Radiology Today Interview: Rebecca Smith-Bindman, MD — Dose
Reduction in Medical Imaging
Vol. 11 No. 4 P. 16
Rebecca Smith-Bindman, MD, is a professor at the University of California, San Francisco (UCSF); a member of the UCSF Helen Diller Family Comprehensive Cancer Center; and a visiting research scientist at the Radiation Epidemiology Branch of the National Cancer Institute (NCI) within the National Institutes of Health in Washington, D.C. She is also the lead author of “Radiation Dose Associated With Common Computed Tomography Examinations and the Associated Lifetime Attributable Risk of Cancer,” which appeared in the December 14/28, 2009, issue of the Archives of Internal Medicine. The study found wide fluctuations in radiation dose among the 11 most common types of diagnostic CT scans, as well as within each type of scan. It used the Biological Effects of Ionizing Radiation 7 model to examine the impact of dose on cancer risk and estimate the range of cancer risk that may be expected based on the type of scan, the radiation dose, and the age and sex of the person being scanned. The abstract of that paper is available online at http://archinte.ama-assn.org/cgi/content/abstract/169/22/2078.
Radiology Today (RT): An estimated 70 million to 72 million CT scans were performed in the United States in 2007. Based on your research, how many new cancer cases and deaths would you expect to see with current CT usage and cancer cure rates?
Smith-Bindman: This question is outside the scope of our research. Amy Berrington de Gonzalez, PhD, a researcher at the NCI, in the same issue of the Archives of Internal Medicine in which our paper was published, estimated 29,000 cancers and 15,000 cancer deaths would result from CT imaging that occurred in 2007.
RT: What are some of the variables that affected your results?
Smith-Bindman: That’s an incredibly important question. I think one of the things we really need to do is to start collecting data on a large number of patients so we know what’s going on. We can then figure out why there’s so much variation. For our study, I can explain some of the variation we found; but for some of it, I cannot. One of the choices that a radiologist makes is scan length. If you scan a larger area, the patient receives a higher dose. Another parameter is how many times you do repeat scanning, multiphase scanning. Any multiphase study increases the dose. There are also programs available—and different radiologists or technologists choose to use them or not use them—that will influence dose. So using programs that vary milliampere seconds can have an impact on dose. The peak kilovoltage and the milliampere seconds that you choose to set it at, if you’re not using automatic exposure, will impact the dose. So all of those things impact the dose. Whether technologists at certain facilities use those parameters or don’t use those parameters—or use them for certain patients and don’t use them for certain patients—all impacts the dose.
RT: Why do you think your study found higher doses than what are typically reported?
Smith-Bindman: I would say the main reason is that there have been very few studies that have recorded actual doses that patients encounter. For example, the FDA collects information about typical doses used in CT and for that study they ask a technologist from each facility how they usually set up a study, but they don’t actually go to see what happens and measure actual dose used in actual patients. The American College of Radiology, as part of their accreditation process, asks radiologists to submit a small number of selected studies; thus, they also do not get to see what happens for a large number of patients who undergo CT. The only large studies that have collected dose data on a large number of actual patients have been for cardiac scanning, and those studies have found the same kind of dramatic variation in dose that we found.
RT: What steps would you recommend to standardize radiation dose?
Smith-Bindman: I would say the first step is to start keeping track of dose. I think unless you can start measuring dose, there’s no way you can even understand the magnitude of the problem. And the National Institutes of Health recently said in the ACR Journal that they’re going to start requiring any manufacturers who sell them equipment to export easily into their electronic medical record the dose that the patient received. So No. 1, we need to start collecting actual dose information; No. 2, we have to then as a community start setting very clear targets of what are acceptable doses.
The amount of variation that we found in our study is not acceptable. It’s exposing patients to risk needlessly since we have techniques to lower the dose dramatically from the levels that we found. I think without having made that dose information widely available and with what’s currently going on, it makes it much harder to lower those doses. But I think we need to have some oversight about what’s acceptable, and then I think we’ll be able to reach those doses.
To put this into context, we’ve seen the same kinds of variations for many tests. The problem here is that the levels of doses are so much higher for CT—one hundred-fold higher, five hundred-fold higher than an x-ray—that we can’t afford that kind of variation for CT scans. It’s one thing to have variation in a low-dose study; it’s not acceptable to have variation in a high-dose study. So collecting dose is one way to start. Personally, I think the guidelines should come from the FDA taking a much more active role in helping radiologists reach these more desirable targets.
RT: What about the ACR’s guidelines?
Smith-Bindman: The ACR has two programs in play. One of them is an accreditation program where they ask radiologists who want to be accredited to send in a few examples of their studies. One would hope they send in their best studies, not their average or worst studies. Even given that format of the accreditation program, a significant minority of facilities fail. I don’t know the number, but I’m hoping that publication comes out soon. It’s interesting that a significant minority fail even when they’re asked to send in their best studies and even when the guidelines for “passing” are extremely explicit. What the ACR does, as I understand it, is they change the threshold they consider allowable based on the distribution in the dose data that were submitted rather than by applying more evidence-based standards of what is desirable. Thus, when a large number of radiologists sent in dose information higher than their originally allowable thresholds, they raised the threshold. My understanding is that these allowable thresholds, for example, for head CT, are considerably higher than for most European countries. Additionally, because the ACR collects dose information on only a very small number of cases, the ACR accreditation program does not measure what’s actually happening in routine clinical practice. So I think it’s certainly a way to begin, but it doesn’t have enough bite to really get doses as low as I’d like to see them. I think it may help get rid of some outliers—some really high doses—but we need much more thorough and comprehensive guidelines.
