Evaluation of contributions to the uncertainty of cholesterol determination in human serum by means of gas chromatography/(quadrupole-) mass spectrometry

 The uncertainty budget of the determination of cholesterol in a reference material (freeze-dried human serum) by isotope dilution mass spectrometry (IDMS) was evaluated. Measurements were run on a GC/quadrupole MS. A nested experimental design combined with a corresponding model for the analysis of variance (ANOVA) revealed the components making up the total uncertainty of the determinations. The coefficient of variation (CV) of the sole GC/MS-step was 0.47%, while that of the sample preparation including hydrolyzation, extraction and derivatization was 0.84%. This made the in-laboratory (overall) CV 0.96%. In addition, there was a vial-to-vial effect of 1.00% probably due to both inhomogeneity of the pooled material and lack of reproducibility of the filling procedure (including freeze drying) of the supplier as well as the reconstitution prior to use in the laboratory. These contributions sum up to an overall CV of a single determination of about 1.4%. Received: 17 October 1996/Revised: 14 November 1996/Accepted: 18 November 1996