AJR 2003; 181:1491-1493
© American Roentgen Ray Society
Multiinstitutional Computer Database for Recording Nonvascular Imaging-Guided Interventions
William W. Mayo-Smith1,
Mahesh V. Jayaraman,
Roger S. Han,
Damian E. Dupuy and
Jonathan S. Movson
1 All authors: Department of Diagnostic Imaging, Brown Medical School, Rhode
Island Hospital, 593 Eddy St., Providence, RI 02903.
Received March 31, 2003;
accepted after revision May 30, 2003.
Address correspondence to W. W. Mayo-Smith
(wmayo-smith{at}lifespan.org).
Abstract
OBJECTIVE. We sought to describe a multiinstitutional database
created for tracking CT and sonographically guided interventional
procedures.
CONCLUSION. The database, created using commercially available
software, has been placed on the secure hospital internal network for easy
access from two institutions. More than 1,000 separate interventions have been
added. The data may be queried and filtered for quality assurance and research
purposes.
Introduction
Imaging-guided intervention for both diagnostic and therapeutic purposes
has increased over the past decade
[1–4].
Procedures must be tracked for patient follow-up, quality improvement, and
research purposes. Although computer databases have previously been described
for tracking interventional procedures
[4,
5], we know of no reports of a
database that has been applied across a network at more than one institution.
The purpose of this report is to describe a database for tracking nonvascular
interventional procedures that was created using commercially available
software and placed on a server accessible from multiple institutions.
Materials and Methods
We created a computer database using Access (Microsoft, Redmond, WA).
Separate fields were included for general demographic information, including
the patient's name, date of birth, institution, medical record number, date of
procedure, location of patient (inpatient or outpatient), type of procedure,
referring physician, attending radiologist, imaging fellow, and resident.
Patient history relevant to the procedure was also documented, including
history of carcinoma and surgery. The type of anesthesia used, procedure room
time, and CT fluoroscopy time were also documented. For all procedures, both
immediate and delayed complications were documented as predefined fields. For
example, data on complications related to lung biopsies might include whether
the patient had a pneumothorax, whether the pneumothorax required chest tube
placement, and whether the patient was admitted to the hospital for
observation.
Imaging-guided interventions were divided into one of four
procedure-specific categories: biopsy, abscess drainage or aspiration,
radiofrequency ablation, and "other." Each of the four major
categories included multiple subcategories relevant to that procedure. For the
biopsy category, we documented the organ biopsied, the size and number of
lesions, the type and gauge of needles, the technique of the biopsy (coaxial
versus tandem), the number of passes, and the presence of an on-site
cytopathologist. In addition, the operator's prebiopsy suspicion of malignancy
was recorded. Fields were also created for pathologic and cytologic results of
the biopsy. New fields can be added without significant reprogramming and do
not affect data already residing in the database. Angiographic procedures were
not included in this database because they are performed in a division
separate from the abdominal imaging and interventional section at our
institution.
For the abscess drainage category, fields were created to prospectively
record the size and location of the collection, the type and size of catheter,
the technique (Seldinger vs trocar), the volume of aspirate, and the
predrainage likelihood of infection. A field was created in the database to
document the microbiologic results of the drainage and the duration of
catheter placement.
For radiofrequency ablation procedures, the organ treated and the size,
location, and number of lesions treated were recorded. In addition, the type
of radiofrequency electrode, number of treatments, duration of treatment,
current, wattage, impedance, and maximum posttreatment temperature were
documented.
The physician performing the procedure entered all data on a hard-copy data
sheet by circling relevant categories. The sheet was designed to minimize
free-text entry. Circling the appropriate fields on the data sheet took
approximately 20 sec of physician time at the conclusion of the procedure, and
this extra time requirement had no subjective effect on workflow. One standard
sheet served for all patients and procedure types. The hard-copy sheets were
stored in a three-ring binder and were subsequently entered into the
electronic database at weekly intervals by a technologist aide who had been
instructed in data entry. Data entry by the aide took approximately 2 min per
procedure. Data acquired after the procedure (delayed complications,
pathologic and microbiologic results) can be added to the database at any
time.
In designing the database, we chose Access because it offers relational
database capabilities and has built-in multiuser capabilities. The database
was created in approximately 40 hr by one of the authors who has extensive
experience with computer programming and Access. In our design, patient
information was in one table, and details of aspiration or drainage, biopsy,
and ablation were in separate tables. Links were then created among the
relevant tables. This design is more efficient than traditional flat-file
databases and minimizes duplication of data entry. Searches are also more
efficient with a relational database, and more complex queries can be
performed. In addition, the built-in security and network capabilities of the
software allow multiple users to enter and search data simultaneously across
the wide-area network. The major disadvantage of relational databases is the
complexity of setting them up and maintaining them. However, the initial time
investment is recovered later, when complex searches can easily be performed
that would not be possible on a flat-file database.
