Zworth Reading
EMMax’s EM Weekly Update
Highlight of the Week
JAMA · Multicenter adaptive-allocation RCT (LOE 2)
ICECAP randomized 1158 comatose survivors of out-of-hospital cardiac arrest to therapeutic hypothermia at 33°C for durations ranging from 6 to 72 hours, using a Bayesian adaptive design. For both shockable and nonshockable rhythms, 6 hours of cooling was as good as anything longer. There was no difference in 90-day neurological outcome, mortality, or any secondary outcome across any duration.
The framing here is what matters. This adds to a growing body of literature suggesting that normothermia is adequate. Cooling does not equal better patient-oriented outcomes, and now we have good evidence that cooling for longer does not either. Worth remembering that the 2025 AHA guidelines already say essentially this — maintain a temperature between 32 and 37.5°C. My current practice in the ED is strict normothermia: avoid fever above 37.5°C, but do not aggressively cool below 36°C.
Bottom line: After cardiac arrest, protect against fever, and put your effort into the rest of post-arrest care.
FOAM Radar
EMCrit 431 – OB-EMCrit – Life-Threatening Post-Partum Hemorrhage (PPH) with Cho EspinozaKnowledge Translation
EMCrit · LOE 5 — expert knowledge translation
Scott Weingart and Dr. Cho Espinoza work through catastrophic postpartum hemorrhage for the situation where you are the only resuscitationist in the room. The high-yield content: uterotonic choice, dosing, and routes; bimanual uterine compression and aortic compression to temporize; TXA and early coagulopathy management; and when to activate massive transfusion. It is aimed squarely at the ED or prehospital physician trying to buy time before OB or IR can take over.
A few things I took away. This is a really useful reference to bookmark for medication dosing, as some of the uterotonics are hard to remember when you actually need them. The author’s advice is to give all of the medications that are not contraindicated. TXA has a signal for decreased mortality, so give 1 g upfront. A reminder that the EM doc should be sweeping the uterus for retained tissue. And remember to think about fibrinogen and calcium while you are transfusing.
Bottom line: Definitely worth reading the whole post.
ToxCard: Diphenhydramine ToxicityKnowledge translation / clinical review
emDocs · LOE 5 — expert summary of case reports and pharmacology
A concise review of diphenhydramine toxicity covering anticholinergic toxidrome, sodium channel blockade causing QRS widening, and seizure risk. Emphasizes that massive ingestions can cause wide-complex dysrhythmias mimicking TCA overdose and may require sodium bicarbonate.
Bottom line: Useful refresher for a common OTC ingestion. The key teaching point: diphenhydramine in large doses is not a benign antihistamine. It is a sodium channel blocker and anticholinergic agent that can be deadly. Treat wide QRS with bicarb, manage agitation and seizures aggressively.
A Slow RetreatCommentary/Opinion
EMCrit · LOE 2 evidence discussed (TTM trials), but post is interpretive commentary
Good compliment to the highlight of the week. Rory Spiegel traces the history of targeted temperature management after cardiac arrest, from the original 2002 trials through TTM, TTM2, and HYPERION. He argues the practice was built on a shaky foundation that the field has been slow to let go of.
Bottom line: I like reading Dr. Spiegel. The discussion of methodology is a bit dense, but I always enjoy a narrative of how we arrive at certain practices in medicine and the trials that got us there.
Adjacent Specialties
Corticosteroids in ARDS: old controversies, new insights, and future directionsLOE 5 (Narrative review/expert opinion)
Intensive Care Medicine · Critical Care
Review of corticosteroid use across ARDS etiologies and phenotypes.
Steroids appear beneficial in moderate-severe ARDS, particularly COVID-19 ARDS and severe CAP. Evidence in influenza ARDS remains conflicting. Long-term harms (ICU-acquired weakness, metabolic complications) are real but incompletely characterized.
This is a narrative review, not a systematic review or meta-analysis, so it reflects the authors' interpretation of the literature. The phenotype-based approach is intellectually appealing but not yet clinically operationalized. We can't easily identify hyperinflammatory vs hypoinflammatory patients at the bedside in real-time. The COVID-19 steroid data is solid; extrapolation to other ARDS etiologies is less certain.
Bottom line: Steroids for COVID-19 ARDS and severe CAP: yes. For other ARDS: probably reasonable in moderate-severe disease, but the evidence is weaker.
Methodology Flag
Chest
Flagged not because the study is bad but because "target trial emulation" is showing up more and more, and it is worth knowing what this design can and cannot tell you.
- What it is: the authors write down the protocol of the randomized trial they wish they could run (who is eligible, what the two strategies are, when the clock starts, what the outcome is) and then find patients in existing records who match it. Here the imagined trial is: adults in the ICU whose glucose hits 10–13.9 mmol/L, randomized to insulin within 6 hours or not.
- Why not just run the trial: most guidelines already tell you to treat glucose at that level, so deliberately withholding insulin to see what happens is hard to justify. The emulation asks the question using care that already happened.
- What the design fixes: writing the protocol first forces a single "time zero" and sorts everyone into a group at that same moment. Weaker observational studies quietly credit the treated group with survival time that elapsed before treatment started. This study used a prespecified 6-hour landmark to prevent exactly that, across 5,755 patients.
- What it cannot fix: it is still not randomization. The patients who got early insulin were actually sicker (SOFA 8 vs 6, more diabetes, more comorbidity), and the raw mortality rates showed no difference between groups at all. The authors say plainly that early insulin may be standing in for something they could not measure, like staffing levels or how closely a unit follows its protocols.
What it does contribute: A well-run emulation is the best available answer when the randomized trial is never going to happen. Read it as a strong hypothesis and a good argument for running the real trial, not as an answer.
Bottom line: Do not change your insulin practice on this. Do get comfortable with the design, because you will be reading more of them. The questions are always the same: when did the clock start, who ended up in each group, and how much work did the adjustment have to do to produce the result?