Cultural Context

Alternative medicine has a long tradition in the United States, starting with the herbalist and mineralist movements in the eighteenth and nineteenth centuries. Indeed, scientifically based medicine struggled to win the high ground of cultural authority in the first part of the twentieth century [2]. Organized medicine succeeded greatly in achieving this goal, in part because of standardization of medical training and the technologic advances rooted in the scientific method, as well as the restrictive politics of the profession, which excluded ersatz professionals or quacks. Over the past two decades, the reputation of physicians has suffered. This deterioration started with malpractice attorneys. More recently, it has been fueled by restrictions imposed on patients by the insurance and managed care industries and, by extension, the employers who use them, as managed care substituted lay administrators and clerks for physicians in the realm of health care decision making.

Such resulting distrust of the health care system resonates with the innate cultural American character trait of distrust of any professional paternalistic authority, a trait ingrained in our cultural and political history. Embedded in the Declaration of Independence is the notion of egalitarianism, a notion that all Americans treasure. However, the derivative flip side of this ultrademocratic tenet is antiintellectualism and, by extension, distrust of authority figures and major institutions. This latent American sentiment is particularly activated when such entities betray the trust of individuals repeatedly, as has occurred with managed care.

It is in this historical and cultural context that the phenomenon of self-referral for CT screening should be considered.

The Role of Targeted CT Screening

The rapid development of CT technology in the past 2 years has allowed its application to screening. The early entrepreneurial screening facilities concentrated on detecting coronary calcification as a marker for coronary artery disease in individuals who are still asymptomatic [3,4,5,6,7].

The accumulated data in this field suggest that both electron beam CT scanners and modern multidetector CT scanners can readily detect coronary artery calcification and, with quantification of calcium load, can stratify patients into risk groups for future coronary events. Despite existing controversies in this arena (the scope of this article precludes their definitive analysis), the fact is that a large number of patients have been referred by their physicians for such studies, and self-referral for such studies is clearly popular as well. The American Heart Association and the American College of Cardiology consensus statement supported the use of the coronary artery calcium score as a validated test, adding an additional risk factor to the traditional ones for estimating the presence of silent heart disease and for predicting future events [8].

At nearly the same time, the use of CT for lung cancer screening has evolved. Perhaps the most widely publicized study—the Early Lung Cancer Action Project (ELCAP)—documented that low-dose CT can greatly improve the likelihood of detection of small nodules and thus of lung cancer in an earlier and potentially more curable stage [9]. A follow-up report from the same group indicated that the cost-benefit analysis of such lung cancer screening is quite favorable and that with follow-up tests one can readily separate benign nodules (“pseudodisease”) from early cancer [10]. Despite corroborative evidence from other mass screening programs such as the one in Japan [11], skepticism persists regarding the value of such screening. The Society of Thoracic Radiology consensus statement did not recommend mass screening for lung cancer but did encourage trials [12]. The controversy centers on the question of whether early detection of lung cancer translates to decreasing overall disease-specific mortality (elimination of lead-time bias and length-time bias being key).

Many physicians are beginning to think that the delay imposed by a randomized clinical trial (perhaps a decade) makes such a trial's results an unacceptable Holy Grail. Exemplifying this attitude is the opinion of a chest surgeon who also discounts the fear of overdiagnosis bias [13]:

Lung cancer is a true monster...[it] consumes 160,000 people in our country each year. We should be hot on the track of this very real and deadly killer with timely screening and research based upon the promise of the ELCAP experience using low-dose, noncontrast, spiral CT scans. We must not delay a further decade to perform a randomized prospective trial designed to chase an apparition, overdiagnosed lung cancer, while millions die of under diagnosed advanced-stage lung cancer.

Such an endorsement of screening CT for lung cancer is based on the existing scientific evidence that screening CT can detect cancer in a much earlier stage, thus making it potentially curable. On the other hand, the current 5-year survival rate for detected lung cancer is 14%, a statistic that has not significantly improved for the last 25 years. Lung cancer is the most common killer in our society; most recent American Cancer Society statistics show nearly 170,000 new cases and nearly 160,000 deaths per year. Lung cancer kills more individuals than the next three most common cancers combined. We already know that untreated stage I lung cancer has a dismal short-term prognosis [14, 15].

