Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
APPENDIX A CASE STUDY OF MAMMOGRAPHY B Morris F. Co lien INTRODUCTION Mamnography provides a good case study of equipment-embodied technology, since it exemplifies many of the problems considered in this committee's report. It is used for diagnostic purposes and involves a significant capital investment, special facil- ities, equipment, and trained personnel. It has sufficiently diffused into the practice of medicine to have a significant impact on the diagnosis of breast cancers (the most common can- cer in women). The currently evolving policies for its use are controversial, since it has been difficult to evaluate its cost- effectiveness. Sufficient data and experience are now available on the use of mamtnography in older women, so that a reasonably good evaluation can be made of its effectiveness and benefits. The evaluation methodology presented in this sutdy is that of cost-effectiveness, i.e., the comparison of the costs of alter- native methods for achieving the specific objective of the de- tection of breast cancers. Cost-effectiveness analysis is usually the most appropriate method for evaluating equipment- embodied technology. However, a cost-benefit analysis, although more difficult to complete, would require policymakers to con- sider all the various benefits in patient outcomes that result from the different technologies. As will be seen in this case study, mammography can provide increased benefits to patients from earlier detection of cancer, but this is only partly recog- nized in a cost-effectiveness study. l0l
l02 DEFINITIONS AND PURPOSES Definitions Mammography is an X-ray examination of the breast that uses special roentgen equipment, films, and procedures to provide adequate photographic details of the soft tissues of the breast and expose the patient to a relatively low dose of radiation. Screening mammography is the term applied to the procedure when used for cancer detection in asymptomatic women. It usu- ally includes two views (cephalocaudal and lateral) of each breast. Diagnostic mammography is the term applied when patients with a breast abnormality are referred to a radiologist for mam- mography and may include three views. Xeroradiography is a modified X-ray procedure that records an electrostatically charged image on (a) a selenium-coated alumi- num plate from which it is then printed by electrostatically charged powder, or (b) directly onto an electrostatically charged plastic film that, after exposure, is dusted with electrostati- cally charged powder. It produces high-contrast, good-quality pictures of the breast tissue. Clinical examination is the physical examination of the breast by visual inspection and manual palpation by a physician (or other trained health professionals). It is the most commonly used method for breast examination; however, it can only detect palpable cancers and those with visible skin abnormalities. Purposes Generally, women seek advice from a physician when they discover a lump in the breast or have other breast symptoms, or they become anxious when they learn of someone else who has breast cancer. In the past decade, publicity by various media has motivated women to undertake periodic breast cancer screening by self- examination and/or visits to cancer detection programs. Mammography is used for examination of the breast for both benign and malignant disease; however, its primary use is for the detection and diagnosis of breast cancer. This case study will evaluate its cost-effectiveness for breast cancer detection. Since breast cancer is the most prevalent cancer in women and since the major decrease in mortality is achieved by its early detection before the cancer has spread to areas outside the breast, the primary goal of any program for breast cancer control should be its earliest detection.
l03 When a woman discovers a mass in her breast by self- examination, or when a mass is detected in a clinical examina- tion by a physician, then the patient is usually referred to a radiologist for a diagnostic mammography. Principles of Operation Mammography provides pictures of the breast tissue in which cer- tain abnormalities of the glandular tissue can be visualized. The characteristic variations from normal, which are interpreted as being suspicious for cancer, include a mass or density with irregular borders, microcalcifications, skin thickening, altera- tion of blood vessel or glandular duct patterns, or a variation in architecture as compared to the same area in the other breast. In postmenopausal women (which includes women age 50 and over), the normal increase in fatty tissue in the breast provides more contrast in the mammograms and improves the sensitivity and spec- ificity of the test. This, plus the higher prevalence of breast cancer in older women, explains the greater cost-effectiveness of screening mammography for postmenopausal women as compared to premenopausal women. Certain attributes increase the risk of breast cancer in women,9,15 and these include: â¢ increasing age; â¢ chronic cystic mastitis, single or multiple nodules, or irregularities in the breast; â¢ nipple discharge; â¢ history of cancer in the other breast; â¢ family history of breast cancer; â¢ no history of pregnancy before 30 years of age; â¢ early onset of menstruation (prior to age l2). Accordingly, the cost-effectiveness of a single screening examination can be improved by selective testing of only high- risk cases. This short-term view has great implications for the long-term cost-effectiveness, as will be discussed later, since low-risk cases who later detect breast cancer on self- examination are more likely to have axillary node involvement and will have increased costs of care. (See pp. l06-l20.) The most important principle that has evolved from studying the course of breast cancer is that early detection while the cancer is still limited to the breast produces the highest long- term survival rates. Most studies show that about one-half of women with breast cancer with axillary node involvement will have a recurrence of their cancer within 5 years.1,7,14
