Medical Audit of Diagnostic Mammography Examinations: Comparison with Screening Outcomes Obtained Concurrently

Transcription

1 Katherine E. Dee 1,2 Edward A. Sickles 1 Received July 3, 2000; accepted after revision September 12, Presented in part at the annual meeting of the American Roentgen Ray Society, Washington, DC, May, This work was supported by Cancer Surveillance Consortium cooperative agreement No.1U01 CA from the National Cancer Institute. 1 Department of Radiology, Box 1667, University of California Medical Center, San Francisco, CA Present address: Department of Radiology, Box , University of Washington Medical Center, 1959 N.E. Pacific St., Seattle, WA Address correspondence to K. E. Dee. AJR 2001;176: X/01/ American Roentgen Ray Society Medical Audit of Mammography Examinations: Comparison with Outcomes Obtained Concurrently OBJECTIVE. We performed a medical audit of our diagnostic mammography practice and compared clinical outcomes with those of screening mammography examinations performed concurrently. MATERIALS AND METHODS. We analyzed 46,857 consecutive mammography examinations (10,007 diagnostic, 36,850 screening) from 1997 to 2000, including data on demographics, image interpretation, and biopsy (including size, nodal status, and cancer stage). RESULTS. The mean age at diagnostic mammography was 55.8 years (mean age at screening mammogram, 59.1 years; p < ). Among patients who underwent diagnostic examinations, 14.7% had a strong or very strong family history of breast cancer (screening, 11.6%; p < ). Examination findings were interpreted as abnormal in 14.4% (screening, 5.2%; p < ). Biopsy was performed in 11.9% (screening, 1.4%; p < ). Forty-six percent of the biopsies were positive for malignancy (screening, 38%; p < ). The cancer detection rate was 55 per 1000 (screening, 5/1000; p < ). Of cancers found, 74.4% were stage 0 or I (screening, 89.3%; p < ), average size was 18.0 mm (screening, 12.9 mm; p < ), and axillary nodes were positive for malignancy in 19.9% of invasive cancers (screening, 6.3; p < ). Differences between diagnostic and screening outcomes were attributable predominantly to the subgroup of diagnostic examinations performed for evaluation of palpable masses. CONCLUSION. Medical auditing of diagnostic mammography examinations yields substantially different results compared with those of screening examinations, including different patient demographics; higher number of positive biopsies; higher cancer detection rates; and larger, more advanced-stage cancers. and screening data should be segregated during auditing, or if this is not possible, analysis of combined results should be based on known differences between diagnostic and screening outcomes. A ll mammography practices in the United States are required by the Mammography Quality Standards Act to perform an annual medical audit of selected clinical outcomes [1]. In addition to facilitating regulatory compliance, the audit is also a valuable tool to measure the success of a mammography practice in detecting clinically occult, early-stage breast cancer, and to suggest the presence of any deficiencies in technical performance and image interpretation [2 5]. The audit may also increase compliance with screening guidelines among patients and referring clinicians by offering convincing evidence of the success of mammography within a given practice [2 4]. On the basis of published mammography audit data from large screening mammography practices and populationbased screening mammography programs [2 4, 6 14], desirable goals have been put forth for the detection of breast cancer in asymptomatic women [5, 9, 15 16]. The typical mammography practice includes a mix of screening and diagnostic examinations, all of which are subject to auditing requirements [1]. mammography is performed for a variety of problem-solving indications, including workup of screening examinations with abnormal findings, evaluation of abnormalities found on clinical examination, and short-term follow-up examinations for probably benign lesions ( Imaging Reporting and Data System [BI-RADS] category 3) and for patients with cancer who have been recently treated with breast preservation. AJR:176, March

