Abstracts

Rationale: Estimation of individual pharmacokinetic parameters is important in achieving and maintaining desired drug levels, especially in medically urgent acute settings. Drug accumulation and elimination can be quite variable between individuals. In the acute inpatient setting obtaining formal kinetic estimations is difficult or impractical. The electronic medical record (EMR) can permit acquisition of doses, times, and levels. This report describes a matrix solution to parameter estimation that does not require formal kinetics estimations. Methods: As part of an IRB-approved study of seizures, trauma and intracranial hemorrhage in 18 infants seen in the PICU of the Children's Healthcare of Atlanta at Egleston between 6/12/2010-5/18/2014, the date, time, and values of 799 phenobarbital (PB) doses and 109 PB levels were recorded. The rise in concentration (C_est) after a dose (D) was estimated by the equation C₁ = C₀ + (D / kg) / V_d and the subsequent elimination by the formula C₂ = C₁ * exp (-k * ΔT), where k = fractions eliminated per 24 hours and ΔT is the elapsed time. For each patient a matrix was created with a wide range of values possible for V_d and k. Each possible combination of parameters was used to estimate the levels (C_est) which would result from known doses. These C_est were compared to actual levels (C_act) to determine the parameter pair giving the lowest absolute deviation from C_act. Results: The best fit Vd and k pair was strikingly different from other possible values (figure 1), suggesting this pair fairly estimates the patient's parameters. Across patients, average V_d = 0.69 liters/kg (range 0.5-0.96), and average k = 0.2783 fractions/day (range 0.2356-0.9379), equivalent to half-life ranges of 17.7-70.6 hours (both within published ranges). For these 18 patients, the deviation of C_est was within 4.2% of C_act, well within lab error (figure 2). Conclusions: The great variability in V_d and k or half-life accounts for the difficulty often found in achieving and maintaining desired drug levels. This method demonstrates that these parameters can be accurately estimated from known nonsteady state doses and levels without formal kinetics measurements in real patient care settings. Such estimates may help to more quickly achieve and maintain desired levels, may help to explain puzzling or difficult cases, may estimate steady state levels for a particular dose, and could be used in research applications to determine such things as the relationship of seizures to peak-trough variability of levels.