|  
I'm sure everyone has heard the terms sensitivity,
specificity, positive predictive value (PPV),
negative predictive value (NPV), accuracy,
etc. They all sound alike and boring to
me, and I think most of us, have to stop and think
for a couple seconds before deciding which is
which. Let me refresh your memory.
The
main principle to keep in mind is that
most diagnostic tests are
not 100% reliable. You may
be tested once and diagnosed with the disease,
(even though you really don't have the disease),
and if you repeat the test, it may very well be
negative. For any diagnostic test to be
useful, it has to have a high sensitivity, specificty,
PPV, NPV and accuracy. Unfortunately, there
is only a limited number of tests which meet these
requirements and for this reason, we only have
a few tests recommended for screening, such as,
mammogram or colonoscopy.
Sensitivity
Probability
of finding people with disease.
Sensitivity
= True Positive / (True Positive + False Negative)
or if you prefer: TP / (TP
+ FN)
Specificity
Probability
of finding people with no disease.
Specificity
= True Negative / (True Negative + False Positive)
or TN / (TN + FP)
Positive
Predictive Value (PPV)
Probability
that if the test diagnoses you as having the disease,
it is really true... stop for a second and think
about the difference between PPV and sensitivity...
both sound similar at first, right?
PPV
= TP / (TP + FP)
Negative
Predictive Value (NPV)
This
is the exact opposite of PPV - the probability
that if test shows you as not having the disease,
you are really disease free.
NPV
= TN / (TN + FN)
Accuracy
We
all know what accuracy is, but do you remember
what the formula for accuracy is?
Accuracy
= (TP + TN) / (TP + TN + FP + FN)
|
