| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Breast Imaging |
1 From the Guttman Diagnostic Center, Memorial Sloan-Kettering Cancer Center, 55 Fifth Ave, 12th Floor, New York, NY 10003 (K.N.F., D.M.K., P.M.T.); and Department of Radiology, Breast Imaging Center, University of Virginia at Charlottesville, Charlottesville, VA (M.A.C.). From the 2004 RSNA Annual Meeting. Received August 16, 2005; revision requested October 17; revision received December 29; accepted February 1, 2006. Address correspondence to K.N.F. (e-mail: feigink{at}mskcc.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Materials and Methods: The study received a waiver of authorization from the institutional review board, informed patient consent was not required, and the study was compliant with HIPAA regulations. The records of 60 027 consecutive asymptomatic patients who underwent screening mammography were retrospectively reviewed. CBE was performed on all patients by a nurse practitioner. Patients with positive CBE findings were required to convert from screening to diagnostic evaluation; the number of cancer diagnoses that resulted was determined. The reports, four-view mammograms, or both of patients requiring conversion to diagnostic evaluation were reviewed to determine those patients likely to undergo diagnostic imaging on the basis of screening mammographic findings alone. The cost of CBE was calculated and divided by the number of cancers detected solely with CBE to determine the cost of CBE per additional cancer detected.
Results: Four hundred seventy-four (age range, 32–95) of 60 027 asymptomatic patients had positive CBE findings which required conversion to diagnostic evaluation. Forty-six cancers in 44 patients were subsequently diagnosed; 32 would have been detected with mammography alone, whereas 14 were imperceptible at screening mammography. The cost of CBE was $122 598 per cancer detected solely with positive CBE findings.
Conclusion: CBE performed by nurse practitioners led to the diagnosis of 14 cancers in 13 patients with mammographically occult tumors (0.02% of the screening population and approximately 3% of all cancers diagnosed at the facility during this study). The cost of detecting these additional cancers is estimated to be $122 598 per cancer.
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Despite widespread advocacy of CBE for breast cancer screening, to our knowledge there are no data showing that CBE is an effective screening tool for reducing breast cancer mortality. Specifically, to our knowledge no randomized trials of CBE as a sole screening method have been performed. In the absence of direct evidence of the effectiveness of screening CBE, indirect evidence has been used to support its widespread use (3–5). Many researchers and advocates agree, however, that "[as] a growing proportion of women are receiving regular mammograms, the relative contribution of CBE to early breast cancer detection and its cost-effectiveness warrant renewed attention" (6). Thus, the purpose of our study was to retrospectively evaluate the cost of CBE and its contribution to screening mammography in the detection of breast cancer.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Patients and CBE
There were 60 027 consecutive asymptomatic patients who presented for screening mammography between January 1, 1997, and December 31, 1998, and between January 1, 2001, and December 31, 2002. Data from screening visits in 1999 and 2000 were not included, because this was a period of transition to computerized data collection at our facility, and the necessary study data may not have been reliably recorded. All patients underwent CBE performed by one of six nurse practitioners (NPs), five of whom had master's degrees in science and nursing and certified by the American Nurses Credentialing Center. Training of the remaining NP predated these requirements, but she held a New York State NP license. These NPs spent most of their time conducting CBEs, and they had 1–4 years of experience as NPs and 0–6 months of experience performing CBEs at the time of the earliest CBE.
Imaging and Evaluation
Mammography of patients with positive CBE findings was "converted" from a screening to a diagnostic imaging evaluation, which the radiologist tailored to evaluate the clinical finding(s). Image-guided breast biopsy was performed for all lesions considered category 4 and 5 according to the Breast Imaging Reporting and Data System (7). These biopsies were performed by one of four radiologists (including M.A.C., D.M.K.) with 4–25 years of experience. In most cases, an imaging correlate (eg, cyst, solid mass, prominent fatty lobule) to the clinical finding was identified by the radiologist. In the rare cases in which no imaging correlate to the clinical finding could be identified at diagnostic examination, the patient was referred to a breast surgeon for clinical evaluation and biopsy. Retrospective review of the records of these patients requiring conversion to diagnostic evaluation, which included patient questionnaires, clinical examination notes, and radiologic and pathologic reports, was performed together by two authors (K.N.F., P.M.T.) to determine the number of converted cases, the nature of the clinical findings, and the outcomes of the evaluations. The number of patients with a subsequent diagnosis of breast cancer was determined, and the clinical finding, size, and pathologic status of these cancers were determined. We also determined whether these patients with cancer had previously undergone mammography.
