(a) pooling data from a database in which measurements of the analyte from a reference population have been stored, wherein the reference population comprises both healthy and unhealthy subjects;

(b) plotting cumulative frequencies of data against the measurements of the analyte from the reference population to form a distribution;

(c) transforming the plotted data to normalize the distribution if the mean of the plotted data is located in the first or fifth quintile of the distribution, thereby resulting in transformed plotted data;

(d) calculating a linear regression of the plotted data, or if transformed, of the transformed plotted data, resulting in a curve and a value for maximum residual error, wherein the linear regression is constrained so that the maximum residual error of the linear portion of the curve is below a predetermined threshold to account for a within-individual biological variation for the analyte; and

(e) determining the reference interval for the analyte in the reference population by selecting a range that corresponds to the linear portion of the curve.