Errors in radiation oncology: a study in pathways and dosimetric impact.

Eric E. Klein, Robert E. Drzymala, James A. Purdy, Jeff Michalski

Research output: Contribution to journalArticlepeer-review

51 Scopus citations


As complexity for treating patients increases, so does the risk of error. Some publications have suggested that record and verify (R&V) systems may contribute in propagating errors. Direct data transfer has the potential to eliminate most, but not all, errors. And although the dosimetric consequences may be obvious in some cases, a detailed study does not exist. In this effort, we examined potential errors in terms of scenarios, pathways of occurrence, and dosimetry. Our goal was to prioritize error prevention according to likelihood of event and dosimetric impact. For conventional photon treatments, we investigated errors of incorrect source-to-surface distance (SSD), energy, omitted wedge (physical, dynamic, or universal) or compensating filter, incorrect wedge or compensating filter orientation, improper rotational rate for arc therapy, and geometrical misses due to incorrect gantry, collimator or table angle, reversed field settings, and setup errors. For electron beam therapy, errors investigated included incorrect energy, incorrect SSD, along with geometric misses. For special procedures we examined errors for total body irradiation (TBI, incorrect field size, dose rate, treatment distance) and LINAC radiosurgery (incorrect collimation setting, incorrect rotational parameters). Likelihood of error was determined and subsequently rated according to our history of detecting such errors. Dosimetric evaluation was conducted by using dosimetric data, treatment plans, or measurements. We found geometric misses to have the highest error probability. They most often occurred due to improper setup via coordinate shift errors or incorrect field shaping. The dosimetric impact is unique for each case and depends on the proportion of fields in error and volume mistreated. These errors were short-lived due to rapid detection via port films. The most significant dosimetric error was related to a reversed wedge direction. This may occur due to incorrect collimator angle or wedge orientation. For parallel-opposed 60 degrees wedge fields, this error could be as high as 80% to a point off-axis. Other examples of dosimetric impact included the following: SSD, approximately 2%/cm for photons or electrons; photon energy (6 MV vs. 18 MV), on average 16% depending on depth, electron energy, approximately 0.5 cm of depth coverage per MeV (mega-electron volt). Of these examples, incorrect distances were most likely but rapidly detected by in vivo dosimetry. Errors were categorized by occurrence rate, methods and timing of detection, longevity, and dosimetric impact. Solutions were devised according to these criteria. To date, no one has studied the dosimetric impact of global errors in radiation oncology. Although there is heightened awareness that with increased use of ancillary devices and automation, there must be a parallel increase in quality check systems and processes, errors do and will continue to occur. This study has helped us identify and prioritize potential errors in our clinic according to frequency and dosimetric impact. For example, to reduce the use of an incorrect wedge direction, our clinic employs off-axis in vivo dosimetry. To avoid a treatment distance setup error, we use both vertical table settings and optical distance indicator (ODI) values to properly set up fields. As R&V systems become more automated, more accurate and efficient data transfer will occur. This will require further analysis. Finally, we have begun examining potential intensity-modulated radiation therapy (IMRT) errors according to the same criteria.

Original languageEnglish
Pages (from-to)81-94
Number of pages14
JournalJournal of applied clinical medical physics / American College of Medical Physics
Issue number3
StatePublished - 2005


Dive into the research topics of 'Errors in radiation oncology: a study in pathways and dosimetric impact.'. Together they form a unique fingerprint.

Cite this