Ten Criteria for Effective Screening: Their Application to Multislice CT Screening for Pulmonary and Colorectal Cancers
Multislice or multidetector CT is a relatively new technology that offers great potential as a screening tool because it is fast, potentially very accurate, and realtively noninvasive. During the last few years, many institutions in the United States have established screening programs for pulmonary cancer and colorectal adenomatous polyps using single and multislice helical CT. Establishing an effective screening program, however, is not a trivial task. We cannot assume that early detection of disease translates into improved patient outcome. In fact, early detection of disease can lead to higher morbidity and mortality rates. The issues involved in effective screening are complex and multifaceted.
We researched the literature on the epidemiology of screening and have formulated 10 criteria for evaluating screening programs, which we summarize in the Appendix. These criteria encompass the characteristics of the disease, the screening test, and treatment. We apply these criteria to screening for pulmonary and colorectal cancer using multislice CT. Using our findings, we compared multislice CT for the detection of lung and colorectal cancer with mammography for the detection of breast cancer (Table 1). Although mammography is clearly not an ideal screening model [1], it is the only imaging screening test in widespread use. Our evaluation of these screening programs using the proposed 10 criteria allowed us to identify strengths and weaknesses of the programs, shortcomings of our literature, and directions for future research.
| Criteria | Lung Cancer | Colorectal Caner | Breast Cancer |
|---|---|---|---|
| Disease | |||
| 1. Consequences | No. 1 cause of cancer deaths | No. 2 cause of cancer deaths | No. 2 cause of cancer in women |
| 2. Prevalence of detectable preclinical phase | In smokers, 2-4% | In 50-year-old persons, 3% | 0.6-1.0% |
| Screening test | |||
| 3. Pseudodisease | Negligible | 2.5 polyps per 1000/year progress to cancer; small polyps often regress | About half of DCIS becomes invasive |
| 4. Accuracy for detectable preclinical phase | ? | ? | Sensitivity, 83-95%; false-positive rate, 0.9-6.5% |
| 5. Critical point | ? | Polyp at 1-2 cm | Stage I |
| 6. Morbidity | No short-term, radiation exposure | Patient discomfort, radiation exposure | Patient discomfort, radiation exposure |
| 7. Affordability | Out-of-pocket | Out-of-pocket | Insurance |
| Availability | Yes | Yes | Yes |
| Treatment | |||
| 8. Existence | Resection | Polypectomy | Resection |
| 9. Preclinical effectiveness | ? | Reduction in mortality, 16% | Controversial |
| 10. Risks and toxicity | Mortality, 3.7% | Mortality, 0.01% | Mortality, 0% |
| Note.—DCIS = ductal carcinoma in situ, ? = no data available. | |||
We now discuss our 10 criteria for effective screening.
1. Disease Has Serious Consequences
Screening should target diseases with serious consequences such as mortality or severe or prolonged morbidity [2]. The value of screening for such diseases is easy to justify from both an emotional and a cost-effectiveness standpoint.
Both pulmonary and colorectal cancer are serious diseases, being the first and second leading causes of cancer death in the United States [3], respectively. Breast cancer is the second leading cause of cancer death in women. Thus, all three cancers have serious consequences.
2. Screening Population Has High Prevalence of Detectable Preclinical Phase
To justify the cost of screening, the detectable preclinical phase of the disease should have a high prevalence among people who are screened [2]. The natural history of disease is divided into the preclinical and clinical phases [2, 4, 5] (Fig. 1). The preclinical phase is the time from the onset of disease to the first appearance of signs and symptoms. The preclinical phase ends when the patient seeks medical care [2], which depends on the population's awareness of the disease and the patient's access to health care [5]. The detectable preclinical phase is a component of the preclinical phase; it is the interval of time when the disease is detectable by the screening test.
Prevalence is the proportion of patients in the detectable preclinical phase among all patients screened. If the prevalence is low, then screening will not detect many cases of disease, so it may not be cost-effective. Also, if the prevalence is low, then even if the screening test is accurate, the probability of disease after a positive test result will still be low. For example, using Bayes' theorem, if the prevalence is 1% and the test's sensitivity and specificity are both 95%, then the probability of disease after positive test results is only 16%. In contrast, if the prevalence is 5%, then the probability of disease after positive test results is 50%.
