Computed tomography (CT) is a tremendously valuable tool. In fact, if you asked me to pick only one of all the modalities at my disposal, it would undoubtedly be CT because it gives me the most flexibility in diagnosing a wide range of diseases and disorders that affect children. True, CT use has been increasing rapidly, but it provides valuable information for many situations, such as evaluating appendicitis, cardiac disease and for renal stone protocols. However, I think we have a responsibility to understand the cost of using this tool.
As radiologists, we have a responsibility to protect the safety and welfare of our patients. Dose control is not the responsibility of the ER physician; it is not the responsibility of the technologist; and it is not the responsibility of the family practitioner. We are directly responsible for delivering a material and the potential side effects of this material may be seen decades down the line. This is not the same case as delivering an antibiotic where the patient may develop a rash, or delivering chemotherapy where the patient may have an acute response. So we have the responsibility to account for the material we deliver and its long-term effects, making sure that our patients, children in my case, are well cared for.
To demonstrate the issues involved with CT, we performed a test with a phantom on one of our 64-slice CT scanners. For this demonstration, we simply turned up all of the settings: we used the lowest pitch, the highest kVp, the highest tube current, and a 1 sec rotation time. In short, we made the scan as dose-inefficient as possible. We generated nearly 120 mSv from a single exam. That was a somewhat frightening amount of radiation to deliver to the phantom. More alarming was the fact that the CT scanner itself never provided any indication that we were exceeding ACR guidelines or ACR accreditation standards for CT. We look at this as a real opportunity for the vendor community to incorporate some of the published standards onto the scanner so automated alerts can be delivered if an exam far exceeds published standards or diagnostic reference levels (DRLs). There should be a greater effort between radiology experts and manufacturers, to make sure that the equipment provides simple ways to arrive at correct doses.
We must also be cognizant that some patients are exposed to multiple CT scans. In an article in the April issue of Radiology, Sodickson and colleagues examined 22 years of CT data in >30,000 patients.1 Interestingly, about one third of their patient population had >5 CT scans in the 22-years prior to 2007. One out of 20 (5%) had 22 to 132 CT examinations. That is a tremendous amount of CT. When we look at effective dose, 15% of those patients had an effective dose >100 mSv. Remember, 100 mSv is a threshold for what we consider to be a moderate level of radiation. There is scientific proof that the risk increases after that point. One out of 25 patients (4%) had 250 mSv to 1375 mSv and 1 of 100 patients (1%) had 400 mSv.
Evaluating radiation dose in children
Dose estimation becomes a critical topic. But understand that effective dose has nothing to do with children. I wish we had conversion factors for organ doses. The only tools available are an anthropomorphic phantom with non–gender-related conversion. We do not know individual patient dose. I think everyone is trying to move towards organ dose, because that is a better way to establish and define risk. But we are years away from defining organ dose.
So how do we determine CT radiation dose in children? We can use qualitative measures: saying that a procedure may have a relatively high, medium or low dose. Or we can do more numerical evaluation with published estimations, physical measurements in anthropomorphic phantoms, Monte Carlo coefficients, and look-up tables such as dose length product (DLP).
Mettler et al. recently conducted a meta-analysis of dose ranges reported in the literature for various adult CT examinations.2 From this type of analysis, we can derive reference levels for adult CT examinations. These may serve to define what is reasonable for a 1-year-old head CT examination, for instance.
In terms of look-up tables, we tend to use the dose-length product which is then multiplied by a conversion factor to derive our effective dose in mSv. We take our pediatric-adjusted conversion tables from a paper by Thomas and Wang.3
The problem is that all of these methods to determine effective dose are generic. They employ standard body sizes. We have been developing a way to assess actual organ doses.
Using a Duke-developed software application, CT images of patients were converted into a full-body computer model to enable dose estimations for organs both inside and outside the image volume. Some organs and structures were created by morphing an existing male or female full-body adult model to match the framework defined by the segmented pediatric organs.
We have been able to determine, through voxelized methodology, the organ geometry for a range of children that are about 2 to 6 years of age. Now, based on weight we can take a cross-sectional area and prospectively determine what the organ doses are going to be to the colon, to the spleen, to the liver and to the kidneys. And then we can calculate an effective dose using those conversion factors.
