Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Screening Clinical Breast Examination: How Often Does It Miss Lethal Breast Cancer?

Screening Clinical Breast Examination: How Often Does It Miss Lethal Breast Cancer? Abstract Background: Although most American women regularly receive screening clinical breast examination (CBE), little is known about CBE accuracy in community practice. We sought to estimate the rate of cancer detection (sensitivity) of screening CBE performed by community-based clinicians on women who ultimately died of breast cancer, as well as to identify factors associated with accurate detection. Subjects and Methods: We evaluated CBE accuracy among asymptomatic female health plan enrollees in five states (WA, OR, CA, MA, and MN) who received a CBE within 1 year of breast cancer diagnosis and who died of breast cancer within 15 years of diagnosis (N = 485). Sensitivity was estimated as the proportion whose exam was abnormal. Bivariate and logistic regression analyses identified patient characteristics associated with cancer detection. Results: An abnormality was noted on screening CBE in one of five women who ultimately succumbed to breast cancer (sensitivity = 21.6%; 95% confidence interval [CI] = 18.1% to 25.6%). The odds of a true-positive screening CBE (sensitivity) were decreased among women using estrogen (odds ratio [OR] = 0.23; 95% CI = 0.07 to 0.80), receiving a Pap smear during the same visit as CBE (OR = 0.45; 95% CI = 0.27 to 0.72), and with increasing chronic disease comorbidity (Ptrend = .08). Conclusion: Screening CBE as performed in the community may be insufficiently sensitive to detect most lethal breast cancers. Low sensitivity of screening CBE in community practice may be partly attributable to its performance alongside time-consuming clinical tasks such as Pap smear screening or chronic illness care. Seventy percent of U.S. women older than 40 years regularly receive clinical breast examination (CBE) as a screening test for breast cancer (1), representing 22 million CBEs performed per annum (2). Proponents of CBE screening emphasize that in some clinical trials of breast cancer screening, many breast cancers were missed by mammography but detected by CBE (3). Others have suggested that the effectiveness of CBE screening can be inferred from one clinical trial in which mammography reduced breast cancer mortality no more than did a carefully conducted, standardized 5- to 10-minute CBE (4). The results of clinical trials, however, may reflect care delivered under ideal circumstances. Although the sensitivity of screening CBE in clinical trials was approximately 54% (5), three studies of community-based screening programs have reported much lower sensitivity, ranging from 28% to 36% (6–8). These studies are limited by the inclusion of women with breast symptoms (who therefore received “diagnostic” examinations) (6) or have been based on patient samples from novel screening mammography programs that may not reflect primary care practice (7,8). Studies of predictors of CBE sensitivity have either examined a limited range of patient-level factors (9) or have been based on exams performed by nurses (7). Failure to diagnose breast cancer with CBE is a leading cause of malpractice claims against primary-care clinicians (10), so greater insight is needed regarding current levels of screening CBE accuracy among primary-care clinicians. We evaluated the cancer detection rate of CBE among asymptomatic women who ultimately died of breast cancer and determined patient characteristics associated with CBE accuracy. Because longer exam duration has been associated with improved lump detection in studies of silicone breast models (11), we hypothesized that screening CBE would be less sensitive when performed alongside time-consuming clinical tasks that may shorten CBE duration, such as Pap smear screening or chronic illness care. METHODS Setting and Subjects This study was conducted within the Cancer Research Network (CRN), a National Cancer Institute–supported consortium of nonprofit health maintenance organizations. The CRN consists of the research programs, enrollee populations, and databases of 11 integrated health care organizations that are members of the HMO Research Network. The health care delivery systems participating in the CRN are as follows: Group Health Cooperative, Harvard Pilgrim Health Care, Henry Ford Health System/Health Alliance Plan, HealthPartners Research Foundation, the Meyers Primary Care Institute of the Fallon Healthcare System/University of Massachusetts, and Kaiser Permanente in six regions (Colorado, Georgia, Hawaii, Northwest [Oregon and Washington], Northern California, and Southern California). The 11 health plans have nearly 10 million enrollees. The CRN conducts collaborative research on variations in cancer prevention and treatment policies and practices. The subjects were female enrollees from six large health plans in five U.S. states (WA, OR, CA, MN, and MA) whose medical records were reviewed for a case–control study assessing the effectiveness of breast cancer screening. Cases were diagnosed with breast cancer between ages 40 and 65 years in 1983–1993 and subsequently died of breast cancer. Breast cancer cases were identified from local tumor registries or the regional Surveillance, Epidemiology, and End Results registry, and cause of death was ascertained with standardized review of registry data and medical charts. We included in this study all cases who received screening CBE within 1 year of the date of breast cancer diagnosis (N = 485). We defined screening CBE as any CBE performed on an asymptomatic subject that was not performed to evaluate another positive test (e.g., mammography). The study methods were approved by the institutional review boards of the six participating health plans. Data Abstraction and Definitions The methods of chart abstraction and data quality monitoring have been described previously (12). In brief, trained assistants abstracted the following data from medical records: exposure to CBE during the 3-year period prior to initial suspicion of breast cancer; family history of breast cancer; history of breast biopsy; history of natural or surgical menopause; use of estrogen therapy; the Charlson Comorbidity Index (13) modified to include an additional point for hypertension; and receipt of a Pap smear on the date of CBE. We defined a positive family history of breast cancer as chart documentation of breast cancer in any relative. CBE results were coded as follows: 1) normal, 2) abnormal benign (e.g., fibrocystic changes), 3) indeterminate (e.g., abnormality noted requiring follow-up evaluation), or 4) suspicious of cancer. We defined a screening CBE result as positive if it was coded as either indeterminate or suspicious of cancer and negative if it was coded as normal or abnormal benign. Abstractors noted each date estrogen use was mentioned and whether estrogen was started, stopped, or continued on that date. We classified a woman as an estrogen user if her screening CBE occurred: 1) during a period defined by two separate chart notes signifying