Up to six weeks of antibiotics for left-sided infective endocarditis has always been consensus rather than evidence. POET II randomised 508 adults with endocarditis caused by Staphylococcus aureus, Enterococcus faecalis or streptococcal species. Everyone had already completed 2 to 4 weeks of appropriate therapy and met prespecified stabilisation criteria before randomisation; those assigned to tailored therapy then stopped, while the standard group completed a total of 4 to 6 weeks.
Days alive without antibiotics over six months were 183 with tailored therapy versus 169 with standard, a difference of 13 days (95% CI 12 to 13, p<0.001 for superiority). The safety composite of death, unplanned cardiac surgery or symptomatic embolism occurred in 21 patients (8.2%) versus 27 (10.7%), difference -2.4 percentage points (95% CI -7.7 to 2.7), meeting noninferiority against a 7.5 point margin.
Relapse is the catch. Bacteraemia or endocarditis recurred in 13 patients (5.1%) on tailored therapy and 4 (1.6%) on standard, p=0.04. In six months of follow-up that did not translate into more deaths or operations, but six months is short for a relapsing endovascular infection.
For an ID service the decision is now explicit rather than reflexive. Two weeks less of intravenous therapy means fewer line days, fewer catheter-related bloodstream infections, earlier discharge and substantially lower cost — which in Indian practice, where the bill is often paid directly by the family, is a material argument. Against that sits a roughly one-in-twenty relapse rate. The strategy is only defensible where follow-up is guaranteed, because it depends on catching relapse rather than preventing it.
- Restrict early stopping to patients genuinely stabilised after at least 2 to 4 weeks of appropriate therapy, as the trial required
- Consent explicitly on relapse: about 5% in six months versus about 2% with the full course
- Book the follow-up before stopping, with a clear instruction on when to return and repeat blood cultures
- Do not extend this to prosthetic valves, right-sided disease, or organisms outside S. aureus, E. faecalis and streptococci
- Count the line as a risk in its own right — two extra weeks of intravenous access is not a neutral option
The statistics, in plain English
Two endpoints, two different statistical questions. The efficacy endpoint was tested for superiority and won cleanly. The safety endpoint was tested for noninferiority against a margin of 7.5 percentage points, meaning the design accepted in advance that the shorter course could be up to 7.5 points worse on death, surgery or embolism and still pass — a wide allowance for outcomes this serious. The observed difference happened to favour the short course, but the interval reaches 2.7 points the other way. The relapse finding rests on 17 events and a p value of 0.04 on a secondary endpoint, so as a statistic it is fragile; as a mechanism, less antibiotic causing more relapse is entirely expected, which is why it should be believed.
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.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for infectious diseases, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free