This is very disconcerting, given that quality assurance (QA) has such a solid foundation in radiation therapy, resulting at least in part from its quantitative nature. Unlike many other medical specialties, we can measure the therapeutic agent that we administer highly accurately and can be confident that each delivery is consistent with previous deliveries.

So why do patients still receive incorrect radiation doses? During the past few years, a number of significant radiation therapy errors have been reported. It has been suggested that the errors could result from the introduction of, and increasing dependence upon, advanced technologies such as intensity-modulated and image-guided radiotherapy. Some reported errors seem to have arisen from the introduction of the technologies themselves, with several publications postulating that the introduction of advanced equipment has permitted errors to occur that might otherwise have been detected. There are also indications that the demands of advanced technologies on department resources have drawn resources away from simpler or basic functions.

The New York Times articles described one of the accidents as having been discovered during a routine audit by the Radiological Physics Center (RPC). A new treatment machine had been installed at a hospital in Florida in 2005. At the time of commissioning, a physicist calibrated the accelerator output, but made a mathematical error that resulted in the beam delivering 50% more radiation than intended. The physicist then set up the instrument to be used for daily QA, and dedicated it for use with that particular accelerator. This procedure unfortunately ensured that the 50% error continued to exist undetected. It was not until the visit by the RPC that an independent measurement was made, and the error was identified.

Independent overview

In these cases and others, a key factor seems to have been a failure to implement the QA programme appropriately.3 A QA programme in radiation oncology must include clinical, physical and administrative components. It is important that each component contributes; the QA activities of one component cannot function adequately without the contributions of the other two.

A QA committee must supervise the programme and should have responsibility for co-ordination and oversight of all QA activities. The committee clearly must include staff who represent each of the contributing components. These staff members must be sufficiently senior to be able to determine the direction of the QA programme, order changes in the programme or the staff performing specific activities, and allocate resources. A key member of the QA committee is a senior radiation oncologist who must demonstrate a commitment to the programme.

One critical aspect of any QA programme is independence; that is, the QA procedures conducted to assure the quality and accuracy of radiation therapy delivery must be independent of the delivery itself. The failure to establish independence can lead to the risk that the QA systems merely mimic the performance of the parameters being measured, masking any error or change. An example of this occurred at the Florida hospital mentioned earlier. Had an independent measurement been performed (as was the case when the error was ultimately detected by the RPC), or had the QA device been used with a different treatment machine, the error would have likely been found earlier.

An independent measurement of output constancy (as well as other parameters) can be obtained from a number of sources. The RPC, funded by the US National Cancer Institute (NCI) and housed at the MD Anderson Cancer Center in Houston, TX, audits institutions that participate in NCI-sponsored clinical trials.4 The first level of audit consists of annual irradiation of a mailed thermoluminescent dosimeter. Institutions that do not participate in NCI-sponsored trials, or who wish monitoring more frequently than annually can purchase a similar service from Radiation Dosimetry Services, also located at MD Anderson. Other sources of such independent measurements are available.

The RPC's mission is to assure the NCI and co-operative groups that institutions participating in clinical trials deliver prescribed radiation doses that are clinically comparable and consistent. It does this by assessing the institutions' radiotherapy programmes, helping the institutions implement remedial actions, assisting the study groups in developing protocols and QA procedures, summarizing the results and describing the lessons learned to the community.

Addressing the issues

In medicine, we tend not to fully appreciate that any staff member, regardless of training, qualifications or experience, is susceptible to error. As such, we often fail to design and implement procedures that could detect and mitigate such errors. We also tolerate several common and error-prone situations that exist in radiation oncology departments today.

For example, many departments are understaffed, while existing staff frequently work in stressful situations. Measurements are postponed, consoles are not attended to and second checks are not made – in part, because everyone is too busy. Relegating important QA measurements to the end of the day, when staff are tired and eager to go home, leads to rushing and inattention to detail.

Secondly, there may need to be a change in culture in some departments. Observations of the working environment show that, in some departments, the radiation therapists are permitted to read magazines and eat or drink at the treatment consoles. These activities can distract the therapist from his or her duties, and allow errors such as those described by the New York Times to occur. Physicists may not always pay sufficient attention to the clinical work, and sometimes postpone important measurements until a more convenient time.

What are the solutions to these problems? Several authors have pointed to the airline industry as a model that could be adopted by the medical field. In fact, some medical practices have already adopted techniques used by airlines and other industries, with notable success.5 The use of checklists has been described; even where staff believe that they understand the procedures well, a checklist helps to make sure that even when rushed, important steps are not overlooked.6 A scheduled "time out" is also employed in some surgical practices. Here, all participating staff stop all other activities and focus on the procedure to be done, with the intent of discovering potential errors before they occur.

Such practices might need to be employed in radiation oncology. It certainly appears appropriate to re-evaluate the number of staff members required, and perhaps even to publish standards for staffing, to replace publications that became irrelevant years ago.7 Other standards should be considered, particularly one requiring that new linear accelerators be checked by an independent measurement before clinical use.