l04 HISTORY OF DEVELOPMENT AND DIFFUSION The history of case-finding for breast cancer has been reviewed by Breslow,1 Seidman,14 and others who reported on the stability of breast cancer mortality in the United States from l930 through the mid-l960's. More recently, Gilbertsen7 reported some im- provements in case survival rates from physical examination alone. Through the mid-l960's, early case-finding of breast cancer was principally carried out by periodic clinical examinations by phy- sicians and by teaching women self-examination of their breasts. In the late l950's, mammography was first advocated for breast cancer screening by Gershon-Cohen and Egan. Since the mid-l960's, X-ray and thermography techniques for breast cancer detection have been increasingly used. In the l940's, about 63 percent of women with breast cancer had axillary node involvement at the time of diagnosis, in the l950's about 57 percent, and in the l960's about 50 percent.1 Since the mortality rate from breast cancer did not change in 40 years from clinical examination alone, it is evident that more sensitive methods for earlier detection of breast cancer are needed. Two large studies, supported by grants from the U.S. Public Health Service, evaluated in the early l960's the effectiveness of maminography for breast cancer screening of asymptomatic women. One study was conducted by Shapiro and Strax at the Health In- surance Plan of New York,18"24 and the other by Griesbach and Eads at the Kaiser-Permanente Plan in Oakland and San Francisco. Both studies showed generally similar prevalence rates of breast cancer and effectiveness of mammography. Subsequently, routine screening mammography was continued by Strax at the Guttman Breast Diagnostic Institute in New York City and in the Oakland and San Francisco Kaiser-Permanente Multiphasic Health Checkup program for women age 48 and over. Over the past l0 years, the technology of maminography has been modified to improve the quality of images for more effective can- cer detection, to decrease radiation dosage and to decrease costs. The National Cancer Institute (NCI) and the American Cancer Society (ACS) are currently sponsoring 27 Breast Cancer Detection Demonstration Projects (BCDDP) in the United States to evaluate mammography, xerography, and thermography for breast cancer de- tection and control, each center screening annually at least 5,000 asymptomatic women for 2 years plus a 5-year followup. In March l977, the National Cancer Institute issued guide- lines that do not endorse mass screening mammography for women under age 50, unless they have a personal or family history of breast cancer.3 This was primarily because of the risk that irradiation may increase future breast cancer rates in this age group.15
l05 CURRENT STATE OF DEVELOPMENT AND DIFFUSION Diagnostic marranography is now generally available in most radi- ologists' offices. Screening mammography is now widely used for breast cancer detection and is being evaluated in the NCI-ACS BCDDP mentioned above. Present advanced technology for screening mammography uses ar. X-ray tube with a molybdenum target, a vacuum-packed rare earth fluorescent screen and film, and breast compression devices. This provides high-contrast images with good detail and exposes the breast tissue to a relatively low X-ray dosage. Although mammography is no longer considered to be experi- mental and its effectiveness for breast cancer detection has been established, it has the disadvantage of exposing examinees to X-rays. The hazard of future cancer from these X-rays themselves is a small risk, and the epidemiological data from Hiroshima and Nagasaki have recently provided some measures of this risk. Sev- eral committees recently reported to the National Cancer Insti- tute on this issue and concluded that periodic mammography could expose women aged 35 to 50 to significant X-ray dosage during their lifetime and potentially increase the incidence of breast cancer in their later years; but the consensus was that for women age 50 and over, the risk was not significant. Accordingly, it is current generally recommended policy3 that screening mammog- raphy , if done, be provided only to asymptomatic women age 50 and over; and only diagnostic mammography be available for symp- tomatic or high-risk women under age 50. This case study there- fore will evaluate the cost-effectiveness of periodic screening of women age 50 and over. IMPACT OF PUBLIC POLICY ON DEVELOPMENT AND DIFFUSION In the past 5 years, the newspaper publicity generated by the discovery of breast cancer in a President's wife and a Vice- President's wife resulted in a sudden increase in the use of mammography by the public and its widespread adoption in roent- genology services. The fear of having breast cancer was a powerful motivating force that essentially established a public policy. Currently, public policy on screening mammography is being generally set by the National Cancer Institute and the American Cancer Society, and the results of their ongoing early Breast Cancer Detection Demonstration Projects will probably establish policy for the future.