2 Dee and Sickles special breast problems, such as the presence of implants or the evaluation of extent of disease of a known malignancy also may represent indications for diagnostic mammography. Because diagnostic examinations involve different patient populations from that of screening mammography, one might expect the clinical outcomes as measured by a medical audit to be different for diagnostic versus screening examinations. However, to our knowledge, only one audit has included data on diagnostic mammography, and that series involved only about 1000 examinations [8]. Thus, there is a paucity of diagnostic mammography outcomes data to use as a benchmark either for assessing the performance of a given practice or for deriving more general performance goals. In this study, we present detailed audit data involving slightly more than 10,000 consecutive examinations from our diagnostic mammography practice, in comparison with concurrently obtained screening data. We also separately analyze our diagnostic mammography outcomes by major clinical indication to provide insight into the mechanisms by which diagnostic outcomes differ from those encountered at screening. Materials and Methods From January 1997 through August 2000, data were collected for all screening and diagnostic mammography examinations that were performed at fixed-site facilities in our institution. In our practice, screening examinations involve asymptomatic women, consist of standard mediolateral oblique and craniocaudal projection mammograms of each breast, are interpreted in batches twice daily, and are compared with two previous examinations (when available). On the other hand, we perform diagnostic mammography as a problemsolving examination that uses the full spectrum of mammographic projections, which are tailored for each specific case to elucidate unresolved imaging features identified at screening or unexplained clinical signs and symptoms. Each diagnostic examination is interpreted immediately after imaging is completed. Data for all mammography examinations were stored in a computerized database designed at our institution and patterned on principles described previously [17]. The radiologist recorded the indication for each examination during image interpretation, according to the classification scheme listed in the Appendix. The radiologist also recorded standard BI-RADS assessment categories separately for each breast [16]. examination findings were considered to be abnormal if either breast was assessed as BI-RADS category 0 (incomplete: need additional imaging), category 4 (suspicious), or category 5 (highly suggestive of malignancy). examination findings were considered to be abnormal if either breast was assessed as BI-RADS category 4 or category 5. For all examination findings interpreted as abnormal, we searched the pathology database at our institution to determine whether biopsy (fine-needle aspiration, core, or surgical biopsy) was performed, and for those cases not identified in this manner, obtained the information directly from the referring clinician [17]. For biopsies resulting in the diagnosis of malignancy (defined herein as either ductal carcinoma in situ or any invasive carcinoma), we also recorded the size, grade, nodal status, and stage, which was based on the American Joint Committee on Cancer staging system [18]. Data tabulations and statistical computations were accomplished using the S-plus programming software (Mathsoft, Seattle, WA). The Student s t test was performed on normal distributions and the chi-square test was performed for comparison of proportional data. A p value of less than 0.05 was considered statistically significant. Results Patient Population This study included 46,857 consecutive mammography examinations, of which 10,007 were classified as diagnostic and 36,850 as screening examinations. Table 1 shows the age distribution of our diagnostic and screening mammography populations. The mean age of diagnostic mammography patients (55.8 years) was 3.3 years younger than the mean age of women undergoing screening (59.1 years) (p < ). Data on family history of breast cancer are shown in Table 2. Among our diagnostic examinations, 14.7% of patients had a strong or very strong family history of breast cancer, slightly greater than the 11.6% rate observed for screened women (p < ). Mammography Interpretation Results of mammography interpretation are shown in Table 3. examination findings were interpreted as abnormal (biopsy recommended) in 14.4% of cases. This rate is almost three times that observed for screening interpretations (5.2%) (p < ), despite the fact that screening examinations were considered to be abnormal not only if biopsy was recommended, but also if additional (diagnostic) imaging was requested. It is noteworthy that the abnormal interpretation rate among diagnostic cases varied greatly according to the indication for examination. For