[Skip to Content]
[Skip to Content Landing]
Citations 0
Editor's Correspondence
June 28, 1999

Being Sensitive to the Specifics of Predictive Values in the Diagnosis of Tuberculous Pleuritis

Arch Intern Med. 1999;159(12):1370. doi:

Light has been responsible for defining many aspects of the classification of pleural effusions. However, we thought that his editorial1 on tuberculous pleuritis confused the statistical terminology. In the editorial, Light comments on test specificity but actually describes positive predictive value. Sensitivity and specificity are properties inherent to the test itself and are usually considered to be independent of prevalence. This independence may not be completely true in practice, since the disease severity in patients whose results are selected to demonstrate the sensitivity and specificity of a test may be different than that typically encountered (spectrum bias).2 However, the predictive value (both positive and negative) of a test is highly dependent on the prevalence of disease. This can be demonstrated using 2×2 tables with the test characteristics described by Light.