Stopping endocarditis antibiotics on response saved days but tripled relapse

Design

International open-label randomised trial

Population

508 stable adults with left-sided endocarditis (S aureus, E faecalis, streptococci)

Primary outcome

Days alive without antibiotics at 6 months; safety composite of death, unplanned surgery or embolism

Effect

+13 days (95% CI 12–13); safety 8.2% vs 10.7%, difference −2.4 points (−7.7 to 2.7)

POET II, published in NEJM on 28 August, randomised 508 adults with left-sided infective endocarditis due to Staphylococcus aureus, Enterococcus faecalis or streptococci. All had completed at least a prespecified two to four weeks of antibiotics and met clinical stabilisation criteria. They were then assigned to stop antibiotics (response-tailored) or continue to a standard total of four to six weeks.

Tailored therapy gave a median of 13 more days alive and off antibiotics over six months. Death, unplanned cardiac surgery or symptomatic embolism occurred in 8.2% versus 10.7%, which met non-inferiority. But relapse of bacteraemia or endocarditis was 5.1% with tailored therapy against 1.6% with standard duration.

Six weeks has always rested mostly on expert opinion, so a randomised test is welcome. The relapse signal is the price, and relapse in endocarditis often means another admission, another course and sometimes surgery.

The trial is open-label and the population was selected: stable, with specific organisms, after a defined initial course. It argues for individualising duration in carefully chosen patients, not for routinely shortening it.

• Keep standard-duration therapy as the default for left-sided endocarditis until guidelines change.
• Consider shorter courses only in stable patients with the organisms studied, under infectious diseases or endocarditis team review.
• Explain the trade-off to patients: fewer antibiotic days against a higher chance of relapse.
• Arrange follow-up blood cultures and a low threshold for review after any shortened course.

Why it matters

It challenges the six-week convention while showing exactly what shortening costs.

Don't overread it

Non-inferiority on a safety composite does not mean shorter courses are free of harm; relapse was higher.

The statistics, in plain English

The safety composite difference was −2.4 percentage points with a confidence interval of −7.7 to 2.7, inside the 7.5-point non-inferiority margin. Relapse was a secondary outcome, and the P value of 0.04 is close to the conventional threshold, but the direction is clinically important and consistent with a shorter course.

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.