For pulmonary cancer, the target population for screening is asymptomatic men and women who are 40 years old or older and who have a history of smoking of at least 10 pack-years. The prevalence of detectable pulmonary cancer in this population is 2-4% [6,7,8]. For colorectal cancer, the target population is asymptomatic men and women who are 50 years old or older and who do not have a known risk factor for colon cancer such as familial polyposis or ulcerative colitis. The prevalence of a 1-cm or larger adenomatous polyp in a 50-year-old person is 3% and in an 80-year-old person is 3% and in an 80-year-old is 5-6% [9]. Compared with breast cancer, for which the prevalence in the screening population is only 0.6-1.0% [10,11,12], both pulmonary and colorectal cancers have a high prevalence in the detectable preclinical phase.
3. Screening Test Detects Little Pseudodisease
Pseudodisease, by definition, is not disease. It appears to be disease on the screening test, but it does not and will never negatively affect the patient's life. Two types of pseudodisease have been described. In the first type, the disease never progresses and, in fact, may regress naturally. This has been termed type I pseudodisease [4]. Screening tests cannot distinguish type I pseudodisease from true disease, so the patient undergoes unnecessary and sometimes risky tests and procedures. In type II pseudodisease, the disease progresses so slowly that the patient never develops symptoms and dies from another cause. Type II pseudodisease is common in diseases with long detectable preclinical phases or among patients with short life expectancies [5]. Patients with type II pseudodisease also undergo unnecessary tests and treatment but derive no benefit from the treatment. Screening tests that detect a high frequency of pseudodisease cannot be cost-effective.
No direct data exist on the frequency of pseudodisease of pulmonary carcinoma. In particular, no good data exist on the cause-specific death rate for patients with pulmonary cancer. Similarly, no data exist on the incidental occurrence of pulmonary cancer in large autopsy series. Intuitively, however, the frequency must be small, because 80-100% of untreated asymptomatic patients with detected stage I pulmonary cancer die within 5-10 years [13, 14].
With colorectal cancer, not all adenomatous polyps progress to invasive carcinoma. Evidence shows that many small (<1 cm) polyps regress [15]. The rate of adenomatous polyps progressing to cancer has been estimated at about 2.5 polyps per 1000 individuals per year [16]. The presence of pseudodisease for colorectal cancer is supported by a large autopsy study [17]. In this series, colorectal cancer was present in 0.5% of 50-60 year olds, 1% of 60-70 year olds, and 1.5% of 70-80 year olds. None of these individuals was symptomatic, and the cause of death was unrelated.
Similarly, not all breast ductal carcinoma in situ progresses to invasive carcinoma. In one study of 15 patients with untreated breast ductal carcinoma in situ, eight developed invasive carcinoma [18]. In another study of 25 women with untreated breast ductal carcinoma in situ, seven developed invasive carcinoma; the remaining women were followed up for at least 8 years without evidence of invasive tumors [19]. From autopsy studies of women who died of causes other than breast cancer [20,21,22], the prevalence of breast ductal carcinoma in situ was 6-14%. Considerable work has been done in correlating the different subtypes (i.e., architectural, histologic) of breast ductal carcinoma in situ with rates of subsequent invasive carcinoma in an effort to determine the preferred treatment (i.e., lumpectomy versus mastectomy) [23]. However, the presence of pseudodisease in screening for both colorectal polyps and breast cancer limits the effectiveness of these screening programs.
4. Screening Test Has High Accuracy for Detecting the Detectable Preclinical Phase
The screening test must have good sensitivity and specificity. To detect more true-positive cases than false-positive cases when the prevalence of disease is less than or equal to 5% (which covers most screening populations), a screening test must have sensitivity exceeding 95% if the specificity is less than or equal to 95% and vice versa (specificity must be >95% if the sensitivity is ≤95%). Most screening tests do not meet this high standard, which means that the screening program must absorb the costs of many false-positive results.
Increasing the specificity of a screening test will increase the cost-effectiveness of screening. However, it is not always cost-effective to increase a screening test's sensitivity [5]. An increase in sensitivity might mean an increase in the detection of pseudodisease or an increase in the detection of disease after the critical point in the natural history (i.e., after the primary tumor metastasizes). Both these situations are detrimental to screening.
