Pitfalls and problems in the interpretation of AFP data--myths and misuse of the "normal range"

Drawing a line more intelligently for AFP data means using as a denominator two fairly well-defined populations (abnormal and normal), instead of one ill-defined normal population. The line is drawn at the intersect between two curves to minimize the error of misclassification; this replaces an arbitrarily defined cutoff level (multiples of the standard deviation unit). Moving the line to the right reduces the false positive error when screening for NTD, whereas moving the line to the left provides a sensitive alert signal for routine screening of maternal sera by AFP determinations for fetal distress. Adjustment of the line for cost and benefit factors and the use of probability statements that include the prevalence of specific conditions all help sharpen medical classification and decisions. Cross-sectional analysis is separated from longitudinal analysis. A benefit from this discussion may be the replacement of the normal range with a simple plot of AFP results or a percentile distribution, which would be more meaningful and more easily understood by all.