| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Editorials |
1 From the Department of Radiology, University of Virginia Health System, PO Box 800170, Lee St, Charlottesville, VA 22908. From the 2002 RSNA scientific assembly. Received December 16, 2002; accepted December 21. Address correspondence to the author (e-mail: bjh8a@virginia.edu).
Index terms: Cancer screening • Editorials
The use of x-ray computed tomography (CT) for screening presents a paradoxical challenge to American health care consumers. The extraordinary spatial sensitivity of CT would seem to promise earlier disease detection and improved patient outcomes, but there are characteristics of the technology, the diseases at which CT screening is directed, and our health care financing system that complicate our ability to use CT effectively for this purpose.
To see why this is the case, it is necessary to first consider what is meant by the term "screening." Screening is the systematic testing of individuals who are asymptomatic with respect to some target condition. The goal of screening is to prevent, interrupt, or delay the development of advanced disease in the subset of the screening population that has a preclinical form of the target disease (unpublished report of the American College of Radiology, Methods Subcommittee of the Task Force on Screening Technologies). It follows, then, that there are some essential requirements for successful screening. First, screening must advance the time of diagnosis relative to that for individuals who have the condition and do not undergo screening. With screening, the disease is identified in the preclinical presymptomatic phase. While this is necessary, it is not sufficient. Second, there must also be a treatment available that is more effective for presymptomatic disease than for symptomatic disease—one that incurs less morbidity and results in a better health outcome. Finally, from a societal perspective, screening must produce greater benefit than it does harm, and it must do so at a cost that society can afford in the context of other demands on its resources. The question is whether these qualifications are fulfilled with respect to CT screening.
During the past several years, CT screening has been popularized in the media. It has been marketed by a small number of "technology leaders," and it is being adopted by an increasing number of radiologists and entrepreneurs. As a result, an increasing number of CT screening examinations are being performed nationally. All of this is occurring with little or no data to support the practice. It also is happening without third-party reimbursement. This means that individuals who seek CT screening are paying out of pocket for their examinations. Is there anything wrong with this? I believe most Americans subscribe to the belief that the individual has the right to spend his or her discretionary income according to preference. I also think, however, that they would agree to this only with the corollary that the individual who makes the expenditure should not only derive benefit from the purchase but should also assume any associated personal or financial risk or harm.
This is not the case with CT screening. While the individual makes the payment for screening out of his or her own resources, all subsequent screening-related costs that follow on as a result of indeterminate or positive findings are paid by public or private health insurance. As I will discuss later, these "follow-on" costs dwarf the initial screening expenditure. The fundamental economic problem with screening is what economists call "moral hazard." Patients neither know nor for the most part care about their health care costs because they are protected with third-party insurance from the financial consequences of seeking medical care. Thus, the decision of an individual to seek screening may result in serious financial consequences for society in the form of higher expenditures for public insurance (Medicare and Medicaid) and higher costs for employer-based health insurance.
Outcomes of CT Screening
A number of societal factors are impelling the diffusion of CT screening. First, it is intuitive that the discovery of disease earlier in its course is better for the patient than is discovery after disease becomes more advanced. Second, CT screening addresses the key health concerns of the American population, principally cancer and coronary artery disease. Third, there is a revolution occurring in health care that goes by the name of "consumerism." This revolution is being led by the baby boomer generation. Boomers are more preoccupied with health and longevity than were preceding generations; they wish to learn more about medicine and health and intend to use that knowledge to participate more fully in decisions about the health care they receive. Furthermore, a sizable fraction of the baby boomers have discretionary income to spend on their health. Fourth, we humans tend to be more heavily influenced than we should by the "powerful anecdote"—or what the sociologists call the "availability heuristic." This means that we give more weight to a striking story—such as highly publicized cases of CT screening depicting a cancer early and "saving a life"—than is rational. Finally, there has been considerable marketing of CT screening, both by entrepreneurs and the media, that has emphasized the positive outcomes while ignoring the negative outcomes of screening.
Unfortunately, the things than can go wrong with screening have been ignored. An analogous situation is reflected in a quotation from one-time University of Texas (Austin) football coach, Darrell Royal, who said, "Three things can happen when you pass the football, and two of them are bad." (Available at: www.darrellroyal.org/quotes.html. Accessed December 15, 2002). CT screening is more complicated than football: Four things can happen in screening. Consideration of the ramifications of each possible outcome is essential to determine whether CT screening fulfills the requirements for successful screening discussed at the beginning of this article.
First, consider the true-negative outcome. This means that test results in an individual who is healthy are negative for disease. This is the best possible outcome. The patient is reassured. He or she will probably experience less anxiety and have a better quality of life. We can postulate that a negative screening result may encourage other valuable health-seeking behaviors such as eating more healthfully or exercising regularly. Alternatively, however, a negative screening result may instead reinforce destructive behaviors, such as smoking.
