Two recent trials challenged the historical preference for anti-staphylococcal penicillin. For MSSA bacteremia, cefazolin should be first-line, and when the isolate is penicillin susceptible, benzylpenicillin is the preferred narrow-spectrum option

In a proof-of-concept AI second opinion study of 76 ED visits, an LLM exceeded 2 independent MD reviewers on diagnostic performance at 3 touchpoints (ER triage, ER physician & admit) Whether it outperforms physicians in real-world practice is yet to be determined

In a double-blind RCT of 108 obese AUD pts motivated for sobriety, semaglutide reduced heavy drinking days by 13% (absolute) vs placebo Reduction in drinking strongly correlated w/ weight los Suggests shared mechanism or functional unblinding, which might exaggerate benefits

Do patients on a low dose of levothyroxine truly have hypothyroidism? In a multicenter longitudinal descriptive study, 64% on a levothyroxine dose of 50 µg/d or lower were able to successfully discontinue treatment, suggesting that very low doses more often reflect overtreatment

Does hydration reduce kidney stones? A behavioral intervention ↑ fluid intake ~300 mL (w/ slight ↑ urinary symptoms) but did NOT ↓ recurrence or improve outcomes Intake difference may have been too small, as a prior small trial showed benefit w larger separation (~2.5 vs 1L)

Minocycline for acute ischemic stroke? In a multicenter RCT, oral minocycline within 72h of an ischemic stroke (NIHSS 4–25) improved functioning at 90 days (mRS 0–1: 52.6% vs 47.4%; ~5% absolute; RR 1.11), supporting a modest neuroprotective benefit

In a head-to-head trial for VTE treatment, apixaban was safer than rivaroxaban with similar effecacy at 90 days, with ~4% less clinically relevant bleeding (RR 0.5) Since the safety advantage was confined to the loading-dose phase, its not relevant for atrial fibrillation

The SOHO trial confirms the use of high-flow nasal cannula for acute hypoxic respiratory failure with less intubation (NNT ~17; benefit by day 1, plateaus by day 4) and less dyspnea (NNT ~7; ~5-point improvement on a 100-point scale)

Should longer vanc courses be prescribed for first episodes of C diff? In a Bayesian RCT, a 4-week pulse+taper reduced recurrence vs standard 2-week course at 1 month (RRR 43%, 99% probability of superiority) and at 2 mos (RRR 16%, 74% prob) But 3-month recurrence was similar

Does correcting chronic hyponatremia improve outcomes? In an RCT of hospitalized adults with Na <130, a targeted correction algorithm did not improve 30-day death/rehospitalization (20.5% vs 21.8%) or neurocognitive outcomes...

Treating addiction is hard In a pragmatic RCT of patients with nonfatal opioid overdose, ED peer navigators did not reduce 1-year outcomes This builds on a prior step-wedge cluster RCT: addiction consult increased med use, but 6-month retention was low & similar (2.4% vs 3.2%)

Mild predialysis hyperkalemia may not warrant treatment Among ESRD patients with a pre-dialysis K≥5.5, Lokelma improved long-term potassium control but did NOT reduce arrhythmia-related cardiovascular events (8.8% vs 8.9%) The trial was stopped early for low event rates

In a nationally representative sample of adults 41–76, almost all had abnormal shoulder MRI findings, including ~51% of asymptomatic individuals with partial-thickness rotator cuff tears Like spine, knee, & hip imaging, this high prevalence suggests these are normal age-related

Long-acting monthly injectable cabotegravir/rilpivirine was superior to oral ART in the LATITUDE trial among people with nonadherence, with an 18% absolute reduction in regimen failure A prior small economic analysis from a safety-net clinic suggests it is cost-effective

Up to 40% of hospitalized adults discharge to post-acute care at home or in a facility This review demystifies post-acute care with a practical framework to match the right setting to a patient’s needs and plan for a smoother transition

In an open-label RCT of HIV-related sepsis in endemic TB regions, immediate anti-TB therapy reduced 28-day mortality in the subgroup with later microbiologically confirmed tuberculosis This suggests, as with sepsis, that time to initiate anti-TB therapy might also matter

Consider single-dose amphotericin for cryptococcal meningitis In a US safety-net hospital, the AMBITION regimen established in LMICs (single-dose liposomal ampho+flucy/fluc) achieved same outcomes with fewer severe adverse events in a pre-post analysis vs standard 14-day course

Antibiotics hold up for more than half of patients with uncomplicated appendicitis, but decision to avoid upfront surgery is in the eye of the beholder At 10-year follow-up, 44% eventually underwent an appendectomy Greatest risk was in the first 12-18 months

A 2023 narrative review challenges the centrality of dual-process models, arguing that diagnostic expertise matters more Improving clinical reasoning may depend less on debiasing & more on richer problem representations, illness scripts & better access to the right knowledge

For older adults w/ newly diagnosed afib during hospitalization the 1-yr stroke risk w/o anticoag was low: 0.7% for CHA2DS2VASc 1-4 & 1.8% for 5-8 Given consensus of ≥2%/year, anticoag may be less compelling in hospital-detected AF but should still be individualized by risk

Best of Hospital Medicine in 2025 In article format...

Age-adjusted d-dimer cutoffs can safely exclude DVT A multicenter prospective cohort found that 7% of imaging could be avoided by using this strategy over the standard d-dimer cutoffs

Prebiopsy steroids don’t reduce diagnostic yield in lymphoma (DLBCL) A retrospective study confirmed biopsy type mattered most for yield (excisional > core >>> FNA)

Shunting in idiopathic NPH helps gait but with notable risks In a placebo-controlled RCT of patients who improved with CSF drainage, shunting improved gait & reduced falls at 3 mos (24% vs 46%) BUT more positional headaches (NNH 3), subdural bleeds (NNH 10) & hygromas (NNH 17)

In a small but compelling proof-of-concept RCT in AUD, semaglutide reduced alcohol consumed vs placebo in a controlled setting where participants were presented with their preferred beverage Also reduced several measures of self-reported consumption Confirmatory trials needed

In a sham-controlled trial with an active control, compression garments did not significantly reduce recurrent vasovagal syncope (orthostasis excluded) But a meaningful benefit wasn't excluded (underpowered?) & fewer syncopal episodes occurred when participants actively wore it

In a pragmatic trial, RSV vaccine reduced RSV hospitalizations for respiratory illness in older adults by 83% But was rare (NNT ~1800) with similar total hospitalizations & death (numerically favored placebo) DANFLU-2 showed no benefit of high-dose vs standard influenza vax

Aspirin’s decline continues In high-risk stable CAD patients already on anticoagulation (all post-PCI, 70% with prior MI; mostly on DOACs for Afib), adding aspirin increased death, CV events, and major bleeding in a multicenter trial, prompting early termination

For provoked VTE, standard therapy is 3 months, but what about patients with ongoing risk (ie autoimmune disease, obesity)? In a single-center RCT, extended low-dose apixaban (2.5mg BID) reduced recurrent VTE vs placebo at 1 year (HR 0.13, NNT=10), with KM curves still diverging

Is GDMT for heart failure coming full circle? In a double-blind multicenter trial of advanced HFrEF, digitoxin—a safer digoxin analog not available in the US—reduced both mortality and hospitalizations A Dutch trial of digoxin is underway, with results expected soon