Piersante and group,
I would like to weigh in on this discussion by saying that I think you
are all right. Ok, so that sounds like a bad comedian's joke. Here is
the way I would look at these studies.
There are two kinds of studies of diagnostic tests. In the first, a
group of patients with the disease are compared to a group without and
both groups (we hope) are given the test being studied and the gold
standard test. Here we are starting with the disease (outcome) and
ending up with the risk factor (how many in each group are positive on
the test vs. the gold standard). This is clearly a case control study
(begin with disease and see who has the risk factor). However, for each
patient, the two tests (the one being tested and the gold standard) are
done (essentially) at the same time, making it a "cross sectional" study
also. Let's agree to leave that part out.
The second and stronger type of study follows a group of patients who
are all suspected of having the disease and having the entire group (we
hope) get both the test being studied and the gold standard test. Here
we are starting with the risk factor (how many in each group are
positive on the test vs. the gold standard) and ending up with the
disease (outcome or correct diagnosis). This is clearly a cohort study
(begin with risk factor and see who has disease). However, again for
each patient, the two tests (the one being tested and the gold standard)
are done (essentially) at the same time, making it a "cross sectional"
study also. Let's agree to leave that part out of this also.
I propose that we therefore call diagnostic test studies either case
control or cohort depending upon which study design the researchers use.
Clearly each has it's problems (selection of patients to be studied,
definitions of positive tests, selection of gold standard, measurement
biase, etc.) and in the end, I think that a cohort design is inherently
stronger and makes more sense clinically as long as the cohort is
sensibly defined (as those patients in whom the test ought to be done in
a real live clinical setting).
Hope this helps,
Cheers,
Dan
****************************************************************************
Dan Mayer, MD
Professor of Emergency Medicine
Albany Medical College
47 New Scotland Ave.
Albany, NY, 12208
Ph; 518-262-6180
FAX; 518-262-5029
E-mail; [log in to unmask]
****************************************************************************
>>> Andrew Jull <[log in to unmask]> 11/01/2004 3:54:54 PM >>>
Piersante
Like Paul Glasziou, my preference is for diagnostic cohort study;
intuitively it makes more sense as you are gathering a cohort of
participants.
The only rationale I can discern for using the term cross-sectional is
that
a diagnostic study is about a cross-section of disease presentation,
but it
will take some time to get adequate numbers even of consecutive
patients
enrolled and then followed up.
Another term I like but is little used is diagnostic case-control - not
a
good design but one that is often seen especially in early evaluations
of
diagnostic tests. Using the terms cohort and case-control
distinguishes
these designs in ways that cross-sectional, observational etc does
not.
Andrew Jull RN MA(Appl)
Research Fellow
Clinical Trials Research Unit
School of Population Health
Tamaki Campus
University of Auckland
Private Bag 92019 Tel: +64 9 373-7599 Ext
84744
Auckland Fax: +64 9 373-1710
New Zealand Email:
[log in to unmask]
-----Original Message-----
From: Piersante Sestini [mailto:[log in to unmask]]
Sent: Tuesday, 2 November 2004 7:50 a.m.
To: [log in to unmask]
Subject: Re: Diagnostic test studies
At 10.41 01/11/2004 -0500, Doggett, David wrote:
> Really, "diagnostic study" is a class unto itself. It should not
be
> encumbered with the stigma that"observational", "cross-sectional",
or
> other non-RCT terms
I mostly agree with what you say, but just "diagnostic study" would be
too
generic for my students to accept. They would argue that also a
case-control study could be used as a "diagnostic study".
Furthermore, "diagnostic study" here (and probably elsewhere) is
sometimes
used as a synonymous for "diagnostic test", so it may be even more
confusing
Paul Glasziou wrote:
>Though I like the name "diagnostic cohort" study, most other folk
working
>in diagnostics are unhappy with the time frame it implies. So
>cross-sectional analytic study is the most accepted term for the
>reasonable design of taking a series of patients with the same
presenting
>complaint and the applying some agreed gold standard to all cases.
What I don't understand is the meaning "analytic" here, as opposed to
"conventional" cross sectional studies. Laboratory and statistical
analyses
are performed in both kinds of study. What am I missing?
thanks
Piersante Sestini
|