Portal de Periódicos da CAPES

QA procedures in radiation therapy are outdated and negatively impact the reduction of errors

2011; Wiley; Volume: 38; Issue: 11 Linguagem: Inglês

10.1118/1.3605472

2473-4209

Howard Amols, Eric Klein, Colin G. Orton,

Advanced X-ray and CT Imaging

Arguing against the Proposition is Eric E. Klein, Ph.D. Dr. Klein is Professor of Radiation Oncology at Washington University, where has been for 21 years. He has published over 80 papers, half as first author and many on quality assurance issues. He is certified by the ABR in Therapeutic Radiologic Physics. Dr. Klein is very active in the AAPM and ASTRO, previously serving as chair of AAPM's Quality Assurance and Outcome Improvement Subcommittee and TG-142 (Linear Accelerator Quality Assurance). He was called as an expert witness to a Congressional hearing on the use of radiation in medicine in 2010. Dr. Klein has been involved with CAMPEP's Residency Review Committee since 1995 and directs the longest standing accredited residency program. Imagine if automobile maintenance was still performed as it was 25 years ago. Changing tires, spark plugs, and distributor points every 10 000 miles, repainting rusty chassis, and adding water to the battery weekly. Starting your car on a cold morning means “adjusting the choke,” pumping the gas before turning on the ignition, and idling for 10 min before shifting into gear. We'd be pumping the brakes and spinning steering wheels whenever we went into a skid. But nobody does this anymore because it isn't necessary. Cars are built better and have different maintenance problems. Taking your car into the shop now usually means replacing a computer chip or sensor that didn't exist 25 years ago. Linacs also are built better than they were 25 years ago, but we haven't changed our QA procedures accordingly. We still routinely check “cGy/mu,” isocenter accuracy, laser drift, etc. Sure, we've added new QA procedures for modern accessories (EPIDs, MLCs, CBCT, etc.), but we never subtract. We never redesign the process to reflect the characteristics of modern equipment. We just increase the workload. How many patients have been mistreated recently because a laser drifted or a linac dose rate changed between Monday and Tuesday? None! Mistakes made during commissioning: Wrong dosimeter to calibrate a stereotactic radiosurgery (SRS) beam; corrupted computer file or software bug; not understanding a “hidden equation” underlying an Excel spread sheet, or the data format required for a computer program. Being rushed or complacent and not following existing procedures. Outdated QA procedures are fixated on 1–2% or mm drifts in things that rarely drift anymore. Why not spend less time testing “outdated” things and more time testing things that have been injuring patients recently? The “name of the game” for new technology is acceptance testing, accuracy, commissioning, interconnectivity, and training; yet outdated QA procedures focus on labor-intensive measurements of precision. Case in point: the SRS beam that was miscalibrated because an inappropriate dosimeter was used to calibrate the small fields. I suspect that daily, monthly, etc., checks of linac output, etc. were precise and in compliance with recommended QA procedures, but the equipment was miscalibrated from day one, and no one checked accuracy (e.g., did an external audit). At my own institution, like most others, we used to check laser alignment for every SRS patient. Recently, however, we changed our procedures and patient setup is now adjusted and confirmed using CBCT. Further, our new linac has isocentric accuracy (“star shots” geometric/mechanical measurements, etc.) of <0.5 mm. So why care if the lasers are properly aligned? Why waste time checking this when mistreatment of SRS patients is now more likely to result from computer errors? Space does not permit a detailed listing of QA report recommendations that should no longer top the list of everyday checks, but, hopefully, I have made my point. Let us eliminate some “historical” QA tests to reflect the quantum changes in equipment design and focus more on QA tests that are more likely to prevent “modern” treatment errors. Three Mile Island, Chernobyl, Fukushima. Names synonymous with nuclear power plant catastrophes. Riverside OH, Tyler TX, Indiana PA, Cleveland (Plain Dealer), Epinol, Panama, Glasgow, Zarragosa, New York (Times),1 etc. Names synonymous with radiotherapy events for which patients were either injured or killed. Events for which QA procedures were not followed. Perhaps QA procedures are outdated, but they are not negatively impacting error reduction. In fact, they are still preventing catastrophic errors. Somehow, with limited resources, physicists must maintain current procedures while embracing a paradigm shift prioritizing error reduction.2 The AAPM TG-142 report3 accomplishes this by suggesting tests with more frequency and scrutiny for greater risk procedures (IMRT, SBRT), and less frequency for mundane tests that are benign with regard to error impact. Simultaneously, the long awaited AAPM TG-100 report4 will guide the physicist on how to develop QA programs using failure mode and effects analysis (FMEA). This is not an easy task. TG-100 was constituted in 2004 and has yet to produce a final report. Ford et al. published an FMEA based manuscript5 describing analysis of the 269 steps of taking a patient from consult to treatment. This was an expansive undertaking involving consultants. There is an effort to analyze the impact of human factors and culture on errors and error