Shorter antibiotic courses are winning almost every trial that tests them, across pneumonia, bacteraemia, bone and joint infection and now endocarditis. What those trials share is a structure that rarely survives translation into a busy ward: a defined stabilisation checkpoint, and systematic follow-up that catches relapse.
The relapse signal in the endocarditis data makes the point. Roughly one in twenty patients whose antibiotics were stopped early had bacteraemia or endocarditis return, against one in sixty who completed the standard course. In a trial, every one of those was detected because follow-up was protocolised. On a ward, a patient discharged early with no arranged review is a patient whose relapse presents late, somewhere else, often sicker.
So make the follow-up part of the decision, not a consequence of it. Before stopping short, write down what would count as relapse for this patient, when they will be seen, who will chase the cultures, and what the patient has been told to watch for. If you cannot answer those four questions, the shorter course is not the one the trial tested — it is just a shorter course.
- Name the stabilisation criteria you are relying on and document that they were met.
- Book the review appointment and the repeat blood cultures before discharge, with a named person responsible.
- Tell the patient specifically what relapse feels like — recurrent fever, sweats, new weight loss — and where to present.
- Give the patient a written record of organism, susceptibilities and total days treated; the next clinician will need it.
- Where reliable follow-up genuinely cannot be arranged, complete the standard course instead.
The statistics, in plain English
Non-inferiority trials are designed to answer whether a less intensive strategy is not unacceptably worse on a defined primary endpoint, usually death or a major event. They are not designed to rule out smaller harms, and relapse is exactly the kind of outcome that shows up as a secondary finding with a borderline p value. When a shorter strategy passes non-inferiority but carries a relapse signal, the safe reading is that the strategy works within the follow-up system the trial provided — which is a condition of the result, not a detail of it.
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