The Best Studies of 2025 for Hospital Medicine
What changed practice, what confirmed it, and what absolutely should not
Adam and I delighted to publish this roundup of the year’s big studies in Hospital Medicine from UCSF academic internist Dr. Anil Makam. Anil also has a thoughtful feed on X. JMM
Every year brings a flood of studies that seek to “change practice.” Most don’t. A few genuinely should make us change what we’re doing or at least confirm what we think we should be doing.
Below is a tour of the most important studies for hospital medicine this past year.
“Yippee ki-yay”: Practice-Changing Studies
Olezarsen is a new drug to treat severe hypertriglyceridemia that prevents pancreatitis
A novel triglyceride-lowering therapy dramatically reduced acute pancreatitis in patients with severe hypertriglyceridemia, with large and immediate benefit in reduction of acute pancreatitis (RR ≈ 0.15; NNT=20). Absolute benefit was greatest in those at highest baseline risk (NNT ≈ 4 at 1 year among patients with TG > 880 mg/dL and prior pancreatitis)
Teaching moment: Relative risk reduction was consistent across doses and risk strata. Despite no true heterogeneity of treatment effect, absolute benefits differ by baseline risk. This is how “personalized medicine” actually works.
The HiPRO trial: Extend anticoagulation for provoked VTE with ongoing risk
In patients with provoked VTE but persistent risk factors, extended low-dose apixaban (2.5 mg BID) markedly reduced recurrent symptomatic VTE (HR ≈ 0.13; NNT ≈ 10 at 1 year) over placebo, with curves that separated early and kept diverging.
Teaching moment: When KM curves continue to diverge, ARR, and consequently the NNT, will continue to get better with over longer time horizons chosen.
The ALONE AF trial: Anticoagulation can be safely stopped after successful AF ablation
Among patients without AF recurrence for ≥ 1 year after ablation, discontinuing DOACs led to slightly better overall outcomes, driven by fewer major bleeds. About 10% experienced AF recurrence and restarted anticoagulation.
Teaching moment: Always inspect which component drives a composite endpoint, and assess the event rate. Alone AF found extremely low annual stroke risks in both groups, suggesting the CHADSVASC scoring system may not apply to patients with successful AF ablation.
The AQUATIC Trial: Stop aspirin in stable CAD when patients are anticoagulated.
In high-risk stable CAD patients already on anticoagulation (mostly DOACs, nearly all post-PCI), adding aspirin was harmful, including higher mortality.
Teaching moment: A subgroup showed difference by sex but unlikely to be true. Be skeptical of subgroup claims unless prespecified, biologically plausible, precise, and externally consistent. Most aren’t.
The UK SWHSI-2 trial: Wound vacs do not improve healing for surgical wounds
Despite widespread adoption, negative-pressure wound therapy did not speed healing in a high-risk population with surgical wounds.
Teaching moment: Unblinded trials usually favor the intervention. The fact that this study showed no benefit, and that masked photographic adjudication confirmed the result, strengthens confidence in the null finding.
The PARTHENOPE RCT: Personalize DAPT duration after PCI using the DAPT Score
Risk-stratified DAPT duration outperformed guideline-mandated defaults. Higher-risk patients benefited from longer DAPT with fewer MIs and revascularizations and no excess bleeding or death.
Teaching moment: Applying EBM principles to tailor therapy beats one-size-fits-all care. (Editor’s Note: post-coronary stent antiplatelet decisions are about as complex as can be. Variables include patients conditions, coronary anatomy, stent size and risk of bleeding.)
The Post-MI Beta-blocker Meta-analysis for patients with normal LVEF : Beta-blockers did not improve outcomes after MI with preserved EF
A meta-analysis of five contemporary trials, including two published in 2025. found no benefit of beta-blockers after MI in patients with preserved EF.
Teaching moment: Evidence can expire with advances in care. See John Mandrola post.
Use an LLM for a second opinion for clinical reasoning
In the first of its kind, an LLM nearly matched the clinical reasoning prowess of a master clinician for an NEJM case.
Teaching moment: Rarely can you witness transcendence. Read the case, watch both videos, and give an LLM a try for a second opinion!
Doctors can strive for diagnostic excellence by maximizing both sensitivity and specificity
Given that physicians vary by aptitude, education, experience, and intentional practice, summary estimates cannot convey meaningful differences across doctors or the maximal level of performance that can be achieved. More skilled doctors can excel in both detecting and excluding disease simultaneously without the typical tradeoff for diagnostic tests.
