Special B. Leonard
Holman,
MD
The Research That Based on a Survey To determine the nature of radiologic research as practiced by both radiologists and nonradiologist physicians, original imaging-related reports published in the clinical literature during the first 6 months of 1989 were classified and compared. Most of the research by radiologist first authors involved technology development: invention (12%), standardization (35%), and description (32%). Original reports by nonradiologist-physician first authors involved description (28%), standardization (16%), validation (12%), assessment of other technologies (17%), and assessment of pathophysiology (19%). There were few patient outcome or cost-related artides in either category. The greatest difference between the categories was the larger number of reports by the nonradiologists that assessed pathophysiology by functional rather than morphologic criteria. The author concludes that the research performed by radiologists is primarily technology development, that assessment of pathophysiology with radiologic techniques is performed primarily by nonradiologists and is described more often in the nonradiology literature, and that academic radiology programs need to extend their research efforts to include evaluation of technology and functional assessment of disease process. Index
terms:
partmental
Radiology
and
management
ologists,
history
research
#{149} Special
Radiology
.
1990; the
Department
Women’s 75 Francis
02115.
Received
quested cepted
March 8; revision April 9. Address
author. RSNA,
January
1990
and
and
radi-
radiologists,
Radiologists of the Literature’
T
character
HE
demic
Our
research
ingby
of research
radiology by
Radiologists
are
studies
asked
MATERIALS
to perform
to validate
ence
in laboratory as well?
of building
do,
a new
changes
and
give
we must have what research and how that
resources
culture
If we are to deal
sci-
is it a question
on existing
introducing departments?
with
or
within
our
direction
to them,
foundations
research has been as (a) the research do, (b) research that biologic and physical
of radiation
development
and
and
to the
improvement
of
techniques that use that energy (3), and (c) research that fits into a conceptual frame in which the nature of the work involves in some way the imaging third,
lows
disciplines (4). holistic, perspective
us not
only
It is this that al-
to categorize
the
re-
search that radiologists do but also to compare it with imaging research performed by nonradiobogists, to determine whether there are funda-
appropriateness
of Radiology,
Hospital
and
St.
Boston,
4, 1990; received
reprint
and
Harvard
April
requests
of those
re4; ac-
to the
for future
were were
surveyed.
The
To determine
the
nature
as practiced nonradiologist
cians,
reports
original
first
Radiology,
American
journals
pub-
6 months
of
radiology
1989
journals
Journal
of Roent-
genology, and Investigative Radiology. These journals were selected because they publish articles of general radiologic interest and are the official journals of the three
leading tions. New
American
radiologic
organiza-
The other clinical journals England Journal of Medicine,
nal of the American chives of Neurology,
Medical Cancer,
tion, Gastroenterology, and Joint Surgery,
The
were The The Jour-
Association, ArChest, CirculaJournal
of Bone of Urolo-
and The Journal the general medical Journal
journals and
of Medicine
The Journal of the American Medical Association) contained only 14 articles pertaining
to radiologic
clinical specialty and selected to encompass a wide range of clinical specialties that use imaging procedures as dijournals
research,
were
agnostic,
included
therapeutic,
and
investigational
tools. The clinical specialty journals were also selected because they listed department affiliations and educational degrees for all coauthors. All original reports published during the
6-month
period
involving
radiologic,
imaging, or radiotracer techniques were included in the survey. Original reports were defined as articles that had clearly stated objectives or hypotheses and specifically
articulated
methods
sections. Excluded from review articles, editorials, al and teaching material. The
to first
original and
and
the
reports
author:
thors
were
affiliation
logic research diologists and
the
were
classified
(a) members nuclear
results
survey were and educationas
of radiology,
medicine
depart-
ments with medical degrees; (b) members of other departments with medical degrees; and (c) nonphysicians. The first
boundaries,
prescriptions
during
imaging,
growth.
