The reason your article was rejected has more to do with journal article
organization conventions than evidence-based medicine. The statements in
your "Methods" section are the type of deduced conclusions and advice that
are usually presented in a conclusion/discussion section at the end. What
by convention we expect in the methods section is a detailed description of
the methods you used in carrying out the study that allowed you to make your
deductions. Some of these types of details are inappropriately presented in
your "Discussion/Results" section. For these reasons your article should
have been rejected temporarily pending some organizational revisions.
From the standpoint of EBM the problems with your study are much more
complicated. While you obviously have diagnostic experience and good
instincts, and may well be correct in your advice, you do not show evidence
of adequate training in the formal analysis and presentation of diagnostic
studies. This is clear from the complete absence of the proper terms for
the fundamental parameters of diagnostic testing: sensitivity, specificity,
positive predictive value, negative predictive value, and a "gold" or
reference standard. Unfortunately it is not surprising that your medical
education may not have provided you this training, as good research
training, particularly diagnostic research training, is uncommon in medical
schools and residency programs. Such training, if it exists at all, tends
to come informally from mentors with good research skills. Fortunately it
is never too late to educate oneself with any of the several good books
available on this subject (see references below).
Diagnostic test analysis is more complicated than most practitioners
realize, and your study is not alone - most diagnostic study articles are
inadequate. Even with some self education, it would be wise to consult a
biostatistician with experience in diagnostic test analysis before starting
a study. Older and busier statisticians might want a consulting fee,
younger ones might be happy to consult just to get their name on the
publication.
As for the specifics of your study, it is a retrospective analysis of
archived data. While this design is not considered as reliable as a
prospective randomized controlled trial, it can nevertheless provide fairly
convincing evidence if it is carried out and described properly. I see two
major problems with your study. Your implication is that rib fractures were
missed in the symptomatic hemithorax films. But how do we know that these
alledgedly missed fractures really existed. There needs to be some way to
follow up all the cases to determine the true fractures; in short, a gold
standard. There could well be selection bias if most of the three-film
diagnoses were made by one or a few experienced sports injury doctors who
see alot of atheletes and rodeo riders who actually have alot of rib
fractures. Most of the symptomatic hemithorax films could have been ordered
by GPs who see alot of weekend handymen who have more bruised ribs than
fractures. In other words the less reliable tests may have been done
because more reliable tests were not perceived to be needed (not that this
is good medicine, it's just that it could explain your results). One way to
investigate this possibility would be to carry out multivariate analysis in
which some of the variables would be the practice center, the individual
doctor ordering the films, the specialty of the doctor, and the type of
patient. This is fairly complicated analysis and would require the help of
a biostatistician.
These are just some cursory, superficial observations concerning your study.
A more thorough examination might reveal other problems as well as strengths
of your study. Your conclusions may well be correct; but a higher level of
analysis and reporting would make them more convincing. Presenting
inconclusive data that raises more questions than are answered wastes the
data, wastes your time, and wastes the readers time. You appear to be onto
something here, and I encourage you to consult with a biostatistician,
rewrite and resubmit.
David Doggett,
Philadelphia, PA
Some good books on diagnostic test analysis (these are physician-level
books, not statistics text books):
Fletcher RH, Fletcher SW and Wagner EH (eds.) Clinical Epidemiology, the
Essentials; Williams & Wilkins, Baltimore, 1988.
Sox HC Jr., Blatt MA, Higgins MC and Marton KI (eds.) Medical Decision
Making; Butterworth-Heinemann, Boston, 1988.
-----Original Message-----
From: [log in to unmask] [mailto:[log in to unmask]]
Sent: Thursday, June 14, 2001 10:10 AM
To: [log in to unmask]
Subject: On laments about evidence-based medical practice
Worries about the lack of list-serve debate and doubts about
the worth or even the reality of evidence-based medical
practice are most disheartening.
In an attempt to fuel list discussion as requested by one of
you, is the following abstract a legitimate example of
EBM?
Dealing with the clinico-radiologic diagnosis of minor
chest trauma, the full manuscript was rejected by a free-
subscription journal because it showed "no methodology".
This may be so, but is it evidence-based by your lights?
While it is not clear in the abst ract, every claim, observ-
ation and technique mentioned in the MS is referenced.
With best regards....Karl Dockray
On Finding More Rib Fractures
I have never seen rib x-rays confirming that
bone has been taken by the Almighty to make his
first Adam a mate (Ref 1, Genesis,2:22). But,
I always have trouble in diagnosing simple
fractures with conventionally-made rib surveys.
Exposures following old teachings and hoary
custom are hard to interpret(2,3). Such studies
are limited to "the side of the chest that
hurts" and use filming techniques that black out
upper ribs and leave lower ones underexposed.
Examination times rise along with radiation
dose as exasperated doctors order repeat film
after repeat study to find out what is wrong.
Expenses climb even more as those same doctors
add regularly-exposed postero-anterior chest films to
each rib study. They do this because they are
taught that only a standard PA view with its
lighter exposure can show catastrophic lung
collapse.
Methods
To improve diagnostic accuracy, reduce patient
irradiation, save examination time and cut labor
costs, try the following: ask for three-film rib
series that guarantee coverage of all the ribs in
both sides of the chest. Insure that the upper-
most ribs are always in view as well as the
lowest ones. Demand exposure techniques that do
not black out those upper ribs made lucent by
the air of the lungs or leave lower ones hidden
by the thicknesses of the upper abdominal viscera.
Discussion/Results
A consecutive series of 121,055 patients film-
ed in several rural and semi-urban hospitals over
ten years held 1853 rib examinations. Seven-hundred
seventy of these were expressly ordered for one side
of the chest (i.e. the symptomatic hemithorax).
Reviews of 100 of these single-sided examin-
ations showed only thirty-nine fractures.
One thousand and eighty-three rib examintions
covered all of the chest from top to bottom and
side to side with three-film techniques. Sixty-
eight breaks were demonstrated in a one-hundred
case sample for a 57% improvement in fracture
detection.
Lastly, the 120,000 patient data-base was
checked for the raw totals of pneumothoraces(4).
One hundred and eighty-six cases of lung collapse
were listed only three of these occurred in the
1800 requested rib series. This paucity of cases
suggests that doctors bypass time-consuming rib
studies for profoundly-injured patients. Only
clients with minor chest complaints get full
rib examinations. It appears then, that these
people do not need supplemental PA views to
find lung punctures that are seldom present.
Conclusions
Routine three-film rib studies displaying
all the bone arcs in both sides of the thorax
with tailored exposure techniques reveal 57%
more fractures then do single-sided surveys.
Supplemental PA chests are not necessary to
find seldomly-occuring pneumothorax.
The new strategies reduce filming expense,
production times, radiation dose and the costs
of repeated examinations.
References
1. Genesis 2:22, p.9 in The Holy Bible, New
International Version, 1973 by the International
Bible Society published by the Zondervan Corpor-
ation, Grand Rapids, MI. pp 1950.
2. Bontrager KL and Anthony BT: Textbook of
Radiographic Positioning and Related Anatomy,
2nd ed, St. Louis, MO. C.V. Mosby Company, 1987.
3. Ballinger PW and Frank ED (eds): Merrill's
Atlas of Radiographic Positions and Radiologic
Procedures,St. Louis, MO, C.V. Mosby Company,1999.
4. Dockray, KT: Solo practice management:value
of a computerized reporting system. AJR 162:1439-
1441, 1994....and 15 others.
For those interested, the original draft and its
details are available at 1808 19th Street,
Lubbock TX USA 79401.
|