Volume 137, Issue 1 , Pages 256-257, January 2009
Referent values and equipoise: Editors' notes
Article Outline
CTSNet classification: 11
To the Editor:
Lim and Dusmet,1 in a letter to the editors critiquing an article by Marra and colleagues,2 make 4 statements about referent values (accuracy, sensitivity, specificity, positive predictive value, and negative predictive value)3 that require clarification.
Statement 1: “Specificity (ability to rule in disease) is not the major focus for surgeons.”
Referent value specificity is the measure of the ability of a test to rule out disease. It is defined as the number of patients without the disease who are identified by the test (true-negative results) divided by total number of patients without disease (true-negative plus false-positive results).
Statement 2: “What is critical is sensitivity (the ability to rule out disease).”
Referent value sensitivity is a measure of the ability of a test to rule in disease. It is defined as the number of patients with disease who are identified by the test (true-positive results) divided by the total number of patients with the disease (true-positive and false-negative results).
Statement 3: “Remediastinoscopy is associated with 100% specificity. That in itself is expected … when tumor is seen in a biopsy it is invariably seen on the reference test.”
We believe these statements indicate the authors are confusing specificity with sensitivity. In general, screening tests have low sensitivity but high specificity. As work-up continues and the population is increasingly “enriched” with disease, sensitivity increases and specificity decreases.
However, for the surgeon dealing with this “enriched” population and who should not operate if there is persistent N2 disease after induction therapy, the referent value, negative predictive value, is most important. It is defined as the number of patients without N2 disease who are identified by the test (true-negative results) divided by the total number of patients identified by the test to be without N2 disease (true-negative plus false-negative results). A high negative predictive value (few false-negative results) permits the surgeon to confidently use the results of remediastinoscopy to plan further surgical intervention.
Statement 4: “Sensitivity truly starts at 50% (a test of no sensitivity or equipoise), not 0%.”
A referent value has merit throughout its range. A test with 0% sensitivity is unable to identify any patient with disease; one with 100% sensitivity identifies every patient with disease. Equipoise is equality of distribution, equilibrium, or counterbalance. Although we cannot verify this, we believe the writers might be confusing the referent value sensitivity with the area beneath the receiver operator characteristic, expressed as the C statistic, which is a graph of sensitivity versus 1 minus specificity. An area of 0.5 (C = 0.5) might be interpreted as “equipoise” because this is the expected relationship between groups whose testing has produced completely random results.
References
PII: S0022-5223(08)01640-1
doi:10.1016/j.jtcvs.2008.10.001
© 2009 The American Association for Thoracic Surgery. Published by Elsevier Inc. All rights reserved.
Volume 137, Issue 1 , Pages 256-257, January 2009
