The Record & Verify system is actually incorrectly named! It is actually a Verify & Record system.
Initially the system serves an important function within the use of a linear accelerator as a method for catching human error. In the early days there was a 'simple' system consisting of a hardware intercepts on the linac for measuring things such as jaw position, etc, a database that contained three main parts
- the intended field settings
- these settings were entered by RTs/dosimetrists when the completed plan had been checked for accuracy. In the early days this transfer was manual and needed to be checked.
- the delivered field settings
- these setting were delivered from the hardware intercepts in real time.
- the signatures of those responsible along with a date/time stamp
There was also some software associated with the database that:
- cross-checked the intended settings with the actual settings
This system enables the checking of some parts of the radiation delivery process. It could not check whether the right side was being treated. I could check whether the wedge was in the right way, but not whether the bolus had been placed on the field. Although there were demographic identifiers for the patient, the patient's idenity still had to be verified by the therapists.
How has the system changed with time?
As always happens, the system has become more complex. It became clear very quickly that just as hardware intercepts could measure the jaw settings, the addition of jaw activators in a feedback loop permits the software to actually set the intended field sizes without human intervention (which surely would be better? right?).
The system has developed to push back the points of human interaction. It is now routine to pass the planning system parameters to the R&V electronically using a protocol called the DICOM format, rather than have human fingers/brains do the job. Indeed as the data needed for such treatments as 3DCRT and IMRT have become more complex, the sea of numbers have increased.
More methods for verifying the radiation treatment have been introduced, such as the imaging system. This strictly is not part of the R&V software but has an ancillary role to play.
How has this changed the safety of the radiation delivery process?
The process is made safer by the modern automated R&V but there are some caveats. Processes will always fail at the interfaces - and the worst interface is the human-machine/machine-human interface. So what are the new problems?
* the belief that the computer is always right
* the belief that since a computer is there, humans can stop checking
* if the wrong data has been entered once, it is likely to be perpetrated all the way down the line
* systematic errors are likely to be applied across the whole treatment
* the small errors that humans can make can become really major problems (take the wrong patient in the room?)
* that it can be harder to see more simple and gross errors
* computer error can go unseen easily
The reports on Radiation accident here provides a salutary reminder of just what we are capable of doing if something stuffs up - machine (Therac20) or man and machine. And if you think it couldn't happen to you - well, good luck, you have just set up the pre-conditions for the occurrence.
If this is your feeling, might I suggest that you read J. E. Gordon's book where in the chapter on "A Chapter of Accidents" he eloquently states:
 |
"In nearly all accidents we need to distinguish tow different levels of causation. The first is the immediate technical or mechanical reason for the accident; the second is the underlying human reason. It is quite true that design is not a very precise business, that unexpected things happen, that genuine mistakes are made and so forth; but much more often the "real" reason for an accident is preventable human error.
It is rather fashionable at present to assume that error is one of those things for which it is not really fair to blame people, who, after all were "doing their best" or are the victims of their upbringing and environment, or the social system — and soon and so on. But error shades off into what is now very unpopular to call "sin." In the course of a long professional life spend, or misspent, in the study of strength of materials and structures I have been forced to the conclusion that very few accidents just "happen" in a morally neutral way. Nine out of ten accidents are caused, not by more or less abstruse technical effects, but old-fashioned human sin — often verging on plain wickedness. Of course I do not mean the more gilded and juicy sins like deliberate murder, large-scale fraud or Sex. It is squalid sins like carelessness, idleness, won't learn and don't need to ask, you can't tell me anything about my job, pride, jealousy and greed that kill people. " |
…. and then seek out a course on Root Cause Analysis. Quickly!
The best analogy I have come across is that of 'swiss cheese' - the layers of error-checking in your department are like layers of swiss cheese - none are perfect, they all have holes. Disasters occur when you line up the holes and the patient slips right on through to reach a disaster [like this].