Types of Errors
Random Errors occur due to unpredictable variations with a measurement. Random errors occur in many circumstances; a precise measurement device minimises the magnitude of random errors. In radiotherapy, patient immobilisation is an important factor in reducing random errors in patient positioning.
Systematic Errors are more of a concern, as they represent a constant inaccuracy in a measurement relative to reality. Systematic errors could include the incorrect documentation of a patient position, so that for every treatment they are 5 cm away from the expected position.
Errors may occur in many situations during the simulation, planning and treatment of a patient.
Clinical Judgement
A doctor must decide whether a patient is appropriate for treatment. It is always possible that an 'error' could be introduced at this stage!
Simulation
Simulation is the procedure where patient data is transferred from their physical body into an electronic format, usually by acquisition on a CT scanner.
Random errors in this circumstance could occur due to limited resolution of measuring instruments, small movements in patients during data acquisition or from limitations of immobilisation devices.
Systematic errors can be generated in simulation, through inaccurate tattooing of the patient, incorrect transcribing of measurements/immobility devices used (eg. a different head rest could be recorded, leading to alternate patient positioning).
Minimisation of Errors
Simulation is always carried out by two therapists, with additional staff as required:
- The radiation oncologist may be present for complex setups, or when markup of the patient is required (breast, skin, head and neck)
- Physicists and biomedical engineers may be involved if there are complex issues with bolus, compensators or setup positions
The number of staff, particularly the therapists, ensure that (as far as possible) accurate data is recorded.
Planning
Modern treatment planning uses computer software almost exclusively. This has the advantage of significantly speeding up calculations and allowing more complex plans to be used. There is also the potential for new errors to occur relative to the old 2D planning style.
Random errors can occur:
- When contours and beamed are placed on the patient volume (potentially large error)
- If computers use rounded numbers to perform calculations (very small error)
Systematic errors are more concerning, and occur in situations where:
- The dose distribution algorithm is not well designed, or deals with certain situations (eg. inhomogeneity boundaries) poorly.
- The dose distribution algorithm is using physics data that is incorrect
- Contours and beams could be placed incorrectly, leading to errors in dose distribution within the patient.
Minimisation of errors
Data imported into the treatment planning system should be accurate, which depends on an accurate physics quality assurance process as well as accurate data transfer. The data should be checked on phantoms before computerised treatments begin on actual patients. Improved staff training in contouring and use of treatment planning systems is also a useful way of reducing error.
Treatment Delivery
Types of Errors
Random errors can occur during any part of treatment delivery. With patient positioning, due to the nature of patients, it is hard to be more exact than 5 mm at best without thermoplastic masks or stereotactic frames. Other factors, such as the patient's respiratory rate or a change in the dilatation of their large bowel, are hard to correct for without the appropriate technology. Random errors can also occur in the positioning of radiotherapy equipment, such as the linac collimator, gantry or couch. These errors are minimised to < 2 mm, < 3% or < 1o by regular quality assurance checks.
Systematic errors are also a problem in treatment delivery. These may have been generated in an earlier stage of simulation or planning and not recongised. A patient may have a mole mistaken for a tattoo, leading to inaccurate setup. Data imported from the planning system may be incorrect due to human error in entering it. All of these factors lead to an ongoing error in patient setup that can lead to underdosing of tumour and/or overdosing of normal tissues.
Minimisation of Errors
Random errors can be minimised by improving technology (such as gating or cone beam CT) that allow more accurate positioning relative to patient movement. For critical areas, a thermoplastic mask or stereotactic technique might be appropriate.
Record and verify systems have been implemented to compare the setup of the patient with 1) the setup for that patient's previous treatment and/or 2) the set up in planning. These systems collate data on the couch, gantry and collimator positions, as well as the positions of the jaw, multileaf collimators and beam energy for each beam used. This is designed to assist staff in delivering the right treatment and reduce the risk of systematic errors; however the potential for worse errors is also present, as staff have less to do with the physical setup of each beam and rely on the computer to do so. If the computer is in error there may be no way for staff to tell there is a problem; or alternatively staff may not pay enough attention to the treatment and simply trust the computer to deliver the treatment accurately.
