Mathematical modeling links benefits of short and long antibiotic treatment to details of infection

Antibiotics are the major tool for treating bacterial infections. With rising antibiotic resistance in microbes, strategies that limit further evolution and spread of drug resistance are urgently needed, in individuals and populations. While classical recommendations favor longer and aggressive treatments, more recent studies and clinical trials advocate for moderate regimens. In this debate, two axes of aggressive treatment have typically been conflated: treatment intensity and treatment duration, the latter being rarely addressed by mathematical models. Here, by using a simple mathematical model of a generic bacterial infection, controlled by host’s immune response, we investigate the role of treatment timing and antibiotic efficacy in determining optimal duration of treatment. We show that even in such simple mathematical model, it is impossible to select for universally optimal treatment duration. In particular, short (3 day) or long (7 day) treatments may be both beneficial depending on treatment onset, on the criterion used, and on the antibiotic efficacy. This results from the dynamic trade-off between immunity and resistance in acute, self-limiting infections, and uncertainty relating symptoms to the start of infection. We find that treatment timing can shift the trend between resistance selection and length of antibiotic exposure in individual hosts. We propose that major advances in predicting impact of antibiotics on bacterial infections must come from deeper experimental understanding of bacterial infection dynamics in humans. To guide rational therapy, mathematical models need to be constrained by data, including details of pathology and symptom thresholds in patients, and of host immune control of infection.