UNITED STATES
NUCLEAR REGULATORY COMMISSION
OFFICE OF NUCLEAR MATERIAL SAFETY AND SAFEGUARDS
WASHINGTON, D.C.  20555

September 12, 1994


NRC INFORMATION NOTICE 94-65:POTENTIAL ERRORS IN MANUAL BRACHYTHERAPY DOSE
CALCULATIONS GENERATED USING A
COMPUTERIZED TREATMENT PLANNING SYSTEM


Addressees

All U.S. Nuclear Regulatory Commission medical licensees.

Purpose

NRC is issuing this information notice to alert addressees to an event
involving errors in the calculation of dose tables and isodose curves for
several brachytherapy treatments and to the potential for errors in
computer-generated dose calculations for brachytherapy treatment plans.  It
is expected that recipients will review the information for applicability to
their operation and consider action, as appropriate, to avoid similar
problems. However, information contained in this notice does not constitute
new NRC requirements; therefore, no specific action nor written response is
required. Description of Event

Recently, NRC was notified of several events, at two NRC-licensed facilities,
involving multiple patients receiving manual cesium-137 (Cs-137) brachytherapy
treatments.  The radiation dose administered to eleven patients was calculated
to be in excess of the dose prescribed by the authorized users, by 5 to 30
percent.

The licensees used a Theratronics Theraplan/TP11 treatment planning system to
perform the dosimetry calculations for linear Cs-137 sources.  Through a
detailed review of backup versions of the data files and discussion with
Theratronics, the licensees determined that at some point before September
1992, the treatment planning system users re-entered data for Cs-137 as linear
sources because a data file had been inadvertently deleted from the system.
The Theraplan system, as supplied, contains sample data for radium-226
(Ra-226), iodine-125, and palladium-103 as linear sources, but no similar data
for Cs-137.  The user must enter 10 parameters to characterize linear sources
other than those supplied by the vendor.  The user manual and software offers,
as an alternative, direct entry of a data table from a published source.
However, due to a known "bug" in the version of the program used by this
licensee, that option was not available.  The computer-generated dose tables
that were computed using erroneous data were in error by as much as 20 to
25 percent.

When the treatment planning system was originally set up, the vendor's
representative assisted the licensee in entering the treatment parameters, and

9409070084. IN 94-65
September 12, 1994
Page 2 of 5


the correct values were used.  However, based on information provided by
Theratronics, review of data files and testing of the software, it appears
that when the users re-entered data for the Cs-137 source files in 1992, they
entered the value of zero for the filter attenuation coefficient (to correct
for attenuation by the source capsule).  The software recognized the keyboard
entry of zero as selection of an option to use an existing non-zero default
value.  The users relied upon the screen prompts to reenter the data.

In the case at hand, the software program displayed the value "0.0" both to
the user on the screen and in the hard copy printout of the linear source data
table entered by the user.  The information displayed to the user on the
computer monitor for this program did not indicate that the "0.0" was, in
fact, a default value.  The instructions in the technical manual regarding the
use of default values were in a chapter on external beam treatment planning.
The user was not aware that a default value selection option existed in this
portion of the program nor that the value zero would not be accepted as a
valid numerical entry.

Data tables for radium with a platinum filter are commonly used for non-radium
sources at some facilities when the sources or treatments are described in
milligrams-radium-equivalent (mgRaeq).  The Cs-137 line sources with stainless
steel filters (source encapsulation wall material) are comparable to Ra-226
sources filtered with 0.5 mm platinum.  In this case, the users entered a
gamma constant that was equivalent to Ra-226, as filtered by 0.5 mm of
platinum.  However, because of the default value selected for the attenuation
factor when "0.0" was entered, a non-zero value was selected.

The users erroneously checked the computer-generated dose tables against
along-and-away dose tables (Quimby tables) developed for Ra-226 sources
filtered by 1.0 mm of platinum.  There was a failure to identify the error in
the computer-generated dose values that were lower than would have been
generated using correct parameters to characterize the Cs-137 sources.  If the
correct table (for Ra-226 sources filtered by 0.5 mm of platinum) had been
used to check the calculations, the discrepancy in the computer-generated dose
values would have been identified and could have been adjusted by adjusting
the data elements.  Coincidentally, the net result of the user's data re-entry
and use of an incorrect data table for verification, was that the dose rate
.IN 94-65 September 12, 1994
Page 3 of 5


tables calculated by the treatment planning system appeared accurate within
approximately 5 percent.

One of the licensees discovered, through routine verification of a patient
treatment plan, that the computer-generated dose table had values lower than
expected when compared with published dose tables for Cs-137 (Krishnaswamy
tables).  After subsequent review and investigation, the licensee detected
the error in the selection of one parameter used to define the physical
characteristics of the Cs-137 sources.

Although both licensees had a written Quality Management Program (QMP), the
treatment planning system that was involved in these events was owned and
located at a cancer center that is jointly financed by the two licensees.
Because the individual who performed the calculations was employed by the
cancer center, neither licensee had detailed procedures to address the
treatment planning as part of the QMP.  The licensees assumed that the
authorized users had responsibility for oversight of all aspects of the
patients' treatments and that the licensee need only be aware of those aspects
of the treatment actually conducted at the licensee's facility.

