14 What is the sensitivity, specificity and positive
predictive value of an abnormal exercise test?
Vic Froelicher
While sensitivity (% of those with disease who have an abnormal
test) and specificity (% of those without disease who have a
normal test) are relatively independent of disease prevalence they
are reciprocally related and dependent upon the cut point or
criterion chosen for diagnosis. The positive predictive value of an
abnormal test (% of those with an abnormal test that have
disease) is directly related to the prevalence of disease. Another
way to compare the diagnostic characteristics of a test is by use of
predictive accuracy that is the percentage of total true calls (both
negative and positive). While it is affected by disease prevalence,
since diagnostic testing is usually only indicated when the pre-
test probability is 50% (i.e. a disease prevalence of 50%) this
measurement is a simple way of comparing test performance.
Meta-analysis of the exercise test studies with angiographic
correlates has demonstrated the standard ST response (1mm
depression) to have an average sensitivity of 68% and a
specificity of 72% and a predictive accuracy of 69%.^1 But most
of these studies have been affected by work up bias that means
that those with abnormal tests were more likely to be entered
into the studies to be catheterised. When work up bias is
removed by having all patients with chest pain undergo
catheterisation different results are obtained though the
predictive accuracy remains the same. In such a study we found
a sensitivity of 45% and a specificity of 85%.^2 It appears that
this is how the test performs in the clinic or doctor’s office.
However, the inclusion of clinical and other test results in
scores can increase the predictive accuracy of the standard
exercise test to nearly 90%.^3
RReeffeerreenncceess
1 Gianrossi R, Detrano R, Mulvihill D et al. Exercise-induced ST
depression in the diagnosis of coronary artery disease: a meta-analysis.
Circulation1989; 8800 : 87–98.
2 Froelicher VF, Lehmann KG, Thomas R et al.The ECG exercise test in a
population with reduced workup bias: diagnostic performance,
computerized interpretation, and multivariable prediction. Veterans