British colleagues found no scientific evidence for “completing your course of antibiotics”. Nothing new, but in the absence of competing news (the White House has become a daily soap) they opened Pandoras’ box for the lay press, with patients being recommended to stop their antibiotics, whenever they want. The birth of yet another inconvenient truth, as we cannot translate our knowledge into daily medical practice, and patients get even more confused. The good news: a new research agenda.

No comments on the BMJ piece, although I had always interpreted Flemings’ warning as related to underdosing rather than too short treatment courses. The obvious challenge is to reduce selection of resistant bacteria through unnecessarily long antibiotic treatment in the community where 85% of all antibiotics are used. But what is the potential gain? Indeed, antibiotic exposure creates selective pressure (a no-brainer), but what part of our resistance problem can be modulated by patients stopping to take their pills when feeling better? What part of the antibiotic course creates what part of selection? Is it linear? Every day adds the same risk. Or non-linear? 95% of selection reached by day 3? The current prevalence of carriage with ESBL-producing bacteria in the “open population” is 4-5% (Netherlands). What would it be if the average treatment course would reduce from 6 to 4 days? Could we expect a 33% reduction in ESBL-carriage? Does anyone know?

And what is the potential risk? What will happen with the unused pills? Returned to the local pharmacy or stored in the bathroom, for when symptoms return? The latter – storage fors elf-medication – seems common practice in southern and eastern European countries, see. If selection occcurs non-linearly, during the first days of treatment, self-management, indeed, may even increase antibiotic resistance, as the selection force is mobilized more than once.

So, what to do? We all know that the currently recommended  durations (5, 7, 10 or 14 days) are more biblical than scientific. Yet, minimum durations have hardly been determined, and will most probably differ between infections and between patients (the adult affected once every 5 years versus the 75+ with multiple comorbidities affected yearly). Furthermore, shorter treatment should be as good as current practice, necessitating non-inferiority studies with relevant clinical endpoints, for each indication and for each patient group. Using the “patients’ feeling of improvement” as a “biomarker” to stop treatment might be as heterogeneous as patients tend to be. Tempting to think of home-based biomarker testing to stop antibiotics, but that is yet an unexplored area.

Although, I fully agree that current duration recommendations lead to overuse of antibiotics, changing this practice is easier said than done. It may take decades before we can replace the lack of evidence for what we are used to do by personalised medicine for treatment of community-acquired infections. Untill then, we have nothing better to recommend then to complete your course of antibiotics.