l06 EVALUATION OF MAMMOGRAPHY FOR BREAST CANCER SCREENING Criteria for Evaluation The evaluation of a diagnostic technology requires information on its yield rates and costs. Yield rates depend upon test sensitivity and specificity and disease prevalence. A good test sensitivity is critical since it measures the ability of the test to detect patients with cancer and is repre- sented by the proportion of test positive patients who actually have cancer (i.e., the ratio of true-positives to all of the women with cancer in the screened population). (See Table l.) More serious is the effect of a poor test sensitivity, which is represented by the cases missed (i.e., false-negatives). Test specificity is measured by the proportion of patients with negative tests who actually do not have the disease (i.e., the ratio of true-negatives to all the women without breast can- cer) . Poor specificity gives a high proportion of false-positive tests and increases the costs of the program. The prevalence of the condition in the target population must be determined or estimated. The higher the prevalence, i.e., the more diseased people in the group being tested, generally the higher the yield and the more cost-effective will be the di- agnostic technology. The unit costs of the test must be established, as well as the costs of the followup tests and procedures necessary to identify true-positives and true-negatives. If it is desired to add to the evaluation the impact of the diagnostic technology on the desired ultimate outcome of the patients, then additional information is required on the alter- native treatments likely to be provided, the probable results of each treatment, the resources used for treatment, and the unit costs of all treatment procedures. Alternative Methods for Breast Cancer Screening In order to detect breast cancer early (i.e., while still local- ized to the breast), the following alternative methods for breast cancer detection will be considered: l. Clinical examination. The physical examination of the breast by visual inspection and manual palpation by a physician (or trained nurse) is the most commonly used method. 2. Mammography. X-ray examinations of the breast are in- creasingly being used in breast cancer screening programs because mammography is a more sensitive test than clinical examination,
l07 TABLE 1. Sensitivity and Specificity of Breast Cancer Screening Modes (Projected Cumulative 5-Year Experience for Women Age 50+) Test Result Total Test Results Cancer Present Cancer Not Present Positive (+) Total Tests (+) True(+) False (+) Clinical examination 467 70 397 Mammography l80 90 90 Clin. exam. & mammo. 156 97 59 Negative (-) Total Tests (-) False (-) True(-) Clinical examination 9,533 30 9,503 Mammography 9,820 10 9,810 Clin. exam. & mammo. 9,844 3 9,841 Total Total Totals Total Tested Cancers Noncancers 10,000 100 9,900 Clinical Clin. Exam. Examination Mammography & Mammo. -cn-itivitv Truel+'s> 100 0-70 70 90 n in 97 Total cancers 100 - 0'90 100 = Â°'97 -occificitv Truel-'s> 9,503 9,810 9,841 Total noncancers 9,900 9,900 9,900 since it can detect some nonpalpable cancers. However, some solitary dominant masses of the breast will not be detected by mamroography,23 especially in premenopausal women. 3. Clinical examination and mammography. Some breast cancer detection programs provide both a screening mammogram and a clin- ical examination of the breasts by physicians or trained nurses. Since there is less controversy in the screening of women over age 50 due to their relatively high rate of breast cancer, the lower risk of radiation-induced cancer, and the increased sensitivity of mammography in this age-group, this analysis will assume that four groups each of l0,000 women age 50 and older were randomly selected and three will be tested by one of the above modes and one group will serve as controls. Accordingly, Figure l shows the predicted 5-year experience of a control group of l0,000 women age 50 or more who are not in- vited to participate in a breast cancer screening program.
l08 Figure 2 shows the expected 5-year experience of a similar group of l0,000 women who receive an initial and three subsequent an- nual clinical examinations of the breast. Figure 3 is the antic- ipated flow diagram for the group that receives an initial and three annual screening mammograms. The participants in the group in Figure 4 receive both screening mammograms and clinical examinations, for their initial and three annual reexaminations. The costs of any test mode could be decreased further by pre- screening out women with lower than average risks of breast can- cer (i.e., examining only those with certain types of benign breast disease, with early onset of menses, no pregnancies, with a family history of breast cancer, etc.). However, not screen- ing, lower risk women eliminates the possibility of early detec- tion of many cancer cases. Periodic reexaminations for breast cancer would provide lower yield rates than would the initial examination, depending upon the interval between reexaminations. Reexaminations yearly would yield significantly fewer positives after the first exam- ination, and would probably yield about the same number of posi- tives each reexamination. Accordingly, estimates could be made of the most cost-effective interval between examinations for various combinations of examination modes, e.g., manual palpa- tion annually with mammography every 2 years or manual palpation and mammography the first and second year, then manual palpation every 2 years with mammography every 4 years, etc.5 In this example, it is assumed that periodic examinations are annual, but for purposes of simplification the results of the second, third, and fourth reexaminations in the 5-year study period are assumed to be similar and are combined. Assumptions of Case Study In this analysis, the assumptions made were based upon the studies from screening asymptomatic women in prepaid group prac- tices by Shapiro and Strax17"24 and Griesbach and Eads,8 with modifications in estimates of sensitivity and specificity based upon improved current mammography technology as reported from BCDDP15,30 and others.2,4,5 In different populations with dif- ferent examiners the results may be significantly different. The calculations in this case study are not meant to be defini- tive and are shown primarily to demonstrate the evaluation meth- odology for equipment-embodied diagnostic technology. The following assumptions have been made for this case study: l. For each examination mode, the initial examination was the first breast cancer detection examination for each woman.
