Estimation of Nornal Range by the Method of Curve fitting to the Histogram

Abstract

A calculation method was described for the estimation of the normal range of laboratory tests from patients data in the clinical laboratory. Two calculation procedues were applied for the normal range estimation. Logarithm of histogram frequencies are plotted against measured variable, and the central portion of this histogram was fitted to the 2nd order regression curve using the least square method. The regression curve corresponds theoretically to the normal distribution curve, and the lower and the upper limits of the normal range were calculated from the parameters of the normal distribution curve (L-method). The other method is that logarithm of histogram frequencies are plotted against logarithm of measured variables and the 2nd order regression curve fitted to this histogram corresponds theoretically to the log-normal distribution curve (LL-method). The upper and the lower limits were estimated from the parameters of the distribution curve These calculation methods are applied to several model distributions and to clinical laboratory data, and accuracy and precision of the calculation methods are checked. After general evaluation of the methods, it was concluded that LL-method can be used for the estimation of the lower limit of the normal range, and that L-method can be used for the estimation of the upper limit.