> Basically my interpretation a few weeks down the line and how I now use
> them is: suspected PE/DVT: D Dimer is useful for its negative predictive
> value if your suspicion for PE/DVT clinically is low to moderate.
> So: D Dimer has a good negative predictive value when used in those with
> low to moderate probability.
> Sam Waddy
> A&E SHO Oxford
Hi,
Although these affirmations are not false, the logic behind is a bit loose.
Main caracteristics of any test are sensitivity and specificity, which are
relatively independant of prevalence, and not NPV or PPV, which are fully
dependent of prevalence.
Although specificity and sensivitity do vary for a specific disease
depending of the severity of the disease (eg: stress test do find more
easily coronarian disease for a severe three-vessel disease than for mild
circonflex disease, or cerebral CT is more sensitive for grade IV SHA than
for warning leak), it's an inherent caracteristic of the test, by which
different tests can be compared.
NPV and PPV, are, quite the contrary, fully dependent of prevalence.
Extreme example: if you have a 0% prevalence of a disease in a given
population, then NPV will be 100% no mather what's the test. For example,
if you got a 0% prevalence of malaria in some population, then the test:
"conducting a red car" will have 100% NPV -- quite good. Any person NOT
conducting a red car wont neither have malaria. Quite usefull.
So when talking about test, it's probably more logical to talk about them in
terms of sens and spec.
Cheers,
Alain VAdeboncoeur MD
ED coordinnator
Montreal Heart INstitute
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|