l09 For each group of l0,000 asymptomatic women age 50 or older, all who were invited complied with the initial and three annual re- examinations and with any advised followup care. 2. For each l0,000 women in this age-group, l00 cases of cancer of the breast would occur in the 5-year period. This as- sumes an expected rate of 40 per l0,000 on the initial examina- tion and 20 per l0,000 for each of three subsequent annual reexaminations. Early data from NCI's BCDDP estimates for women age 50-59 an initial screening rate of 57 per l0,000 and a second screening rate l year later of 26 per l0,000.1! To simplify this analysis, second cancers occurring in the other breast were not considered. 3. For each single screening examination, the sensitivity of the test made is such that clinical examination detects 60 per- cent of breast cancers, current mammography technology detects 85 percent (preliminary BCDDP data exceed 85 percent sensitiv- ity ), and clinical examination plus mammography detects 95 percent. Of those screened as "positive" by a clinical examination (i.e., dominant mass palpated) and/or mammography (i.e., sus- picious for cancer), all will receive a diagnostic surgical bi- opsy. For those biopsied for a dominant mass after a positive clinical examination alone, l5 percent will be positive. For those biopsied after a positive mammogram alone, 50 percent will be positive. (Preliminary BCDDP data indicate about 90 percent true-positive mammograms.30) For those biopsied following both clinical examination and mammography, 90 percent will be positive for cancer in one-half of the patients who have both a palpable dominant mass and a mammogram suspicious for cancer. For the remaining one-half the percentages are as for either test cited above, or an overall average of 62 percent of biopsies for this group will be positive. The 5-year cumulative experience from four examinations will show projected sensitivity and specificity of the alterna- tive modes as presented in Table l. 4. All false-negatives (i.e., women with cancer of the breast who were not so detected on the screening examination) returned in the intervals between examinations and within the 5 years when they detected a lump in the breast; they then required a surgical biopsy. 5. For the control group that was not invited to receive screening examinations, during the 5 years (although perhaps three-fourths may see a physician who will do a routine clinical examination including the breasts without additional charge) l0 percent would seek a conventional clinical examination for a breast complaint and incur a separate cost thereby.
ll0 6. Axillary node involvement will be found: in 50 percent of the breast cancer cases in the control group,1, 17~24 and in the false-negative and interim cases; in 30 percent of cases with a dominant mass detected by clinical examination; and in 20 percent of cases detected by mammography.*3 7. "Early" cases are defined herein as those with no axil- lary node involvement and cancer localized to the breast only; and in these 90 percent will not have a cancer recurrence within 5 years. "Late" cases are those with axillary node involvement and 50 percent of these patients will return within 5 years for more medical care and hospitalization. This analysis does not consider other benefits to patients associated with the varying outcomes. 8. In women age 50 or older, the long-term effects of ex- posure to X-rays from mammography are negligible. (In women under age 50, the risk of increasing the future incidence of cancer from X-ray exposure must be considered in any evaluation.) 9. Average costs used in these calculations for patients in l977 for their examinations and treatment procedures are repre- sentative. It is important to emphasize that the calculations that follow are shown primarily to demonstrate the evaluation methodology. Since treatment regimens for breast cancer are even less standardized than are diagnostic procedures, each pro- gram should determine its own costs and then can use this model to calculate its cost-effectiveness. The costs used herein are generally based upon the California Relative Value Studies, and are as follows: Range ($) Average ($) Clinical examination of breast and axillae, by M.D. l0-20 l5 Clinical examination of breast and axillae, by R.N. 5-l0 5 Average screening cost for clinical examination l0 Mammography screening, bilateral (two views each breast) 5-62 20 Breast biopsy and associated workup for negative biopsy case 750-l,l50 930 (biopsy, anesthetic, operating room, frozen section, one hospital day at $l25) Surgery and care for early cancer case 3,500-5,500 4,500 (biopsy, mastectomy, anesthesia, operating room, 6 hospital days and 5-year followup visits)
Ill Range ($) Average ($) Surgery and care for late cancer cases 4,500-6,500 5,400 (biopsy, surgery as for early case plus 5-year followup with chemotherapy and/or radiation therapy) Late care costs for cancer recur- rence 4,000-8,000 6,000 (recurrent hospitalization for l4 days and chemotherapy and/or palliative X-ray therapy) Projected Results Figure l shows the projected experience for breast cancer care for the 5-year period for the control (unscreened) group. It is predicted that for the l,000 women who seek care for a breast complaint (at an average cost of $l5 an examination), 700 will receive a clinical workup and surgical biopsy; and 600 will have a negative biopsy and will stay in the hospital for only l day with total costs ranging from $750 to $l,l50 with an average cost of $930 each patient. (Note that the figures show costs for biopsies for only biopsy-negative cases since for biopsy- positive cases the biopsy