The accuracy of a screening test is difficult to estimate [2]. One common but erroneous approach is to measure a screening test's accuracy among patients who have symptomatic disease. This approach will be misleading because the sensitivity of a test for detecting the preclinical phase will often be less than its sensitivity for detecting clinical disease. A better approach is to apply the screening test to asymptomatic people, then follow up those individuals for a sufficient length of time to determine their true disease status [2]. The thoroughness of the follow-up and the duration of follow-up are critical. If only those who test positive on the screening test are followed up, then the estimate of sensitivity will tend to be falsely high and the estimate of specificity will tend to be falsely low. If the duration of follow-up is too short, then some cases of disease will be missed; if the followup is too long, then new cases of cancer will incorrectly be assessed as false-negative.
No good studies indicate the accuracy of multislice helical CT for detecting either pulmonary cancer or colorectal polyps in asymptomatic people. For pulmonary cancer, three recent studies [8, 24, 25] compared the yield of helical CT and chest radiography, but no studies have been published addressing the sensitivity and specificity of CT in a screening population. In contrast, a number of studies have reported the accuracy of mammography in screening populations. A meta-analysis that was published in 1998 reported a sensitivity range of 83-95% for mammography and a range in the false-positive rate of 0.9-6.5% [26].
5. Screening Test Detects Disease Before Critical Point
For most diseases, a critical point occurs in the natural history of the disease; treatment is more effective before this point and less effective after this point (Fig. 1). For most cancers, the critical point occurs when the primary tumor metastasizes [5]. If the critical point occurs before the detectable preclinical phase, then screening cannot be effective. If the critical point occurs soon after the start of the detectable preclinical phase, then screening will often be too late. Furthermore, the test's sensitivity may be too low at the beginning of the detectable preclinical phase, as with very small lesions. Finally, if the critical point occurs after symptoms appear (i.e., in the clinical phase), then screening cannot be cost-effective.
Three recent studies [8, 24, 25] suggest that CT can detect stage I pulmonary cancer is asymptomatic people. In these three studies, 71-93% of people with pulmonary cancer detected by screening had stage I disease. These numbers sound encouraging, yet the critical point in the natural history of pulmonary cancer has often passed by the time stage I cancer is detected. The 5-year survival rate of patients diagnosed with stage I pulmonary cancer is currently only 49% at best [27]. Compare that with the 5-year survival rates of stage I breast and colorectal cancers—97% and 92%, respectively [27]. With respect to all stages of pulmonary cancer, the 5-year survival is only 14% (versus 86% for breast and 62% for colorectal cancer) [27]. Furthermore, this survival rate has changed little in the past 30 years: in 1970, the 5-year survival rate among all cases of pulmonary cancer measured only 10% [27].
For colorectal cancer, the critical point is usually defined as progression of an adenomatous polyp to larger than 1 cm. The risk of carcinoma in a 1-cm adenomatous polyp is less than 1%; for polyps of 1-2 cm, the risk is 5-10%; and for polyps larger than 2 cm, the risk is 10-50% [15]. Metastases from malignant polyps smaller than 1 cm are rare.
For breast cancer screening, the survival rate decreases directly with increasing cancer stage [28]. For stages 0 and I disease, the 5-year survival rates are 92% and 87%, respectively. At stage II, the 5-year survival rate decreases considerably: 78% and 68% for stages IIA and IIB, respectively. Thus, the critical point is at stage I, before metastasis to the regional lymph nodes.
One way to increase the chance of detecting disease before the critical point is to offer screening at an optimal age and with multiple, optimally spaced, screenings [2]. The optimum interval is the one that provides the most benefit to patients relative to costs (both monetary and patient outcome costs). In general, the interval between screenings should be relatively short when screening is inexpensive, the cost of workup of false-positive cases is low, and the time from when the disease is first detectable to the critical point is relatively short.
Other important factors include the sensitivity of the test for the detectable preclinical phase, the prevalence of the disease, and the range of duration of the detectable preclinical phase before the critical point (rather than just the average duration) [2]. A number of reports address the determination of the optimum screening age [29] and the optimum interval between screenings [29,30,31,32,33].
6. Screening Test Causes Little Morbidity
The screening test must not inflict mortality or significant morbidity on those screened. At the time of screening, a person's risk of death or serious morbidity from the target disease is relatively small. Thus, even a small adverse effect to many of the screened persons is likely to offset any substantial benefit of screening afforded to a few [5].
