In-house method validation: application in arsenic analysis

A detailed procedure for in-house method validation was proposed and applied to the validation of a method for determination of arsenic in bovine liver. Linearity and matrix effects were assessed by solvent and matrix-matched curves. Data were fitted by ordinary least squares method (OLSM). Violations of assumptions underlying OLSM residuals (normality, independence and homoscedasticity) were evaluated. F-tests were undertaken to check the adjustment to the model by evaluation of regression and lack of fit significances. The slopes and interceptions were compared for significant differences by t-test. Linearity was demonstrated and significant matrix effects were not observed between 0.5 and 8 ng mL(-1). Selectivity, trueness, precision and experimental limits of the method were established by assays with blanks and spiked samples. Recovery (means of 88.9-99.9%) was calculated and analysis of variance was carried out to estimate the standard deviation under repeatability (3.7-7.2%) and within-reproducibility (3.7-7.3%) conditions. The spiking experiments indicated precision and no lack of trueness. Experimental limits of detection and quantification were 15 and 30 mu g kg(-1), respectively. The procedure was found to be suitable for in-house validation with respect to the evaluation of efficiency parameters, ease of application and cost-benefits. (c) 2006 Elsevier Inc. All rights reserved.