cost is included in the cost for mas- tectomy.) The remaining l00 will be found to have a biopsy positive for breast cancer. Of these l00 patients, 50 will not have axillary nodes involved and their total hospital and surg- ical care with 5 years of routine followup visits will cost $3,500 to $5,500, with an average of $4,500 per patient. The remaining 50 patients will have axillary node involvement and require, in addition to surgery, followup care for chemotherapy and/or radiation therapy for about an added $900, giving an average total cost per patient of $5,400. Of the 50 patients with axillary nodes involved, 25 will require subsequent care for cancer recurrence with additional chemotherapy and/or radia- tion therapy and an average of 2 weeks hospitalization over the 5 years for a cost ranging from $4,000 to $8,000, or for an average cost of $6,000 per case. Of the 50 patients without axillary node involvement, only 5 will have a recurrence of cancer and require similar late hospital care. It is expected that of the l00 breast cancer patients in the group, 30 will have a recurrence or extension of cancer in the 5-year period. The total projected costs for the 5-year period for breast cancer are calculated to be $l,248,000. This can be expressed as a
0 Ul 5 *J tn in O 0 O 0) </> O G 0) 0) It â¢H ia h O) t o a a) X in iu 0) id c a) 0) >i ^ I id in ft t) C a) -H ^J u in eo 0) â¢n M O 3 M O> 0,-H b H 6 o m o Di cO rt â¢* â¢H CN /0 Ul c -u .2 8 m o â¢H u M S3 OJ â¢â¢ in o 0) a) H .H O nJ m v w o Â« in O 0) H u> h n) ll2
ll3 cost of $l2,480 per cancer patient or $l25 per woman in the group of l0,000 for the 5 years. Figure 2 presents the predicted experience for the group that receives annual breast clinical examinations from a screening program assumed to be able to provide such examinations at a cost of $l0 per examinee. Over the 5 years, 39,790 screening examinations will be thus provided at a cost of $l00,000 for the initial examination and $297,900 for the reexamination. These screening examinations will generate 467 clinical workups and surgical biopsies for a dominant mass palpated in a breast, and 70 will be found to have a positive biopsy for cancer with the followup experience shown in Figure 2. Over the 5-year period at times other than the annual screening examinations, it is predicted that 230 women will seek care from their physicians for a breast complaint and will undergo a surgical biopsy, and 30 will be found to have breast cancer. Of the total l00 breast cancer patients in this group, 23 will have some cancer recur- rence in the 5-year period. The total projected costs for breast cancer care for this group of l0,000 women receiving annual breast clinical examinations are $l,573,5l0, or $l5,735 per can- cer patient, or $l57 per examinee for the 5-year period. Figure 3 shows the predicted experience for the 5-year period for the group who receive only annual screening mammograms. It is expected that l80 patients will be found to have mammograms suspicious for cancer from the initial and subsequent annual ex- aminations, and 90 will be proven to have breast cancer by surg- ical biopsy. Twenty patients will seek care and receive mammograms at times other than the screening examinations, and l0 will be found to have breast cancer. Over the 5-year period, l8 of the l00 patients with breast cancer will receive late care for cancer recurrence. The cost for providing screening mammog- raphy, two views of each breast, varies considerably from $5 in the Oakland Kaiser-Permanente multiphasic screening program to $62 by fee-for-service hospital radiologists. Excluding the l0,000 initial and 29,730 reexamination mammograms, the costs associated with all the breast cancer care in this group are projected at $67l,700, which allows $576,300 that could be ex- pended for annual mammograms and still not exceed the total costs of $l,248,000 for the control (unscreened) group. Accordingly, up to $l4.5l could be spent per patient examination for mammog- raphy and not exceed the 5-year costs of the control group. The total costs of $l,466,300 for the 5 years (shown in Figure 3) are calculated on the basis of a mammography unit cost of $20, which should be achievable by any well-organized mammography screening program. At Kaiser-Permanente's costs of $5 per mammogram (which includes radiologist's interpretation, for a screening load of more than l0,000 women a year in the Oakland and San Francisco
ll7 programs), the total costs for the group for 5 years would be only $920,750; this can be expressed as a cost of $9,208 per can- cer patient, or $92 per examinee for the 5 years. Preliminary BCDDP data indicate a 90 percent true-positive experience with manunography.30 Using this rate of true-positives in this example would decrease the number of false-positive bi- opsies done by 89 and decrease the total 5-year cost for this group by $82,770. Figure 4 presents the projected experience for the group who would receive annually both breast clinical examination and mam- mography, at an average cost of $30 per examinee. From a recent survey, Oldfield12 estimated $35 per patient to be a comfortable figure for the cost of examining 25 patients a day for a compre- hensive breast screening program with mammography, xerography, and thermography. The end results over the 5-year period would be to predict l7 late recurrent cancers. The total costs would be $l,8l9,290, or $l8,l93 per cancer patient, or $l82 per ex- aminee for the 5 years. The increased sensitivity of four annual examinations in 5 years as compared to a single examination in this period is shown in Table l. The percentage of breast cancers detected in the 