For pulmonary cancer screening, the CT study is performed without IV contrast material, so short-term toxicity is not a problem. The long-term adverse effect could be related to radiation exposure. However, in screening for pulmonary cancer low-dose CT is performed, which uses only 20% of the conventional dose (i.e., 20% of 0.4 rads). The increased risk of developing fatal cancer is estimated to be one in 25,000 as a result of a single screening pulmonary CT examination, although this has not been studied directly [34,35,36,37].
For colorectal cancer screening, there is preparation discomfort and dehydration, as well as procedure discomfort (distention, cramping, abdominal pain). The risk of perforation is negligible because the rectal tube is soft and smaller than it is when an air contrast barium enema is performed. The long-term effect of screening is low-dose radiation exposure. As with low-dose chest CT, the longterm risk is negligible.
For breast cancer screening, the short-term effect is patient discomfort. A conservative estimate of the long-term effect is based on the radiation dose from annual two-view mammography beginning at age 40 and continuing to age 75; the estimated excess risk of breast cancer death is one in 7246 [38]. Thus, the risks associated with pulmonary and colorectal cancer screening are no greater than with mammography.
7. Screening Test Is Affordable and Available
The diagnostic test must be affordable and available to the target population. If the test is available only at large urban medical facilities, or is not affordable for patients, or both, then screening cannot be effective.
At our institution, the charge for bilateral screening mammography is approximately two thirds that of screening pulmonary CT. The charge for CT colorectal cancer screening has yet to be determined but will also modestly exceed that of mammography. The charge for pulmonary cancer CT screening is paid by the patient out-of-pocket, although mammography screening is generally covered by most medical insurance. Although this may change if CT screening is found to be cost-effective, the current direct cost of CT screening may prevent many patients from undergoing the examination. CT is probably as widely available as mammography, and many radiology departments have helical CT. Multislice scanners are being placed in radiology departments throughout the United States.
8. Treatment Exists
An effective treatment for the disease must exist for screening to improve patient outcomes. Detection of disease alone is not cost-effective. This may seem a trite criterion for screening, but it is important because many common diseases (e.g., Parkinson's disease, multiple sclerosis, Alzheimer's) have no treatment. Although it may be possible to detect these conditions preclinically, screening cannot be cost-effective if no treatment exists.
For both pulmonary cancer and colorectal polyps, resection and in some cases radiation or chemotherapy is the acceptable treatment. The approach to treatment for breast cancer is similar.
9. Treatment Is More Effective When Applied Before Symptoms Begin
For screening to be cost-effective, treatment must be more effective or less toxic when applied during the detectable preclinical phase, as compared with treatment applied after symptoms begin [2]. Screening cannot be cost-effective if the disease can be treated successfully after symptoms appear.
It is surprisingly difficult to demonstrate the benefit of early treatment, even when the treatment is known to be effective once symptoms appear. Four problems exist when comparing survival of screened patients with that of unscreened patients: lead-time bias [39,40,41,42], length bias [42, 43], overdiagnosis bias [5], and stage migration bias [44]. We briefly describe each of these biases.
Lead-Time Bias
For unscreened patients, disease-specific survival is measured as the length of time from clinical diagnosis of disease to death from the disease. For screened patients, disease-specific survival is the length of time from disease detection by the screening test to death from the disease. The lead time is the length of time between disease detection and the first appearance of signs or symptoms [2, 39,40,41,42] (Fig. 1). Even if early treatment has no benefit, the survival of screened patients is greater than that of unscreened patients by the addition of the length of the lead time.
Length Bias
Length bias occurs because not all patients' disease progresses at the same rate [2, 42, 43]. Cases of slowly progressing disease have a longer detectable preclinical phase than cases of quickly progressing disease. The slowly progressing disease is easier to detect; thus persons with slowly progressing disease will tend to be overrepresented in the screening cohort. Because the disease is progressing slowly in these patients, their survival is naturally longer, regardless of the effectiveness of treatment. However, the longer survival is usually wrongfully attributed to early treatment.
Overdiagnosis Bias
Overdiagnosis bias occurs when one does not adjust for the occurrence of pseudodisease in the screened cohort. Patients with pseudodisease do not die from the specific disease under study. The survival of these patients is often erroneously attributed to early treatment [5].
