BACKGROUND: The interpretation of semi-quantitative methods has always been difficult, because the different kitmanufacturers use varying concentration values and there is a considerable overlap between kit-defined concentrations within the same kit (Kit: 'Ready-to-use' measuring system specific for each manufacturer's product). MATERIAL: More than 2000 private practitioners and laboratories participated in three external quality control surveys on urine-glucose performed with a total of six control materials with known concentrations. METHOD: The ordinal scale model for evaluation of dichotomous methods based on rankit transformation of fractions of positive results (Petersen et al. Scand J Clin Lab Invest 2008;68:298-311) has been extended and modified to handle semi-quantitative data. Here, the percentages of results larger than the kit-concentration is calculated for each control sample and applied as a dichotomous method. Thereafter, these percentages are separated into all the defined kit-concentrations. RESULTS: A total of eight kits had more than 50 measurements on at least four control materials which made them eligible for the calculations of logarithmic mean and standard deviation and thereby geometric mean and coefficient of variation for each of the kit-concentrations of each kit. Based on these parameters, the true concentration for selected percentages of each kit-concentration could be estimated. Moreover, the percentages of the different kit-concentrations could be calculated for each known true concentration. CONCLUSIONS: The present model is a powerful tool for improved characterization of semi-quantitative kits, which makes it possible to evaluate and validate kits and to optimize external quality control.
BACKGROUND: The interpretation of semi-quantitative methods has always been difficult, because the different kitmanufacturers use varying concentration values and there is a considerable overlap between kit-defined concentrations within the same kit (Kit: 'Ready-to-use' measuring system specific for each manufacturer's product). MATERIAL: More than 2000 private practitioners and laboratories participated in three external quality control surveys on urine-glucose performed with a total of six control materials with known concentrations. METHOD: The ordinal scale model for evaluation of dichotomous methods based on rankit transformation of fractions of positive results (Petersen et al. Scand J Clin Lab Invest 2008;68:298-311) has been extended and modified to handle semi-quantitative data. Here, the percentages of results larger than the kit-concentration is calculated for each control sample and applied as a dichotomous method. Thereafter, these percentages are separated into all the defined kit-concentrations. RESULTS: A total of eight kits had more than 50 measurements on at least four control materials which made them eligible for the calculations of logarithmic mean and standard deviation and thereby geometric mean and coefficient of variation for each of the kit-concentrations of each kit. Based on these parameters, the true concentration for selected percentages of each kit-concentration could be estimated. Moreover, the percentages of the different kit-concentrations could be calculated for each known true concentration. CONCLUSIONS: The present model is a powerful tool for improved characterization of semi-quantitative kits, which makes it possible to evaluate and validate kits and to optimize external quality control.