Stopping endocarditis antibiotics early bought two weeks — and more relapses

The standard treatment for infective endocarditis on the left side of the heart is up to six weeks of antibiotics. That duration is not derived from trials — the POET II investigators state that the recommendation rests largely on expert consensus [s1]. Six weeks is a long time to hold a bed, a cannula and a patient, so the question of whether it can be shortened is a real one.

POET II, published in the New England Journal of Medicine on 28 August and presented at ESC Congress 2026, tested it. The answer has two halves that point in different directions [s1].

What the trial did

POET II was an international, open-label, randomised trial in adults in stable condition with infective endocarditis caused by Staphylococcus aureus, Enterococcus faecalis, or streptococcus species [s1].

Crucially, everyone got treated first. Before randomisation, all patients received at least a prespecified 2 to 4 weeks of therapy and had to meet criteria for clinical stabilisation [s1]. Only then were they assigned: the tailored-therapy group stopped antibiotics, and the standard-therapy group continued to a total duration of 4 to 6 weeks [s1].

There were two primary endpoints, tested against different standards. The efficacy endpoint was days alive without antibiotic treatment for infective endocarditis or bacteraemia within six months of randomisation, tested for superiority [s1]. The safety endpoint was a composite of death from any cause, unplanned cardiac surgery, or symptomatic embolic events within six months, tested for noninferiority with a margin of 7.5 percentage points [s1]. Relapse of bacteraemia or infective endocarditis was a key secondary endpoint [s1].

A total of 508 patients were randomised: 255 to response-tailored therapy and 253 to standard duration [s1].

What happened

On the efficacy endpoint, stopping early did what stopping early does. Median time alive without antibiotic treatment was 183 days (IQR 181 to 183) with tailored therapy and 169 days (IQR 166 to 171) with standard therapy — a Hodges-Lehmann estimated difference of 13 days (95% CI 12 to 13, P<0.001 for superiority) [s1].

On the safety endpoint, a primary safety event occurred in 21 patients (8.2%) with tailored therapy and 27 (10.7%) with standard therapy, an absolute between-group difference of -2.4 percentage points (95% CI -7.7 to 2.7, P<0.001 for noninferiority) [s1]. Noninferiority was met.

And then the secondary endpoint. Relapse of bacteraemia or infective endocarditis occurred in 13 patients (5.1%) with tailored therapy and 4 patients (1.6%) with standard therapy (P = 0.04) [s1].

The three results have to be read together

The efficacy result is close to tautological. If you randomise one group to stop antibiotics and the other to continue them, the group that stops will have more antibiotic-free days. A difference of 13 days with a confidence interval of 12 to 13 is a very precise measurement of an intervention that was defined to produce it [s1]. It quantifies the saving; it is not evidence of clinical benefit.

The safety result is the substantive noninferiority finding, and it is genuinely reassuring on the outcomes it covers: death, unplanned cardiac surgery and symptomatic embolic events were not more common with early stopping, with a confidence interval whose upper bound of 2.7 percentage points sits well inside the 7.5-point margin [s1].

The relapse result is the one that complicates the picture, and the authors say so directly in their conclusion: the strategy met the criterion for noninferiority with respect to safety but was associated with a higher incidence of relapse [s1].

Relapse was a secondary endpoint, and 13 events against 4 is a small number on which to build a firm estimate. But relapse of endocarditis is not a minor event, and its exclusion from the primary safety composite means the composite's reassuring result does not cover it.

Where this sits in a longer argument

This is the second time this line of research has tested a piece of endocarditis orthodoxy. The original POET trial, published in 2018, randomised 400 stabilised patients with left-sided endocarditis to continue intravenous antibiotics or switch to oral treatment, and found the primary composite occurred in 24 patients (12.1%) in the intravenous group and 18 (9.0%) in the oral group — a difference of 3.1 percentage points (95% CI -3.4 to 9.6, P = 0.40), meeting noninferiority [s2].

The pattern across both trials is the same: in patients who have already stabilised on treatment, the conventional regimen contains slack. What POET II adds is a boundary — the slack is not unlimited, and the price of taking too much of it appears to be relapse rather than death.

Limits

The trial was open-label [s1]. Patients enrolled only after 2 to 4 weeks of therapy and clinical stabilisation, so the findings say nothing about patients who do not stabilise [s1]. Only three pathogen groups were studied [s1]. Follow-up was six months, which is the conventional window for endocarditis relapse but not an indefinite one.

The trial was funded by Sygeforsikringen "danmark" and others, and registered as NCT03851575 [s1].

What to watch

Whether guideline committees adopt a response-tailored strategy, and if they do, whether they pair it with a defined surveillance schedule for relapse. The trial's own numbers make the case for the pairing: a strategy that is noninferior on death, surgery and embolism while producing 13 relapses against 4 is one that needs follow-up built into it, not simply a shorter course [s1].

This article describes trial results. It is not advice about any medicine, and nothing here should be used to start, stop or change treatment.

Sources

• Response-Tailored or Standard-Duration Antibiotic Treatment for Infective Endocarditis, New England Journal of Medicine, 28 August 2026
• Partial Oral versus Intravenous Antibiotic Treatment of Endocarditis, New England Journal of Medicine, 29 August 2018