External validation of a model informed precision dosing software for cefotaxime for treating critical ill infants...

Critically ill infants and children often experience subtherapeutic cefotaxime exposure when administered licenced doses. Dose individualization guided by model informed precision dosing (MIPD) software is emerging, however, external validation of MIPD performance before implementing into clinical practice is necessary. This in silico study assessed the accuracy of an MIPD software in predicting cefotaxime concentrations among critically ill infants and children using the a priori ('empiric') and the a posteriori (Bayesian forecasting) approaches.

The study population was between 1 month and 12 years of age receiving cefotaxime in a paediatric intensive care unit. Essential data was inputted into the MIPD software ID-ODSTM to perform the a priori (4 h and 6 h post dose) and the a posteriori (one- and two-measured serum cefotaxime concentrations) predictions. The concentrations predicted by ID-ODS were compared against measured concentrations, using bias, precision and linear regression methods.

Thirty infants and children with 115 concentration measurements produced 194 simulations for analysis. The median age, weight and serum creatinine concentration were 22.8 months, 10.6 kg and 23 µM/L, respectively. The a priori approach showed high biases, low precisions, and significant underprediction [weighted mean prediction error (WMPE) -48.48%, -85.63%, weighted root mean squared prediction error (WRMSE) 98.86, 87.60% at 4 h and 6 h, respectively]. The a posteriori approach exhibited improved accuracy (WMPE 0.28%, -21.54%, WRMSE 76.41%, 58.98% for one and two samples, respectively). MIPD-guided dose adjustments achieved pre-defined PK-PD targets in >90% of simulations using the a posteriori approach.

MIPD incorporating measured blood samples of cefotaxime increases the likelihood of achieved target drug concentrations in treating severe infections among critically ill infants and children.