Dr. Liu
is a Professor at California State University, Northridge, CA;
Dr. Berman
is the Director of Radiology, and
Dr. Gray
is Chief, Department of Neuroradiology, Midway Hospital Medical
Center, Los Angeles, CA.
Conflict of interest disclosure: All authors are cofounders
of StructuRad LLC.
Portions of the data in this article were presented by Gerald
Berman, MD and Richard Gray, MD at the Integrating the Healthcare
Enterprise (IHE) Symposium of the Radiological Society of North
America (RSNA) 2001 Annual Meeting November 25Â30, 2001. The
results described in this article influenced the development and
commercial availability of the StructuRad Reporting tool. For a
complete abstract or copy of this paper and presentation, contact
the authors directly at info@structurad.com.
The use of structured reporting has long held the promise of
closing the digital and productivity divide in radiology that
exists between advances in modern computer-aided studies and
procedures and healthcare informatics and information technology
(IT) systems. The authors have undertaken an 8-year journey to
improve and replace their traditional dictation/transcription
process with a modern solution that delivers the productivity,
economic, and quality benefits of a complete digital reporting
solution. This article presents an analysis of the authors'
experience producing structured radiology reports at Midway
Hospital Medical Center, a full-modality, 150-bed community
hospital in Los Angeles, CA. The authors began developing
algorithms and report templates for generating structured reports
in 1995. The motivation for developing templates was to improve the
interpretation and reporting abilities of the non-neuroradiologists
who sometimes interpreted these studies. By 1996, we had developed
a general template structure for all modalities, although the
detailed work focused on structured reports for chest radiographs,
since they make up approximately 40% of the workload at our
hospital.
In 1997, we began to implement our report templates with
software from our picture archiving and communication system (PACS)
vendor, ALI Technologies (now McKesson Corp., San Francisco, CA).
From mid-1997 until early 1998, we continued to develop and deploy
templates for use in all 12 of the hospital's modalities. By
mid-1998, the development was complete, and, since that time, all
radiology reports have been produced with the "structured
reporting" system, for a total of more than 200,000.
Currently, Midway Hospital is in production use with the
third-generation of a structured reporting tool based upon the
experiences and ongoing productivity gains in report generation,
improved accuracy and consistency, report coding and billing, and
IT systems integration. This article provides a case history of
this process and a summary of the benefits of a structured
reporting system used for a wide range of modalities in a clinical
setting in a community hospital.
History of radiologic reporting systems
The most widely used method of radiology reporting is
transcription, in which a radiologist dictates findings that are
then transcribed by a transcription service. When the report is
returned from the transcription service, the radiologist verifies
the accuracy of the report and signs it. The report is then
distributed to other departments in the hospital. In some cases,
reports must be returned to the transcription service to correct
transcription errors. With transcription, the final output is a
printed report, or perhaps a text file, that is distributed outside
the radiology department.
An alternative approach is to capture findings as structured
data when the radiologist is observing study results. In this case,
there is no separate step for transcription. The radiologist enters
findings directly, enabling software to capture the findings as
structured data. The structured reporting system used in this study
is one of the radiology reporting systems that capture structured
data at the point of observation. Several other structured
radiology reporting systems are available that provide for
menu-based data entry, including ProVation MD (ProVation Medical
Inc., Minneapolis, MN), eDictation (eDictation, Inc., Marlton, NJ),
and SPIDER (Medical College of Wisconsin, Madison, WI). There are
also some integrated reporting systems available, such as the Rex
workstation (PointDx, Inc., Winston Salem, NC), which combines an
image viewing workstation with a structured reporting system; and
the reporting features within the PenRad Mammography Information
System (PenRad Inc., Plymouth, MN).
Description of the structured reporting system at Midway
Hospital
Environment
The radiology department at Midway includes 2 radiologists and
20 technologists who produce approximately 40,000 reports per year
for 12 modalities: X-ray, mammography, fluoroscopy, ultrasound,
computed tomography (CT), CT angiography (CTA), magnetic resonance
(MR), MR angiography (MRA), nuclear medicine, conventional
angiography, ultrasound angiography, and interventional radiology.
Radiology findings are printed as paper reports and are input
directly into an electronic patient record system.
Usage
The reporting process begins when a technologist enters
demographic data into the reporting tool system. This creates a
file for the study, which appears in the radiologist's inbox and
provides for the necessary worklist and workflow functionality. The
radiologist periodically checks the inbox, which may be sorted by
the attributes of the studies. The inbox contains new studies for
which no reports have been generated, as well as works-in-progress
for which reports have been started, but not yet completed.
The reporting system contains templates for the modalities and
anatomic sites for which examinations are performed in the
hospital. Each template is a complete knowledge base of possible
findings that are appropriate for the given modality and anatomic
site. Therefore, the reporting task for the radiologist is to
navigate the menus of possible findings and select the items that
correspond to observations from the given study.
Most of the data entry occurs by selecting items in a sequence
of cascading menus. Thus, selecting one menu item will cause
another submenu to be displayed with additional options. Any of
these menu items may contribute standard text to the report. Some
items cause an input box to pop up, so that the radiologist can
enter a value. For example, the radiologist might be prompted to
enter the size of a mass in centimeters. As the radiologist makes
selections and enters values, the system keeps a running record of
the corresponding report that is generated from the menu
selections. Unlike dictation, this provides immediate feedback of
the exact representation of the resulting report. The radiologist
may save a partially completed report. When the report is complete,
the radiologist approves the report, an electronic signature is
applied and the report may be distributed at that time (Figure
1).
Deployment
Prior to 1997, all radiology reports were produced through a
conventional transcription process following radiologist dictation
of findings. The dictation was transcribed by an outside
transcription service. When the report was returned from the
transcription service, the radiologist verified the accuracy of the
report, which was then distributed to other departments in the
hospital. In some cases, reports were returned to the transcription
service to correct transcription errors.
