Healthcare and Medicine Reference
about one-third of patients fall into this undesirable category. By using
highly specific tests, we want to exclude from further care individuals who
do not have the disease and who would be treated then for something that
they do not have, that is, we try to avoid “treatment for nothing.”
When a test is highly sp eciic, its p ositive result rules in the diagnosis.
(The mnemonic SpPin reflects this property of a diagnostic test.) 16 A highly
specific test will rule out by its negative result patients who do not deserve
further care. Positive test results, whatever their sensitivity might be, make us
more certain that the patient has the disease. If the patient did not have the
disorder, a negative test result would confirm it.
A highly sensitive test would find by its positive result most of those
individuals who do have the disease of interest. This is a highly desirable
property in screening programs for disease; community medicine and public
health specialists aim for the maximum of cases to be detected to control
the problem. On the other hand, a clinician facing an individual patient may
be sure that, when using a highly sensitive test yielding a negative result in a
particular patient, the patient in question does not have the disease. In other
terms, when using a highly s ensitive test, its n egative result rules out the diag-
nosis. (The mnemonic SnNout reflects this property of a highly sensitive test.) 16
Predictive value of a positive test result tells the clinician what is
the probability that the patient really has the disease. We want to
be certain before the action, before doing something (treatment) that
would follow diagnosis. We want to be sure that the patient really
has the disease if tested positive for it. We are concerned by possible adverse
effects and cost, time, and other requirements for needless procedures in
patients who do not need these procedures or for whom they are eventu-
ally contraindicated. In our example, more than one-half of positively tested
patients would fall into this situation.
Predictive value of a negative test result indicates the probability that the
patient really does not have the disease if the patient is tested negative. The
clinician wants to have certainty before deciding not to take action (to treat).
In our example, the physician may be almost certain that doing nothing was
a good decision (95.4% in our case).
In clinical epidemiology terms, the Bayesian reassessment of the predic-
tive values of diagnostic and screening tests is based on linking the preva-
lence of the problem of interest (i.e., before testing) to the sensitivity and
specificity of the test in order to yield revised probabilities of the positive
and negative test results' predictive values.