Second, false-negative CT screening results mean that a person who actually has an important or life-threatening disease receives the incorrect diagnosis of being healthy. Usually only anecdotal information exists at this time, but there is concern based on experience with diagnostic unenhanced CT (the most common practice in whole-body CT screening) that many abnormalities will be missed in the abdomen. Individuals who receive a false-negative diagnosis will be reassured incorrectly that they are healthy. They may ignore later signs and symptoms, which will result in later diagnosis of their condition and, perhaps, an ultimately poorer health outcome.
Third, a false-positive screening result occurs when an abnormality is detected at CT that does not really exist. Anecdotal comments from screening centers indicate referral of 20% to more than 90% of individuals to undergo additional studies. Major disease would be expected to be present in only 1%–3% of an asymptomatic middle-aged and elderly population; therefore, false-positive diagnoses are by far the most frequent outcome of CT screening. False-positive screening results turn healthy people into "patients" and engender follow-on diagnostic procedures. These follow-on procedures carry implications for morbidity, mortality, and cost. Findings in some of them will be false-positive and will result in unwarranted therapies. As noted earlier, not all of these costs are paid by the individual; some of the costs are paid by public and private health insurers. The consequences can be severe, as reported by a radiologist who underwent CT colonography with negative results but with a number of renal, hepatic, and pulmonary abnormalities depicted on his CT colonographic images (1). The ultimate diagnosis was dormant histoplasmosis, but his positive screening result was followed on with a number of invasive procedures that cost his insurance company more than $50,000. He lost 2 weeks of work and experienced an extended period of pain and untold anxiety for himself and his family (what the lawyers would call "pain and suffering").
Fourth—and perhaps the most interesting situation—is the true-positive screening result. One possible outcome after such a result is what makes proponents encourage widespread use of CT screening: Disease is found earlier than it would have been if screening had not been performed, the patient experiences less morbidity during treatment, and life is prolonged. However, there is an alternative possible outcome—namely, even though the disease is found earlier, there is no change in the patient’s outcome. The patient dies exactly when he or she would have died if screening had not been performed. This occurs because the biology of disease is heterogeneous. First, imagine the case of a highly aggressive disease for which our current therapeutic capabilities are insufficient, even when the disease is found during its presymptomatic phase. In the kind of single-arm observational or cohort studies in which CT screening has been studied to date, it appears that the patient has lived longer because diagnosis occurred earlier, even though there has been no actual benefit derived from screening. This is referred to as the "lead-time bias" of such observational studies. Similarly, consider the situation of very slow growing, indolent disease. CT screening preferentially depicts indolent disease compared with aggressive disease because of its much longer preclinical phase. Thus, the sample of individuals portrayed in observational trials is skewed to those longer living patients and, again, screening appears to convey a benefit in longevity that does not truly exist. This is termed "length bias." At the extreme of this situation is disease that grows so slowly that the affected individuals do not die of the target disease but of some other cause. Such disease has been termed "pseudodisease."
Whole-Body versus "Targeted" Screening
For purposes of further discussion, it is useful to categorize CT screening into two broad groups: whole-body and targeted. Currently, targeted CT screening is directed at lung cancer, coronary artery disease, and colon disease.
Not much is known about whole-body CT screening beyond anecdotal or sparse single-institution experiences (2). Proponents argue that CT screening is simply a more precise version of an annual physical examination, that it is like other screening examinations that have proven effectiveness (eg, mammography), or that there is a competitive need to provide the technology even though we do not know much about it. While some of these rationalizations seem mystifying (particularly the analogy with mammography [to me, there seems to be no relationship]), there is not much factual information that supports opposing views. Opponents of CT screening argue that it does more harm than good, that the known high false-positive rate will overburden the health care system, and that CT screening further promotes a multitiered health care system that disenfranchises poorer patients. The fact is, it will be difficult to conduct a reasonable multicenter trial of screening to define the parameters of cost and benefit. The number of patients required and the expense generated would be prohibitive. With that in mind, during 2002, the American College of Radiology commissioned a decision analysis to be conducted by the Massachusetts General Hospital DATA Group. The final results of this decision analysis are pending.
For targeted screening, it is useful to consider lung cancer CT screening as an example because it is currently of greatest interest to radiologists and because more observational trials directed at patient outcome have been conducted than those directed at other targeted screening examinations. Proponents offer several rationales for lung cancer CT screening: First, lung cancer is the most common cause of cancer death. Second, there is an identifiable population at risk; thus, not everyone needs to undergo screening but only those with the risk factors. Third, findings in observational studies have shown that CT aids in detection of more and smaller lower stage cancers than are detected with chest radiography or that would be detected without screening (3,4). Fourth, patients who are treated for stage I cancers have a better prognosis than they would if no treatment occurred (5). Finally, lung cancer is always aggressive, so length bias and pseudodisease are not relevant concerns in the interpretation of results of observational studies.
To the contrary, opponents argue that tumor size is not related to mortality within stage I disease. It is unclear whether CT screening truly results in a downward shift in the stage of tumors discovered or if screening simply depicts smaller, more slow-growing tumors. The longer survival shown in observational studies is, at least in part, related to the screening biases detailed earlier, and study results of increased mortality in patients with untreated lung cancer are partly due to selection and validation biases.