The ACR is also beginning a dose registry project that’s focused on collecting large amounts of dose data. I think this type of effort is crucial to the development of realistic guidelines. However, this is a very small, very underfunded effort. They’re currently collecting dose information from six machines, so this project needs to dramatically expand if it’s to have an impact on the quality and safety of CT imaging use.
RT: So do you think it’s better to have somebody outside of the radiology community making sure that the guidelines are being met rather than self-regulating?
Smith-Bindman: I think we’ve had self-regulation, and I’m not sure that’s really helped us very much. There’s no data out there to tell us how we’re doing with self-regulation. My studies were done at some highly regarded institutions [UCSF; Alta Bates Summit Medical Center in Berkeley, Calif.; Marin General Hospital in Marin County, Calif.; and California Pacific Medical Center in San Francisco], and the doses were higher than generally known. It can be tricky, if not impossible, for an individual radiologist to figure out how to set up those parameters to try to get a dose as low as possible. Manufacturers change the default settings on their machines to give us the most beautiful pictures possible because, as radiologists, we like beautiful pictures. They’re very important to us. But those beautiful pictures come at a dose cost, and there’s no one in that equation—manufacturer or radiologist—who really cares about dose the way a patient or patient group would care about dose. So having the industry by itself monitor dose, I’m not in any way opposed to that, but it’s currently not leading to the minimization of radiation doses that I would like to see happen.
The manufacturers, to my understanding, have the tools to lower dose immediately. And they’re selling these software products to lower dose. The price point is such that most radiologists are not purchasing these products and, thus, patients continue to receive higher exposures than necessary. So I question whether this is where we should turn to make sure that our patients are as safe as possible. Is that where you want to go? I don’t get why the manufacturers aren’t saying, “Oh my goodness, doses are too high! We’re going to give everyone this software free tomorrow.” My area of expertise is not how we get some systems in place to ensure safety. If we could self-regulate and get that product out to the patients so it’s possible to make scans safer, that’s great. If not, I think we need alternative systems of quality assurance in play. The Mammography Quality Standards Act is the only area where we have legislation to oversee the quality of a radiology test. And I think as a result of that legislation, we’ve seen really dramatic improvements in the quality of mammography.
RT: What are some dose-reduction techniques that are effective and can be instituted right now?
Smith-Bindman: There are tools out there to help radiologists lower dose. And the best work that I can cite for you is work that the Image Gently campaign is doing. And they basically are trying to make it easier to lower doses, specifically in children. And they’ve provided a list of protocols freely online that you can download to help you lower dose. I wish it were as easy as pressing a button and getting the best study, but it’s not always as easy as that. So learning about ways to do that, for example, going to Image Gently’s Web site. and downloading its program, is one way that you can immediately lower your doses within a particularly challenging group [children] by basically targeting the settings [peak kilovoltage, milliampere seconds, scan length] to the size of your patients.
RT: You mentioned that repeat scanning is particularly problematic, and radiology benefits managers (RBMs) have been implemented, besides to control cost, to look at how often scans are used. Do you believe RBMs are an effective tool for reducing the number of scans, or do you believe they don’t do enough?
Smith-Bindman: If you want to reduce a population’s exposure to radiation, there are two ways to do it: You can lower the number of tests they have, and you can lower the dose per test. Those are your only two choices. Lowering the dose per test is, I think, the quickest way.
In terms of reducing tests, there are a whole bunch of attempts out there to try to do that, and the simple answer is we don’t know yet what the impacts of those efforts are going to be. In general, most of those products report a 10% decrease in imaging. I wish they were a little bit more evidence based so they could focus on reducing inappropriate imaging. I’m not positive they’re reducing inappropriate imaging, but I think some of them certainly do. And we [radiologists] certainly can and should identify tests that we shouldn’t repeat. But the impact of those products on utilization is not yet known.
RT: In your study, you recommend that doctors and patients be informed about the potential risks CT poses. What are some ways doctors and patients can be educated about these potential risks?
Smith-Bindman: I think part of the increased use of imaging comes from patients. Patients believe that all imaging is good, and their belief is often that quality care means receipt of a new imaging test such as CT or MRI. While I wouldn’t disagree with that general view, imaging comes at a cost, and I think it’s really important to consider those costs, including the potential risks of radiation, alongside the benefits. So there needs to be much more discussion between patients and physicians about the risks and benefits of CT and, specifically, whether the information that would be obtained will help make a diagnosis that will change care.
I recently took my son to the emergency room after he took a fall skiing, and I didn’t go in thinking he needed a CT. I went in thinking I need a physician to look at my son and tell me if he looks OK or not. And the emergency room doctor checked him and said, “I think he has a small concussion, but he looks too good to have any kind of bleed. I think he’s fine. I don’t think we need to image, but I will image if you want me to.” And I said, “No, I don’t want you to. I want you to do what you think makes sense.” And I was really comforted by the doctor’s expertise, and my son didn’t undergo any imaging. Whereas a lot of patients might come in thinking, “Of course I want the imaging test. What’s the downside?” I think emergency physicians, who are very pressed for time and have a lot of patients who are very sick on their hands, in general, they image everyone who walks in before they even assess them fully. It turns out it may be faster to image before fully evaluating the patient’s history or consulting a specialist, and that will result in some unnecessary imaging. Often, imaging is done primarily for triage purposes, and that should not happen unless the patient will likely receive benefit from the test.
And so it’s not so much that we need to be told what the specific risk is but rather we need to understand that there is some risk and that consideration of the decision to undergo imaging needs to be balanced. We do a lot of repeat imaging where there is no consideration of the fact that the findings are unlikely to lead to any change in management. So that discussion is not happening and needs to start happening. I think the understanding that CT is a great test that we have to use wisely and prudently needs to come into play a little bit more than it currently does.