At our facility, a converted examination begins with acquisition of the same mammographic views that are obtained at screening, namely, full craniocaudal and mediolateral oblique views of each breast; tailored mammographic views and ultrasonographic (US) scans are obtained as necessary. If an imaging correlate to the CBE finding was mentioned in the radiologic report, the initial craniocaudal and mediolateral oblique views from screening were independently reviewed by three radiologists (K.N.F., D.M.K., M.A.C.) with 2–25 years of experience in breast imaging. This review was performed to determine how many of these patients would have been recalled for diagnostic evaluation on the basis of abnormal findings on the screening mammogram in the absence of a positive CBE finding. If the diagnostic imaging evaluation, however, failed to demonstrate a correlative imaging finding, these screening mammograms were presumed to be negative and were not reviewed. Members of the review panel were blinded to the patients' findings at diagnostic imaging evaluation, ultimate diagnosis, and findings at subsequent examinations, but they had access to older studies in order to simulate a screening environment. The location of the relevant CBE finding detected by an NP was usually indicated on the standard full craniocaudal and mediolateral oblique views with a metallic skin marker.
Each radiologist rated the corresponding mammographic finding(s) as warranting recall (possibly clinically important lesion or obvious abnormality) or as not warranting recall (no visible finding, benign finding, or very subtle finding). If none or one of the three radiologists would have recalled the patient for a diagnostic workup on the basis of the screening mammographic views alone, we considered this a finding detected solely by using CBE. If at least two of the three radiologists would have recalled the patient for a diagnostic workup on the basis of the screening mammographic views alone, we considered this a finding demonstrated by using both CBE and mammography. We calculated the percentage of mammographically occult cancers diagnosed at our facility during the 4-year study period by dividing the number of mammographically occults cancers by the total number of cancers diagnosed at our facility during the same period (n = 453).
Cost
We estimated the cost of CBE by determining the cost of diagnostic workups generated solely because of positive CBE findings (ie, excluding those converted cases in which diagnostic imaging evaluation would have been required anyway on the basis of an abnormal screening mammographic finding) by using contemporaneous Medicare reimbursement rates (Table). The costs of all additional diagnostic tests—including diagnostic mammography, US, ductography (galactography), fine-needle aspiration biopsy, core biopsy, and surgical excisional biopsy—and associated image-guidance and pathologic costs were included. We subtracted the cost of screening mammography from the total cost of the diagnostic imaging. We added this to the cost of employment of our NPs. We then divided by the number of cancers detected solely at CBE to calculate the cost of CBE per additional cancer detected.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Cancers
Of 474 patients requiring conversion to diagnostic evaluation, 44 (9%) had breast cancer in the same quadrant. Two patients each had two palpable masses at CBE, both of which were malignancies. Therefore, we diagnosed 46 cancers in 44 patients with CBE, which yielded a cancer detection rate of 0.77 cancers per 1000 patients screened. Three (7%) of the 44 patients in whom cancer was detected at CBE presented for a baseline screening examination, and 41 (93%) had previously undergone mammographic screening evaluations. The CBE finding in 45 of the 46 cancers was a palpable mass; associated nipple retraction was noted in one case, and associated skin dimpling was noted in two cases. The solitary cancer detected without a palpable mass manifested as nipple and areolar thickening.
Forty-four of 46 cancers were measured at pathologic examination, and the mean tumor size was 1.7 cm (range, 0.3–5.5 cm). Of the remaining two cancers, one was described as "DCIS [ductal carcinoma in situ that] focally involves an intraductal papilloma and is present as two minute foci," and the other was described at CBE as a 6-cm mass that at core biopsy yielded the pathologic diagnosis of "poorly differentiated infiltrating mammary carcinoma with lobular features." This patient received neoadjuvant chemotherapy and was then lost to follow-up.
At final pathologic examination, seven tumors (including the aforementioned core biopsy case) were invasive lobular carcinomas, one was invasive carcinoma with both ductal and lobular features, three were ductal carcinomas in situ arising in papillomas, one was ductal carcinoma in situ alone, and the remaining 34 were invasive ductal carcinomas.