example, shortterm interval follow-up examination findings were interpreted as abnormal in only 94 cases (3.6%), compared with 286 cases (20.6%) interpreted as abnormal among patients with palpable masses, and 689 cases (34.0%) interpreted as abnormal among patients having additional workup of abnormal findings on screening examinations. Biopsy As shown in Table 3, the rate at which biopsy was performed was almost nine times greater for diagnostic mammography (11.9%) than for screening mammography (1.4%) (p < ). The pathology results of all biopsies are summarized in Table 4. Malignancy was found in 46% of biopsies for diagnostic mammography, more than the 38% positive biopsies for screening mammography (p = ). As also shown in Table 3, this translates to a cancer detection rate of 55 per 1000 cases for diagnostic mammography versus only five per 1000 cases for screening mammography (p < ). Table 4 shows that the positive biopsy rate for diagnostic mammography varied greatly Patient Age: TABLE 1 Versus Mammography Age (yr) < , , ,400 2,643 11, ,341 2,935 13, ,658 1,741 9, ,886 1,266 8, , ,807 36,850 10,007 46,857 Mean Median Family History of TABLE 2 Cancer: Versus Mammography History None 29,793 (80.8) 7,831 (78.3) Strong a 2,611 (7.1) 900 (9.0) Very strong b 1,665 (4.5) 573 (5.7) Not recorded 2,781 (7.5) 703 (7.0) a First-degree relative (mother, sister, daughter) with unilateral breast cancer. b First-degree relative with either premenopausal or bilateral breast cancer, or more than one first-degree relative with any breast cancer. 730 AJR:176, March 2001

3 Medical Audit of Mammography Examinations according to the indication for examination. The positive biopsy rates for patients with cancer having 6-month follow-up after breast preservation surgery (74%) and for patients with palpable masses (63%) were much greater than those for examinations performed for workup of abnormal findings on screening examinations (38%) (p < ). However, it is not surprising that the 38% TABLE 3 positive biopsy rate for patients having workup of abnormal findings on screening examinations was almost identical to the rate observed for our screening population. Overall Clinical Outcomes: Versus Mammography Outcome No. of Patients % No. of Patients % findings Biopsy performed Cancer detected Note. The category cancer detected is a subset of the category biopsy performed, which itself is a subset of the category abnormal findings. TABLE 4 Biopsy Results: Versus Mammography a Lobular carcinoma in situ or any type of atypical hyperplasia. b Ductal carcinoma in situ or any type of invasive carcinoma. 273 (53) 291 (52) 23 (82) 7 (15) 97 (33) 128 (48) 546 (46) Premalignant a 46 (9) 55 (10) 3 (11) 5 (11) 10 (3) 24 (9) 97 (8) Malignant b 197 (38) 216 (38) 2 (7) 35 (74) 184 (63) 114 (43) 551 (46) TABLE 5 Tumor Size: Versus Mammography Tumor Size DCIS Invasive DCIS Invasive 1 5 mm mm mm >20 mm Mean (mm) Note. DCIS = ductal carcinoma in situ. TABLE 6 Tumor Size Tumor Size for Mammography by Indication for Examination 1 5 mm mm mm >20 mm Mean (mm) Characteristics of Cancers Tables 5 and 6 summarize the data on tumor size. The mean size of cancers identified on diagnostic mammography was 18.0 mm, considerably larger than the 12.9-mm mean size of cancers detected at screening (p < ). This difference was found to be statistically significant for invasive tumors (p = ) and for intraductal tumors (p = 0.046). Cancers found in patients with palpable masses were even larger (mean, 26.4 mm), approximately twice the size of cancers found on screening (p < ). However, it is important to note that the size of cancers identified on diagnostic mammography performed for short-interval follow-up and for workup of abnormal findings on screening examinations (Table 6) was very similar to that observed for screening-detected cancers (Table 5). There were no cases in our database in which ductal carcinoma in situ was associated with axillary nodal metastasis. Table 7 summarizes the data on nodal status for the invasive cancers. Axillary nodes were positive for metastasis in 20% of invasive cancers found on diagnostic mammography, a rate more than three times greater than the 6% rate observed for screeningdetected cancers (p < ). The rate of nodal metastasis was even greater (33%) for patients with palpable invasive cancers (p < ). However, the nodal metastasis rate for invasive cancers identified on diagnostic mammography performed for short-interval follow-up and for workup of abnormal findings on screening examinations was very similar to that observed