Scott Richardson wrote:
>
> 5. What other questions arise in the acts of clinical diagnosis that are
> NOT well answered by 'usual' cross-sectional test accuracy studies? I
> would like to hear from others on this list.
I wouldn't call them "usual', at least in my setting (respiratory
diseases). Perhaps "optimal' describes it better.
As you already pointed out, in many cases there aren't optimal studies
available.
But I think that you are missing my point, which was not which type of
question is more effective to get the clinical answer (we can probably
get to the conclusion that they are all equivalent at the end), but
which one is more effective in catching the interest of medical students
(still retaining effectiveness).
Of course this could be something to be addressed experimentally.
>
> 6. Should foreground questions about diagnostic tests always include a
> specification of pretest probability in the 'P' segment? An example: "in
> patients with leg swelling and low to intermediate probability of DVT ..."
This is a good point.
This is similar to have the severity of disease in the "P" for a
therapeutic question.
I usually don't, unless it is clear that the stage of disease may affect
the effectiveness of the test or treatment, so I expect that such
patients are addressed by separate studies. However, I admit that I
always keep it on my mind, to get over it later in the "application"
step: can I apply this evidence to my patient?
So, having it explicitly on the query since the beginning would be a
good idea, provided that you are prepared to temporarily simplify your
query if there is a chance that having something like "low risk" in the
search strategy would increase the chances of missing relevant studies,
coming back to the full question later.
cheers,
Piersante
|