reduction, with training available through workshops and annual meetings. But no matter what strides are made, vendors will continue to sell complex equipment to facilities with understaffed and/or undertrained physics personnel. Engineers will continue to adjust machinery without involving physicists for proper review. And software will still be written that is not robust enough to prevent unforeseen user misappropriation. It is no wonder that of the top ten worst software bugs ever reported, two involved radiotherapy.6 An example for which a procedural upheaval is immediately needed is IMRT QA. Every evening around the country, thousands of physicists (or QA technicians, residents, fellows, students) perform time-consuming measurements to validate that the correct dose is being delivered to a “box,” a poor surrogate for the actual patient. In terms of prioritization, physicists should be spending time performing on-screen plan reviews to unmask potential problems that would go undetected with simplistic phantom QA. This includes incorrect image fusion, erroneous hot spots, improper beamlet segmentation, etc. These problems could still go undetected with alternative QA techniques such as independent Monte Carlo calculations or leaf position-timing analysis. Which brings up a final point vital to existing and future quality assurance procedures training. Physics residency programs, which are growing rapidly in number, must include teaching of human engineering and process analysis, beam modeling and critical review of treatment plans and, despite how outdated they may seem, routine quality assurance of imaging, planning, and delivery systems. Dr. Klein agrees that “QA procedures are outdated” but asserts “they are not negatively impacting error reduction.” Wrong! If they are outdated then doing them wastes time. Every minute wasted means something more important either doesn't get done on time, or is rushed by the now overworked physicist, increasing the chances of a serious mistake. Dr. Klein and I do agree that QA tests need to be prioritized such that “mundane tests” (his words) with low error impact should be done less frequently, while high error impact tests should be done more frequently. We also agree that medical physics training programs must do a better job teaching QA. I would include courses on the history of medical physics disasters. Dr. Klein, for example, refers to “Riverside, Tyler, Glasgow, etc.” How many new medical physicists even know what he is referring to (i.e., sites where horrific medical physics errors occurred)? Physicists must truly understand the consequences of their mistakes. I am also concerned that QA has become a cookbook kind of exercise. Do whatever it says in TG-51: put a lead sheet in the beam, get kQ from Table I, etc. Don't think about it, just follow the directions! Cookbooks tell you how to make a soufflé or calibrate a linac, but they don't tell you how to not mess up. And that is what physicists are supposed to know. We are teaching people how, but not why, and there is an old expression: “the person who knows how will always be working for the person who knows why.” A physicist who knows why QA tests are done and how equipment really functions, will also know when those tests are no longer necessary and, more importantly, what can go wrong and how to prevent it. Finally, antiquated QA manuals “advising” us to make unnecessary tests are misinterpreted by regulatory agencies that then turn them into laws. Let's get these counterproductive anachronisms off the books! I agree with Dr. Amols that mundane non-beneficial quality assurance procedures must be reduced. TG142 promotes that concept and TG100 will soon enable physicists to accomplish this. Regarding the fact that linear accelerators are built better than they were 25 years ago, this is obvious. However, not all facilities possess new machinery. Some accelerators are old and/or not maintained well. Some hospitals are buying new machines for purposes of competition, without having proper maintenance in place or, more importantly, trained physicists with sufficient time to do the necessary quality assurance or implement new available procedures. As new ancillary devices are being added to accelerators, I concur that the workload has increased. But if there are not enough physics FTEs allocated for this new complex machinery, isn't the increased workload a battle to be negotiated with administration. We have to remember that staffing levels and tools available at a large academic center are much more comprehensive in number and versatility compared to a small practice with a solo physicist. If we can't win the battle to have the appropriate number of physics FTEs in place, then how can we maintain safety? I agree that QA of improperly commissioned planning and delivery systems is worthless. Perhaps manufacturers should be required to supply an expert to work with the local physicist to perform end-to-end tests as part of final commissioning. This is especially important for the imaging aspects of localization devices which tend to be technologically more volatile. As mentioned in my opening statement, routine quality assurance procedures would have stopped most of the horrific events that have happened in radiation therapy. Allocation of time to simply check output and flatness (sensitive indicator of energy) and a sliding window DMLC or SMLC output check is time well invested each day. Finally, as I own a hybrid car made by a Toyota based company, I somehow can't help routinely checking the brakes and floor mat placement.