continuing estrogen use, 2) more than 30 days after starting estrogen therapy and continuation or discontinuation of estrogen was subsequently noted, or 3) within 90 days of an isolated note signifying estrogen continuation. Women were otherwise classified as nonusers or uncertain. Women were considered peri- or postmenopausal if: 1) menopausal status was specifically stated in the chart, 2) the woman had undergone surgical oophorectomy, or 3) symptoms of menopause (e.g., vasomotor instability, irregular menses) were recorded in the chart without a specific diagnosis of menopause. For each subject, we determined the dates of breast cancer diagnosis and initial stage (14) from either the medical record, local tumor registry, or by linkage with the regional Surveillance, Epidemiology, and End Result registry. Outcome Measures The primary outcome was the accuracy of each woman's single, most recent screening CBE before the date of cancer diagnosis. Sensitivity was defined as the number of women whose screening CBE was positive divided by the total number of women with breast cancer receiving a screening CBE. Positive examinations among women with cancer were defined as true-positive. Data Quality Assessment Randomly selected charts were reabstracted by a second reviewer who was masked to the subjects' cancer status, and there was excellent interrater reliability on abstraction results related to screening and diagnostic mammography (κ ranged from 0.76 to 0.91) (12). Three clinicians (MBB, SWF, JGE), also masked to cancer status, reviewed records of women for whom coded events might be questionable (e.g., a screening CBE within 9 months of a prior indeterminate/suspicious screening CBE) and resolved ambiguities by consensus. Coding remained unchanged for 93% of those labeled as screening CBEs and 92% of those labeled as diagnostic CBEs. Statistical Analyses We performed descriptive analyses to characterize the subjects demographically and clinically. We estimated standard errors and 95% confidence intervals (CIs) for sensitivity using the exact binomial distribution. Chi-square tests were used to identify demographic and clinical characteristics associated with sensitivity. When expected values in table cells were 5 or less, we tested for association using either Fisher's exact test or an extension of Fisher's exact test for r × c contingency tables (15). We used logistic regression to identify independent simultaneous patient-level predictors of sensitivity. The binary outcome for these analyses was whether women with cancer had a true-positive screening CBE. Regression models of sensitivity included variables predictive of the outcome in bivariate analyses (P≤.20) and indicator variables for three levels of the modified Charlson comorbidity index, age in 5-year categories, breast cancer stage, and health plan. None of 10 women with stage I breast cancer had a true-positive screening CBE, so the reference group for breast cancer stage included women with either stage I or II. We tested for a linear trend in sensitivity associated with increasing comorbidity by substituting the modified Charlson index as a single ordered categorical variable (coded 0, 1, or ≥2). Similarly, we tested for linear trend in sensitivity with increasing breast cancer stage by including breast cancer stage as a single ordered categorical variable (coded 0 for stage I or II, 1 for stage III, 2 for stage IV, and 3 for unstaged). Hypothesis tests were two-sided with a level of significance of 0.05. RESULTS Subject Characteristics The subjects were aged 40–65 years with a mean age of 51 years; 26% were nonwhite. Nearly one-quarter of subjects (24%) had a family history of breast cancer, and 20% had had a prior breast biopsy. More than one-third (39%) had at least one chronic disease. The number of subjects from each health plan ranged from 20 to 177. Most women with cancer (86%) had either stage II or III at the time of diagnosis, whereas 12% had stage IV. Women survived a mean of 4.7 years (standard deviation = 2.6) from the date of screening CBE. Sensitivity of Screening Clinical Breast Examination The mean interval between screening CBE and breast cancer diagnosis among cases was 151 days (median = 140 days). Of the 485 screening CBEs, the examination was interpreted as suspicious for cancer for 20 women (4.1%), indeterminate for 85 women (17.5%), abnormal benign for 56 women (11.6%), and normal for the remaining 324 women (67.0%). Thus, the screening CBE was either “suspicious for cancer” or “indeterminate” for 105 of 485 women, yielding a sensitivity of 21.6% (95% CI = 18.0% to 25.6%). Predictors of Sensitivity Sensitivity was statistically significantly lower among women with a family history of breast cancer, who were using estrogen at the time of the examination, and who received a Pap smear on the date of screening CBE (Table 1). Sensitivity was lower among women with either stage IV or unstaged cancer at diagnosis and with a comorbidity index of 2 or greater, though these associations did not reach statistical significance. Sensitivity of the breast examination did not differ significantly by age, ethnicity, history of breast biopsy, menopausal status, or year of examination. Table 1.  Sensitivity of screening clinical breast examination within 1 year of cancer diagnosis by demographic and clinical characteristics of women who died of breast cancer (N = 485) Characteristic  N  No. with positive exam  Sensitivity (%)  P  Age, y              40–44  132  22  16.7  .23      45–49  149  35  23.5        50–54  50  11  22.0        55–59  68  12  17.7        60–65  86  25  29.1    Ethnicity              African American  61  15  24.6  .67      White  360  79  21.9        Other  34  7  20.6        Unknown  30  4  13.3    Family history of breast cancer              No  367  87  23.7  .05      Yes  118  18  15.3    No. of prior breast biopsies              None  390  84  21.5  .99      1  55  12  21.8        ≥2  40  9  22.5    Menopausal status*              Premenopausal  154  36  23.4  .34      Peri- or postmenopausal  243  47  19.3        Unknown  88  22  25.0    Estrogen use on date of examination              Nonuser or uncertain  442  102  23.1  .01      Estrogen user  43  3  7.0    Charlson comorbidity index†              0  296  67  22.6  .40      1  113  26  23.0        ≥2  76  12  15.8    Pap smear concurrent with examination              No  245  70  28.6  <.001      Yes  240  35  14.6    Breast cancer stage              I  10  0  0  .12      II  143  39  27.3        III  269  57  21.2        IV  56  8  14.3        Unknown  7  1  14.3    Year of examination              1982–1984  69  13  18.8  .44      1985–1987  152  34  22.4        1988–1990  169  42  24.9        1991–1993  95  16  16.8    Characteristic  N  No. with positive exam  Sensitivity (%)  P  Age, y              40–44  132  22  16.7  .23      45–49  149  35  23.5        50–54  50  11  22.0        55–59  68  12  17.7        60–65  86  25  29.1    Ethnicity              African American  61  15  24.6  .67      White  360  79  21.9        Other  34  7  20.6        Unknown  30  4  13.3    Family history