This Week in Emergency Medicine — Sep 8, 2026

Welcome to This Week in Emergency Medicine. This week we're covering 10 notable papers spanning time-critical prehospital and vascular care, resuscitation physiology, and the ongoing search for risk scores that actually change what we do at the bedside. Let's dive in.

We'll start with the drugs and blood we give before the patient ever reaches definitive care. In the European Journal of Emergency Medicine, Wu and colleagues pooled seven studies — two randomised trials and five observational cohorts, almost eleven thousand patients with ST-elevation myocardial infarction heading for primary angioplasty [3]. Giving unfractionated heparin in the ambulance, rather than waiting for arrival at the catheterisation-capable hospital, raised the odds of an open infarct-related artery before the procedure by roughly a half, and there was no signal of increased major bleeding. The mortality picture is more nuanced and worth stating plainly: thirty-day all-cause mortality was not significantly reduced. A composite of thirty-day or in-hospital mortality did reach significance, cutting the odds by about a third, but with only two randomised trials in the pool this remains association rather than proof. The practical read is that prehospital heparin looks safe and improves preprocedural patency, and if your system is considering it, the bleeding fears appear unfounded — but don't promise a mortality benefit that the data haven't delivered. Alongside that, Prehospital Emergency Care published work by Warren and colleagues asking a very concrete question: which trauma patients actually need blood in the ambulance [2]. Applying a proposed set of transfusion criteria retrospectively to more than twenty-five thousand trauma patients in Los Angeles County, just over nine hundred patients met at least one criterion, and about half of those received blood within four hours of arrival. The best-performing trigger was the combination of a systolic pressure under ninety with a heart rate of a hundred and ten or higher, with a positive predictive value of around fifty-six percent; the other criteria, including systolic pressure under seventy and paramedic-witnessed traumatic arrest, clustered in the low fifties. So a coin flip, essentially — which is arguably acceptable for a low-harm intervention in haemorrhagic shock, but it tells us these criteria are inclusive rather than precise, and that pairing hypotension with tachycardia beats hypotension alone.

Staying with time-critical vascular care, Internal and Emergency Medicine published a meta-analysis from Zhang and colleagues on tenecteplase versus alteplase specifically in patients aged eighty and over — a group that makes up a large share of our stroke presentations and is systematically under-represented in trial subgroups [1]. Five randomised trials, just over thirteen hundred very elderly patients. Tenecteplase was associated with better functional outcomes at ninety days, and symptomatic intracranial haemorrhage, parenchymal haematoma and mortality were all statistically indistinguishable between the two drugs. But the authors are refreshingly honest about fragility: the benefit on excellent functional outcome disappeared entirely when a single trial was removed from the analysis, heterogeneity was high, and every contributing trial was designed to test non-inferiority, not superiority. The conclusion for practice is reassurance rather than revolution — tenecteplase appears at least as safe as alteplase in the very old, which matters when you're weighing thrombolysis in an eighty-eight-year-old, but nobody should be claiming superiority from these data.

Resuscitation brings two papers that both challenge assumptions about what we see during arrest. Howard and colleagues examined cardiopulmonary resuscitation-induced consciousness — patients who move, groan, open their eyes or actively resist compressions while still pulseless — across more than forty-three thousand out-of-hospital arrests in Victoria over sixteen years [6]. It occurred in about three percent of arrests, and its incidence has risen steadily year on year, likely reflecting better bystander response, mechanical compression devices and higher-quality cardiopulmonary resuscitation. Crucially, after adjustment, these patients had nearly three times the odds of surviving to hospital discharge, and that held for both the interfering and non-interfering phenotypes. So when the patient starts fighting you mid-arrest, that is a marker of favourable physiology, not a nuisance — the response should be sedation and analgesia while continuing high-quality resuscitation, never a reason to slow down. The companion paper, from Wong and colleagues using the Get With The Guidelines-Resuscitation registry, asks whether we're right to lump pulseless electrical activity and asystole together as simply non-shockable in children [7]. Across more than nine thousand paediatric in-hospital arrests, asystole carried about thirty percent lower odds of return of spontaneous circulation compared with pulseless electrical activity. But the survival-to-discharge difference largely evaporated once neonatal intensive care events were excluded — survival was around thirty-five percent in both rhythm groups — and age-stratified analysis showed the overall difference was driven by neonates. For the paediatric resuscitation in front of you outside the neonatal unit, initial rhythm alone should not shape your prognostic expectations or your willingness to persist.

