Objective: To compare two methods in the estimation of the uncertainty in laboratory quality control.
Methods: A computerized simulation is performed to compare the delta method (suggested by the International Organization for Standardization and the Entidad Nacional de Acreditación) and a bootstrap-based method. The simulation includes several situations with different environmental conditions and different relationships between the analyzed variables.
Results: The mean in the coverage obtained by the estimated confidence intervals is higher and closer to the nominal using the bootstrap than using the delta method. The most important differences are observed in the coverage percent distribution: while using the bootstrap, a great number of simulations obtain coverage near the nominal of 95%; using the delta method the coverage are more dispersed, including coverage of 100% in some occasions and lesser than 80% in others. The bootstrap offers very similar results under different conditions, including in the presence of unknown and unmeasured variables or when the analyzed variables are correlated. The delta method shows poorer results in both situations.
Conclusion: The uncertainty in the laboratory quality control can be estimated more accurately with bootstrapping than with the delta method.