of breast cancer              No  367  87  23.7  .05      Yes  118  18  15.3    No. of prior breast biopsies              None  390  84  21.5  .99      1  55  12  21.8        ≥2  40  9  22.5    Menopausal status*              Premenopausal  154  36  23.4  .34      Peri- or postmenopausal  243  47  19.3        Unknown  88  22  25.0    Estrogen use on date of examination              Nonuser or uncertain  442  102  23.1  .01      Estrogen user  43  3  7.0    Charlson comorbidity index†              0  296  67  22.6  .40      1  113  26  23.0        ≥2  76  12  15.8    Pap smear concurrent with examination              No  245  70  28.6  <.001      Yes  240  35  14.6    Breast cancer stage              I  10  0  0  .12      II  143  39  27.3        III  269  57  21.2        IV  56  8  14.3        Unknown  7  1  14.3    Year of examination              1982–1984  69  13  18.8  .44      1985–1987  152  34  22.4        1988–1990  169  42  24.9        1991–1993  95  16  16.8    * P value after excluding women with “Unknown” menopausal status. † Modified to include one point for hypertension. View Large In a multiple logistic regression analysis adjusting for age, health plan, cancer stage, family history of breast cancer, and chronic disease comorbidity, the odds of a true positive screening CBE were reduced among women using estrogen (OR = 0.23; 95% CI = 0.07 to 0.80) or who received a Pap smear concurrently with CBE (OR = 0.45; 95% CI = 0.27 to 0.72) (Table 2). In the same analysis, there were non–statistically significant trends toward decreased sensitivity among women with a family history of breast cancer (P = .051), increasing chronic disease comorbidity (Ptrend = .08), and more advanced breast cancer stage (Ptrend = .09). Table 2.  Factors associated with reduced odds of true-positive screening clinical breast examination during the year prior to breast cancer diagnosis among women who died of breast cancer (N = 485) Characteristic  Adjusted OR (95% CI)*  Family history of breast cancer  0.56 (0.31 to 1.00)  Estrogen use on examination date  0.23 (0.07 to 0.80)  Charlson comorbidity index†        0  1.0 (ref)‡      1  0.83 (0.47 to 1.46)      ≥2  0.52 (0.25 to 1.08)  Concurrent Pap smear  0.45 (0.27 to 0.72)  Breast cancer stage        I or II  1.0 (ref)§      III  0.83 (0.51 to 1.37)      IV  0.55 (0.23 to 1.32)      Unknown  0.24 (0.03 to 2.19)  Characteristic  Adjusted OR (95% CI)*  Family history of breast cancer  0.56 (0.31 to 1.00)  Estrogen use on examination date  0.23 (0.07 to 0.80)  Charlson comorbidity index†        0  1.0 (ref)‡      1  0.83 (0.47 to 1.46)      ≥2  0.52 (0.25 to 1.08)  Concurrent Pap smear  0.45 (0.27 to 0.72)  Breast cancer stage        I or II  1.0 (ref)§      III  0.83 (0.51 to 1.37)      IV  0.55 (0.23 to 1.32)      Unknown  0.24 (0.03 to 2.19)  * OR = odds ratio; CI = confidence interval. Adjusted for age in 5-year categories, health plan, and all listed variables. † Modified to include one point for hypertension. ‡ Ptrend = .08, two-sided. § Ptrend = .09, two-sided. View Large DISCUSSION Among a community-based population in five U.S. states, screening CBE failed to detect breast cancer in nearly four of five women who ultimately died of breast cancer. The sensitivity of screening CBE among these women was substantially lower than in clinical trials of breast cancer screening. Sensitivity was statistically significantly lower among women who were using estrogen or who received a Pap smear along with CBE. Sensitivity of the screening CBE was also low in three other community-based studies (6–8). Among low-income women receiving examinations within the National Breast and Cervical Cancer Early Detection Program, CBE had a reported sensitivity of 59% (6). Yet nearly half of cancer cases had breast symptoms at the time of CBE, so many of these examinations were performed for diagnosis rather than screening (6,9). When only asymptomatic women were considered, sensitivity of the screening CBE was 36% (6). Sensitivity was similarly low among women with invasive breast cancer receiving CBE from nurse examiners within one of the six health plans from which our study population derived (7). One urban academic screening program reported a sensitivity of 28% (8). Alongside these studies, the low rate of breast cancer detection in our study suggests that the performance of CBE by community-based clinicians differs substantially from CBE performance in clinical trials. In studies using silicone breast models, the strongest predictor of lump detection was exam duration (11). In the Canadian trial, examiners used a standard protocol with an average duration of 5 to 10 minutes, which likely contributed to the achievement of 69% sensitivity (16). In contrast, a study in an ambulatory setting found that the average duration of clinical examination of both breasts was less than 2 minutes (17). If the typical CBE in community practice is relatively brief, clinicians may be more likely to miss subtle breast cancers than examiners in clinical trials. We found that receiving a Pap smear concurrently with screening CBE was associated with reduced sensitivity, and there was a trend toward lower sensitivity among patients with greater chronic disease comorbidity. It is possible that providers may perform the screening CBE more quickly during visits that include other time-consuming clinical tasks, such as Pap smear screening or chronic illness care. Although failure to detect breast cancer during CBE is a common cause of malpractice allegations, time constraints in usual primary-care practice may not allow the performance of thorough, high-quality screening. Screening CBE was significantly less sensitive among women using estrogen. Hormone replacement therapy was also associated with reduced CBE sensitivity among women screened at one urban academic program (8). Hormone replacement reduces the sensitivity of screening mammography (18), seemingly through its effect on breast density (19). Although a similar effect on CBE sensitivity seems biologically plausible, our findings conflict with a study from one of the participating health plans, which reported increased CBE sensitivity among women using hormone replacement (7). Also, our analysis did not account for progesterone usage, which may mediate the effect of hormone replacement on breast density (20). Cautious interpretation of our results is warranted because our sample of women with breast cancer may not be representative of the broader population of women who are diagnosed with breast cancer following CBE. All the women in our sample died of breast cancer. The women in our study may have had relatively aggressive cancers, which may have been more difficult to detect at screening than a population composed of women with a broader spectrum of cancer. Selection of more aggressive cases may explain the surprising trend toward reduced sensitivity among women with more advanced breast cancer stage at diagnosis, in contrast to previous studies that have found higher sensitivity with advanced stage (7,9). Nonetheless, the sensitivity of CBE in a separate report (7) among a general