[Control survey data analysis with smoothed distribution curve].

Abstract

Assay data and normal range collected in quality control surveys were analyzed with a smoothed distribution curve, in order to perform appropriate analysis of the data. In the control survey, the committee set a relative allowable limit for each test. A variable was changed continuously through the data range, and the data within the allowable limit were counted. In the computer program, the variable was changed step wise in logarithmic scale. These data were plotted on a graph with log-scale on X-axis and per cent of relative data in Y-axis. The graph thus obtained shows smoother curve than the histogram of the data, and Y-axis scale shows actual data within the allowable data at the variable scale. Normal ranges and assay data of several chemistry assays were analyzed with this method. Distribution of normal range of enzyme assays showed plural peaks in each test. Especially, in choline esterase assay, more than ten different units were used in laboratories in Japan with the same unit name, IU/1. In, amylase assay, nevertheless the normal range was separated in three peaks, the data of sample assay did not separate in peaks correspond to the normal range peaks. Thus, the smoothed distributions curve is useful for data analysis in quality control survey.