IntroductionEstablishing analytical performance requirements for in vitro diagnostic (IVD) assays is a challenging process. Manufacturers try to optimize analytical performance by choosing amongst many combinations of different product performance characteristics. Sigma metrics and method decision charts can be helpful aids in choosing appropriate analytical performance requirements. The objective of this research was to demonstrate the use of Sigma metrics and method decision charts to help establish analytical performance requirements and to optimize analytical performance at medical decision concentrations for an IVD assay.Materials and methodsA range of possible Sigma metrics were determined using three sources for total allowable error (TEa) and hypothetical total PSA assay results. Method decision charts were created for each TEa source and used to identify the maximum precision and bias that the assay could have to maintain sigma level performance of at least 3.ResultsTo achieve a sigma performance level of at least 3 for a hypothetical total PSA assay, the maximum allowable coefficient of variation ranged from 5.0% to 11.2% depending on the TEa source. To achieve a sigma performance level of at least 6, the maximum allowable coefficient of variation ranged from 2.5% to 5.6% depending on the TEa source.ConclusionsUsing Sigma metrics and method decision charts when establishing analytical performance requirements can help manufacturers choose product requirements that will optimize IVD assay product performance.
Glycated hemoglobin (A1C) is a key biomarker for glycemic control in diabetes and has an important relationship with glucose. The objective of this study was to evaluate the relationship of real-world laboratory results for patient samples with both glucose and A1C. 213,698 paired glucose and A1C measurements from 7laboratories in 39 countries were collected January to October 2021. Values within the measuring interval 70 to 400 mg/dL for glucose and 4% to 13% (NGSP) for A1C were analyzed. The pairs were analyzed in twenty equal-sized groups by composite rank order. The summed absolute relative differences (SARD) were compared between best-fit nonlinear and linear models. The group mean values presented a clear nonlinear relationship, with an SARD seven times lower than the best-fit linear model. The nonlinear Michaelis constant for glucose and glucose transporter 1 (GLUT1) was 4mg/dL, which agrees with literature values. The 95% prediction intervals at the diabetes diagnosis levels of 6.5% A1C and 126 mg/dL glucose were narrower for the nonlinear model. This model-based formula allows better prediction of the relationship of glucose and A1C which can lead to improved personalized clinical decisions.
Figure: Nonlinear (left) and linear (right) best-fit models, N = 213,698. Each point represents the mean of approximately n=10,680 paired measurements. Dotted lines are 95% prediction intervals.
Disclosure
T.Dunn: Employee; Abbott. Y.Xu: Employee; Abbott Diabetes. S.Schneider: None. S.Gawel: Employee; Abbott Diagnostics. R.Gopinath: Employee; Abbott Diagnostics. M.Berman: None. A.P.Orzechowski: Employee; Abbott.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.