Stage Migration Bias
Stage migration bias, or the “Will Rogers phenomenon” [44], can occur when the disease-specific survival rates of the screened and unscreened patients are compared according to the stage at which the disease was detected. Stage migration occurs when a cancer's metastases are detected before any symptoms of metastasis appear. The early detection of metastases shifts the TNM stage from stage I or II to stage II or III. When this migration occurs, the survival in the lower stages of the screened cohort appears greater than in the unscreened cohort because the patients with the silent metastases have been removed from the lower stages. Similarly, the survival in the higher stages of the screened cohort appears greater than in the unscreened cohort because early cases of metastasis, with a naturally longer survival time, have been added. The overall survival rate of the screened and unscreened patients could actually be identical, yet the survival rates in each stage falsely suggest that screening is effective.
Because of these potential biases, disease-specific survival is not a useful measure when studying the effectiveness of early treatment. A better measure, although not ideal, is to compare disease-specific mortality [4]. The disease-specific mortality rate is computed as the number of deaths from the specific disease divided by the number of people at risk. This measure is not subject to the same biases; however, this measure is not especially sensitive to some types of treatment benefits. For example, if early treatment truly does increase the duration of survival but does not prevent death, then this type of benefit will be overlooked. Another problem is that even with large study sample sizes, it is difficult to observe enough disease-specific deaths for meaningful statistical comparisons among cohorts.
Two studies have shown a promising trend of increased survival from pulmonary cancer as the size of the tumor decreases [45, 46]; however, to our knowledge, all published studies are potentially flawed because of biases associated with using disease-specific survival.
In 1998, Towler et al. [47] published a meta-analysis of randomized controlled trials of colorectal cancer screening with fecal occult blood testing. These researchers used disease-specific mortality rates to assess the benefit of screening and found that early detection reduces mortality from colorectal cancer by 16%.
Many large randomized trials have assessed the benefit of early treatment for women screened by mammography. The conclusions from these studies are highly controversial [1].
10. Treatment Is Not Too Risky or Toxic
Treatment cannot be so risky or toxic that it offsets its long-term benefits. This is particularly important when many false-positive cases or many cases of pseudodisease undergo treatment; these patients derive no benefit from treatment, only its side effects. It's also important to realize that early treatment means that the detrimental effects of treatment are felt earlier and often for a longer period of time than when treatment begins with the onset of symptoms [2]. Because most people put greater value on their next few years of life than on future years, these early detrimental effects of treatment are a powerful negative effect on the patient's quality of life.
As with the detection of a lesion on mammography, detection of a pulmonary nodule on CT may prompt percutaneous biopsy before surgical resection to minimize the number of false-positive findings at surgery. The 30-day mortality rate after pulmonary resection is 3.7% [48]. In addition, both minor and major complications are associated with surgery (e.g., atelectasis, pneumothorax, bronchopleural fistula, myocardial infarction atrial arrhythmia, pulmonary embolism, prolonged pulmonary insufficiency, chylothorax, and chronic thoracic pain). For patients whose disease is unresectable or inoperable, radiation therapy is used, with complications of lethargy, fatigue, and skin and lung toxicity.
The mortality rate associated with endoscopic polypectomy is only 0.01% [49]. The complication rate (e.g., bleeding) is 1-2.3%; the perforation rate is 0.3%. These are acceptable rates. Similarly, the mortality rates from lumpectomy and mastectomy are effectively zero.
Conclusions
The advantage of screening for pulmonary cancer is that the disease has serious consequences and probably very little pseudodisease. That is, any pulmonary cancer detected is likely to be of serious clinical consequence within 5-10 years. A potentially critical shortcoming of CT screening is the apparent inability to detect the disease before the critical point in its natural history. Also, the risks and complication rates associated with pulmonary cancer treatment are much higher than those for breast and colorectal cancer. No good studies indicate the accuracy of screening CT or the effectiveness of early treatment; these areas need further research.
The strengths of colorectal cancer screening are that it is a serious disease with a long detectable preclinical phase before the critical point in the natural history. Good data are available on the benefit of early treatment (i.e., polypectomy). A potential shortcoming is the presence of polyps that never progress to invasive carcinoma. The literature is insufficient to evaluate the significance of pseudodisease on the cost-effectiveness of screening. Insufficient data exist as to the accuracy of CT for detecting adenomatous polyps.