MA
revision
to write
METHODS
clinical
lished
gy. Because (The New England
these
a clean definition of in radiology should be definition differs from
reality. Radiobogic defined variously that radiologists is related to the
during a recent 6 months of and compared.
AND
Representative
so-
new
biomedical
If we
literature the first classified
reduced
greater reliance marriages.
techniques and, furthermore, are cabled to task when their assessment studies are deemed inferior (1). Research in radiology departments appears to be largely oriented toward technology development and assessment (2). Should we encourage
growth
Perspective
the clinical time period, 1989, were
are increas-
sharply
federal support and on industrial-academic phisticated
in aca-
is changing.
programs
affected
Do:
mental differences in the way the two groups perceive the boundaries of their research, to determine the
176:329-332
and School,
de-
reports
‘From Medical
radiologists,
#{149} Radiology
Radiology
Brigham
Reports
classified
and
on
educational
the
basis
degrees
au-
of the
listed
of radioby both raphysi-
published
in
Abbreviation:
AUR
=
Association
of Univer-
sity Radiologists.
329
in
the
article,
available. confirmed
when
this
Otherwise, by means
the Radiological ca or by direct
information
Society
was of
of North
telephone
physicians were excluded. Original reports written by these radiologists that appeared in the January through June 1989 issues of the Science Citation Index and in journal issues published during that same period were surveyed. The artides were classified as to journal origin:
was
classification of the directory
Amen-
contact.
The original reports were classified as either assessment or development of technology or the assessment of pathophysiology with radiologic techniques. Development of technology was further classified into its major components: invention, description, and standardization; assessment
into
of technology
was
its components: therapeutic
outcome,
assessment
of other
technologies.
components
were
invention:
the
defined
new
the
cost,
development
the
interpretation or
RESULTS
of a new
with
qualitative
the
proto a
determination
values
for
the
determination
tissue,
ance,
the development and
normal
of classification of quality
and
author diobogy months,
Validation: the determination of the diagnostic value of a procedure as measured by sensitivity, specificity, accuracy, or predictive value; Patient outcome: the effect of a diagnos-
ports logic
tic
physician author
test
on
patient
management,
or quality of life; Therapeutic outcome:
survival,
the
effectiveness
of
a therapeutic radiologic procedure; Cost: the financial impact of the test or cost-benefit
ratio,
cost per year procedure; Assessment
procedure,
of life added
by the
of radiologic
procedures
the
value,
of other
or the
test or
technologies:
the
effectiveness,
or efficacy
the new
techniques
or thera-
to break
in identifying
the
biologic,
ical, or pathologic
consequences
anisms of disease. Original reports
involving
the
of nonra-
radiologic
ground
use
to determine
diologic techniques, procedures, pies; Assessment of pathophysiology:
use of chem-
or mechmore
than
one category were tallied multiply. For example, an original report describing the invention and validation of a new technique was both categories.
also
classified
collected:
recorded Original
by the
morphology,
ogy or metabolism), or physics/computer. Arbitrary selection
separately reports
type
of information
function treatment, of the
(physiolchemistry,
journals
to be
for such cited
cent tion
a bias,
the liter-
including all journals by the Science Citation Index. Ten perof the membership of the Associa-
of University selected randomly listing.
330
we also surveyed
randomly,
Radiation
Radiology
#{149}
Radiologists (AUR) from an alphabetized oncologists
and
non-
was
June
1989,
was
the
a radiologist
was
the
contained
315 original
pertaining topics. The
to imaging first author
in 19 reports,
physician
coauthors)
in 204,
ra6
were
few
outcome-
or radiowas a ra-
radiologist
a nonradiobogist
inThere
or cost-related
an-
tides written by nonradiologist first authors. Reports by nonphysician first authors involved invention (37.5%) and standardization (72.5%) almost exclusively.
