Stopping antibiotics once endocarditis stabilises: fewer antibiotic days, more relapse

Design

International, open-label randomised trial with non-inferiority safety testing

Population

508 stable adults with left-sided endocarditis due to S aureus, E faecalis or streptococci

Primary outcome

Days alive without antibiotics at 6 months; death, unplanned cardiac surgery or embolism

Effect

183 vs 169 antibiotic-free days, difference 13 (95% CI 12–13); safety 8.2% vs 10.7%, difference −2.4 points (−7.7 to 2.7); relapse 5.1% vs 1.6%

POET II randomised 508 stable adults with left-sided infective endocarditis caused by Staphylococcus aureus, Enterococcus faecalis or streptococci. All had received at least 2 to 4 weeks of antibiotics and met criteria for clinical stabilisation. One group stopped antibiotics at that point; the other completed the standard 4 to 6 weeks. The trial was published at the end of August 2026.

Response-tailored therapy gave a median 183 antibiotic-free days out of six months, against 169, a difference of 13 days. The safety composite of death, unplanned cardiac surgery or symptomatic embolism was 8.2% against 10.7%, which met non-inferiority. But relapse of bacteraemia or endocarditis was 5.1% against 1.6%.

This is a genuine trade-off, not a clean win. The six-week course rests largely on expert opinion, and this trial shows many patients can safely stop earlier. It also shows that about one in 30 extra patients relapses. For most clinicians, the practical change is to stop treating six weeks as fixed and to decide duration with the infection team, taking into account organism, valve and follow-up.

• Discuss antibiotic duration with the endocarditis team rather than defaulting to six weeks
• Confirm clinical stabilisation criteria are met before considering an earlier stop
• Arrange close follow-up with repeat blood cultures after stopping
• Tell patients to return promptly with fever after antibiotics end
• Weigh the relapse risk more heavily where follow-up is uncertain

Why it matters

It challenges a six-week standard built on consensus, while showing exactly what shortening costs.

Don't overread it

Non-inferiority on a wide margin is not proof that shorter treatment is equally safe, especially given the higher relapse rate.

The statistics, in plain English

The safety difference of −2.4 points (95% CI −7.7 to 2.7) sits within the 7.5-point non-inferiority margin, but that margin is wide — the upper limit is compatible with about 3 more serious events per 100. Relapse at 5.1% versus 1.6% (P = 0.04) was a secondary end point, so it is less certain than the primary results, but it points in a clinically important direction.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.