population of women diagnosed with breast cancer within one of the participating health plans (including women who did not die) was very similar to the sensitivity among patients in our study from the same plan (χ2 [1 degree of freedom] = 0.09; P = .77) (data not shown to preserve health plan confidentiality). Moreover, our findings are consistent with those of two other community-based studies that found low screening CBE sensitivity (6,8). Our results may not be generalizable to women younger than 40 years or older than 65 years or to uninsured populations. Although examinations in this study were performed between 1982 and 1993, we detected no secular improvement in sensitivity during this period, so we suspect that our estimate of CBE accuracy is reflective of current community practice. Strengths of this study include rigorous methods of chart abstraction and data quality assessment (12), which make us confident that the CBEs we have studied were performed for screening rather than diagnosis. Also, we studied examinations performed by many community-based clinicians in five U.S. states and evaluated the impact of a rich variety of demographic and clinical factors on examination outcomes. Our findings suggest that CBE fails to detect breast cancer in many women who ultimately succumb to the disease. Thus, missed breast cancer with CBE appears to be a regrettably common outcome of current standards of community practice, leaving primary-care physicians who commonly perform CBE screening in a medico–legal quandary. Because neither the practice environment nor patient expectations are conducive to careful, high-quality CBE, physicians may need to choose whether to proceed with usual, brief CBE screening or to omit CBE screening altogether, yet neither choice will clearly attenuate physicians' risk of tort claims for missed breast cancer. We acknowledge, however, that even if practice environments facilitated more thorough CBE, physicians may require training in conducting high-quality CBE, and educational efforts may be needed to modify patient expectations regarding optimal CBE content and duration. CBE achieved a high sensitivity in clinical trials (16), a finding that indicates that physical examination could play an important role in breast cancer screening. However, the low detection rate in this community study suggests that the screening CBE in clinical practice frequently misses breast cancers that ultimately prove to be lethal. Clinicians and patients should be informed that a high-quality CBE likely requires minutes rather than seconds. Although clinicians should be encouraged to conduct the screening CBE in a deliberate fashion, improved CBE performance may require organizational interventions that support the delivery of high-quality preventive services in primary care. Supported by the National Cancer Institute (grant U19CA79689, Edward H. Wagner, principal investigator). Dr. Fenton was a Robert Wood Johnson Clinical Scholar at the time of this work. The views expressed in this article are those of the authors and not necessarily the Robert Wood Johnson Foundation. We thank Sarah Greene, Kevin Beverly, Gene Hart, and the data abstractors for their efforts on this project. J. Elmore has served as an expert witness for medical malpractice cases. References (1) Meissner HI, Breen N, Yabroff KR. Whatever happened to clinical breast examinations? Am J Prev Med  2003; 25: 259–63. Google Scholar (2) United States Bureau of the Census. American Fact Finder. Available at: http://factfinder.census.gov/servlet/QTTable?_bm=y&-geo_id=01000US&-qr_name=DEC_2000_SF1_U_QTP1&-ds_name=DEC_2000_SF1_U&-_sse=on [last accessed: August 19, 2004]. Google Scholar (3) Smith RA, Saslow D, Sawyer KA, Burke W, Costanza ME, Evans WP 3rd, et al. American Cancer Society guidelines for breast cancer screening: update 2003. CA Cancer J Clin  2003; 53: 141–69. Google Scholar (4) Mittra I. Breast screening: the case for physical examination without mammography. Lancet  1994; 343: 342–4. Google Scholar (5) Barton MB, Harris R, Fletcher SW. The rational clinical examination. Does this patient have breast cancer? The screening clinical breast examination: should it be done? How? JAMA  1999; 282: 1270–80. Google Scholar (6) Bobo JK, Lee NC, Thames SF. Findings from 752081 clinical breast examinations reported to a national screening program from 1995 through 1998. J Natl Cancer Inst  2000; 92: 971–6. Google Scholar (7) Oestreicher N, White E, Lehman CD, Mandelson MT, Porter PL, Taplin SH. Predictors of sensitivity of clinical breast examination (CBE). Breast Cancer Res Treat  2002; 76: 73–81. Google Scholar (8) Kolb TM, Lichy J, Newhouse JH. Comparison of the performance of screening mammography, physical examination, and breast US and evaluation of factors that influence them: an analysis of 27825 patient evaluations. Radiology  2002; 225: 165–75. Google Scholar (9) Bobo JK, Lawson HW, Lee NC. Risk factors for failure to detect a cancer during clinical breast examinations (United States). Cancer Causes Control  2003; 14: 461–8. Google Scholar (10) Physicians' Insurers Association of America. Breast cancer study. 3rd ed. Washington (DC); 2002. Google Scholar (11) Fletcher SW, O'Malley MS, Bunce LA. Physicians' abilities to detect lumps in silicone breast models. JAMA  1985; 253: 2224–8. Google Scholar (12) Reisch LM, Fosse JS, Beverly K, Yu O, Barlow WE, Harris EL, et al. Training, quality assurance, and assessment of medical record abstraction in a multisite study. Am J Epidemiol  2003; 157: 546–51. Google Scholar (13) Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis  1987; 40: 373–83. Google Scholar (14) Singletary SE, Allred C, Ashley P, Bassett LW, Berry D, Bland KI, et al. Revision of the American Joint Committee on Cancer staging system for breast cancer. J Clin Oncol  2002; 20: 3628–36. Google Scholar (15) Mehta CA, Patel NR. A network algorithm for performing Fisher's exact test in r × c contingency tables. J Am Stat Assoc  1983; 76: 427–34. Google Scholar (16) Baines CJ, Miller AB, Bassett AA. Physical examination: its role as a single screening modality in the Canadian National Breast Screening Study. Cancer  1989; 63: 1816–22. Google Scholar (17) Kahn KL, Goldberg RJ. Screening for breast cancer in the ambulatory setting. Clin Res  1984; 32: 649A. Google Scholar (18) Laya MB, Larson EB, Taplin SH, White E. Effect of estrogen replacement therapy on the specificity and sensitivity of screening mammography. J Natl Cancer Inst  1996; 88: 643–9. Google Scholar (19) Carney PA, Miglioretti DL, Yankaskas BC, Kerlikowske K, Rosenberg R, Rutter CM, et al. Individual and combined effects of age, breast density, and hormone replacement therapy use on the accuracy of screening mammography. Ann Intern Med  2003; 138: 168–75. Google Scholar (20) Greendale GA, Reboussin BA, Slone S, Wasilauskas C, Pike MC, Ursin G. Postmenopausal hormone therapy and change in mammographic density. J Natl Cancer Inst  2003; 95: 30–7. Google Scholar © The Author 2005. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png JNCI Monographs Oxford University Press

Loading next page...