The third category of targeted screening that is offered for self-referral by free-standing CT screening centers is CT colonography (virtual colonoscopy). A number of publications have suggested the usefulness of colonography for the detection of early colorectal cancer [16,17,18,19,20]. The interest in virtual colonoscopy has been fueled by the attention to colonic cancer from major media personalities such as Katie Couric of the Today show. Although periodic screening for colorectal cancer in individuals older than 50 years is widely recommended by physicians, the time-honored testing of feces for occult blood, even when combined with proctoscopy or sigmoidoscopy of the distal colon, suffers from poor sensitivity, with a 24% failure rate [21].

The gold standard screening examination (endoscopic colonoscopy) is one that suffers from poor patient compliance, with many patients refusing it. Indeed, physicians and nurses invited for free screening colonoscopy had an acceptance rate of fewer than 15% [22]. Colonoscopy has a sensitivity of 73% for polyps smaller than 5 mm, 87% for polyps of 6-9 mm, and 94% for polyps larger than 10 mm [23]. The procedure requires IV sedation, is costly, and is incomplete in 5-10% of cases; it also shows localization accuracy of only 86% for colorectal cancer [23]. Complications of perforation, bleeding, and even death (one in 500, three in 1000, and two in 10,000, respectively) occur. CT colonography is perceived as a minimally invasive and therefore more acceptable substitute for screening aimed at detecting colonic cancer in its early stages. The evidence is preliminary, but it is becoming more compelling. The early statistics of CT colonography indicate that polyp detection, particularly for those polyps most likely to harbor cancer (i.e., those > 8 mm), approaches 90% [24]. Thus, CT colonography places itself between air-contrast barium enema and conventional endoscopic colonoscopy in terms of early detection of colorectal cancer. Not surprisingly, a number of individuals have suggested that CT colonography may take a prominent role in screening for colorectal cancer [25].

The Role of Whole-Body Screening

Given this context, whole-body CT scanning in self-referred individuals is not truly “screening” in the realm of medical epidemiology. That word can be applied to this service only in the vernacular. It has been suggested that a set of common-sense criteria be used to evaluate true screening programs [26]. The first three criteria relate to the characteristics of the disease: The disease should have serious consequences; the disease should have a high prevalence of detectability in the preclinical phase (time from the onset of disease to the first appearance of signs and symptoms); and, finally, there should be little pseudodisease in the screening population. The next four criteria relate to the screening test itself: The test should have high accuracy for detecting the preclinical phase. The test should detect the disease before the critical point, that point at which treatment becomes relatively ineffective (e.g., when a primary tumor metastasizes). The test should cause little morbidity, and it should be affordable and widely available. The last three criteria relate to treatment: Treatment for the disease should exist, treatment should be more effective before symptoms begin, and treatment should not be too risky or toxic.

It is likely that lung cancer, and possibly colonic cancer, are suitable diseases for screening when the these criteria are scrutinized. Using such criteria for coronary calcium detection with CT may also be valid. Studies have certainly shown that preclinical coronary disease is common, and that intervention (particularly with statins) in early stages of coronary artery disease lowers the incidence of future coronary artery events.

Given the huge detrimental impact of coronary artery disease today, with more than 725,000 deaths annually, its detection with screening CT and the resultant improvement in the treatment of high-risk patients may be the biggest payoff of screening CT. Because the chest, abdomen, and pelvis are covered by the targeted screening studies, the other organs in the abdominal cavity come under scrutiny by default, thus providing a wholebody CT study. The term “screening” may therefore indeed apply to whole-body scanning even when patients self-refer without a standardized epidemiologic consensus. Obviously, guidelines for such self-referral patterns would be desirable. Currently, the entrepreneurial nature of many centers makes it difficult to exclude any individual who is willing to pay for the study. Most centers attempt to exclude individuals who are younger than 40 years old, but the financial incentive clearly tempts some to lower the threshold.

The Growing Debate

Partly because of inadequate data on cost—benefit analysis, organized radiology has not endorsed whole-body scanning or CT screening for any specific purpose. The notion that earlier detection of lung, or any other cancer, positively affects disease-specific mortality has not been proven. This higher threshold (prolonging life) has been applied by the American College of Radiology (ACR) to body CT screening [27], as compared with mammography, which is endorsed by the ACR with the statement that [28]:

ACR believes that clinical trials have shown that by having screening mammograms every year...breast cancers are found at an earlier stage. The earlier breast cancers are detected, the better are the chances for improved treatment results.

