Predicting unbound phenytoin concentrations in the critically ill neurosurgical patient

OBJECTIVE: To assess the correlation of measured unbound phenytoin concentration (dphF) to estimated unbound concentration (dphEF) using the Sheiner-Tozer equation in critically ill patients in the neurosurgical intensive care unit. DESIGN: The dphF and total phenytoin (dphT) trough serum concentrations were measured during the first week of therapy in 17 consecutive patients with albumin concentrations less than 3.5 g/dL. Serum albumin concentrations were measured within 24 hours of serum phenytoin concentration measurement. SETTING: A university-affiliated urban teaching hospital. PARTICIPANTS: The study population consisted of 17 neurosurgical patients who were at least 18 years old. MAIN OUTCOME MEASURES: The predictability of the Sheiner-Tozer equation was tested by measuring dphF, dphT, and serum albumin concentrations. Measured phenytoin concentrations were compared with phenytoin concentrations calculated from the Sheiner-Tozer equation. To estimate correlation between variables linear regression was calculated. Mean absolute value of error and mean error were estimated to assess precision and bias between measures, respectively. RESULTS: The mean +/- SD dphT was 13.05 +/- 5.15 mu g/mL. The measured dphF was 1.89 +/- 0.80 compared with 2.00 +/- 0.8 mu g/mL for the dphEF (NS). Regression analysis for dphEF versus dphF revealed a significant correlation (r(2) = 0.94, p = 0.001). The mean absolute value of error for the Sheiner-Tozer equation to predict dphEF was 0.167, which was 9% of the mean value of dphF (1.89). CONCLUSIONS: These results indicate that, in this population, the Sheiner-Tozer estimate of dphEF provides an unbiased, precise clinical estimate of dphF in patients for whom measured dphF is unavailable or impractical.