 
/lp/oxford-university-press/screening-clinical-breast-examination-how-often-does-it-miss-lethal-2sWKExy7fi

References (28)

Publisher
Oxford University Press
Copyright
© The Author 2005. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org.
ISSN
1052-6773
eISSN
1745-6614
DOI
10.1093/jncimonographs/lgi040
pmid
16287888
Publisher site
See Article on Publisher Site

Abstract

Abstract Background: Although most American women regularly receive screening clinical breast examination (CBE), little is known about CBE accuracy in community practice. We sought to estimate the rate of cancer detection (sensitivity) of screening CBE performed by community-based clinicians on women who ultimately died of breast cancer, as well as to identify factors associated with accurate detection. Subjects and Methods: We evaluated CBE accuracy among asymptomatic female health plan enrollees in five states (WA, OR, CA, MA, and MN) who received a CBE within 1 year of breast cancer diagnosis and who died of breast cancer within 15 years of diagnosis (N = 485). Sensitivity was estimated as the proportion whose exam was abnormal. Bivariate and logistic regression analyses identified patient characteristics associated with cancer detection. Results: An abnormality was noted on screening CBE in one of five women who ultimately succumbed to breast cancer (sensitivity = 21.6%; 95% confidence interval [CI] = 18.1% to 25.6%). The odds of a true-positive screening CBE (sensitivity) were decreased among women using estrogen (odds ratio [OR] = 0.23; 95% CI = 0.07 to 0.80), receiving a Pap smear during the same visit as CBE (OR = 0.45; 95% CI = 0.27 to 0.72), and with increasing chronic disease comorbidity (Ptrend = .08). Conclusion: Screening CBE as performed in the community may be insufficiently sensitive to detect most lethal breast cancers. Low sensitivity of screening CBE in community practice may be partly attributable to its performance alongside time-consuming clinical tasks such as Pap smear screening or chronic illness care. Seventy percent of U.S. women older than 40 years regularly receive clinical breast examination (CBE) as a screening test for breast cancer (1), representing 22 million CBEs performed per annum (2). Proponents of CBE screening emphasize that in some clinical trials of breast cancer screening, many breast cancers were missed by mammography but detected by CBE (3). Others have suggested that the effectiveness of CBE screening can be inferred from one clinical trial in which mammography reduced breast cancer mortality no more than did a carefully conducted, standardized 5- to 10-minute CBE (4). The results of clinical trials, however, may reflect care delivered under ideal circumstances. Although the sensitivity of screening CBE in clinical trials was approximately 54% (5), three studies of community-based screening programs have reported much lower sensitivity, ranging from 28% to 36% (6–8). These studies are limited by the inclusion of women with breast symptoms (who therefore received “diagnostic” examinations) (6) or have been based on patient samples from novel screening mammography programs that may not reflect primary care practice (7,8). Studies of predictors of CBE sensitivity have either examined a limited range of patient-level factors (9) or have been based on exams performed by nurses (7). Failure to diagnose breast cancer with CBE is a leading cause of malpractice claims against primary-care clinicians (10), so greater insight is needed regarding current levels of screening CBE accuracy among primary-care clinicians. We evaluated the cancer detection rate of CBE among asymptomatic women who ultimately died of breast cancer and determined patient characteristics associated with CBE accuracy. Because longer exam duration has been associated with improved lump detection in studies of silicone breast models (11), we hypothesized that screening CBE would be less sensitive when performed alongside time-consuming clinical tasks that may shorten CBE duration, such as Pap smear screening or chronic illness care. METHODS Setting and Subjects This study was conducted within the Cancer Research Network (CRN), a National Cancer Institute–supported consortium of nonprofit health maintenance organizations. The CRN consists of the research programs, enrollee populations, and databases of 11 integrated health care organizations that are members of the HMO Research Network. The health care delivery systems participating in the CRN are as follows: Group Health Cooperative, Harvard Pilgrim Health Care, Henry Ford Health System/Health Alliance Plan, HealthPartners Research Foundation, the Meyers Primary Care Institute of the Fallon Healthcare System/University of Massachusetts, and Kaiser Permanente in six regions (Colorado, Georgia, Hawaii, Northwest [Oregon and Washington], Northern California, and Southern California). The 11 health plans have nearly 10 million enrollees. The CRN conducts collaborative research on variations in cancer prevention and treatment policies and practices. The subjects were female enrollees from six large health plans in five U.S. states (WA, OR, CA, MN, and MA) whose medical records were reviewed for a case–control study assessing the effectiveness of breast cancer screening. Cases were diagnosed with breast cancer between ages 40 and 65 years in 1983–1993 and subsequently died of breast cancer. Breast cancer cases were identified from local tumor registries or the regional Surveillance, Epidemiology, and End Results registry, and cause of death was ascertained with standardized review of registry data and medical charts. We included in this study all cases who received screening CBE within 1 year of the date of breast cancer diagnosis (N = 485). We defined screening CBE as any CBE performed on an asymptomatic subject that was not performed to evaluate another positive test (e.g., mammography). The study methods were approved by the institutional review boards of the six participating health plans. Data Abstraction and Definitions The methods of chart abstraction and data quality monitoring have been described previously (12). In brief, trained assistants abstracted the following data from medical records: exposure to CBE during the 3-year period prior to initial suspicion of breast cancer; family history of breast cancer; history of breast biopsy; history of natural or surgical menopause; use of estrogen therapy; the Charlson Comorbidity Index (13) modified to include an additional point for hypertension; and receipt of a Pap smear on the date of CBE. We defined a positive family history of breast cancer as chart documentation of breast cancer in any relative. CBE results were coded as follows: 1) normal, 2) abnormal benign (e.g., fibrocystic changes), 3) indeterminate (e.g., abnormality noted requiring follow-up evaluation), or 4) suspicious of cancer. We defined a screening CBE result as positive if it was coded as either indeterminate or suspicious of cancer and negative if it was coded as normal or abnormal benign. Abstractors noted