|
Fracture Detection: A Possible Method to
Aid in Diagnosis and Improve Reporting Accuracy.
Authors: Joseph A. Gagliardi, M.D. (1),
Stacey M Nunberg, M.D. (2) Thomas Fisher, M.D. (1)
April 03, 2001
- St. Vincents Medical Center, Bridgeport, CT
- Long Island Jewish Medical Center, New Hyde Park, NY
Background: Providing accurate and immediate
interpretations of the radiological studies performed in the emergency
department (ED) is a challenge for many radiology residency programs.
Although the radiology resident’s preliminary reports are reviewed
periodically by attending radiologists, they are often the only
review available while the patient is being evaluated and treated
at the hospital.
During a quality improvement review at Saint Vincent's Medical
Center (SVMC), the question of unacceptable rates of missed fractures
and/or misleading preliminary reports in the ED was raised. A review
of the preliminary reports with missed fractures found that the
majority of missed/misleading diagnoses were due to radiology residents’
misinterpretations. A committee of Emergency Department attendings,
support staff, radiology attendings, residents, and technicians
met to address this issue. Possible contributing factors were identified
and solutions were suggested. In particular, the radiology residents
felt that the clinical histories provided on the requisitions were
often inadequate; they suggested that this served as a major causative
factor in missed/misleading diagnoses .
A quality-of-care audit published by radiology staff members from
the Massachusetts General Hospital reached a similar conclusion:
there was an increase in the number of significant misreadings of
radiographs in the ED when the clinical history provided was inadequate
(1). This study divided the errors into three levels of significance:
- high (having an important effect on patient care),
- moderate (probably having an effect on patient care), and
- low (having little effect on patient care).
Among all ED cases reviewed in this study, errors graded as "high"
increased from 20% to 27%, and errors graded as "moderate"
increased from 29% to 40%, when the histories provided were inadequate.
Furthermore, when a continuous quality improvement intervention
plan was implemented at Louisiana State University Medical Center
that mandated inclusion of pertinent histories on all radiograph
request forms, patient recall for misread radiographs decreased
by 42.9% (2).
Objective:
This study was designed to review the adequacy of the clinical
history for radiographic interpretations done by residents and emergency
department attending physicians. We also assessed the use of markers
to improve the accuracy of preliminary interpretations of skeletal
radiographs in the four anatomical areas that were most frequently
imaged for fracture detection (hand, wrist, foot, and ankle).
Methods:
We reviewed 114 consecutive request forms from the four most frequently
encountered anatomical areas with suspected fractures. These four
areas accounted for approximately 67% of all musculoskeletal radiographic
exams performed in our ED over a three month period. Each
history was evaluated by the first author to determine if they were
adequate. The percentage of fractures missed in the ED by residents
or EPs was also assessed. At the first author’s institution,
all ED radiographs are initially interpreted by radiology residents
from 8:00 a.m. until 11:00 p.m. After 11:00 p.m., the emergency
department attending physician interprets the radiographs.
In an attempt to lower our missed fracture rate in the four "high
frequency" imaging areas and further evaluate the impact of
clinical history on the detection of fractures, an additional 277
consecutive patients were asked to place a small 1.5 millimeter
(mm) radio-opaque X-spot marker (The Beekley Skin Marking System)
over the site of maximal tenderness. The hypothesis was that these
markers would allow one to focus on the site of maximal pain, thus
increasing fracture detection regardless of the histories provided.
After the marker intervention, preliminary interpretations of
skeletal radiographs performed in the ED were compared to the final
interpretation by a staff attending radiologist. In the case of
a discrepancy, two attending radiologists had to concur on the final
diagnosis. Patients were excluded if they had casts in place, fracture
or dislocation reduction studies without new trauma, or inadequate
radiographs due to patient positioning or filming techniques.
Results:
The first author reviewed the clinical history data on the 114
unmarked patients. Twenty-eight patients had acceptable histories
while 86 patients had unacceptable histories. Unacceptable histories
failed to target the site of maximum tenderness and neglected to
include reports of potentially contributory diseases. Examples of
unacceptable histories included words or phrases such as "pain",
"rule out fracture", "motor vehicle accident",
and "trauma."
The accuracy of the preliminary readings of emergency skeletal
radiographs before and after marker intervention are shown in Table
1. Prior to marker intervention, the efforts of the emergency
department physicians and radiology residents produced a 4.3 % false
negative rate (fractures missed were confirmed when the radiographs
were interpreted by a board certified staff radiologist, usually
on the following morning). Fractures missed in the first reading
often occurred in uncommon areas or had little displacement (Fig
1 and Fig
2).
