Abstracts

Very little information is available on antiepileptic drug pharmacokinetics in elderly nursing home residents. Contemporary population pharmacokinetic approaches enable the use of sparse data to characterize pharmacokinetics and relate them to patient-specific covariates. The objective of this study was to identify covariates that affect VPA clearance in elderly nursing home residents. Data were collected from June 1, 1998 to December 31, 2000. Entry criteria included residency in a nursing home for at least two months, aged 65 years or older, a stable dosing regimen of VPA for at least 4 weeks, VPA concentration(s), and complete dosing information including time, date, and dose of the last four VPA administrations. Apparent oral VPA clearance (CL/F) was analyzed by NONMEM. A one-compartment model with first-order absorption and elimination was used. Both volume and absorption rate constant were fixed (14 L and 1 hr^-1, respectively). Information on 23 covariates was available for testing. Dichotomous covariates present in less than 10% of the population were not included, leaving 15 covariates tested. Covariates were tested by forward-inclusion and backward-elimination. Interindividual variability in clearance was estimated using an exponential error model and expressed as a coefficient of variation (CV%). Residual error was estimated using a combined additive and constant CV error model. An objective function decrease of 7.9 ([chi]², p[lt]0.005) for forward-inclusion and increase of 10.8 ([chi]², p[lt]0.001) for backward elimination were used to determine significance. The study consisted of 405 observations from 146 (52 men, 94 women) elderly nursing home residents. The population CL/F was 0.843 L/hr with CL/F being 1) 27% lower in female residents; 2) 41% greater when the resident was on concomitant carbamazepine (CBZ) or phenytoin (PHT) co-therapy; and 3) 25% greater when syrup formulation was used. Therefore, the final model was CL/F (L/hr) = 0.843 x 0.729 (if female) x 1.41 (if on CBZ or PHT co-therapy) x 1.25 (if formulation is syrup). Variability in CL/F was 32.9%. CV and standard deviation of the residual error were 18.2% and 10.6 mg/L, respectively. The mean population CL/F of VPA is lower in elderly nursing home women than in men. Patients taking a metabolic inducer (CBZ or PHT) along with VPA or taking the syrup formulation have a higher clearance than patients without CBZ or PHT co-therapy. The increased CL/F in patients taking VPA syrup may be due to a decreased bioavailability rather than an increased CL. This could be associated with pathology requiring use of syrup rather than an inherent property of the drug formulation. (Supported by NIH NINDS P50-NS16308)