for screening-detected cancers. We observed no significant difference between the grade of tumors found at screening versus that of diagnostic mammography, either for ductal carcinoma in situ or for invasive cancers. On diagnostic mammography, 46.9% of ductal carcinoma in situ cases were found to be high grade (52.1% at screening) and 24.1% of invasive tumors were classified as high grade (19.0% at screening). Table 8 summarizes the data on cancer stage. Of cancers found on diagnostic mammography, 74.4% were stage 0 (ductal carcinoma in situ) or stage I, a considerably smaller percentage than the 89.3% observed for screening-detected cancers (p < ). Only 46.2% of cancers in patients with palpable masses were stage 0 or stage I (p < ). However, paralleling previously described observations for cancer size and axillary node status, the stage distribution for cancers identified on diagnostic mammography performed for short-interval follow-up and for workup of abnormal findings on screening examinations was very similar to that observed for screening-detected cancers (i.e., approximately 90% of cancers were stage 0 or stage I in these groups of patients). AJR:176, March

4 Dee and Sickles TABLE 7 Result Nodal Status of Invasive Cancer: Versus Mammography a Most patients with ductal carcinoma in situ, and a few patients with small low-grade invasive cancerous tumors, did not undergo axillary node sampling or sentinel node biopsy because probability of nodal metastasis was judged to be extremely low. Discussion The comprehensive medical audit of a mammography practice is a powerful tool in assessing the ability of mammography to detect breast cancer. The audit serves primarily as a self-assessment device, revealing both successes and deficiencies in the practice, thereby facilitating enhancements that improve patient care. Many large-scale audits have been reported, from academic and community-based practices as well as from population-based mammography programs [2 4, 6 14]; however, the peer-review literature on audits almost exclusively involves screening mammography. There is a paucity of published data on diagnostic examinations, which all practices in the United States must audit by government regulation, albeit on a limited basis. This study reports in detail on many important clinical outcomes from a large series of consecutive diagnostic mammography examinations compared with screening outcomes obtained concurrently. Results of the study show major differences in diagnostic versus screening outcomes, Negative or assumed negative a 118 (94) 124 (93) 2 (100) 20 (95) 110 (67) 53 (82) 309 (80) Positive 8 (6) 10 (7) 0 (0) 1 (5) 54 (33) 12 (18) 77 (20) TABLE 8 Stage Cancer Stage: Versus Mammography Note. DCIS = ductal carcinoma in situ. Numbers in parentheses are percentages. 0 (DCIS) 71 (36.0) 82 (38.0) 0 (0) 14 (40.0) 20 (10.9) 49 (43.0) 165 (29.9) I 105 (53.3) 115 (53.2) 2 (100.0) 17 (48.6) 65 (35.3) 46 (40.4) 245 (44.5) IA 11 (5.6) 19 (8.8) 1 (50.0) 3 (8.6) 5 (2.7) 9 (7.9) 37 (6.7) IB 53 (26.9) 52 (24.1) 0 (0) 9 (25.7) 19 (10.3) 14 (12.3) 94 (17.1) IC 41 (20.8) 44 (20.4) 1 (50.0) 5 (14.3) 41 (22.3) 23 (20.2) 114 (20.7) II 21 (10.7) 18 (8.3) 0 (0) 4 (11.4) 86 (46.7) 18 (15.8) 126 (22.9) III 0 (0) 1 (0.5) 0 (0) 0 (0) 8 (4.4) 1 (0.9) 10 (1.8) IV 0 (0) 0 (0) 0 (0) 0 (0) 5 (2.7) 0 (0) 5 (0.9) which can aid the individual mammography practice in analyzing its own audits. The demographics of patients undergoing diagnostic mammography appear to be somewhat different from those of screened women. patients are, on average, several years younger. This is due, at least in part, to the more widespread use of diagnostic mammography among patients in their 20s and 30s, many of whom have palpable masses. mammography patients also are slightly more likely to have a strong or very strong family history of breast cancer. The prevalence of cancer among diagnostic mammography patients is much greater than that in screened women. In our series, malignancy was found in 55 per 1000 diagnostic mammography examinations, 11 times more than the five per 1000 cancer detection rate observed in our screening population. We also found that the cancers identified at diagnostic mammography are larger (both for ductal carcinoma in situ and invasive carcinoma), more likely to be associated with axillary nodal metastasis, and more advanced in stage than those detected at screening. There are a variety of clinical indications for performing diagnostic mammography. One common reason is to work up screening-detected abnormalities. This subset of