5- year period increases from 60 percent for a single clinical exam- ination to 70 percent from four annual clinical examinations. For mammography this would increase from 85 percent for a single to 90 percent from four annual mammograms. For both clinical exam- ination and mammography, this would change from 95 percent for a single examination to 97 percent from four annual examinations. In other words, the least sensitive test, a single clinical examination in 5 years, will miss 30 breast cancers in l0,000 women; whereas the most sensitive testing alternative, annual clinical examinations and mammography, will miss only three breast cancers in 5 years in l0,000 women. The four groups show the following comparative predicted out- comes for their l00 patients with breast cancer over a 5-year period: No Re- Axillary Late currence Nodes + Cases in 5 Years Unscreened group 50 30 70 Annual clinical examinations 36 23 77 Annual mammograms 23 l8 82 Annual clinical exam. S mammo. 2l l7 83 The primary advantage of mammography over clinical examination from the viewpoint of effectiveness for breast cancer screening
ll8 (see Table l) is its better sensitivity and specificity, thus decreasing the number of biopsies performed on false-positive cases without breast cancer, increasing the number of early cases detected, and increasing the number of patients who have no re- currences in 5 years. Discussion This evaluation model did not consider estimates of "lead time," i.e., the time between diagnosis with the screening program and the usual time of diagnosis under current medical practice,6,10,15 since it does not compare survival rates over this 5-year period. However, if a l0-year cost-effectiveness study were made, it might affect the time that late cases appeared in years 6-l0. Without consideration of any benefits to the patient from de- creased disability and added years of life, that is, strictly from a cost-effectiveness viewpoint, ideally the total costs of care with screening should be less than the total costs of care without screening. It must be emphasized again that the costs and yield rates given in the calculations for this case study, although probably representative, are used primarily to demonstrate the methodology for evaluation of this diagnostic technology. In Figures 2 and 3, the cost of a mammography examination is assumed to be twice that of a clinical examination, and any change in this ratio will significantly alter the final 5-year costs. Similarly, the sen- sitivity of mammography as compared to clinical examination is very critical in determining the higher rate of detection of early cases from mammography. Each mammography program should determine its own unit costs and yield rates and then can use this evaluation model to calculate its own cost-effectiveness. If the unit cost for mammography were higher than $20 or if the sensitivity of the test were less than 85 percent, then the costs and patient outcomes from the screening program would be different from the example given in this case study. The costs for care of patients with axillary node involvement and for late care for re- current cancer are extremely variable since these are less stan- dardized and will need to be individualized in accordance with the prevailing medical and surgical practice of each community. Based upon the assumptions presented in this study, for women age 50 or more with an expected 5-year rate of breast cancer of l00 per l0,000 or more, health care costs for 5 years from an- nual screening mammograms can be projected to be less costly than: (l) annual clinical examinations, (2) annual mammograms plus clinical examinations, and (3) less costly than not screen- ing if the cost per mammogram does not exceed about $l4.50. This
ll9 conclusion is based upon the estimated comparable costs for the nonscreened group, and the predictions that annual mammograms will result in patient outcomes over 5 years comparable to good current medical experience for the detection and treatment of breast cancer. The yield rate of breast cancer will vary, of course, with the population being studied. By this evaluation model, for women ages 35 to 50 years, the yield rates from screening mammography will be insufficient to make the procedure cost- effective, since the prevalence of breast cancer is about one- half that of women age 50 or greater and the sensitivity of mammography is less in premenopausal women (since there is less fatty tissue in the breast, which makes it more difficult to visualize early cancer). Furthermore, as already mentioned, the cumulative X-ray exposure to the breast by periodic mammog- raphy in younger women introduces the hazard of the increased incidence of breast cancer in later life due to radiation. Mention should be made of xeroradiography (see p. l02), which is very competitive with mammography for breast cancer screening in women age 50 or more. It may cost slightly more, and yields slightly more false-positives than current mammography; but it provides slightly less X-ray exposure than mammography, and it is easier to interpret. The dense glandular breasts of young women are demonstrated in better detail by xerography than on mammography. In addition, thermography is an alternative procedure wherein the breast surface is scanned with an infrared camera and the infrared radiation emitted from the skin is recorded on a photo- graphic film. After a prior l0-l5 minute cooling period of the patient in a temperature-controlled room, a cancer nodule in the breast may show a localized warmer area on the picture. Thermog- raphy is less sensitive than mammography for breast cancer detec- tion in postmenopausal women. It will be less cost-effective than mammography for women over age 50, since it is less sensi- tive and less specific than mammography (that is, it will have fewer true-positives and true-negatives, and more false- positives and false-negatives). 