each date estrogen use was mentioned and whether estrogen was started, stopped, or continued on that date. We classified a woman as an estrogen user if her screening CBE occurred: 1) during a period defined by two separate chart notes signifying continuing estrogen use, 2) more than 30 days after starting estrogen therapy and continuation or discontinuation of estrogen was subsequently noted, or 3) within 90 days of an isolated note signifying estrogen continuation. Women were otherwise classified as nonusers or uncertain. Women were considered peri- or postmenopausal if: 1) menopausal status was specifically stated in the chart, 2) the woman had undergone surgical oophorectomy, or 3) symptoms of menopause (e.g., vasomotor instability, irregular menses) were recorded in the chart without a specific diagnosis of menopause. For each subject, we determined the dates of breast cancer diagnosis and initial stage (14) from either the medical record, local tumor registry, or by linkage with the regional Surveillance, Epidemiology, and End Result registry. Outcome Measures The primary outcome was the accuracy of each woman's single, most recent screening CBE before the date of cancer diagnosis. Sensitivity was defined as the number of women whose screening CBE was positive divided by the total number of women with breast cancer receiving a screening CBE. Positive examinations among women with cancer were defined as true-positive. Data Quality Assessment Randomly selected charts were reabstracted by a second reviewer who was masked to the subjects' cancer status, and there was excellent interrater reliability on abstraction results related to screening and diagnostic mammography (κ ranged from 0.76 to 0.91) (12). Three clinicians (MBB, SWF, JGE), also masked to cancer status, reviewed records of women for whom coded events might be questionable (e.g., a screening CBE within 9 months of a prior indeterminate/suspicious screening CBE) and resolved ambiguities by consensus. Coding remained unchanged for 93% of those labeled as screening CBEs and 92% of those labeled as diagnostic CBEs. Statistical Analyses We performed descriptive analyses to characterize the subjects demographically and clinically. We estimated standard errors and 95% confidence intervals (CIs) for sensitivity using the exact binomial distribution. Chi-square tests were used to identify demographic and clinical characteristics associated with sensitivity. When expected values in table cells were 5 or less, we tested for association using either Fisher's exact test or an extension of Fisher's exact test for r × c contingency tables (15). We used logistic regression to identify independent simultaneous patient-level predictors of sensitivity. The binary outcome for these analyses was whether women with cancer had a true-positive screening CBE. Regression models of sensitivity included variables predictive of the outcome in bivariate analyses (P≤.20) and indicator variables for three levels of the modified Charlson comorbidity index, age in 5-year categories, breast cancer stage, and health plan. None of 10 women with stage I breast cancer had a true-positive screening CBE, so the reference group for breast cancer stage included women with either stage I or II. We tested for a linear trend in sensitivity associated with increasing comorbidity by substituting the modified Charlson index as a single ordered categorical variable (coded 0, 1, or ≥2). Similarly, we tested for linear trend in sensitivity with increasing breast cancer stage by including breast cancer stage as a single ordered categorical variable (coded 0 for stage I or II, 1 for stage III, 2 for stage IV, and 3 for unstaged). Hypothesis tests were two-sided with a level of significance of 0.05. RESULTS Subject Characteristics The subjects were aged 40–65 years with a mean age of 51 years; 26% were nonwhite. Nearly one-quarter of subjects (24%) had a family history of breast cancer, and 20% had had a prior breast biopsy. More than one-third (39%) had at least one chronic disease. The number of subjects from each health plan ranged from 20 to 177. Most women with cancer (86%) had either stage II or III at the time of diagnosis, whereas 12% had stage IV. Women survived a mean of 4.7 years (standard deviation = 2.6) from the date of screening CBE. Sensitivity of Screening Clinical Breast Examination The mean interval between screening CBE and breast cancer diagnosis among cases was 151 days (median = 140 days). Of the 485 screening CBEs, the examination was interpreted as suspicious for cancer for 20 women (4.1%), indeterminate for 85 women (17.5%), abnormal benign for 56 women (11.6%), and normal for the remaining 324 women (67.0%). Thus, the screening CBE was either “suspicious for cancer” or “indeterminate” for 105 of 485 women, yielding a sensitivity of 21.6% (95% CI = 18.0% to 25.6%). Predictors of Sensitivity Sensitivity was statistically significantly lower among women with a family history of breast cancer, who were using estrogen at the time of the examination, and who received a Pap smear on the date of screening CBE (Table 1). Sensitivity was lower among women with either stage IV or unstaged cancer at diagnosis and with a comorbidity index of 2 or greater, though these associations did not reach statistical significance. Sensitivity of the breast examination did not differ significantly by age, ethnicity, history of breast biopsy, menopausal status, or year of examination. Table 1.  Sensitivity of screening clinical breast examination within 1 year of cancer diagnosis by demographic and clinical characteristics of women who died of breast cancer (N = 485) Characteristic  N  No. with positive exam  Sensitivity (%)  P  Age, y              40–44  132  22  16.7  .23      45–49  149  35  23.5        50–54  50  11  22.0        55–59  68  12  17.7        60–65  86  25  29.1    Ethnicity              African American  61  15  24.6  .67      White  360  79  21.9        Other  34  7  20.6        Unknown  30  4  13.3    Family history of breast cancer              No  367  87  23.7  .05      Yes  118  18  15.3    No. of prior breast biopsies              None  390  84  21.5  .99      1  55  12  21.8        ≥2  40  9  22.5    Menopausal status*              Premenopausal  154  36  23.4  .34      Peri- or postmenopausal  243  47  19.3        Unknown  88  22  25.0    Estrogen use on date of examination              Nonuser or uncertain  442  102  23.1  .01      Estrogen user  43  3  7.0    Charlson comorbidity index†              0  296  67  22.6  .40      1  113  26  23.0        ≥2  76  12  15.8    Pap smear concurrent with examination              No  245  70  28.6  <.001      Yes  240  35  14.6    Breast cancer stage              I  10  0  0  .12      II  143  39  27.3        III  269  57  21.2        IV  56  8  14.3        Unknown  7  1  14.3    Year of examination              1982–1984  69  13  18.8  .44      1985–1987  152  34  22.4        1988–1990  169  42  24.9        1991–1993  95  16  16.8    Characteristic  N  No. with positive exam  