Mammographic Findings
Thirty-two tumors in 31 of the 44 patients with cancer would have been diagnosed on the basis of mammographic findings alone, even in the absence of a positive CBE finding, as determined according to our retrospective review. Fourteen tumors in 13 patients were judged to be mammographically occult on routine screening views. The 13 patients represent approximately 3% of 453 patients with breast cancer that was diagnosed at our facility during the 4-year study period. Logistic regression indicates that the 31 patients with a diagnosis from CBE findings and mammographic findings and the 13 patients with a diagnosis from CBE findings alone do not differ significantly in age (P = .68) or tumor size (P = .26). Similarly, Fisher exact tests show no significant difference between these two groups with respect to tumor stage (P > .99) or breast density (P = .135). The one variable that differs significantly between groups is histologic type, in that mammographically occult tumors are more likely to be invasive lobular carcinoma than are mammographically evident tumors (P = .01). Although not statistically significant, the probability of the tumor being mammographically evident increases with increasing tumor size and decreases with increasing breast density.
Three hundred twenty-three patients requiring conversion to diagnostic evaluation were confirmed to have benign disease on the basis of biopsy results, negative findings at follow-up mammography, or both (mean duration of follow-up, 3.1 years; range, 0.5–6.8 years). One hundred seven patients requiring conversion to diagnostic evaluation had no evidence of malignancy at diagnostic imaging evaluation but were lost to follow-up. This group includes patients with definitively benign correlative imaging findings, such as simple cysts at US or calcified fibroadenomas at mammography, as well as patients with no recognizable imaging findings. Of the 430 with benign results, 51 would have required diagnostic evaluation because of abnormal screening mammographic findings, even in the absence of a positive CBE finding. Nine cases could not be retrospectively reviewed, because the patients had removed the images in question from our facility and had not returned them.
Cost
The cost of additional workups generated because of positive CBE findings amounted to $15 851 in 1997, $18 895 in 1998, $14 604 in 2001, and $9890 in 2002, totaling $59 240 during the 4-year study period. We determined the cost of employment of our NPs to be their average salary ($72 488 for 1997 and 1998 and $83,124 for 2001 and 2002) multiplied by 1.25 to account for fringe benefits, and then multiplied by 0.90 to account for the percentage of their time devoted to CBE and CBE-related activities. Cost of employment was an average of $81 549 per NP per year for 1997 and 1998 and $93 515 for 2001 and 2002. We employed five full-time NPs during 1997 and 1998 and 4.5 full-time equivalents during 2001 and 2002, for a total cost of employment of $1 657 125 during the 4-year period. By adding the totals for the additional diagnostic workups ($59 240) to the cost of NP employment ($1 657 125), we have a total cost of $1 716 365. By dividing this total cost by the number of cancers that would not have been detected without positive CBE findings (n = 14), we estimate $122 598 as the cost of CBE per additional cancer detected.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Forty-five (98%) of 46 cancers in our population were palpable masses, and only three of these had associated findings, namely, nipple retraction (one case) or skin dimpling (two cases). The single cancer detected without a palpable mass manifested as nipple and areolar thickening. This is similar to the findings of Mahoney and Csima, who found in a series of women referred for CBE that 275 (96%) of 286 breast cancers detected were evident as a palpable mass (9).
Our review panel retrospectively judged 14 (30%) of 46 cancers detected at screening CBE to be inapparent on routine screening views. These 14 cancers occurred in 13 patients, who represent approximately 3% of 453 patients with breast cancer diagnosed at our facility during the 4-year study period. This finding is essentially in keeping with the literature. Reported rates of breast cancers detected at CBE but missed at mammography range from 3% to 45% (3,10). Our calculated percentage is on the low end of this range for several reasons. First, our study population included only asymptomatic patients who were referred to us for routine screening evaluation. Second, the highest reported rate of mammographically occult cancers is from an older study, namely, the Health Insurance Plan, or HIP, of Greater New York's trial (initiated in 1963), in which mammography was technically suboptimal when compared with current methods and when many radiologists had little experience with the technology. Third, our retrospective study design may have been necessarily biased, because all mammograms reviewed by our panel of radiologists for this study were known to have been associated with positive clinical findings. Such findings were indicated on mammograms with metallic skin markers, which perhaps falsely increased the rate at which we perceived correlative imaging findings. We tried to compensate for this by requiring at least two of the three panel radiologists to claim that they would have initiated a diagnostic imaging evaluation on the basis of the screening mammogram alone and by blinding the panel to the subsequent diagnosis.
Newcomer et al (11) found that mammography, breast self-examination, and CBE vary in ability to demonstrate specific histologic types of breast carcinoma. Similarly, we found that mammographically occult tumors were more likely to be invasive lobular carcinoma than were mammographically evident tumors. Invasive lobular carcinoma is a well-known cause of false-negative findings at mammography (12). This error makes sense histopathologically, because invasive lobular carcinoma cells grow in a single-file array around arborizing ducts, often resulting in a lack of a central x-ray beam–attenuating nidus or mass to elicit a radiologist's attention (13).