25-29 However, for women 35 to 50 years of age, screening thermography may be competitive with screening mammography, since in this age-group mammography is also less sensitive; and periodic thermography does not in- crease the irradiation risk of future breast cancers. Thermog- raphy has also been recommended as a method for identifying high-risk women of all ages prior to diagnostic mammography. The National Center for Health Statistics reported11 from health interview surveys that, in l973, 76 percent of females l7 years and over admitted to their ever having a breast examina- tion, of which 63 percent said they had it less than l year,
l20 23 percent in l to 2 years, 6 percent in 3 to 4 years, and 8 per- cent in 5 years or more. It appears that the majority of women in the United States already are receiving clinical examinations of their breasts; the current cost-effectiveness for this mode of breast cancer detection and control is represented by Fig- ures l or 2, and the costs per l0,000 women age 50 and over for 5 years are probably $l.2 to $l.6 millions. Sufficient data are now available for health care planners and policymakers to make the decision as to whether for a similar 5-year cost the greater effectiveness of mammography should make it the method of choice for breast cancer screening of women age 50 and over. The deter- mination of the most cost-effective method of breast cancer con- trol for women age 35-50 can follow this evaluation model when sufficient relevant data have been accumulated. POLICY IMPLICATIONS Evaluations, such as described in the preceding section, carried out by large screening programs so as to provide their own data on costs and effectiveness will help to establish firm guide- lines and policies for mammography in breast cancer detection and control. Especially necessary are more followup data on women age 35-50 who already have been exposed to periodic mam- mography, adjusting for the fact that with current technology (l) the extent of X-ray exposure is much less than it was prior to l975, and (2) the sensitivity and specificity of testing is now improved. Limiting screening mammography to only those with higher risk of developing breast cancer (see p. l03) will decrease the initial screening costs and increase the short-term cost- effectiveness of the programâand this is the current policy guideline for women age 35-50. Data are not yet available to estimate the 5-, l0-, l5-, or 20-year cost-effectiveness of such a policy. For women age 50 and over, data are available (as indicated in the preceding section) to show that the limiting of screening mammography only to high-risk cases will decrease the periodic screening costs. However, since the lower-risk patients not screened will return when they detect breast cancer on self-examination and since their costs will then be higher (and outcomes poorer), it is unlikely that a policy to screen only high-risk women age 50 and over will be cost-effective over a 5-year or longer period. One of the limitations of such a cost-effectiveness model is that it does not consider all the benefits gained by those women who become early rather than late cancer cases.
l2l It is evident from these data that from the viewpoint of the policymaker the differences in costs ($l.25-$l.82 million) for 5 years are not so different since there is clearly a trade-off between early care and late care in the four alternatives con- sidered. The increased benefits from earlier cancer detection in decreasing the numbers of late cases become paramount, since the study shows so little difference in 5-year costs. Therefore, it becomes advisable for policymakers to do a cost-benefit anal- ysis and assign some values to the various increased benefits from screening mammography for women age 50 and over. If one has the responsibility for the continuing care of a defined population, then it is essential to evaluate the cost- effectiveness of any equipment-embodied technology that is di- rected to a potentially postponable disability (such as breast cancer) over a sufficient length of time to include all of the major costs of medical care for the disability. An easy way to decrease immediate short-term costs is to deny access of patients to relevant care resources, but this type of short-sighted econ- omy often results in postponing care to a later date when it can no longer be deferred, and the disability may then be more serious and require more costly care resources. HMO's soon learn that good early care is the best and the most economical care by de- creasing the numbers of costly late, complicated, advanced and neglected cases. ACKNOWLEDGMENTS Norman Walter, M.D., Sr. Surgeon, Hayward Kaiser-Permanente Medi- cal Center, provided data on surgery and followup care costs. Walter Griesbach, M.D., Chief of Radiology, San Francisco Kaiser-Permanente Medical Center; William Eads, M.D., Chief of Radiology, Oakland Kaiser-Permanente Medical Center; and V. H. Voss, M.D., Sr. Radiologist, Santa Clara Kaiser-Permanente Medi- cal Center, all provided advice and assistance on mammography and xerography. Norma M. Oldfield, Executive Director, American Thermographic Society, provided information on thermography and mammography projects. Gary D. Friedman, M.D., M.S., provided epidemiology advice and assistance. Robert Feldman, M.D., and Edmund E. Van Brunt, M.D., offered valuable suggestions and criticisms.
