|
|
Diagnostische Maßzahlen für die Palpation
As for any other diagnostic test, the usefulness of nodule palpation
should be characterized by the diagnostic measures of sensitivity (S),
specificity (C) and the positive and negative predictive value (PPV, NPV).
These measures are derived from the proportions of
true-positive, true-negative, false-positive and false-negative diagnoses
(PTP, PTN, PFP and PFN):
-
The sensitivity is the probability that an infected person is diagnosed "positive":
S=PTP/(PTP+ PFN).
-
The specificity is the probability that a non-infected person is diagnosed "negative":
C=PTN/(PTN+ PFP).
-
The positive predictive value is the probability that a test-positive person is infected:
PPV=PTP/(PTP+ PFP).
-
The negative predictive value is the probability that a test- negative person is not infected:
NPV=PTN/(PTN+ PFN).
The PPV and the NPV depend on the prevalence (V) of infection,
whereas the sensitivity and the specificity depend on the intensity of infection, as follows.
We can derive the sensitivity from the
probability that a nodule is found by palpation,
denoted p in the following.
For some first considerations, we assume that p=30% of nodules can be found by palpation.
The probability to palpate m nodules in a patient
harbouring a total of n nodules is given by the binomial distribution,
and the probability to find none of the n nodules is
The probability to find at least one of the n nodules is the sensitivity, given by
This simple relationship, plotted below, shows that the sensitivity
of nodule palpation for p=30% is very low for weakly infected patients
and exceeds 80% only for patients with five or more nodules.
For a given specificity, we can also derive the predictive values
which depend on the prevalence of infection.
The lower graph shows the relationship between the positive predictive value (PPV)
and the prevalence of infection for a specificty of 90%.
The PPV is less than 50% for a prevalence up to 20%.
If we want our test to yield at least a PPV of 80%, then,
nodule palpation was an adequate diagnosis tool only in regions
with a prevalence higher than 50%. These considerations become more pessimistic
if the specificity is lower, because false-positive diagnoses increasingly
determine the diagnostic result at low prevalence.
|
|
Figure: Diagnostic measures of nodule palpation under the assumption
that only 30% of the nodules are found and that the specificity is 90%.
Positive and negative predictive values also depend on the prevalence of infection.
The positive predictive value of nodule palpation is below 50% for a prevalence under 20%.
If a positive predictive value above 80% is regarded as a reliable measure,
nodule palpation is adequate only in regions where the prevalence of infection exceeds 40%.
(Source: Duerr et al., 2007)
|
|
Related pages:
Nodules,
Diagnosis by palpation of nodules.
Further reading:
Duerr HP, Raddatz G, Eichner M, 2007.
Diagnostic value of nodule palpation in onchocerciasis.
Transactions of the Royal Society of Tropical Medicine and Hygiene.
In press.
|
|
 Return to top of page
|
|
You are the
th visitor of our pages.
|
|
|
|
|