Sensitivity (%)  P  Age, y              40–44  132  22  16.7  .23      45–49  149  35  23.5        50–54  50  11  22.0        55–59  68  12  17.7        60–65  86  25  29.1    Ethnicity              African American  61  15  24.6  .67      White  360  79  21.9        Other  34  7  20.6        Unknown  30  4  13.3    Family history of breast cancer              No  367  87  23.7  .05      Yes  118  18  15.3    No. of prior breast biopsies              None  390  84  21.5  .99      1  55  12  21.8        ≥2  40  9  22.5    Menopausal status*              Premenopausal  154  36  23.4  .34      Peri- or postmenopausal  243  47  19.3        Unknown  88  22  25.0    Estrogen use on date of examination              Nonuser or uncertain  442  102  23.1  .01      Estrogen user  43  3  7.0    Charlson comorbidity index†              0  296  67  22.6  .40      1  113  26  23.0        ≥2  76  12  15.8    Pap smear concurrent with examination              No  245  70  28.6  <.001      Yes  240  35  14.6    Breast cancer stage              I  10  0  0  .12      II  143  39  27.3        III  269  57  21.2        IV  56  8  14.3        Unknown  7  1  14.3    Year of examination              1982–1984  69  13  18.8  .44      1985–1987  152  34  22.4        1988–1990  169  42  24.9        1991–1993  95  16  16.8    * P value after excluding women with “Unknown” menopausal status. † Modified to include one point for hypertension. View Large In a multiple logistic regression analysis adjusting for age, health plan, cancer stage, family history of breast cancer, and chronic disease comorbidity, the odds of a true positive screening CBE were reduced among women using estrogen (OR = 0.23; 95% CI = 0.07 to 0.80) or who received a Pap smear concurrently with CBE (OR = 0.45; 95% CI = 0.27 to 0.72) (Table 2). In the same analysis, there were non–statistically significant trends toward decreased sensitivity among women with a family history of breast cancer (P = .051), increasing chronic disease comorbidity (Ptrend = .08), and more advanced breast cancer stage (Ptrend = .09). Table 2.  Factors associated with reduced odds of true-positive screening clinical breast examination during the year prior to breast cancer diagnosis among women who died of breast cancer (N = 485) Characteristic  Adjusted OR (95% CI)*  Family history of breast cancer  0.56 (0.31 to 1.00)  Estrogen use on examination date  0.23 (0.07 to 0.80)  Charlson comorbidity index†        0  1.0 (ref)‡      1  0.83 (0.47 to 1.46)      ≥2  0.52 (0.25 to 1.08)  Concurrent Pap smear  0.45 (0.27 to 0.72)  Breast cancer stage        I or II  1.0 (ref)§      III  0.83 (0.51 to 1.37)      IV  0.55 (0.23 to 1.32)      Unknown  0.24 (0.03 to 2.19)  Characteristic  Adjusted OR (95% CI)*  Family history of breast cancer  0.56 (0.31 to 1.00)  Estrogen use on examination date  0.23 (0.07 to 0.80)  Charlson comorbidity index†        0  1.0 (ref)‡      1  0.83 (0.47 to 1.46)      ≥2  0.52 (0.25 to 1.08)  Concurrent Pap smear  0.45 (0.27 to 0.72)  Breast cancer stage        I or II  1.0 (ref)§      III  0.83 (0.51 to 1.37)      IV  0.55 (0.23 to 1.32)      Unknown  0.24 (0.03 to 2.19)  * OR = odds ratio; CI = confidence interval. Adjusted for age in 5-year categories, health plan, and all listed variables. † Modified to include one point for hypertension. ‡ Ptrend = .08, two-sided. § Ptrend = .09, two-sided. View Large DISCUSSION Among a community-based population in five U.S. states, screening CBE failed to detect breast cancer in nearly four of five women who ultimately died of breast cancer. The sensitivity of screening CBE among these women was substantially lower than in clinical trials of breast cancer screening. Sensitivity was statistically significantly lower among women who were using estrogen or who received a Pap smear along with CBE. Sensitivity of the screening CBE was also low in three other community-based studies (6–8). Among low-income women receiving examinations within the National Breast and Cervical Cancer Early Detection Program, CBE had a reported sensitivity of 59% (6). Yet nearly half of cancer cases had breast symptoms at the time of CBE, so many of these examinations were performed for diagnosis rather than screening (6,9). When only asymptomatic women were considered, sensitivity of the screening CBE was 36% (6). Sensitivity was similarly low among women with invasive breast cancer receiving CBE from nurse examiners within one of the six health plans from which our study population derived (7). One urban academic screening program reported a sensitivity of 28% (8). Alongside these studies, the low rate of breast cancer detection in our study suggests that the performance of CBE by community-based clinicians differs substantially from CBE performance in clinical trials. In studies using silicone breast models, the strongest predictor of lump detection was exam duration (11). In the Canadian trial, examiners used a standard protocol with an average duration of 5 to 10 minutes, which likely contributed to the achievement of 69% sensitivity (16). In contrast, a study in an ambulatory setting found that the average duration of clinical examination of both breasts was less than 2 minutes (17). If the typical CBE in community practice is relatively brief, clinicians may be more likely to miss subtle breast cancers than examiners in clinical trials. We found that receiving a Pap smear concurrently with screening CBE was associated with reduced sensitivity, and there was a trend toward lower sensitivity among patients with greater chronic disease comorbidity. It is possible that providers may perform the screening CBE more quickly during visits that include other time-consuming clinical tasks, such as Pap smear screening or chronic illness care. Although failure to detect breast cancer during CBE is a common cause of malpractice allegations, time constraints in usual primary-care practice may not allow the performance of thorough, high-quality screening. Screening CBE was significantly less sensitive among women using estrogen. Hormone replacement therapy was also associated with reduced CBE sensitivity among women screened at one urban academic program (8). Hormone replacement reduces the sensitivity of screening mammography (18), seemingly through its effect on breast density (19). Although a similar effect on CBE sensitivity seems biologically plausible, our findings conflict with a study from one of the participating health plans, which reported increased CBE sensitivity among women using hormone replacement (7). Also, our analysis did not account for progesterone usage, which may mediate the effect of hormone replacement on breast density (20). Cautious interpretation of our results is warranted because our sample of women with breast cancer may not be representative of the broader population of women who are diagnosed with breast cancer following CBE. All the women in our sample died of breast cancer. The women in our study may have had relatively aggressive cancers, which may have been more difficult to detect at