variability of sensitivity and specificity

Last reviewed 01/2018

The sensitivity and specificity of a test are not static. This is because the cutoff between normal and abnormal is usually arbitrarily defined.

As the cutoff is made more abnormal, a test's sensitivity and false positive rate decrease. For example, the higher the blood glucose used to define diabetes, the more certain one can be that a positive result indicates disease but the more patients with the disease will be missed.

Conversely, if the cutoff level is less abnormal, the sensitivity improves but so does the false positive rate. More patients with the disease will be found but more without the disease are in danger of being falsely labelled as having the disease.

Most laboratory tests have locally defined ranges with the normal range containing 95% of results.

ref: Principles of Ambulatory Med 4e L.Randol Barker et al