A study from the University of Iowa reviewed ankle radiographs
from 433 patients and reported that fractures were missed in 4.2%
of cases (3). This result is almost identical to the total false
negative rate for all four areas monitored in our evaluation.
The false negative rate for fracture detection after the use of
markers was 3.2 %. This included 4 missed fractures out of a total
of 97 radiographic studies of the hand (4.1% false negative rate),
3 missed fractures out of a total of 72 radiographic studies of
the ankle (4.2% false negative rate) and 2 missed fractures out
of a total of 39 radiographic studies of the wrist (5.1% false negative
rate). No fractures were missed out of a total of 69 radiographic
studies of the foot.
One significant finding, not reflected in the data presented, was
that 25% of patients needed to place more than one marker on the
anatomical region radiographed. Markers were often placed at distant
sites (Fig
3 and Fig
4), thus focusing attention to multiple areas. This underscores
the need to do a thorough examination of the radiographs even after
a fracture is identified, and the potential value of attention to
patient report as part of a useful history.
There were several cases in this study (Fig
5a/b, Fig
6 and Fig
7) in which marker placement helped to avoid misleading conclusions
about the presence or absence of acute findings in patients with
previous trauma. Therefore, the placement of the markers considerably
reduced our false positive rate to 0.7% as compared to 2.6% when
no markers were present. In addition, there were instances in which
the markers helped detect subtle fractures (Fig
8, Fig
9 and Fig
10a/b).
There was a wide spectrum for missed fractures; some were quite
obvious and correctly located by the patient's markings (Fig.
11), while others were extremely subtle on routine radiographs
and had to be confirmed on MRI (Fig
12 a/b). There were no instances in this study where the placement
of a marker obscured a significant finding. At no time were additional
views needed because of the location of the markers.
Discussion:
The ability to attend to detail and focus on all structures radiographed
is critical for imaging studies. We have found that clinicians,
in an ED setting, for whatever reasons, may neglect to provide an
adequate targeted history for radiographic extremity studies when
a fracture is suspected.
An experienced, board certified radiologist should detect almost
all fractures of the bones and joints studied, and our data support
this notion. There were no instances, during this monitoring period,
in which one of the staff radiologists misdiagnosed a fracture.
The question was, how could we improve our resident’s and EP’s
performance?
The placement of a marker is a cheap and simple method to help
the resident or attending focus on the areas of maximal pain for
the patient. At our institution, the Beekley 1.5mm X-spots come
in boxes that contain 150 X-spots at a cost of approximately $60.00
per box. This equates to an additional cost of approximately 40
cents per case. It is our belief that marker use is not associated
with loss of underlying findings and enables a more accurate report
to be generated.
Although the literature supports the theory that adequate histories
are essential for fracture detection, the reality is that a complete
history may not be available at the time of review. Some type of
compromise, therefore, needs to be reached. We believe that short
histories, targeting a specific area, should be mandatory for all
radiographic study requests. Historical information, typically provided
by the emergency physician, often does not help localize the area
in question. More useful data would include, not only location of
maximal tenderness, but also a history of underlying disease such
as cancer or metabolic bone disease. It would seem that a targeted
history combined with the feedback offered by a marking system,
should improve fracture detection in the ED setting.
We invite any comments from our readers regarding different methods
to improve the accuracy rates for reading ED radiographs. Send correspondence
to editor@radiologyweb.com.
References and Selected Readings:
- Rhea JT, Potsaid MS, DeLuca SA. Errors
of Interpretation as Elicted by a Quality Audit of an Emergency
Radiology Facility. Radiology 1979; 132:277-80.
- Preston CA, Marr JJ, Amaraneni KK, Suthar BS. Reduction
of "Callbacks" to the ED Due to Discrepancies in Plain
Radiograph Interpretation. Am J Emerg Med 1998; 16:160-
- Brandser EA, Braksiek RJ, El-Khoury GY, Saltzman CL, Marsh JL,
Clark WA, Prokuski LJ. Missed Fractures on Emergency Room Ankle
Radiographs: An Analysis of 433 Patients. Emergency Radiology
1997; 4:295-302.
- Berlin L. Is a Radiologic Miss Medical Malpractice? An Ominous
Example. In: Malpractice Issues in Radiology. Leesburg Va.: American
Roentgen Ray Society, 1998; 43-46.
|
