Scott Richardson wrote:
> How fun to see this discussion about application of PICO(T) format to
> questions about diagnosis and diagnostic testing! As usual, I've learned
It is great to have you on this thread!
> a lot. Much as I'd like to be agreeable, I find myself disagreeing with
> the idea that there's only one single correct way to structure these.
> Let's try some examples:
Although I certainly agree that every problem may have more than one
solution, I am a bit unsure about your model a:
> P = pts with swollen leg and possible DVT
> I = D dimer test result negative
> C = reference standard
> O = ability to exclude DVT
Here the C and the O are really the same thing: you define DVT by the
reference standard. So you could just say : O - ability to exclude DVT
as defined by a positive reference standard", or just: "ability to
exclude the presence of a positive reference standard" leaving the C for
a comparison, such as not doing the test (i.e. pretest probability) or
having a positive test.
The second point is that it tends to move the "heart" of the problem to
an evaluation of likelihood ratios rather than of probability of
disease. Many of my students are quite epidemiphobic, and they listen
better and may even get interested about pre- and postest probability of
disease than about LRs. Even after understanding what LRs are and how
they are used, still they may need to wait until computing pretest and
postest probability before judging the usefulness of a test. I find this
quite reasonable and on line with available literature about
understanding better natural numbers than intermediate or composite
parameters (Hoffrage U, Lindsey S, Hertwig R, Gigerenzer G. Medicine.
Communicating statistical information. Science. 2000 Dec 22;290(5500):2261-2
The good thing about that model, though, is that it helps to build a
literature search strategy, clearly separating the content of each line:
"swollen leg AND d-dimer AND reference standard AND DVT" (+ all the
stuff that Pubmed clinical queries would append to select the relevant
studies). This could be very helpful.
cheers,
Piersante Sestini
|