Maria Swartling: Model-informed antibiotic dose individualisation in clinical practice

Abstract

Antibiotics are life-saving medications, and optimal dosing for each individual is critical to effective and safe treatment. Dose individualisation is needed when routine dosing results in concentrations that vary more than accepted between individuals from an efficacy or safety point of view. A tool for individualisation is model-informed precision dosing (MIPD), which uses software with integrated population pharmacokinetic (PK) models to interpret measured antibiotic concentrations and to optimise dosing.

This thesis aimed to improve antibiotic treatment by evaluating pharmacokinetic/pharmacodynamic (PK/PD) target attainment for alternative dosing strategies, assessing PK variability between individuals, and developing processes for MIPD in clinical practice.

One way to improve target attainment for beta-lactams without increasing the daily dose is to prolong the infusions. Different beta-lactam infusion durations were evaluated for adult intensive care patients. The predicted attainment of PK/PD targets recommended for critically ill was high for primary pathogen scenarios regardless of infusion strategy, indicating that short infusions (15 min) can be used. In situations when worst-case scenario pathogens are suspected, e.g., hospital-acquired infection, short infusions were insufficient, and 3h extended or continuous infusions improved target attainment. However, when evaluating the implemented routine use of extended infusions in critically ill, a large proportion of individuals was still below target. The large PK variability between individuals suggests a need for early monitoring to identify under and over-treated patients. In patients with sub-optimal exposure, MIPD can provide an added value.

This thesis has brought MIPD closer to clinical implementation and the benefit of patients. Barriers to implementation were addressed, including an assessment of how documentation errors can impact dosing decisions based on concentration monitoring. For vancomycin, simulations indicate that target attainment evaluations are robust with respect to locally observed errors in dose administration time. For meropenem, the impact of errors was evident already at normal renal function and pronounced in augmented renal clearance. Extra measures to promote correct documentation are needed when monitoring beta-lactams like meropenem, particularly in intensive care. Further, a novel MIPD workflow and an implementation plan for vancomycin dosing were developed. This work defined a new role for physicians or pharmacists as MIPD consultants and new tasks for clinical pharmacists. The pragmatic development process can guide other institutions aiming to initiate MIPD, and the developed workflow provides an opportunity to expand MIPD to other treatments.