In 1997, the menu-based computer system was introduced, which
allowed the radiologists to report findings by selecting from menus
of all possible choices. During that year, approximately half of
the reports were produced through transcription and half were
produced through the structured reporting software. The original
software was developed and deployed over a period of 6 months
during 1997 and 1998. As templates for each modality and anatomic
site were developed, transcription was phased out and use of the
reporting application was phased in. The system continues to evolve
and is the basis for the current commercial version of the
StructuRad reporting tool currently available (StructuRad LLC,
Encino, CA). The current version, StructuRad 2.6, has been in
fulltime clinical use since November 2002 at Midway Hospital.
Since 1999, 100% of the radiology reports are produced with the
structured reporting tool software. After the initial deployment,
reporting templates continue to be expanded and improved based on
feedback from referring physicians and on new requirements
discovered by the authors and reporting radiologists at Midway
Hospital. These enhancements include: 1) developing additional
content within existing templates; 2) improving menu navigation
within templates; 3) revising the underlying reporting model to
make the language of the report generated sound more like standard
written English; and 4) better demographics, input, and overall
improved usability.
Summary of quantitative results
The results of our study are based on a comparison of operations
in 1996, when the hospital used only transcription, with operations
in 1999, when the hospital used only the structured reporting
system.
Transcription costs
Introduction of the new system provided direct operational cost
savings to the hospital by eliminating the cost of transcription
(Table 1).
Report turnaround time
The structured reporting tool resulted in significant reductions
in the amount of time required to submit a complete report to the
appropriate locations within the hospital, as well as outside the
hospital (Table 2).
Billing abstraction
The structured reporting system completely removes the coding
step from report processing, since the report templates already
include ICD-9 codes for clinical indications and CPT codes for
radiographic techniques. In addition to eliminating this
time-consuming step (Table 2), coding accuracy improved from 95%
when performed manually to 100% when captured through the
structured reporting system.
Transcription errors
With the elimination of transcription, structured reporting
eliminates the need to edit transcription errors. Errors occurred
in 13% of transcribed reports at Midway; with the structured
reporting system, no such errors occurred.
Incomplete reporting leading to rework
The structured reporting tool resulted in a dramatic elimination
of rework on the part of the radiologist, which had been caused by
incomplete procedure descriptions, inaccurate clinical information,
or incomplete protocol description. A total of 20% of reports from
1996 were found to contain incomplete or inaccurate descriptions
based on radiologist coding errors. No such errors occurred after
the implementation of the structured reporting system.
The consequences of reworking a report are very costly. When a
third-party payer returns a report to the billing company, this
delays payment for the procedure and causes additional work on the
part of the billing company. For >50% of these rejected reports,
the billing company can do the rework and resubmit the report to
the third-party payer. However, in the remainder of cases, which in
1996 was approximately 7% of all reports, the billing company
returns the report to the radiologist for rework, which adds to the
cost of producing that report.
Interruption of radiologist for results
Since the results are reported and distributed much more quickly
than they were using transcription, the structured reporting system
provided an unexpected benefit to radiologists. The number of
interruptions from referring physicians seeking results was reduced
dramatically, thereby saving radiologists' time. Our study revealed
an average of 4 interruptions per hour from physicians seeking
results in 1996, prior to the use of the structured reporting
system, and only 1 interruption every 2 hours (or 0.5 interruptions
per hour) once the system was in use.
Radiologist time to create a report
In comparing structured reporting with transcription, there was
no difference in time spent recording findings, which averages 5
minutes per report for both methods. However, there is a slight
overall advantage to the structured reporting system, because there
is no need to review or sign a report returned from transcription.
This results in an overall gain of 1 or 2 minutes per report.
Admittedly, the radiologist's familiarity with the report
template determines the speed with which menu selections can be
made. However, based on peer use and review, and due to the
intuitive nature of the system, it is our experience that a
well-trained user can record findings as quickly with a structured
reporting system as they can dictate findings for traditional
transcription.
Another factor that reduces the time spent recording findings
using structured reports is that findings are automatically
extracted to the impressions section of the report, so there is no
duplication of data entry. Upon completion of a particular finding,
the radiologist may mark that finding to appear in the impressions
section. The structured reporting tool automatically extracts the
primary information to be included in the impressions section.
Discussion
The authors have analyzed the structured reporting system that
has been in clinical use at a community hospital since 1997. Since
mid-1999, reports for all 12 modalities are produced exclusively
through a structured reporting system. Transcription is no longer
used for any radiology reporting. While it might seem surprising
that the use of a computer-based system is just as fast as
dictating, the authors' experience is that generating structured
reports provides several process improvements that offer this speed
and productivity.
In this environment, our results show that structured reporting
leads to: improved quality and consistency of reports by using
report templates and a menu-driven interface; improved response and
turnaround times by eliminating process steps, so the final report
is released directly from the radiologist; reduced costs by
eliminating transcription without increasing the time required by a
radiologist to record findings; and, finally, automation of the
billing code process for reports by embedding ICD-9 and CPT codes
in the report template, which provides reduced billing costs and
improved cash flow with less time elapsing before reimbursement
requests are sent to third-party payers. These benefits were
achieved even though the reporting system was not tightly
integrated with the hospital information system (HIS) and radiology
information system (RIS) at the time.
Conclusion
In the future, the authors expect any structured reporting tool
to follow evolving reporting standards, so that the reporting
system can be integrated with other hospital systems, improve
distribution times to other information systems through application
integration, and improve diagnoses by supplying context-sensitive
reference information, which can reduce the time for radiologists
to record their findings.
AR