What might come of this is a modified dose-report page, where we report the DLP, the CTDI, the effective dose, and we could conceivably build in a way to convert this to a risk estimate. Specific organ-dose information may allow us to assess dose and determine risk, to a better degree, in children.
Dose tracking is important because there has been a call for the radiology community to be more responsible about archiving, tracking and potentially reporting dose. This is a complex issue. A recent American College of Radiology (ACR) white paper on Radiation Dose in Medicine indicated that we should develop a surveillance mechanism to identify patients with high cumulative radiation dosesdue to repeat imaging. There is no national standard that does this.
An April 2009 American Journal of Roentgenology paper by Griffrey and Sodickson stated that imaging history impacts the risk-to-benefit equation such that clinicians could potentially take this information and use that to fashion their decision. Do we need another CT for a renal stone or not? Can we wait? Having this information available may improve the decision process, from their standpoint.
However, with this kind of accounting, there must be an agreement on what is conveyed. It cannot be an effective dose in one country, an organ dose in another, a procedure history in a third country, and a risk in the United States. The information needs to be meaningful and accurate and it needs to be consistent. This is not something that should solely apply to radiologists. Anyone that uses ionizing radiation has to be accountable for these measures. And the information has to be commutable. It either should be stored in the electronic patient archive, or be taken with that individual patient, so that if they move from one country to another, the information travels with them. And it has to be protected, just like any health information. So the electronic medical record seems to be the best place to store this information.
Dose report cards
Some groups, including the Image Gently organization have advocated dose cards that report procedures after they are completed. However, this method asks parents to be responsible for the healthcare of their children.
Creating a dose record or archive brings other questions to mind. What types of procedures do we archive? Do we have to consider both diagnostic and interventional procedures? Is it going to be for all ages?
Would there to be a cut off for these report cards? Does a 60-year-old patient need a report card? What should be recorded? Should the procedure be recorded? Should we record some type of units or perhaps a more qualitative measure? How is the information archived in radiology? It could be stored in the DICOM header, the PACS, the RIS or does it go into the HIS? It is not as simple as saying we are going to use the DLP from the DICOM information, and have that go into the hospital information system. It is not that easy to do.
It is also necessary to determine who will take responsibility for monitoring the report card. We cannot take this lightly as the report card will potentially affect clinical decisions, as the Griffey and Sodickson paper illustrated. Another article by Bolan in the Journal of the American College of Radiology dealt with the issue of trying to determine whether a study is indicated, after it has been ordered and the patient has shown up.5 This is a very challenging process and it requires us to work on the order-entry side to ensure proper utilization.
So I think the report card will change the quality of care. In an article by Steve Birnbaum in 2008, he explained how he set up a method of tracking CT exams in his hospital group.6 He established an automated system whereby a letter would go out to referring physicians who were ordering CTs for patients <40 years of age, with benign diagnoses, who had received >5 CT procedures. The letter did not decline the CT order but it asked the referring physician to consider some of the potential effects of CT radiation. In personal communications with Dr. Birnbaum, he said the procedure has been met with a great deal of success in his practice. Physicians want to do the right thing, but rather than being told that radiology will not perform the CT, it is better to educate them and inform them that a patient may have had 6 prior CT scans, and given our understanding of risk factors, perhaps another imaging method would be preferable.
Strategies for lowering dose in children
We did two surveys about 5 years apart, where we evaluated scanning parameters in pediatric radiology.7 We found that in 5 years, the peak kVp had decreased significantly for both chest and abdominal scanning. Of course, the technology between 2001 and 2006 had changed considerably, so the obvious question is: Does the tube current from 2001 match the tube current from 2006? To the best of what we could survey, we found substantial differences in technique. For instance, in the 0- to 4-year-old age group, there was a >50% reduction in the tube current that is used for chest scanning, and a similar reduction was seen in the abdomen. I believe that our educational efforts have had a substantial impact in what we do.
Our efforts may be changing CT utilization. I have seen preliminary data fro m the Children’s Hospital of Philadelphia, presented at RSNA, that indicated that there was a negative growth rate for CT procedures in the most recent 2-year period, while procedures like MRI and ultrasound have been increasing. Hopefully these data will be published soon so we can all see the collective impact of our efforts.
So what are the lessons learned from children? Dose assessment needs improvement, and our understanding of the limitations of dose assessment certainly needs improvement. It will be important to develop strategies to optimize dose that involve changing our utilization and our techniques so we can deliver high-quality care to children.