The issues of disease-specific mortality and the beneficial effects of CT screening, be it for cancer or for coronary artery disease, are part of the debate, as are the potential costs, issues of pseudodisease, and even radiation risks. Coronary artery disease more than cancer is a significant killer, often in a silent stage, yet has reasonable, effective therapy proven to be beneficial in preventing life-threatening coronary events, even when applied in the preclinical stages.

Whole-Body CT Screening and Radiation Fears

As an unnamed wit once observed, the problem isn't so much with what people don't know as with what people do know that isn't so.

Because coronary calcium detection with electron beam CT was the earliest form of CT screening, many of the existing centers were started by nonradiologists, and the resulting quality of full-body examinations in such centers suffers from the physical limitations imposed by first-generation electron beam CT technology. The speed of scanning with electron beam CT was suited to whole-body scanning, but its overall image quality suffered from heat limitations of the electron beam target and the amount of photon flux produced. Therefore, diagnostic body CT has generally been relegated to conventional CT scanners. These scanners have evolved now to multidetector technology, which has all but erased the speed advantage of electron beam CT while maintaining the quality of imaging necessary for optimal diagnosis that is expected in most medical diagnostic settings.

Nevertheless, multidetector CT—based outpatient screening centers are still in the minority when compared with ones equipped with electron beam CT, the latter being predominately organized by entrepreneurial ventures and staffed only variably by radiologists. Some newer radiologist-run facilities have concentrated on image quality and patient interaction. These facilities stress the optimal image quality of multidetector CT.

Concerns have been raised about unnecessary radiation exposure. Indeed, the Food and Drug Administration took notice, particularly after a report appeared in the radiology literature regarding high exposure levels with CT in the pediatric population [29]. Additional commentary on radiation exposure from CT appeared last year in the American Journal of Roentgenology [30]. Unfortunately, the context for radiation exposure from CT has not been fully elucidated in such commentary.

Radiation risk was first evaluated scientifically in the 1950s, and the evaluation has been updated by a variety of committees [31,32,33,34,35]. The Committee on the Biological Effects of Ionizing Radiation (BEIR V) estimated the risk of cancer death to be 0.04% per rem of effective body dose (“effective dose” is used to compare nonuniform or partial body exposure to an equivalent wholebody dose). These risk estimates were based on information gleaned from therapeutic uses of high-dose radiation and from survivors of the atomic bombs in Japan. The actual doses received by the survivors and some of the medical treatment groups have had to be approximated because they are not known precisely. For these estimates, a method was used to extrapolate the measured risks at high radiation doses and subsequent cancer rates over 40 years, to arrive at a risk of cancer induction from low radiation doses, such as those in diagnostic radiography, making the assumption that no threshold dose is needed for cancer induction.

A large difference exists between a whole-body single dose of neutron radiation (atomic bomb exposure) and doses to only part of the body. Estimating the risk over a life span, starting with childhood and early adulthood, from such total body exposure, and translating that to risk estimation from partial-body exposure in middle-aged and elderly individuals, is an even further approximation.

Thus, many critics have argued with the calculated risk statistics of low doses of radiation. The ACR's primer on radiation risk contains the following statements [36]:

For a low-dose event this calculation tacitly assumes that detriment, derived by model from high-dose data, can be applied to low-dose events. There is no evidence that such is the case. Conversely there is no evidence that such is not the case. There simply is no evidence...the public, the media, and even some scientists have accepted these models as established scientific fact. Such is not the case. The models can be used only for a crude first order approximation of risk.

Indeed, some studies document no evidence of increased cancer from low-dose radiation exposure to technologists receiving up to 15 rem and perhaps as much as 50 rem per year [37, 38]. At low levels of radiation, human cells (which have evolved over eons in a radiation-laden environment) repair radiation damage rapidly. In fact, a peer-reviewed body of literature exists on the beneficial effects of low-dose radiation. Nevertheless, the conservative approach of risk estimation used by BEIR V, and others, is the accepted one; therefore, it is prudent to put those estimates in the context of overall risk to health from everyday exposure to radiation as well as from other potentially threatening or harmful events.

In discussing radiation risk, this context needs to be kept in mind. Natural background radiation exposes the average American to 360 mrem per year. Those living in Denver experience an even higher dose, approximately 470 mrem, because of the presence of greater background radiation in the Denver plateau and cosmic ray exposure at that altitude [39]. The effective dose delivered for whole-body (chest—abdomen—pelvis, calcium scoring) screening at our facility is 880 mrem (as measured by an independent radiation physicist). Thus, approximately twice the dose of annual background radiation is delivered.