diagnostic patients differs from the overall screening population only in that mammographic abnormalities are present in all cases, thereby increasing the likelihood of biopsy and cancer diagnosis. However, the size, axillary node status, and cancer stage in this group of diagnostic patients, as expected, are observed to be virtually identical to those found in screening mammography populations. Another indication for diagnostic mammography examination is short-interval follow-up, either for mammographic lesions previously assessed as probably benign (BI-RADS category 3) or for more-frequent-than-annual surveillance of patients with cancer treated with breast preservation surgery. Our results reinforce the findings of previous reports that cancer is identified very infrequently in the follow-up of probably benign lesions, and that those few cancers in this group of patients are just as favorable in prognosis as those detected at screening [19 22]. To our knowledge, little has been reported on the detailed clinical outcomes of mammography among patients with cancer treated with breast preservation surgery [23, 24]. It is reassuring that cancers identified by mammographic surveillance in this group of patients have as favorable a prognosis as screening-detected cancers, just as it is readily understandable that cancer is found more frequently in this high-risk group of individuals (18 per 1000 cases in our series) than in screened women, and that the positive biopsy rate is almost twice as great (74% in our series) as that observed in screened women. The other important indication for diagnostic mammography is the presence of clinical symptoms or signs that suggest the possibility of breast cancer. Our results reveal striking differences in all clinical outcomes between the subgroup of patients who have palpable breast masses versus asymptomatic women undergoing screening mammography. The positive biopsy rate for the palpable-mass subgroup (63%) is much greater, the frequency with which cancer is found (184/1385, 133 per 1000 cases) is much greater, tumor size is much larger (mean size, 26.4 mm), the nodepositivity rate for invasive cancers is much greater (33%), and the stage is more advanced (53.8% stage 2 or higher cancers) than those of patients with screening-detected abnormalities. Similar results have been observed in smaller series of patients undergoing mammography for palpable masses [25 27]. 732 AJR:176, March 2001

5 Medical Audit of Mammography Examinations There are many other indications for which diagnostic mammography is performed, the spectrum ranging from indications similar to screening (patients with breast implants or breast pain), to patients in whom carcinoma is a certainty (patients with a known nonpalpable breast malignancy being evaluated for the extent of disease or response to neoadjuvant chemotherapy). These cases are grouped together in our database as other breast problems because we did not query for these specific indications prospectively. Because of the varied mix of indications in our other breast problems group, it is not surprising that our observed outcomes for this group are intermediate between those for screening examinations and those for diagnostic examinations performed to evaluate a palpable mass. Because the observed clinical outcomes for diagnostic mammography are different from those found for screening mammography, we believe that it is important to analyze the audit data from diagnostic mammography examinations separately from screening data. For example, the combined audit of a mammography practice involving 90% screening and 10% diagnostic examinations can be expected to show a cancer detection rate twice as great as that of a screening-only patient population, thus confounding the interpretation of audit results. Indeed, to derive the most meaningful assessment of results, it would be helpful to analyze diagnostic mammography audit data separately according to indication for examination. However, for logistic reasons, many mammography practices do not have the ability to perform audits in the segregated manner that we recommend. Rather, they are able to perform only a combined audit of all mammography examinations. If such a practice can derive a reliable estimate of the relative mix of its diagnostic versus screening examinations (or at a minimum, the percentage of patients with palpable masses at the time of examination), they should find the clinical outcomes reported in our study instructive in understanding (by extrapolation) the combined outcomes that