Characteristics of: | Specific Criterion: |
| Disease | 1. Serious consequences 2. High prevalence of detectable preclinical phase |
| Screening test | • 3. Little pseudodisease • 4. High accuracy for detectable preclinical phase • 5. Detection of disease before critical point • 6. Little morbidity • 7. Affordability and availability |
| Treatment | • 8. Existence • 9. Effectiveness before symptoms begin • 10. Little risks or toxicity |
Acknowledgments
We appreciate the helpful suggestions and direction provided by Michael T. Modic, Richard C. White, and William A. Chilcote on early drafts of this article.
Footnote
Address correspondence to N. A. Obuchowski.
References
1.
Gotzsche PC, Olsen O. Is screening for breast cancer with mammography justifiable? Lancet 2000; 355:129-134
2.
Cole P, Morrison AS. Basic issues in population screening for cancer. J Natl Cancer Inst 1980; 64:1263-1272
3.
Smith RA, Mettlin CJ, Johnston DK, Eyre H. American Cancer Society guidelines for the early detection of cancer. CA Cancer J Clin 2000; 50:34-49
4.
Morrison AS. Screening in chronic disease. New York: Oxford University Press, 1992: 25-42
5.
Black WC, Welch HG. Screening for disease. AJR 1997; 168:3-11
6.
Eddy DM. Screening for lung cancer. Ann Intern Med 1989; 111:232-237
7.
Nesbitt JC, Putnam JB Jr, Walsh GL, Roth JA, Mountain CF. Survival in early stage non-small cell lung cancer. Ann Thorac Surg 1995; 60:466-472
8.
Henschke CI, McCauley DI, Yankelevitz DF, et al. Early lung cancer action project: overall design and findings from baseline screening. Lancet 1999; 354:99-105
9.
Neugut AI, Jacobsen JS, Rella VA. Prevalence and incidence of colorectal adenomas in asymptomatic persons. Gastrointest Endosc Clin N Am 1997; 7:387-399
10.
Braman DM, Williams HD. ACR accredited suburban mammography center: three year results. J Fla Med Assoc 1989; 76:1031-1034
11.
Burhenne LJW, Hislop TG, Burhenne HJ. The British Columbia mammography screening program: evaluation of the first 15 months. AJR 1992; 158:45-49
12.
Linver MN, Paster S, Rosenberg RD, et al. Improvement in mammography interpretation skills in a community radiology practice after dedicated teaching courses: 2-year medical audit of 38,633 cases. Radiology 1992; 184:39-43
13.
Sobue T, Suzuki R, Madsuda M, Kuroishi T, Ikeda S, Naruke T. Survival for clinical stage I lung cancer not surgically treated. Cancer 1992; 69:685-692
14.
Flehinger BJ, Kimmel M, Melamed MR. The effect of surgical treatment on survival from early lung cancer: implications for screening. Chest 1992; 101:1013-1018
15.
Winawer SJ, Fletcher RH, Miller L, et al. Colorectal cancer screening: clinical guidelines and rationale. Gastroenterology 1997; 112:594-642
16.
Anderson LM, May DS. Has the use of cervical, breast, and colorectal cancer screening increased in the United States? Am J Public Health 1995; 85:840-842
17.
Berg JW, Downing A, Lukes RJ. Prevalence of undiagnosed cancer of the large bowel found at autopsy in different races. Cancer 1970; 25:1076-1080
18.
Rosen PP, Braun DW Jr, Kinne DE. The clinical significance of pre-invasive breast carcinoma. Cancer 1980; 46:919-925
19.
Page DL, Dupont WD, Rogers LW, Landenberger M. Intraductal carcinoma of the breast: follow-up after biopsy only. Cancer 1982; 49:751-758
20.
Kramer WM, Rush BF Jr. Mammary duct proliferation in the elderly. Cancer 1973; 31:130-137
21.
Alpers CE, Wellings SR. The prevalence of carcinoma in situ in normal and cancer-associated breasts. Hum Pathol 1985; 16:796-807
22.
Nielsen M, Thomsen JL, Primdahl S, Dyreborg U, Andersen JA. Breast cancer and atypia among young and middle-aged women: a study of 110 medicolegal autopsies. Br J Cancer 1987; 56:814-819
23.
Fechner RE. One century of mammary carcinoma in situ: what have we learned? Am J Clin Pathol 1993; 100:654-661
24.
Kaneko M, Eguchi K, Ohmatsu H, et al. Peripheral lung cancer: screening and detection with low-dose spiral CT versus radiography. Radiology 1996; 201:798-802
25.