To compare the presumed risk from diagnostic imaging, let us consider that the death rate from cancer in the general population is quite high. Approximately 23% of all individuals will die of cancer (540,000 deaths a year); indeed, accepting the conservative estimates (with all the problems already discussed) that cancer induction risk is 0.04% per rem, one can calculate that of every 100,000 people scanned, 40 will have life-threatening cancer induced by radiation during their lifetimes. On the other hand, of the same 100,000 people, 23,000 are likely to die from cancer. Assuming even a 0.005% early detection rate and resulting cure, 115 people may derive the benefit versus the potential 40 who might have cancer induced sometime in their lives (if one subscribes to the downward linear extrapolation hypothesis). This is a rather simplistic analysis, however (much like the first order approximation cited previously).

Another way of looking at risk is to look at the relative risk of one in a million chances of dying of activities common in our society. For example, the relative risk of one in a million chances of dying from activities common in everyday life is smoking 1.4 cigarettes (lung cancer), eating 40 tablespoons (600 mL) of peanut butter, spending 2 days in New York City (air pollution), driving 40 miles (64 km) in a car (accident), flying 2500 miles (4000 km) in a jet (accident), canoeing for 6 min, and receiving 10 mrem of radiation [39].

Finally, comparing radiation exposure from screening CT with some other common radiologic procedures is useful. The effective dose of a barium enema is approximately 500 mrem, and a single-view chest radiograph is approximately 3.2 mrem (the same as a transcontinental airplane trip). Radiography of the lumbar spine (a five-view series) represents an effective total body dose of approximately 200 mrem. For those of us working in radiation environments, the allowable dose is 5000 mrem, and the estimated life expectancy loss from that dose is 51 days (vs risks on the job in all industries averaging a life expectancy loss of 60 days; for agriculture, 320 days; and for construction, 327 days).

Body CT Screening and the Radiologist

As mentioned in the introduction, a considerable amount of debate exists as to whether money spent in identifying early disease through screening programs is of benefit. Critics claim that asymptomatic individuals have a low prevalence of significant disease and that a large number of false-positive findings will result in excessive anxiety for patients, not to mention needless further testing and the resulting increased risk from invasive procedures. Tables 1 and 2 show the experience at my screening center in the first 6 months (January 15-July 3, 2001) with a stand-alone CT screening program. Almost half of the “abnormalities” were pulmonary nodules, and the second report from the Early Lung Cancer Action Project [10] indicates that repeated screening resolves the issue of these “false-positive” findings. Stage I lung cancer detection (Fig. 1), detection of silent renal carcinoma (Fig. 2A,2B), a large unsuspected abdominal aortic aneurysm (Fig. 3), and an unsuspected lymphoma (Fig. 4) serve as examples of the potentially great benefit achieved by the affected individuals subjecting themselves to the service (few would argue that benefit if the findings were made incidentally on CT ordered by a physician for some other purpose). The high prevalence of silent coronary disease (death in 180,000 people each year is the first sign of this killer) makes the potential of CT in early detection particularly exciting, as shown in Figure 5A,5B.

View larger version:

• In this page


• In a new window

• Download as PowerPoint Slide

Fig. 1. —Stage I lung cancer detected in 62-year-old healthy woman who is former smoker. CT scan shows spiculated left upper lobe nodule. Surgical resection 1 week later verified stage IA squamous cell cancer.

View larger version:

• In this page


• In a new window

• Download as PowerPoint Slide

Fig. 2A. —Renal cancer detected in 58-year-old healthy self-referred man. CT scan shows mass in lower pole of right kidney. No evidence of metastasis was found at workup before surgical removal.

View larger version:

• In this page


• In a new window

• Download as PowerPoint Slide

Fig. 2B. —Renal cancer detected in 58-year-old healthy self-referred man. Coronal reformation shows lower pole right renal mass.

View larger version:

• In this page


• In a new window

• Download as PowerPoint Slide

Fig. 3. —Aortic aneurysm detected in 60-year-old man who requested CT screening as adjunct to annual physical examination (which had normal findings). CT scan shows large abdominal aortic aneurysm that was surgically repaired 2 weeks after study.

