Uncertainty of measurement for competitive and indirect ELISAs

Abstract

A method for the estimation of the uncertainty of measurements for Gaussian outcomes of enzyme-linked immunosorbent assay (ELISA) is described using competitive and indirect foot and mouth disease (FMD) ELISAs. Assay repeatability was determined by random effects analysis of variance, and the normality of the residuals was checked. The standard errors of the individual predicted values were transformed into confidence intervals around the corresponding observed values and further transformed into probabilities of being above/below a cut-off. Logistic regression models were subsequently used to interpolate probability values for the whole range of possible assay values. The uncertainty of measurement of a test result was finally defined as the probability of not observing the same qualitative test result when retesting the same sample. For the competitive ELISA any sample with a percent inhibition 4% above the cut-off value had an uncertainty level (probability of a negative result in the case of retest) below 5%. In the indirect ELISA with a cut-off OD of 0.1, the uncertainty was below 5% for any sample with a normalised OD value above 0.22.