Dr. Mirvis is the Editor-in-Chief of this journal and
Professor of Radiology, Diagnostic Imaging Department,
University of Maryland Medical Center, Baltimore, MD.
This familiar expression is often uttered quietly, or not so
quietly, from the shadowy recesses of the Radiology reading room
and expresses the deep frustration of radiologists. Here is a
familiar story. One evening I encountered one of the critical care
attending staff in the hall outside my office. After a polite
acknowledgement and greeting, he asked me what I thought of a
thoracic CT scan obtained on one of his patients. I did remember
the scan from that morning and told him it was of limited technical
quality (due to rapid breathing), but showed no infiltrate or other
reason for impaired oxygenation. He looked surprised and expressed
his desire to know if there was evidence for pulmonary embolism. I
said that the CT was ordered only with the clinical information of
"infiltrate" and therefore a standard non-enhanced CT was
performed. A CT-angiogram (CTA) for pulmonary embolism was a
tailored examination requiring a specific request. He was
frustrated by the failure of "the resident" to request the correct
study. I was also concerned since the wrong test was performed and
valuable time wasted. Ultimately, a pulmonary CTA was negative.
This is a salient example of the common occurrence of "garbage
in, garbage out." Radiologists all know that indications for
requested imaging studies are not uncommonly limited in detail, are
not always appropriate for the study requested, or sometimes are
even frankly erroneous. When the procedure is invasive, a direct
conversation between clinician and radiologist is typically
required to make sure the procedure is indeed required or to
discuss details that may influence the precise examination or
intervention performed. For more routine studies, this direct
"consultative process" is often impossible and the radiologist must
act on the data provided on the study request. Improving access to
patient clinical information through the evolving "electronic
medical record" will help radiologists figure out what's really
going on when that information is not apparent from the request.
For now, in most venues, laboratory data and other radiology
reports usually serve as the only other medical record available
while studies are being reviewed. Given the usual heavy case
volumes, it is too time-intensive and impractical a method for
seeking information useful to the current study.
It would be of great help if more of our clinician colleagues
ordered imaging examinations with the same care they take in
obtaining blood cultures and doing most other procedures. Of
course, it should be done carefully since the results are optimized
in value by the quality of the medical information input. The
radiologist may often "read between the lines" and figure out the
real answer being sought, but it is best not to have to depend on
clairvoyance. Clinicians simply need to tell the imaging specialist
directly what they want the test to tell them following established
appropriate medical indications. Physicians who do not know the
patient should not, by proxy, order their studies. It must be
recognized that "transfer," "admission," "morning rounds,"
"post-op," or "pre-op" alone, among many commonly used indications,
do not merit the effort and cost of the study. Giving a correct,
valid indication can be performed in about the same time that an
erroneous or limited one can be provided. Maximize the utility of
the study! If one cannot think of a valid medical indication, there
is likely no need for the study.
If the clinical indication garbage coming in one direction can
be cleaned up, there will likely be less interpretation garbage
within the radiologists' report going back out. While this
editorial is clearly preaching to the choir, the choir members
should spread the gospel where needed.