute kidney injury, which occurred in about 14% of patients receiving cefazolin compared to nearly 20% of those receiving an antistaphylococcal penicillin. This provides strong evidence to prefer cefazolin for these infections.

Moving from the resuscitation bay to the bigger picture of systems of care, two papers this week highlight major opportunities for improvement in prehospital transport. First, a study in Prehospital Emergency Care examined pediatric trauma transport in the United States using a national EMS database [8]. Among nearly 100,000 injured children who met American College of Surgeons trauma center criteria, over half had geographic access to a designated pediatric trauma center, or PTC. However, of those with access, only 40% were actually transported to one. This finding suggests a substantial gap in getting critically injured children to the most appropriate level of care. For those interested in the 'how' of system improvement, a realist synthesis in The Lancet Global Health offers a framework for improving pediatric emergency care in low-resource settings, focusing on the interdependent mechanisms of legitimacy, capability, and connectedness [6].

This challenge of optimal routing isn't unique to pediatrics. Another study in Prehospital Emergency Care used a simulation model to look at stroke care [2]. Researchers compared three routing strategies: standard AHA consensus recommendations, a personalized model, and an enhanced model that incorporated real hospital-specific performance data on door-to-needle and door-to-puncture times. The results were striking: when real hospital performance data was included, the recommended destination hospital changed in over 63% of the 400,000 simulated patient scenarios compared to standard recommendations. This suggests that incorporating actual hospital metrics into prehospital decision-making could dramatically alter and potentially optimize patient routing for time-sensitive conditions like stroke.

Next, we turn to the growing influence of artificial intelligence in our daily practice. Two papers in Annals of Emergency Medicine explore this from different angles. First, looking at documentation burden, a retrospective study compared ambient AI scribes, human scribes, and no scribes across nearly 200,000 ED encounters [7]. Both types of scribes were associated with reduced physician documentation time. Compared to encounters with no scribe, human scribes reduced documentation time by a median of 3.3 minutes per note, while ambient AI scribes reduced it by 1.6 minutes. Interestingly, there was no difference in clinical productivity, measured as work RVUs per hour, among the three groups.

Beyond documentation, can AI help with clinical diagnosis? Another study in Annals of Emergency Medicine compared a large language model to a traditional, rule-based computable phenotype for identifying opioid use disorder from ED notes [4]. Using physician chart review as the gold standard, the large language model demonstrated slightly lower sensitivity than the rule-based system, at 81% versus 84%. However, the AI model achieved significantly higher specificity, at nearly 99.6%, and a much higher positive predictive value. This suggests that large language models may be a powerful tool for reducing false-positive alerts in clinical workflows.

Finally, a quick but important update on infectious disease from JAMA Internal Medicine provides some of the first effectiveness data for the 2024-2025 COVID-19 vaccine in United States adults [3]. In a large, test-negative case-control study, vaccination was associated with an estimated effectiveness of 26% against COVID-19-associated emergency department or urgent care encounters. Effectiveness was higher for more severe outcomes, estimated at 35% against COVID-19-associated hospitalization and 41% against critical illness. While these numbers are modest and waned over time, they highlight the continued benefit of vaccination in reducing the likelihood of medically attended COVID-19.

If you only have time for one paper this week, make it the trial on cefazolin for MSSA bacteremia from The New England Journal of Medicine [1]. This study provides strong, actionable evidence that cefazolin is not only non-inferior for mortality but is also safer for the kidneys, solidifying it as the preferred agent for a common and serious condition.

Here are the key takeaways from this week in Emergency Medicine. First, in early septic shock, a fluid-restrictive strategy with early vasopressors does not change 90-day mortality or hospital-free days compared to a more liberal fluid strategy, but it does significantly reduce the risk of pulmonary edema. Second, routine use of sodium bicarbonate in in-hospital cardiac arrest is not supported by evidence and does not improve the rate of sustained ROSC. Third, for methicillin-susceptible bacteremia, cefazolin should be the preferred agent over antistaphylococcal penicillins, as it offers non-inferior mortality with a lower risk of acute kidney injury. Fourth, our prehospital systems for routing both pediatric trauma and adult stroke patients are suboptimal, and incorporating real-world hospital performance data could significantly improve destination choices. And finally, AI tools are demonstrating real utility in the ED, both by reducing documentation time and by accurately identifying conditions like opioid use disorder with high specificity.

That's your roundup for This Week in Emergency Medicine. 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 for audio briefings on your own clinical questions and papers, visit audioscholar dot C C.

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