Determining the accuracy of CBE as a screening test is difficult because of a lack of a reference standard. Cancers that are detected at CBE may be mammographically occult and vice versa, so mammography alone cannot represent the reference standard. Histopathologic correlation cannot be obtained in all cases, particularly in those that are negative. Clinical and imaging follow-up, therefore, is generally used as a standard end point to establish true-negative findings. A limitation of our study is the number of patients requiring conversion to diagnostic examination who were lost to follow-up (107 [23%] of 474). Our state lacks a cancer registry, so we cannot calculate the false-negative imaging rate in this group. This percentage attrition, however, is only slightly above average for our practice, which is approximately 15% of screening patients. Some of this added attrition may be explained by the inconvenience, anxiety, and expense incurred by patients with a positive CBE finding and subsequent negative diagnostic imaging evaluation.
The direct medical cost of CBE per additional cancer detected in our analysis was $122 598. This cost reflects the relatively high salaries of our NPs, which in turn reflect the high cost of living in the New York City area. We used Medicare reimbursement rates to make the cost calculation more widely applicable and to distinguish from charges, which may vary widely among providers.
Putting this cost in perspective is a challenge. Most data regarding the cost-effectiveness of various cancer screening methods are expressed in a cost-effectiveness ratio in which the denominator represents the number of life-years gained, which may or may not be adjusted for quality of life. For example, annual Pap smear screening for cervical cancer costs less than $10 000 per year of life gained (in 1997 dollars), and 10-year colorectal cancer screening with colonoscopy of 50-year-old subjects costs $9000–$22 000 per year of life gained, depending on the assumed precancerous polyp dwelling times (14). Because data on the effectiveness of CBE in reducing breast cancer mortality are lacking and our study was not designed specifically to address this issue, we cannot express our results in a cost-effectiveness ratio to facilitate comparison of CBE with other cancer screening methods. Another issue that limits comparison is that other studies of cost-effectiveness do not routinely include the salaries of professionals involved. We chose to include NP salaries in our cost calculation, because they account for a large proportion of the overall expense of offering CBE at our facility and because we do not charge patients for CBEs. One way to reduce the cost of CBEs might be to employ RNs or other less expensive clinical personnel to perform them. Obviously, the cost of cancer detection with CBE will vary with the specific model employed at each facility.
Any evaluation of our calculated cost of CBE, especially if undertaken for the purpose of resource allocation, should ultimately also consider the intangible value of an opportunity for patient education on breast cancer–related topics, possible "downstream" cost savings from earlier detection of breast cancer, the fact that a screening program that excludes screening CBE will necessarily have a lower sensitivity for detection of breast cancer, and legal repercussions of missed cancers. This last point is especially pertinent in our society, where expectations among the general population regarding medical success in general and breast cancer screening in particular are exceptionally high. This is evidenced by the fact that one in seven physicians is sued for medical malpractice every year (15). The most common medical condition of patients who made medical malpractice claims between 1985 and 1998 was breast cancer, and these claims resulted in an average payment of $207 800 and a median payment of $125 000 (16). The relative value of the costs and benefits of screening CBE will certainly be affected by the perspective of the decision maker, be that a patient, a physician, or an insurer.
In conclusion, screening CBE is complementary to screening mammography in demonstrating breast cancer in asymptomatic patients; CBE accounted for 14 (3%) of 453 cancers detected in our study population, and the cost of detecting these cancers was approximately $122 598 per cancer. Physicians, insurers, administrators, women's organizations, and other interested parties must ultimately determine whether this is an appropriate allocation of our country's limited health care resources.
|
ADVANCES IN KNOWLEDGE |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
FOOTNOTES |
|---|
Abbreviations: CBE = clinical breast examination • NP = nurse practitioner
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, K.N.F.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, K.N.F., M.A.C.; clinical studies, M.A.C.; statistical analysis, K.N.F.; and manuscript editing, K.N.F., D.M.K., M.A.C.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
2404-B (2004).This article has been cited by other articles:
|
|
 |
|
  M. Robson and K. Offit Management of an Inherited Predisposition to Breast Cancer N. Engl. J. Med., July 12, 2007; 357(2): 154 - 162. [Full Text] [PDF]
|
|
Read all eLetters
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
