Statistics of patient test values: application to indirect normal range and to quality control.

The use of patient data for calculation of the indirect normal range and for quality control is investigated. The report deals with an empirical study of the properties of various statistics of patient data, and with the selection of appropriate patient statistics for calculation of the indirect normal range and for quality control. In regard to the indirect normal range, it is shown that the mode of the patient data (calculated as the mid-value in a segment of smallest variance in the rank-ordered set) is a useful statistic for characterizing central tendency of the normal range. For the monitoring of systematic error, the median of the patient data is usually the. most sensitive indicator of change.