The Pulse - 2025 In Review

In this special edition, we review the

Top 10+ practice-changing studies for hospitalists in 2025

Plus our #1 Study for 2025 at the bottom

1. CAP-tivating Updates

   The American Thoracic Society has refreshed its playbook on community-acquired pneumonia (CAP), allowing for diagnostic chest POCUS instead of CXRs, and trimming antibiotic durations down to a tidy 3–4 days for stable patients. If your patient’s CAP comes with a positive viral panel, antibiotics get a nuanced treatment: none for healthy outpatients, but still on the menu for anyone sicker or carrying comorbidities. Steroids remain unrecommended in non-severe cases but are now cautiously welcomed in severe pneumonia—unless influenza crashes the party. The bottom line: think shorter antibiotic courses, be selective with steroids, and don’t treat these guidelines like commandments—much of the evidence still has a soft underbelly. Jones BE et al. Am J Respir Crit Care Med 2025 Jul.

   Study Type: co-sponsored guideline from ATS and IDSA

   Link

2. Etomidate on Trial: RSI Implications

   In this large, pragmatic randomized trial of over 2,300 critically ill adults undergoing emergent tracheal intubation, ketamine did not reduce 28-day in-hospital mortality compared with etomidate (28.1% vs 29.1%). The long-suspected mortality penalty of etomidate failed to materialize when tested head-to-head, at least in this setting. However, ketamine was associated with a higher rate of peri-intubation cardiovascular collapse, driven by hypotension or vasopressor escalation. Overall safety outcomes were otherwise similar, suggesting no clear winner—just different trade-offs. Bottom line: mortality is a wash, but hemodynamics may tilt the scale depending on your patient. Casey J et al. NEJM 2025 Dec.

   Study Type: US, multi-center RCT of 2,365 patients intubated in the ED or ICU, randomized to ketamine (1,176) vs etomidate (1,189) for anesthesia induction.

   Link

3. Beta Early Than On Time: The Case for Early β-Blocker Prophylaxis

   This single-center RCT suggests the benefits of nonselective β-blockade in cirrhosis may begin earlier than previously recognized. Among 104 patients with new-onset uncomplicated ascites and no or low-risk varices, early initiation of carvedilol significantly reduced ascites-related complications, the need for large-volume paracentesis, and all-cause mortality at one year compared with standard care alone. Improvements in hepatic venous pressure gradient supported a mechanistic rationale for these clinical gains. Though limited by open-label design and modest sample size, the data hint that carvedilol started at the onset of ascites could slow progression toward decompensation. If validated, this approach may prompt a rethinking of β-blocker timing in cirrhotic management algorithms. Khajuria R et al. Am J Gastroenterol 2025 Jul.

   Study Type: open-label RCT conducted at a single center in India of 104 patients with uncomplicated, new-onset ascites with no or low-risk esophageal varices, randomized to either carvedilol or none in addition to standard treatment.

   Link

4. RENOV(at)E Your Dose: Anticoagulation for High Risk Recurrence of VTE

   When it comes to preventing recurrent VTE, less might actually be more. The RENOVE trial found that a reduced-dose anticoagulation strategy was just as effective as full-dose treatment but with significantly less bleeding—making it the smarter long-term choice for high-risk patients. While the study didn’t quite meet strict noninferiority criteria, the tiny difference in VTE recurrence was overshadowed by a much lower risk of major bleeding. Bottom line? If you're keeping patients on anticoagulation long-term, cutting the dose could be a win-win. Couturaud F et al. Lancet 2025 Mar.

   Study Type: non-inferiority, multicentre, randomized, open-label, blinded endpoint trial of 1678 French patients with PE/VTE (after completing 6-24 months of full dose anticoagulation) considered to be high risk for VTE recurrence. These patients were randomized to full-dose vs reduced-dose (apixaban 2.5 mg BID or Rivaroxaban 10 mg QD) DOAC.

   Link

5. Code Blue or Code Through? New Rules for Calling It Quits

   New research offers a pragmatic, data-driven approach to deciding when to call it quits during in-hospital cardiac arrests. A Scandinavian study identified four key predictors of non-survival—unmonitored status, (no telemetry), unwitnessed arrest, initial rhythm asystole, and ≥10 minutes of resuscitation—wherein patients meeting all criteria had a mere 0.5% survival at 30 days. Applying this rule could allow clinicians to reasonably stop resuscitation efforts after 10 minutes rather than the average 20 minutes (average within the study), sparing futile efforts while ensuring <1% of potential survivors are missed. While promising, broader validation is needed before this rule becomes standard practice. Holmberg MJ et al. JAMA Int Med 2025 Jan.