There are valid points on both sides. Nonetheless, with regard to the qualifications for successful screening outlined at the outset of this article, the fact remains that the results of observational studies have not shown definitively that there is a real benefit to CT screening, nor have they shown that CT screening is affordable on a population-wide basis. The largest single-institution observational study of which I am aware illustrates the latter problem. In a study reported by Swensen and colleagues (6), investigators found 40 cancers in 1,049 subjects who underwent screening. There was a favorable stage distribution, with 25 stage I cancers. However, they also found 2,832 noncalcified nodules; in fact, 69% of subjects who underwent screening had at least one such nodule that required follow-up study. In addition, the investigators found 696 other findings that were unrelated to lung cancer. They believe these findings represented mostly false-positive diagnoses and pseudodisease, but further work-up is required.
Conclusions
Findings in observational studies show that CT can depict smaller cancers and may result in increased survival. However, it should be evident from the foregoing discussion that survival is a flawed measure of the benefit of screening. Any benefit is exaggerated because of the effect of lead time and length biases and the finding of pseudodisease. What we really wish to know is whether and to what extent screening findings can help reduce the mortality (ie, death rate) of such diseases as cancer and heart disease so that we can determine whether the cost is worth the benefit. The current debate has become strident and often nonproductive. I believe that the only way to resolve the dispute is to move forward with randomized clinical trials focused on the specific questions of mortality reduction and cost.
A randomized trial of CT lung cancer screening is already underway. The National Lung Screening Trial is a collaboration of the American College of Radiology Imaging Network, or ACRIN, (trial principal investigator, Denise Aberle, MD) and the Early Detection Branch of the National Cancer Institute (directed by John Gohagen, PhD). In the trial, 50,000 subjects will be enrolled into three annual screening examinations with either multi–detector row CT or chest radiography, with the principal endpoint being lung cancer–specific mortality. Resource utilization (ie, cost) will be measured for a fraction of the subjects. The trial will last 5–9 years, depending on the effect of screening, and will cost as much as $200 million if study is required for all 9 years.
The commencement of randomized trials now is essential to ensure that 10 years from now, we are not in the same position of uncertainty about the value of CT screening that we are in today. Even at a cost of hundreds of millions of dollars, such trials are a small price to pay compared with the cost of tens of billions of dollars for population-based CT screening.
As is nearly always the case, however, the market for CT screening is growing far faster than the technology can be assessed. The expenditure of billions of dollars on an unproven technology threatens not only our ability to pay for other essential aspects of health care but also the broader competitiveness of the U.S. economy. I believe that until we know both the true benefits of CT screening and their cost, something needs to be done to avoid potentially disastrous economic consequences. I believe it is necessary to attack the problem at the level of moral hazard. Insurers, working with employers, are beginning to implement so-called defined contribution plans. With such plans, the insured is still covered fully for catastrophic care but shoulders a larger burden of the cost of the discretionary care that he or she seeks. Such a system is appropriate for CT screening. Individuals should be responsible not only for the cost of their decision to seek screening but also for subsequent follow-on costs related to that decision. When definitive treatment is required for a final diagnosis of life-threatening or otherwise disabling disease, then the third-party payers would take over.
Radiologists have a responsibility to their patients with symptomatic illnesses. The distraction of screening threatens to impose on already scarce radiologic resources and interfere with our ability to deliver needed care. We must be careful not to alienate a society that rewards radiologists’ efforts. Otherwise, to again quote Darrell Royal, "It’s like having a big ol’ lollipop in your mouth, and the first thing you know all you have is the stick." (Available at: www.darrellroyal.org/quotes.html. Accessed December 15, 2002).
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  I. M. Burger, N. E. Kass, J. H. Sunshine, and S. S. Siegelman The Use of CT for Screening: A National Survey of Radiologists' Activities and Attitudes Radiology, July 1, 2008; 248(1): 160 - 168. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. T. Kolber, G. Zipp, D. Glendinning, and J. J. Mitchell Patient Expectations of Full-Body CT Screening Am. J. Roentgenol., March 1, 2007; 188(3): W297 - W304. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. D. Furtado, D. A. Aguirre, C. B. Sirlin, D. Dang, S. K. Stamato, P. Lee, F. Sani, M. A. Brown, D. L. Levin, and G. Casola Whole-Body CT Screening: Spectrum of Findings and Recommendations in 1192 Patients Radiology, November 1, 2005; 237(2): 385 - 394. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. J. Golding Multi-slice computed tomography (MSCT): the dose challenge of the new revolution Radiat Prot Dosimetry, May 17, 2005; 114(1-3): 303 - 307. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. E. Chapman, D. F. Yankelevitz, C. I. Henschke, and D. Gur Lung Cancer Screening: Simulations of Effects of Imperfect Detection on Temporal Dynamics Radiology, February 1, 2005; 234(2): 582 - 590. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. M. Kalish, M. Bhargavan, J. H. Sunshine, and H. P. Forman Self-referred Whole-Body Imaging: Where Are We Now? Radiology, November 1, 2004; 233(2): 353 - 358. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A K Dixon Whole-body CT health screening Br. J. Radiol., May 1, 2004; 77(917): 370 - 371. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