The database was stored on a secure server on the networks of the two
hospitals. For security purposes, the data were not accessible from outside
the hospitals or via the Internet. Access to the database file on the server
was by means of our standard secure network login, which incorporates minimum
password length and other security measures in compliance with the Health
Insurance Portability and Accountability Act (HIPAA) regulations
[6]. All physicians and staff
involved with data access were credentialed at both hospitals. All patients
sign an informed consent form before each procedure that includes a release of
material for educational and scientific purposes.
Patient anonymity is protected by several mechanisms. Each patient
encounter is assigned a unique patient identifier in the database that is
separate from their medical record number and other patient identifiers.
Access to the database is restricted to physicians and staff at the hospital
via the HIPAA security measures just mentioned. Finally, retrospective
research projects that use information from this database require a separate
institutional review board approval from our institution for each project.
Results
Over a 24-month period 1,105 records have been entered from two
institutions. The main data entry form, including patient demographic fields,
is shown in Figure 1.
Procedure-specific fields are shown in Figures
2A,2B,2C.
Queries to the database can be performed using search functions, and reports
can be generated as shown in Figure
3. For example, we can easily search the database for our
diagnostic yield of lung biopsies performed for lesions smaller than 1 cm and
analyze them further for yield by needle type, number of passes, and attending
radiologist. Complications can be analyzed for quality assurance purposes in a
similar fashion by procedure type, such as the incidence of pneumothorax in
patients undergoing lung biopsy.
View larger version (73K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1. —Full-screen image of main data-entry form. Main form is
subdivided in two sections: patient demographics are at top of form in black
background area, and procedure-specific information is located below, in
gray.
|
|
View larger version (31K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2A. —Images of procedure-information screens that are available
for biopsy, aspiration or drainage, and radiofrequency ablation. Fields change
depending on type of procedure performed, preventing clutter and simplifying
data entry. Biopsy form includes pathologic–cytologic information.
|
|
View larger version (53K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2B. —Images of procedure-information screens that are available
for biopsy, aspiration or drainage, and radiofrequency ablation. Fields change
depending on type of procedure performed, preventing clutter and simplifying
data entry. Drainage form has microbacterial result fields.
|
|
View larger version (54K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2C. —Images of procedure-information screens that are available
for biopsy, aspiration or drainage, and radiofrequency ablation. Fields change
depending on type of procedure performed, preventing clutter and simplifying
data entry. Radiofrequency ablation form has information on tumor size and
specific technique.
|
|
View larger version (68K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3. —Image of search and report form. Standard search capabilities
allow user to search database for any combination of factors. Search and
report form shown in this figure lists all lung biopsies performed for lesions
smaller than 1 cm at one site. Double-clicking on entry opens form containing
more detailed information.
|
|
Discussion
We have developed and placed onto our secure internal hospital network a
software program that can be accessed from two institutions. This arrangement
enables the sharing of information on interventional procedures by qualified
personnel from either hospital. Data-entry forms have been simplified to
minimize free-text entries and save physicians' time. Data can be entered into
the computer by support personnel, further increasing physicians' efficiency.
This database will be useful for performing quality assurance and research
projects and is less expensive and more easily customized than commercially
available database programs. A blank template of our program will be made
available by the corresponding author on request.
The design of this system has its limitations, however. Ideally, the
database should be a true client-server database, hosted on an internal
server, with a Web-based front end. This would facilitate data entry from any
workstation and would also include powerful auditing tools. However, the time,
effort, and cost to develop this type of solution are much more extensive and
would require additional resources at a time when many academic departments
are facing physician shortages, financial pressures, and more limited
information technology support. Although the current computer database is
imperfect, we believe it is preferable to not having a computer database.
In conclusion, we described the use of commercially available workstation
database software to produce a multiinstitutional database. This database
allows the tracking of nonvascular interventional procedures performed at two
institutions. The use of databases such as this can facilitate research and
quality assurance.
References
- Hopper KD. Percutaneous, radiographically guided biopsy: a history.
Radiology1995; 196:329
–333[Free Full Text]
- Mueller PR, vanSonnenberg E. Interventional radiology in the chest
and abdomen. N Engl J Med1990; 322:1364
–1374[Medline]
- Sheafor, DH, Paulson EK, Kliewer MA, DeLong DM, Nelson RC.
Comparison of sonographic and CT guidance techniques: does CT fluoroscopy
decrease procedure time? AJR2000; 174:939
–942[Abstract/Free Full Text]
- Hahn PF, Gervais DA, O'Neill MJ, Mueller PR. Nonvascular
interventional procedures: analysis of a 10-year database containing more than
21,000 cases. Radiology2001; 220:730
–736[Abstract/Free Full Text]
- Hahn PF, Lee MJ, Gazelle GS, Forman BH, Mueller PR. A simplified
HyperCard data base for patient management in an interventional practice:
experience with more than 4000 cases. AJR1994; 162:1443
–1446[Abstract/Free Full Text]
- Centers for Medicare and Medicaid Services (CMS) Web site.
Available at:
www.cms.hhs.gov/hipaa/.
Accessed March 5, 2003
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
Top
Abstract
Introduction
