Comparison of four valproic acid free fraction determination methods used in clinical practice.

Objective: Given that hypoalbuminemia tends to result in higher free fraction  concentrations of valproic acid, different methods have been developed to  determine the latter in patients with this condition. The aim of this study is to  assess the reliability of these methods and, if necessary, design a new  estimation method.

Method: A retrospective analysis was carried out by the Pharmacy Department of Severo Ochoa University Hospital of admitted patients with at least one  trough concentration of valproic acid between October 2017 and February  2019. The estimation methods used were those developed by Kodama,  Hermida, Doré, as well as a new method proposed in the study. A total of 17  serum valproic acid concentrations were used to determine the free fraction of  valproic acid with each method; the values obtained were compared with the  results obtained following laboratory determinations. Accuracy and precision  were calculated using mean error and root mean square error, respectively.

Results: The comparison between observed and predicted free valproic acid  values using the methods under investigation showed that the method proposed in this study provides the highest reliability as it presents the  highest accuracy and precision. The worst results were those obtained  using  the Kodama method, which does not consider albuminemia, an essential variable that determines the concentration, therapeutic effect and toxicity of valproic acid.

Conclusions: Given that the method proposed in this study proved to be superior to the other methods analyzed, we believe it can be reliably used to estimate free valproic acid levels in patients with hypoalbuminemia.