Discussion of Root Cause and Contributing Factors

An NRC team consisting of two risk assessment engineers, a medical physicist,
and a radiation oncologist identified multiple causes of this event,
including: the selection of an inappropriate reference table to check the
computer-generated output dose calculations; use of the same incorrect table
during the manual checks of the treatment plans; the inadvertent use of the
default function for the filter attenuation factor used by the planning
computer; lack of an independent review of treatment plans; lack of formal
policies to control or record the modification of the planning software; lack
of clear, formal policies governing the use of the computerized planning
system; lack of oversight by the Radiation Safety Officer (RSO) or Radiation
Safety Committee (RSC) at either hospital over the treatment planning for the
brachytherapy procedures performed at their facilities; and poor software
design and documentation.  The team identified software deviations from many
accepted standards of user-computer interface design.

Root Cause:  The root cause of the brachytherapy events was a failure to
adequately validate the data used in the computer planning system.
Specifically, an error was made in performing an independent verification of
computer-generated dose tables and patient treatment plans.

Software and Erroneous Data Files:  Although the validation error in the use
of an inappropriate reference table was the root cause of the brachytherapy
errors, there was also an error in the data parameter entry such that the
computer-generated dose values were too low.  According to the licensee, this
problem has been corrected in subsequent versions of the software.
.IN 94-65
September 12, 1994
Page 4 of 5


A meeting was held between NRC and FDA to discuss the computer software
problems associated with these events.  At the time of the incident, FDA had
an Import Alert in effect on the Theraplan treatment planning system resulting
from a review of the manufacturer's Good Manufacturing Practices.  However, on
July 26, 1994, FDA cleared the Theraplan V05B/TP-11 V09B treatment planning
system which will allow modifications to be made to existing software, as well
as the import of new software.  In addition, FDA is working with NRC in its
review of this incident.

Other Contributing Factors:  There were several other contributing factors as
described above.  There was no independent review or check of the treatment
plans until the patient treatment in which the error was identified.
Secondly, there were no formal policies and barriers to control or record the
modification of the software or governing the use of the computerized planning
system.  Formal policies on file modification, and password or key protection
may have prevented the loss of the original Cs-137 parameter set.  In
addition, formal policies governing the use of the planning system could have
included mechanisms to handle inadvertent data loss such as use of backup or
archived files or use of the technical manual for data reentry.  Finally,
there was little or no oversight by the RSO or RSC at either hospital over the
treatment planning for the brachytherapy procedures performed at their
facilities.  Furthermore, the inspection findings revealed that the RSO/RSC
for both licensees appeared to act as if the manual-afterloading treatment
planning performed at the cancer center did not fall under their hospitals'
NRC licenses.

General Discussion

This case highlights the importance of verifying data entry and performing
independent checks of computer-generated dose tables and treatment plans.  In
addition, when data files are deleted, by whatever means, users should
consider comparing data entry with older versions of the file, if backup files
are available, to provide a second means of verifying data entry.  This case
also emphasizes the need for the manufacturer to provide adequate training for
the user, regarding the use of the software in order to understand the logic
of programs used in treatment planning systems and the need to be familiar
with the conventions used by any software manufacturer for data entry.
Although there were instructions in the technical manual indicating that when
a default value was available as an option that entering zero or using the
return key would result in use of the default value, they were only described
in a chapter regarding external beam treatment planning.

Licensees using the Theraplan treatment planning system are encouraged to
review computer-generated dose tables for linear sources other than Ra-226, or
other sources for which data files have been manually entered, and to review
the parameters currently in use on the treatment planning system, to verify
their accuracy.  In addition, users should ensure that they are familiar with
the conventions used for accepting default values and with all parameters for
which default values exist.      .IN 94-65
September 12, 1994
Page 5 of 5


Licensees are also reminded of the importance of appropriate oversight of the
licensee's radiation safety program and QMP.  Radiation therapy has changed in
recent years and frequently involves multiple individuals from potentially
separate organizations in the planning and administration of radioactive
material to patients.  In addition, if authorized users treat patients at more
than one facility, each facility might have different QMPs.  It is critical
that licensee management ensure that individuals are properly instructed in
the licensee's written QMP and that there are adequate mechanisms for
oversight in place to determine that the QMP is effectively implemented by the
appropriate individuals to provide high confidence that byproduct material is
administered as directed by the authorized user.

This information notice requires no specific action nor or written response.
If you have any questions about the information in this notice, please contact
the technical contacts listed below, or the appropriate NRC regional office.


                        ORIGINAL SIGNED BY


Carl J. Paperiello, Director
Division of Industrial and
  Medical Nuclear Safety
Office of Nuclear Material
  Safety and Safeguards

Technical contacts:  Patricia K. Holahan, NMSS
                     (301) 415-7847

                     Linda L. Kasner, RIV
                     (817) 860-8213

Attachments:
1.  List of Recently issued NMSS Information Notices
2.  List of Recently issued NRC Information Notices