Mori K, Tominago K, Hirose T, Sasayawa M, Yokoyama K, Moriyama N. Utility of low-dose helical CT as a second step after plain chest radiography for mass screening for lung cancer. J Thorac Imaging 1997; 12:173-180
26.
Mushlin AI, Kouides RW, Shapiro DE. Estimating the accuracy of screening mammography: a meta-analysis. Am J Prev Med 1998; 14:143-153
27.
Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics. CA Cancer J Clin 1998; 48:6-29
28.
Kopans DB. Breast imaging, 2nd ed. Philadelphia: Lippincott-Raven, 1998: 113
29.
Parmigiani G. On optimal screening ages. J Am Stat Assn 1993; 88:622-628
30.
Eddy DM. A mathematical model for timing repeated medical tests. Med Decis Making 1983; 3:34-62
31.
Kirch RLA, Klein M. Surveillance schedules for medical examinations. Manag Sci 1974; 20:1403-1409
32.
Shahani AK, Crease DM. Towards models of screening for early detection of disease. Adv Appl Probl 1977; 9:665-680
33.
Zelen M. Optimal scheduling of examinations for the early detection of disease. Biometrika 1993; 80:279-293
34.
Faulkner K, Moores BM. Radiation dose and somatic risk from computed tomography. Acta Radiol 1987; 28:483-488
35.
Mossman KL. Analysis of risk in computerized tomography and other diagnostic radiology procedures. Comput Radiol 1982; 6:251-256
36.
Renston JP, Connors AF, DiMarco AF. Survey of physicians' attitudes about risks and benefits of chest computed tomography. South Med J 1996; 89:1067-1073
37.
Dixon AK, Dendy P. Spiral CT: how much does radiation dose matter? Lancet 1998; 352:1082-1083
38.
Committee on the Biological Effects of Ionizing Radiation, National Research Council. Health effects of exposure to low levels of ionizing radiation: BEIR V (1990). Washington, DC: National Academy Press, 1990
39.
Hutchinson GB, Shapiro S. Lead time gained by diagnostic screening for breast cancer. J Natl Cancer Inst 1968; 41:665-681
40.
Prorok PC. The theory of periodic screening. I. Lead time and proportion detected. Adv Appl Probl 1976; 8:127-143
41.
Prorok PC. The theory of periodic screening. II. Doubly bounded recurrence times and mean lead time and detection probability estimation. Adv Appl Probl 1976; 8:460-476
42.
Black WC, Welch HG. Advances in diagnostic imaging and overestimation of disease prevalence and the benefits of therapy. N Engl J Med 1993; 328:1237-1243
43.
Zelen M. Theory of early detection of breast cancer in the general population. In: Henson JC, Mattheiem WH, Rozencweig M, eds. Breast cancer: trends in research and treatment. New York: Raven, 1976: 287-300
44.
Feinstein AR, Sosin DM, Wells CK. The Will Rogers phenomenon: stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer. N Engl J Med 1985; 312:1604-1608
45.
Steele JD, Kleitsch WP, Dunn JE, Buell P. Survival in males with bronchogenic carcinomas resected as smptomatic solitary pulmonary nodules. Ann Thorac Surg 1966; 2:368-376
46.
Jackman R, Good CA, Clagett OT, Woolner LB. Survival rates in peripheral bronchogenic carcinoma up to four centimeters in diameter presenting as solitary pulmonary nodules. J Thorac Cardiovasc Surg 1969; 57:1-8
47.
Towler B, Irwig L, Glasziou P, Kewenter J, Weller D, Silagy C. A systematic review of the effects of screening for colorectal cancer using the faecal occult blood test, Hemoccult. Br Med J 1998; 17:559-565
48.
Ginsberg RJ, Hill LD, Eagan RT, et al. Modern thirty-day operative mortality for surgical resection in lung cancer. J Thorac Cardiovasc Surg 1983; 86:654-658
49.
VanDerWerken BS, Wu WC. Endoscopic evaluation of colon polyps. Semin Roentgenol 1996; 31:118-124
Information & Authors
Information
Published In
Copyright
© American Roentgen Ray Society.
History
Submitted: September 26, 2000
Accepted: November 13, 2000
First published: November 23, 2012
Authors
Metrics & Citations
Metrics
Citations
Export Citations
To download the citation to this article, select your reference manager software.