   Study Type: prospective cohort study of 9863 Danish, 12781 Swedish and 1308 Norwegian patients who underwent resuscitation for in-hospital cardiac arrest

   Link

6. Echo-nomics: How POCUS Paid for Itself (and Then Some)

   In this stepped-wedge, randomized quality-improvement study of patients admitted with undifferentiated dyspnea, incorporating cardiopulmonary POCUS into standard hospitalist care led to meaningful efficiency gains. Compared with usual care, POCUS use reduced median length of stay from 12 to 8 days, saved over $750,000 in total costs, and altered management decisions in more than one-third of cases—all without increasing 30- or 90-day readmissions. Despite these outcomes, only 20% of studies were performed independently by hospitalists, highlighting a persistent skills and training gap. The data suggest that structured POCUS implementation could be a major lever for value-based inpatient care, if hospitals can bridge the adoption divide. Maganti K et al. JAMA Netw Open 2025 Sep.

   Study Type: quality improvement study conducted at a tertiary care, US hospital using a stepped-wedge cluster randomized approach over 6 months with 208 patients randomized to either control group of POCUS group.

   Link

7. POTCAST: This Episode Brought to You by Bananas

   Turns out the ICD’s best friend might just be a banana. In the POTCAST trial, a Danish RCT of 1200 ICD patients with potassium ≤4.3 mmol/L, nudging potassium up to a high-normal range (4.5–5.0 mmol/L) cut the risk of ventricular arrhythmias and related bad outcomes. The intervention arm got supplements, MRAs, and potassium-friendly diet advice, with close monitoring that paid off: fewer arrhythmias, ICD shocks, hospitalizations, and deaths (23% vs. 29%). Importantly, the higher-K strategy didn’t increase hypokalemia or hyperkalemia admissions. Bottom line: for patients at high risk of ventricular arrhythmia, living on the high side of “normal” may actually be optimal. Jøns C et al. N Engl J Med 2025 Aug.  

   Study Type: a Danish, multicenter, open-label, event-driven, randomized superiority trial of 1200 patients at high risk for ventricular arrhythmias (ICD placed) and baseline K of 4.3 or lower, randomized in a 1:1 ratio to treatment regimen or standard of care only. 

   Link

8. Keeping Pressure Off the Rocks: Calcium’s Smooth Move in RVR

   Turns out, giving a little calcium before diltiazem might help avoid hypotension. In this Turkish trial, patients with atrial fibrillation and tachycardia were preloaded with either 90 mg or 180 mg of IV calcium chloride before getting IV diltiazem. The 180 mg dose was the MVP, preventing hypotension without messing up rate control. No major side effects popped up, but the study only watched patients for 15 minutes — basically the clinical version of a speed date. While calcium chloride worked, calcium gluconate might be the friendlier option for your patient's veins. Az A et al. Am J Emerg Med 2025 Feb.

   Study Type: prospective, randomized, double-blind study, placebo controlled trial of 217 adults with Afib or Aflutter and a ventricular rate > 120 bpm

   Link

9. Copilot, Not Captain: Letting AI Draft the Discharge

   In a randomized comparison of 100 discharge summaries, GPT-4–generated notes were clearer and more concise than those written by hospitalists—but also more prone to omission errors. While physicians averaged 2 unique errors per summary, AI averaged 3, mostly due to missing clinical details. Importantly, reviewers judged the potential for patient harm from these errors to be similar between AI and human-authored notes. The takeaway? AI isn’t quite ready to write solo, but it makes a solid scribe. Using LLMs as a first draft—with human oversight—might help reduce documentation burden without compromising safety. Williams CYK et al. JAMA Intern Med 2025 May. 

   Study Type: cross-sectional study of 100 randomly selected inpatient hospital medicine encounters of 3 to 6 days’ duration with blinded, duplicate review among 22 human reviewers at a single, USA center

   Link

10. More Tests, More Beds, Same Outcomes: The Great Admission Mirage

    Turns out, some ED docs are a little too admission-happy, and the only thing it’s achieving is fuller hospitals and longer lengths of stay. A study of over 2 million VA patient visits found that physicians working in the same ED varied widely in how often they admitted patients (as much as 24% between 90th and 10th percentiles)—even when those patients had similar health statuses. The kicker? Those who were admitted by the more “intensive practice” docs weren’t any less likely to die within 30 days (or even a year). Instead, they were more likely to get sent home within 24 hours, suggesting that plenty of these admissions were just expensive, unnecessary sleepovers. The study hypothesizes that some docs may just have a lower tolerance for uncertainty, meaning more tests, more admissions, and more costs—without better outcomes. And before you blame medical training, that didn’t explain much either. While there might be some non-mortality-related benefits to admission (e.g. patient peace of mind), this habit of erring on the side of admission isn’t doing much beyond straining hospitals and jacking up costs. The bottom line? More admissions ≠ better survival. The more important question is how to decrease these unnecessary questions. Coussens S et al. JAMA Int Med 2024 Dec.

    Study Type: observational, cross-section study of 2,137,681 patients visiting VA hospital EDs for either chief complaint of chest pain, shortness of breath, or abdominal pain

    Link

Want to see which study we selected as the most important study for 2025?

Upgrade to Premium to get access not only to this study but also to all premium content in 2026. For less than a cup of coffee each month, the most practice-changing studies will be delivered to your email monthly in a easy-to-read and fast-read format, advancing your clinical practice and preserving your work-life balance.