A clinical trial incident: Why simple PK is sometimes not so simple

The antibiotic dose calculator trial in children aims to find out if a web based dose calculator is better than department protocols used in 3 clinical units for achieving target concentrations of amikacin, gentamicin or vancomycin. 

During the trial an incident occurred when the dose calculator dose proposed a clearly inappropriate maintenance dose and dosing interval. The dose calculator had been programmed to show a warning to clinical staff if the daily dose was more than 50% larger than the department protocol. The warning alerted the clinician who decided to use the department protocol dose regimen.

An investigation of the reason for the dose calculator error revealed 1) a programming error that was triggered by an unusual set of clinical circumstances 2) a dose finding algorithm that met the stated clinical target but was unsuitable for maintenance dosing.

An extensive simulation of over 1/2 million patients covering the full range of ages, weights, renal function and other factors such as non-steroidal anti-flammatory use anticipated in the clinical setting showed that the dose calculator had a root mean square error in reaching the target concentrations that was always lower than the department protocol. This simulation mechanism is now used as a validation mechanism for future changes to the dose calculator.