l22 REFERENCES l. Breslow, L. "Early Case-Finding, Treatment and Mortality from Cervix and Breast Cancer." Preventive Medicine l (l972):l4l-52. 2. Croll, J., Macmillan, I., and Ryan, J. "Screening for Breast Cancer by Mammography. An Australian Experience." Medical Journal of Australia l(l977):287-92. 3. Culliton, B. J. "Cancer Institute Unilaterally Issues New Restrictions on Mammography." Science l96(l977):853-57. 4. Dowdy, A. H., Barker, W. B., Lagasse, L. D., Sperling, L., Zeldis, L. J., and Longmire, W. P. "Mammography as a Screening Method for the Examination of Large Populations." Cancer 28(l97l):l558-62. 5. Eddy, D. M. "Mammography: An Analysis." Stanford Univ., June 7, l976. 6. Feinleib, M. "Some Pitfalls in the Evaluation of Screening Programs." Archives of Environmental Health l9(l969):4l2-l5. 7. Gilbertsen, J. A., and Kjelsberg, M. "Detection of Breast Cancer by Periodic Utilization of Methods of Physical Diag- nosis." Cancer 28(l97l):l552-54. 8. Griesbach, W. A., and Eads, W. S. "Experience with Screen- ing for Breast Carcinoma." Cancer l9(l966) : l548-50. 9. Holleb, A. I. "Editorial: Restoring Confidence in Mammog- raphy." CAâA Cancer Journal for Clinicians 26(l976): 376-78. l0. Hutchison, G. B., and Shapiro, S. "Lead Time Gained by Diag- nostic Screening for Breast Cancer." Journal of the National Cancer Institute 4l(l968):665-8l. ll. National Health Survey Series l0, No. ll0. Use of Selected Medical Procedures Associated with Preventive Care. U.S. l973. DHEW Publ. No. (HRA) 77-l538, l977. l2. Oldfield, N. "A Report to the Members of Region I of the IHEA on the Current Status of Early Breast Cancer Detection in the United States. March l975." l3. Sadowsky, N. L., Kalisher, L., White, G., and Ferrucci, I. T. "Radiologic Detection of Breast Cancer." New England Journal of Medicine 294(l976):370-73. l4. Seidman, H. Cancer of the Breast. American Cancer Society, May l969. l5. Seidman, H. "Screening for Breast Cancer in Younger Women." CAâA Cancer Journal for Clinicians 27(l977):66-87. l6. Stark, A. M., and Way, S. "The Screening of Well Women for the Early Detection of Breast Cancer Using Examination with Thermography and Mammography." Cancer 33(l974):l67l-79. l7. Shapiro, S., Strax, P., Venet, L., and Venet, W. "Changes in 5-Year Breast Cancer Mortality in a Breast Cancer Screening
l23 Program." In Seventh National Cancer Conference Proceed- ings. Philadelphia: J. P. Lippincott Co., l972. l8. Shapiro, S. "Screening for Early Detection of Cancer and Heart Disease." Bulletin of the New York Academy of Medi- cine 5l(l975):89-95. l9. Shapiro, S., Strax, P., and Venet, L. "Periodic Breast Cancer Screening in Reducing Mortality from Breast Cancer." Journal of the American Medical Association 2l5(l97l): l777-85. 20. Strax, P. "Control of Breast Cancer Through Mass Screening." Journal of the American Medical Association 235(l976):l600- l602. 2l. Strax, P. "Results of Mass Screening for Breast Cancer in 50,000 Examinations." Cancer 37(l976):30-35. 22. Strax, P. "New Techniques in Mass Screening for Breast Cancer." Cancer 28(l97l):l563-68. 23. Venet, L., Strax, P., Venet, W., and Shapiro, S. "Adequa- cies and Inadequacies of Breast Examinations by Physicians in Mass Screening." Cancer 28(l97l):l546-5l. 24. Strax, P., Venet, L., and Shapiro, S. "Value of Mammography in Reduction of Mortality from Breast Cancer in Mass Screen- ing." American Journal of Roentgenology ll7 (l973):686-89. 25. Byrne, R. R., and Yerex, J. "The Three Roles of Breast Thermography." Applied Radiology, May/June l975. 26. Isard, H. J., Becker, W., Shilo, R., and Ostrum, B. J. "Breast Thermography After Four Years and l0,000 Studies." American Journal of Roentgenology, Radium Therapy and Nuclear Medicine ll5(l972):8ll-2l. 27. Nyirjesy, S., Abernathy, M. R., Billingsley, F. S. and Biers, P. "Thermography and Detection of Breast Carcinoma: A Re- view and Comments." Journal of Reproductive Medicine l8 (l977):l65-75. 28. Revesz, G., and Lapayowker, M. S. "Breast Thermography as a Screening Technique." Cancer 36(l975):2l59-63. 29. Stark, A. M., and Way, S. "The Use of Thermovision in the Detection of Early Breast Cancer." Cancer 33(l974):l664-70. 30. Wolfe, J. N. "Mammography. A Radiologist's View." Journal of the American Medical Association 237(l977):2223-24.