Simple estimation of reference intervals from routine laboratory data

Abstract

Abstract: According to the recommendations of the IFCC and other organizations, medical laboratories should establish or at least adapt their own reference intervals, to make sure that they reflect the peculiar characteristics of the respective methods and patient collectives. In practice, however, this postulate is hard to fulfill. Therefore, two task forces of the DGKL (“AG Richtwerte” and “AG Bioinformatik”) have developed methods for the estimation of reference intervals from routine laboratory data. Here we describe a visual procedure, which can be performed on an Excel sheet without any programming knowledge. Patient values are plotted against the quantiles of the standard normal distribution (so-called QQ plot) using the NORM. INV function of Excel. If the examined population contains mainly non-diseased persons with approximately normally distributed values, the respective dots form a straight line. Very often the values are rather lognormally distributed; in this case the straight line can be detected after logarithmic transformation of the original values. Values, which do not match with the assumed theoretical distribution, deviate from the linear shape and can easily be identified and eliminated. Using the reduced data set, the mean value and standard deviation are calculated and the reference interval (μ±2σ) is estimated. The method yields plausible results with simulated and real data. With the increasing number of results, which do not match with the model, it tends to underestimate the standard deviation. In all cases, where the QQ plot does not yield a substantial linear part, the proposed method is not applicable.