Of the
90 validation
studies
by ra-
diologist first authors, 57% involved comparison of two or more imaging modalities and 68% involved pathologic correlation (the remaining 32% used one of the competing modalities and/or clinical follow-up as the reference standard). Twenty-nine of these studies (32%) involved fewer than 20 patients. Twenty-two studies (24%) involved and included
20 or a control
22 studies,
more patients population.
14 were
prospec-
tive. Only five studies (6%) were prospective, involved more than 50 patients, and included control subjects. Most of the studies in all journals (71%) described morphology (Fig 2).
A substantially
re-
a nonradiobo-
(without
77% (22% of total articles) fewer than six patients.
Of these
first
in 90% of the articles in the journals. During the same the other clinical specialty
diobogist gist
A radiologist
in 486, a nonradiologist in 16, and a nonphysician
larger
number
of the
reports by nonradiologist-physician first authors described function (physiology or metabolism) than those written by radiologist first authors (24% vs 5%). A smaller number
of reports
involved
physics
or chem-
two (Fig 1). Thus, a radiologist was the first author in only 6% of the antides irt the other clinical journals. The vast majority of radiologic research reports involved technology
istry. The number of articles involving therapeutic outcome were similar for both groups. The greatest difference between imaging articles published by radiologist and nonradiobogist-physician first authors concerned the number
assessment
of reports
with a radiologist in 90, and a nonphysician
and/or
as a coin
development
(96%
of original reports in the radiology journals and 81% of original reports in the other clinical journals). Thus, whereas almost one-fifth of the reports in the other clinical journals involved assessment of pathophysiology, only 4% of the articles in the radiobogy journals involved the application of imaging techniques to understanding disease processes. The preponderance of the 505 oniginal reports by thors involved ment (invention
in were
surveyed possibly skewed the results because of bias reflecting the editorial policies and interests of the journals. To test ature
study.
journals
safety;
and
Radiology, American Journal of Roentgenology, and Investigative Radiology contained 540 original reports meeting the inclusion criteria
in 38. Thus,
dis-
of van-
investigations
January
first author physician
of and
Between journals
for the
of morobtained
procedure; the
eased systems,
subjective
findings
the radiologic
Standardization: quantitative
or
categorization
functional
journals.
and
or software of a technique
and
phologic
selected
These
application;
Description:
of Roentgen(b) all radi-
patient
as follows:
procedure, instrument, gram or the adaptation
Journal Radiology;
ology journals; and (c) all other clinical journals. Radiologic research reports were classified into the same categories as for
classified
validation,
outcome,
(a) Radiology, American or Investigative
ology,
tions, volved
radiologist technology [12%],
first audevelopdescription
[32%], and standardization [35%]) (Table). Of the descriptions, 36% (12% of the total articles by radiologist first authors) involved fewer than six patients. Few studies dealt with patient outcome (4%) or cost (less than 1%). The 310 original reports published by nonnadiobogist-physician first authors
involved
description
standardization (12%),
ogies
assessment
(17%),
physiology
and (19%).
(28%), (16%), validation of other technol-
assessment Of
the
of pathodescnip-
assessing
pathophysiobogy
with functional criteria as opposed to morphologic criteria. Virtually all reports (96%) by radiologist first authors assessing pathophysiobogy descnibed morphologic changes, whereas more than half of the reports (56%) by nonradiobogist-physician first authors evaluated function (physiology or metabolism) (Fig 3).
One
hundred
articles
were
written
by AUR radiologists; 79 were published in radiology journals (47 of which were in Radiology, American Journal tive
of Roentgenology,
Radiology),
and
or Investiga-
21 reports
were
in
other clinical journals. The distnibution of research categories was simibar for the reports by AUR radiobogists and for those by radiologist first authors tides involved ization, vention. quencies
ference bution
(Table). Of the 100 journal written by the AUR group,
an40%
description, 28% standard16% validation, and 13% inAll other categories had freof 6% or less. The only dif-
in research category distribetween reports by radioboAugust
1990
is, invention, 00 O
C omr-C
C
a
U U)
Ou
C
80
or
C