A computer program for estimation of reference percentile values in laboratory data.

It has been suggested that the physician would be aided in his interpretation of clinical laboratory results if the data were reported in terms of percentile ranking in a reference population as well as in conventional units. This paper describes a FORTRAN porgram for estimating the necessary reference percentile values by means of regression over 20% of the distribution rather than by interpolation between adjacent points. The regression estimates are shown to have a negligible bias and to be somewhat more stable than those obtained by interpolation.