View larger version:

• In this page


• In a new window

• Download as PowerPoint Slide

Fig. 4. —Lymphoma detected in 57-year-old, self-referred healthy woman. Screening CT scan shows left paraaortic lymphadenopathy (arrow). Lymphoma was found at biopsy.

View larger version:

• In this page


• In a new window

• Download as PowerPoint Slide

Fig. 5A. —Coronary artery stenosis detected in 70-year-old man with atypical symptoms and negative stress test findings who was referred by cardiologist for calcium scoring. IV CT coronary angiogram shows multifocal calcium deposits and stenoses (arrows) in left anterior descending artery.

View larger version:

• In this page


• In a new window

• Download as PowerPoint Slide

Fig. 5B. —Coronary artery stenosis detected in 70-year-old man with atypical symptoms and negative stress test findings who was referred by cardiologist for calcium scoring. Coronary catheter angiographic image that was obtained as result of high calcium score verfies severe multifocal stenoses (arrows) of left anterior descending coronary artery.

The short time from our experience allows availability of a final diagnosis in only the minority of our patient cohort with abnormal findings, but even in those diagnoses available, the cancer detection rate of 1% is telling. These statistics echo similar preliminary results from the cancer screening program at the Mayo Clinic, in which 14% of individuals were found to have clinically significant abnormalities in the body when lung cancer CT screening was extended to the pelvis (Jett J, personal communication). Despite only a limited time with this service to date, all of us providing CT screening have experienced considerable satisfaction from seeing the number of patients who have benefited from it, including, we believe, many with their lives prolonged.

With the advent of screening mammography, radiologists began to be primarily and routinely involved in direct patient contact during diagnostic studies. As imaging becomes more sophisticated, radiology can identify early markers of disease before clinical manifestations are seen. Given the lack of evidence regarding life benefit for many interventions in disease that is already manifest, it is intuitively arguable that diagnostic detection in preclinical stages of disease should be encouraged. Such intuitive arguments are applied to screening mammography. In criticizing the recently released Danish data about lack of proof that mammography saves lives [40], Robert Smith of the American Cancer Society said, “...On a most basic level, we understand very clearly why mammography works. Mammography finds tumors when they are smaller... the simple logic of finding tumors earlier has given [women] confidence in mammography” [41]. Smith is the head of cancer screening of the American Cancer Society and is not a fan of whole-body CT screening [41].

Indeed, as radiologists become more involved in screening programs that involve imaging (be it CT, mammography, or even sonography and MR imaging), they assume a new role. Given their knowledge of the various medical specialties with which they interact daily, their understanding of diagnostic imaging, and their knowledge of the interplay between these disciplines, radiologists may be well positioned to optimize triage for patients who need further medical care. Providing reassurance to those anxious healthy individuals whose screening studies are unremarkable, or whose scans show pseudodisease, and guiding them in proper health maintenance, is not too great a stretch. Primary care physicians do it every day. They also find pseudodisease and may order “unnecessary” tests if their judgment fails. With the limitations of time and resources placed on primary care physicians in the managed care system, it is not unreasonable to provide access to patients interested in health maintenance and wellness through the screening portal. Doing so can only enhance the image of radiologists as true physicians, ones whose particular expertise is diagnostic image interpretation and translation of that information into appropriate patient treatment.

Screening CT may simply be a Trojan horse for other forms of direct patient interaction with diagnostic radiologists. Screening CT is, like mammography, taking the radiologist from the back room of doctor-to-doctor consultation, and placing him or her in the front office of primary patient treatment. Some physicians, including radiologists, fear that their role will be diminished when the consumer is allowed to direct health care. It seems silly to consider that some of us readily accept a woman's right to choose an abortion but not her right to choose screening CT, despite the fact that more women die of lung cancer than breast cancer.

Yet consumers are clearly taking over health care decision making. A recent article in the Journal of the American Medical Association stated [42]:

If the right thing for healthcare is defined as an approach without potential problems of equity, efficiency, and clinical quality, then consumerism fails the test. All other candidates for setting priorities and managing care, including government, employers, insurers and physicians, also fail the test. But if the right thing is defined as the approach most compatible with the nation's social culture and political institutions, the candidate that remains standing after the other contestants are vanquished, then consumerism is not only the likely but indeed the right thing for US healthcare.

The source of the quotation is not a conservative think-tank. The author is a professor of public health at the University of California at Berkeley. The title of the article is telling: “The End of Managed Care.” Radiologists take note. Our future role in the health care system can be redefined, if we care or dare to do so.