Deriving reference values in electrodiagnostic medicine.

Traditional methods for determining nerve conduction reference values have usually involved studying a group of normal controls, and determining the mean +/- 2 standard deviations (SD) for each parameter. Recent studies suggest other factors should be taken into account. Height has a greater effect on conduction velocity than age or temperature. The common assumption that conduction values follow a normal, bell-shaped Gaussian distribution appears unwarranted. The curve for some conduction parameters is significantly skewed, making a mean +/- 2 SD definition of normality inaccurate to a clinically important degree. In developing and using reference values one should consider height, age, and temperature, and calculate the mean +/- 2 SD of the transformed data to remove the effects of skew. Reference values provide only a guide to the probability a given result came from a healthy or diseased individual; one should therefore seek multiple internally consistent abnormalities before diagnosing disease on the basis of electroneurography.