Abriana M Palumbo, Andrew Soliman, Kaysey Cloud, Khang Tran, Saeed K Alzghari, Margarita Taburyanskaya
Of 126 phenytoin levels reviewed, 66 met inclusion criteria. Population-specific CFs demonstrated lower bias and greater precision. In patients with active/breakthrough seizures and BUN <50 mg/dL, CF 0.29 improved accuracy compared with CF 0.2 (reduced mean percentage error [MPE] -59.59 vs -111.99 - indicating overprediction, root mean square error [RMSE] 1.79 vs 2.91), with similar trends when BUN exceeded 50 g/dL. A preliminary finding among adult patients younger than 65 showed better performance with a CF 0.275 than CF 0.2, with lower MPE (-40.43 vs -78.98) and RMSE (1.32 vs 2.16). In the IHD cohort, CF 0.201 outperformed CF 0.1 (MPE -74.40 vs -197.53 and RMSE 2.07 vs 5.32).
BACKGROUND: Phenytoin is a highly protein-bound medication and monitored due to its narrow therapeutic range. Free phenytoin serum concentratinos (fPHTm) are not readily available in many institutions.
OBJECTIVE: To evaluate whether population-specific correction factors (CFs) improve estimated free phenytoin serum concentrations (fPHTe) compared with traditional Winter-Tozer CF.
METHODS: This retrospective review included adults with same-day measured total phenytoin serum concentrations (mPHT), fPHTm, albumin, and blood urea nitrogen (BUN) values. Patients were stratified by active/breakthrough seizures, intermittent hemodialysis (IHD), or adults younger than 65. Patient-specific fPHTm percentages were used in the estimation of fPHTe.
RESULTS: Of 126 phenytoin levels reviewed, 66 met inclusion criteria. Population-specific CFs demonstrated lower bias and greater precision. In patients with active/breakthrough seizures and BUN <50 mg/dL, CF 0.29 improved accuracy compared with CF 0.2 (reduced mean percentage error [MPE] -59.59 vs -111.99 - indicating overprediction, root mean square error [RMSE] 1.79 vs 2.91), with similar trends when BUN exceeded 50 g/dL. A preliminary finding among adult patients younger than 65 showed better performance with a CF 0.275 than CF 0.2, with lower MPE (-40.43 vs -78.98) and RMSE (1.32 vs 2.16). In the IHD cohort, CF 0.201 outperformed CF 0.1 (MPE -74.40 vs -197.53 and RMSE 2.07 vs 5.32).
CONCLUSION AND RELEVANCE: Population-specific CF estimated fPHTe more accurately and precisely than traditional CF. Findings remain preliminary due to limited representation, and further research is needed to improve CF selection across various subgroups.