-
Brief History
- In the early 1970s Atomic Energy of Canada Limited (AECL) went into business with French company CGR building linear accelerators. CGR built the machines that the Theracs were based on & developed the original software
- The Therac-6 was based on the CGR machine the Neptune, & was capable of producing 6 MeV x-rays
- The Therac-20 was based off of the CGR machine the Saggitaire, & was capable of producing electron & x-ray beams up to 20 MeV
- The relationship between AECL & CGR ended in 1981
-
Therac-25 Brought Into Service
- Capable of producing x-ray beams up to 25 MeV & electron beams of varying levels
- Built with software use inherent to its function, unlike the previous models which could be used independently of the software
- Had two shut-down levels in the event of an error:
-Treatment suspend: complete machine restart needed to proceed
-Treatment pause: one button-push needed to proceed with previous settings
-
Kennestone Regional Oncology Center, 1985
- Patient was set to receive 10 MeV treatment to the lymph nodes in the clavicle region
- She reported hearing a noise & feeling like she was burned, but the operator told her that was impossible. The patient developed symptoms of radiation sickness soon after
- It was estimated that she received 15,000-20,000 rad, & eventually she lost the use of her shoulder
-
Ontario Cancer Foundation, 1985
- Patient was receiving her 24th treatment for cervical cancer
- The machine shut down after 5 seconds & displayed "NO DOSE" so the operator overrode the error & proceeded with the treatment; this happened four more times
- Patient reported “electric tingling shock” in the hip area immediately after the procedure. She died several weeks later from the cancer, but the autopsy revealed that a total hip replacement would have been necessary due to the damage cause by the radiation
-
Yakima Valley Memorial Hospital, 1985
- Patient developed erythema, but continued to treat because it wasn't identified as abnormal until after treatments had completed
- The staff couldn't ID the cause of the parallel-striped pattern of the erythema
- AECL responded that malfunction or operator error couldn't have caused it, that overdose by the machine was impossible, that no other patient injuries had ever occurred, & that the upgrades done to the machine in Yakima following the Ontario incident led to a huge safety improvement
-
East Texas Cancer Center, Tyler, March 1986
- Patient was receiving his 9th treatment fraction
- The operator received an error, attempted to restart the procedure, & received another error. Before she could resume treatment again, she heard the patient hitting the door
- The patient felt like he had been shocked by the machine, & knew this was abnormal so he tried to get away. He received a second dose as he was escaping
- The hospital physicist shut down the machine, but could not replicate the problem so it was returned to service
-
East Texas Cancer Center, Tyler, April 1986
- Patient was receiving a treatment to the face, then heard a loud bang & a sizzling noise & he felt like he got punched in the face. He would end up dying 3 weeks later, & it was estimated that he received ~25,000 rad to the face
- This was the same operator as the previous ETCC incident, so she was able to make note of exactly what happened
- The hospital physicist immediately launched his own investigation and is a huge part of how this problem was identified and solved
-
Yakima Valley Memorial Hospital, 1987
- Patient was supposed to have plain films taken & also receive treatment, but the operator forgot to remove the film plates before starting the treatment
- The operator attempted to begin the treatment but received an error, restarted & received another error, then could hear the patient complaining over the intercom of being burned
- The patient developed parallel-striped burns reminiscent of the other patient, so they followed up & found she had developed chronic injuries from the radiation
-
(So Many) Problems Identified
- Myriad different problems in the software coding caused errors to occur in various different situations
- A specific error in the coding was never pointed out, rather, generally poor software design & development were blamed, particularly the inability to test for and replicate errors
- Related problems were found in the Therac-20, but the hardware-based safety locks prevented any similar incidents from occurring
- Please go skim the Wikipedia page for a good read, it's WILD
-
AECL Corrects Problems with the Therac-25
- 02/03/1987 - AECL announces hardware & software updates to address some of the identified problems
- 02/06/1987 - FDA & Canada's Health & Welfare body agree to recommended that ass Therac-25s be removed from service until the updates could be made
- 05/26/1987 - a recall is issued; Test data submitted by AECL to show their progress actually showed the opposite, baffling the FDA reviewer
- 07/21/1987 - fifth & final revision issued and the Therac-25 could resume treatments