Then there's a cluster of papers on risk stratification, and the results are decidedly mixed. In the American Journal of Emergency Medicine, Damar and Yaman retrospectively studied 472 patients aged sixty-five and over on direct oral anticoagulants who presented after blunt head trauma with a normal initial head CT [4]. Delayed intracranial haemorrhage occurred in about five percent of patients. A modified HAS-BLED score discriminated well, and a threshold of three or more captured roughly nine in ten of the bleeds with a high negative predictive value, suggesting a route to selective rather than routine repeat imaging. It's single-centre and retrospective, so this isn't ready to replace your local protocol — but the question it addresses, how long to observe the anticoagulated elderly head injury, is one we answer badly every shift. Contrast that with the gastrointestinal bleeding work, also in Internal and Emergency Medicine, where Yilmaz and colleagues compared twelve pre-interventional scores head-to-head in over two thousand patients [5]. For thirty-day mortality in upper gastrointestinal bleeding, AIMS65 came out on top and NOBLADS performed strongly regardless of bleeding location, hinting it may work as a site-independent mortality predictor. But for predicting intensive care admission, every single score managed only moderate discrimination — a genuine and persistent gap. And in the American Journal of Emergency Medicine, Yorgun and Taş prospectively derived a bedside ankle injury-burden score in two hundred patients meeting Ottawa Ankle Rules criteria [10]. Higher scores tracked with fracture, with greater CT use and orthopaedic consultation, and — more interestingly — with poorer patient-reported function at six weeks. Only eleven percent of the cohort had fractures, so the appeal here is quantifying overall injury burden rather than just answering the yes-or-no fracture question, but this needs multicentre external validation before it goes anywhere near your department.

Two shorter items round out the week. In Internal and Emergency Medicine, Masip and colleagues analysed nearly nineteen thousand acute heart failure presentations from a nationwide registry, using the oxygen saturation to inspired-oxygen ratio to correct for prehospital oxygen [8]. Impaired oxygenation was present in close to two thirds of patients, and frank hypoxaemia in about forty percent of those admitted — across every heart failure phenotype, not just pulmonary oedema. Mortality risk rose progressively as oxygenation worsened, but flattened at the most severe end, implying that outcome is driven by more than respiratory failure alone. And in the Emergency Medicine Journal, Chan and Saravanan reviewed four prospective studies of point-of-care ultrasound for predicting difficult intubation, looking at hyomental distance, tongue thickness and anterior neck soft tissue [9]. Several parameters showed fair to excellent accuracy against Cormack-Lehane grading, but thresholds and techniques varied so widely that the honest bottom line is that ultrasound should supplement, not replace, conventional airway assessment.

If you only have time for one paper this week, make it the cardiopulmonary resuscitation-induced consciousness study in Resuscitation [6]. It reframes a phenomenon most of us find disconcerting into a positive prognostic sign, and it changes how you should manage the patient on the floor tonight.

Here are the key takeaways from this week in Emergency Medicine. First, a patient who moves or resists during compressions has favourable physiology — sedate, and keep going. Second, in children outside the neonatal population, asystole versus pulseless electrical activity should not change your prognostic outlook. Third, prehospital heparin in ST-elevation myocardial infarction improves preprocedural vessel patency without excess bleeding, but the mortality case is unproven. Fourth, tenecteplase looks as safe as alteplase in patients over eighty, though the apparent functional benefit is statistically fragile. And fifth, risk scores remain uneven: they perform reasonably for delayed bleeding after anticoagulated head injury and for mortality in upper gastrointestinal bleeding, but nothing yet reliably predicts intensive care need.

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 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.