they are able to measure. We plan to produce, in a separate report, a primer on how to make such extrapolations, using a wide range of diagnostic versus screening mammography mixes. References APPENDIX: Indications for Mammography Examination 1. Food and Drug Administration. Quality mammography standards: final rule. 21 CFR parts 16 and 900. Washington DC: United States Department of Health and Human Services, Sickles EA, Ominsky SH, Sollitto RA, Galvin HB, Monticciolo DL. Medical audit of a rapidthroughput mammography screening practice: methodology and results of 27,114 examinations. Radiology 1990;175: Sickles EA. Auditing your practice. RSNA categorical course in breast imaging 1995; Sickles EA. Quality assurance: how to audit your own mammography practice. Radiol Clin North Am 1992;30: Linver MN, Osuch JR, Brenner RJ, Smith RA. The mammography audit: a primer for the Mammography Quality Standards Act (MQSA). AJR 1995;165: Linver MN, Paster SB, 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: Rosenberg RD, Lando JF, Hunt WC, et al. The New Mexico Mammography Project: screening mammography performance in Albuquerque, New Mexico, 1991 to Cancer 1996;78: Robertson CL. A private breast imaging practice: medical audit of 25,788 screening and 1,077 diagnostic examinations. Radiology 1993;187: Tabár L, Fagerberg G, Duffy SW, Day NE, Gad A, Gröntoft O. Update of the Swedish two-county program of mammographic screening for breast cancer. Radiol Clin North Am 1992;30: Thurfjell E. Population-based mammography screening in clinical practice: results from the prevalence round in Uppsala county. Acta Radiol 1994;35: Thurfjell E. Mammography screening methods and diagnostic results. Acta Radiol Suppl 1995;395: Thurfjell EL, Lindgren JA. cancer survival rates with mammographic screening: similar favorable rates for women younger and those older than 50 years. Radiology 1996;201: Lynde JL. Low-cost screening mammography: results of 21,141 consecutive examinations in a community program. South Med J 1993;86: Burhenne LJ, Burhenne HJ, Kan L. Quality-oriented mass mammography screening. Radiology 1995;194: Bassett LW, Hendrick RE, Bassford TL, et al. Quality determinants of mammography: clinical practice guideline no. 13. Rockville, MD: Agency for Health Care Policy and Research, Public Health Service, United States Department of Health and Human Services, AHCPR Publication No American College of Radiology. imaging reporting and data system (BI-RADS), 3rd ed. Reston, VA: American College of Radiology, Sickles EA. The usefulness of computers in managing the operation of a mammography screening practice. AJR 1990;155: American Joint Committee on Cancer. Manual for staging of cancer, 5th ed. Philadelphia: Lippincott, Sickles EA. Management of probably benign breast lesions. Radiol Clin North Am 1995;33: Sickles EA. Periodic mammographic follow-up of probably benign lesions: results in 3,184 consecutive cases. Radiology 1991;179: Sickles EA. Nonpalpable, circumscribed, noncalcified solid breast masses: likelihood of malignancy based on lesion size and age of patient. Radiology 1994;192: Varas X, Leborgne F, Leborgne JH. Nonpalpable, probably benign lesions: role of follow-up mammography. Radiology 1992;184: Stomper PC, Recht A, Berenberg AL, Jochelson MS, Harris JR. Mammographic detection of recurrent cancer in the irradiated breast. AJR 1987;148: Orel SG, Troupin RH, Patterson EA, Fowble BL. cancer recurrence after lumpectomy and irradiation: role of mammography in detection. Radiology 1992;183: Bassett LW, Liu TH, Giuliano AE, Gold RH. The prevalence of carcinoma in palpable vs. impalpable, mammographically detected lesions. AJR 1991;157: Rosen EL, Sickles EA, Keating D. Ability of mammography to reveal nonpalpable breast cancer in women with palpable breast masses. AJR 1999;172: Roubidoux MA, Helvie MA, Lai NE, Paramagul C. Bilateral breast cancer: early detection with mammography. Radiology 1995;196: I. II. (any examination not classified as screening) A. Additional workup of abnormal screening examination B. Short-interval (6 month) follow-up 1. benign lesion (BI-RADS category 3) 2. Recent breast preservation surgery for cancer (at our institution, most such patients undergo mammography of treated breast every 6 months for the first 5 years after surgery and mammography of the contralateral breast every year) C. Evaluation of breast problem 1. mass (as reported by patient, mammography technologist, or referring clinician) 2. breast problem (all examinations not classified into one of preceding categories) AJR:176, March