“Talk about déjà vu”: What’s old is new again studies
DIGIT-HF Study: Digoxin is back in the mix for GDMT for heart failure
In a double-blind multicenter trial, digitoxin, a safer digoxin analog unavailable in the US, reduced both deaths and hospitalizations in patients with advanced HF with reduced EF. A Dutch trial of digoxin is nearly completion with results expected soon.
Teaching Moment: Digitalis as a therapeutic agent has been unfairly maligned as dangerous because of biased observational studies (sicker patients receive digoxin in nonrandomized studies). In RCTs, digitalis actually fairs well, and can be a useful adjunct in patients.
“This is the way”: Practice-Affirming Studies
The CloCeBa Trial: Cefazolin should be first line for MSSA bacteremia
Following the topline SNAP trial results, a new multicenter RCT found cefazolin noninferior to cloxacillin for MSSA bacteremia, with fewer adverse events.
Teaching moment: Interpret non-inferiority studies based on effect size and confidence intervals with a Bayesian lens, incorporating prior evidence like SNAP.
Oral over IV antibiotics for serious and complex infections
The COPAT and POvIV studies further support oral antibiotics for complex and serious infections once patients are stable.
Teaching moment: Evidence diffusion is slow. Clinicians don’t have to be. Follow @EvidenceRounds to stay on top.
The GAP study: Say NO to Vitamin G(abapentin) for post-operative pain
In spite of US guideline recs, gabapentin provided no meaningful benefit for pain, opioid use, length of stay , or quality of life after major surgery, confirming a prior meta-analysis.
Teaching moment: Be wary of strong guideline recommendations built on weaker evidence.
Steroids remain standard of care for severe CAP
Despite RE-MAP CAP showing no benefit, a Bayesian meta-analysis confirms a near 100% probability of mortality benefit for severe CAP, as also shown in SONIA in low-resource community hospitals.
Teaching moment: Be wary of a single trial that refutes a body of evidence.
“Stop trying to make fetch happen”: Should NOT Change Practice
The TAP IT trial: Don’t routinely tap pleural effusions in heart failure, just diurese
Routine thoracentesis did not improve outcomes in HF exacerbations.
The POTCAST trial: Don’t replete K to 4.5-5 for patients at high-risk for arrhythmia
A trial of outpatients with ICDs found high normal K target to be better but the benefit hinged more on MRA use than repletion since many more were newly started and this subgroup suggested they drove the benefit. Outcomes were similar whether or not patients met the K target, ECG intervals didn’t differ, and the small reduction in arrhythmia hospitalization was delayed despite immediate K separation between groups.
“We’re gonna be like three little Fonzies”: Cool studies, but not prime time
GLP-1 agonists may also treat alcohol use disorder!
In a small but compelling trial, semaglutide reduced alcohol consumption in a controlled setting where participants were presented with their beverage of choice.
Benzodiazepines for refractory hyperactive delirium
In a small double-blind RCT, lorazepam, either as monotherapy or in combination with haloperidol, outperformed escalating haloperidol for refractory hyperactive delirium.
Summary
“Yippee ki-yay”
Olezarsen is a new drug for hyperTG to prevent pancreatitis
Continue apixaban indef for provoked PE if ongoing VTE risk
Safe to stop anticoag after successful AF ablation (≥1 year)
Stop ASA for stable CAD (even high risk)
For surgical wounds, NPWT isn’t needed to facilitate healing
Tailor DAPT duration after PCI using low vs high DAPT score
If normal EF after AMI, ditch the beta blocker
Consult an LLM for 2nd opinion for clinical reasoning
Doctors are unique diagnosticians that can max both sens & spec
“Talk about déjà vu”
Digoxin (by way of digitoxin) is back as GDMT for HFrEF
“This is the way”
Cefazolin should be first line for MSSA bacteremia
Say NO to Vitamin G(abapentin) for post-operative pain control
Oral is the new IV for complicated & serious infections
Use steroids (hydrocortisone) for severe CAP
“Stop trying to make fetch happen”
No need to tap that pleural effusion in HF exac, just diurese
Normal K goals suffice even if at high risk for arrythmia
“We’re gonna be like three little Fonzies”
GLPs may be a very effective therapy for AUD
Benzos may be preferred in refractory agitated delirium
 | A guest post by
|
Love how concise and rapid fire you summarized those studies! One small suggestion because not all of your readers are cardiologists. (Pharmacist myself) You’ve included a very large number of abbreviations. For the lesser known ones next time add a definition. Your post will appeal to a broader audience. Still, it was well well done!
BOOM. There it is. Awesome