screening than a population composed of women with a broader spectrum of cancer. Selection of more aggressive cases may explain the surprising trend toward reduced sensitivity among women with more advanced breast cancer stage at diagnosis, in contrast to previous studies that have found higher sensitivity with advanced stage (7,9). Nonetheless, the sensitivity of CBE in a separate report (7) among a general population of women diagnosed with breast cancer within one of the participating health plans (including women who did not die) was very similar to the sensitivity among patients in our study from the same plan (χ2 [1 degree of freedom] = 0.09; P = .77) (data not shown to preserve health plan confidentiality). Moreover, our findings are consistent with those of two other community-based studies that found low screening CBE sensitivity (6,8). Our results may not be generalizable to women younger than 40 years or older than 65 years or to uninsured populations. Although examinations in this study were performed between 1982 and 1993, we detected no secular improvement in sensitivity during this period, so we suspect that our estimate of CBE accuracy is reflective of current community practice. Strengths of this study include rigorous methods of chart abstraction and data quality assessment (12), which make us confident that the CBEs we have studied were performed for screening rather than diagnosis. Also, we studied examinations performed by many community-based clinicians in five U.S. states and evaluated the impact of a rich variety of demographic and clinical factors on examination outcomes. Our findings suggest that CBE fails to detect breast cancer in many women who ultimately succumb to the disease. Thus, missed breast cancer with CBE appears to be a regrettably common outcome of current standards of community practice, leaving primary-care physicians who commonly perform CBE screening in a medico–legal quandary. Because neither the practice environment nor patient expectations are conducive to careful, high-quality CBE, physicians may need to choose whether to proceed with usual, brief CBE screening or to omit CBE screening altogether, yet neither choice will clearly attenuate physicians' risk of tort claims for missed breast cancer. We acknowledge, however, that even if practice environments facilitated more thorough CBE, physicians may require training in conducting high-quality CBE, and educational efforts may be needed to modify patient expectations regarding optimal CBE content and duration. CBE achieved a high sensitivity in clinical trials (16), a finding that indicates that physical examination could play an important role in breast cancer screening. However, the low detection rate in this community study suggests that the screening CBE in clinical practice frequently misses breast cancers that ultimately prove to be lethal. Clinicians and patients should be informed that a high-quality CBE likely requires minutes rather than seconds. Although clinicians should be encouraged to conduct the screening CBE in a deliberate fashion, improved CBE performance may require organizational interventions that support the delivery of high-quality preventive services in primary care. Supported by the National Cancer Institute (grant U19CA79689, Edward H. Wagner, principal investigator). Dr. Fenton was a Robert Wood Johnson Clinical Scholar at the time of this work. The views expressed in this article are those of the authors and not necessarily the Robert Wood Johnson Foundation. We thank Sarah Greene, Kevin Beverly, Gene Hart, and the data abstractors for their efforts on this project. J. Elmore has served as an expert witness for medical malpractice cases. References (1) Meissner HI, Breen N, Yabroff KR. Whatever happened to clinical breast examinations? Am J Prev Med  2003; 25: 259–63. Google Scholar (2) United States Bureau of the Census. American Fact Finder. Available at: http://factfinder.census.gov/servlet/QTTable?_bm=y&-geo_id=01000US&-qr_name=DEC_2000_SF1_U_QTP1&-ds_name=DEC_2000_SF1_U&-_sse=on [last accessed: August 19, 2004]. Google Scholar (3) Smith RA, Saslow D, Sawyer KA, Burke W, Costanza ME, Evans WP 3rd, et al. American Cancer Society guidelines for breast cancer screening: update 2003. CA Cancer J Clin  2003; 53: 141–69. Google Scholar (4) Mittra I. Breast screening: the case for physical examination without mammography. Lancet  1994; 343: 342–4. Google Scholar (5) Barton MB, Harris R, Fletcher SW. The rational clinical examination. Does this patient have breast cancer? The screening clinical breast examination: should it be done? How? JAMA  1999; 282: 1270–80. Google Scholar (6) Bobo JK, Lee NC, Thames SF. Findings from 752081 clinical breast examinations reported to a national screening program from 1995 through 1998. J Natl Cancer Inst  2000; 92: 971–6. Google Scholar (7) Oestreicher N, White E, Lehman CD, Mandelson MT, Porter PL, Taplin SH. Predictors of sensitivity of clinical breast examination (CBE). Breast Cancer Res Treat  2002; 76: 73–81. Google Scholar (8) Kolb TM, Lichy J, Newhouse JH. Comparison of the performance of screening mammography, physical examination, and breast US and evaluation of factors that influence them: an analysis of 27825 patient evaluations. Radiology  2002; 225: 165–75. Google Scholar (9) Bobo JK, Lawson HW, Lee NC. Risk factors for failure to detect a cancer during clinical breast examinations (United States). Cancer Causes Control  2003; 14: 461–8. Google Scholar (10) Physicians' Insurers Association of America. Breast cancer study. 3rd ed. Washington (DC); 2002. Google Scholar (11) Fletcher SW, O'Malley MS, Bunce LA. Physicians' abilities to detect lumps in silicone breast models. JAMA  1985; 253: 2224–8. Google Scholar (12) Reisch LM, Fosse JS, Beverly K, Yu O, Barlow WE, Harris EL, et al. Training, quality assurance, and assessment of medical record abstraction in a multisite study. Am J Epidemiol  2003; 157: 546–51. Google Scholar (13) Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis  1987; 40: 373–83. Google Scholar (14) Singletary SE, Allred C, Ashley P, Bassett LW, Berry D, Bland KI, et al. Revision of the American Joint Committee on Cancer staging system for breast cancer. J Clin Oncol  2002; 20: 3628–36. Google Scholar (15) Mehta CA, Patel NR. A network algorithm for performing Fisher's exact test in r × c contingency tables. J Am Stat Assoc  1983; 76: 427–34. Google Scholar (16) Baines CJ, Miller AB, Bassett AA. Physical examination: its role as a single screening modality in the Canadian National Breast Screening Study. Cancer  1989; 63: 1816–22. Google Scholar (17) Kahn KL, Goldberg RJ. Screening for breast cancer in the ambulatory setting. Clin Res  1984; 32: 649A. Google Scholar (18) Laya MB, Larson EB, Taplin SH, White E. Effect of estrogen replacement therapy on the specificity and sensitivity of screening mammography. J Natl Cancer Inst  1996; 88: 643–9. Google Scholar (19) Carney PA, Miglioretti DL, Yankaskas BC, Kerlikowske K, Rosenberg R, Rutter CM, et al. Individual and combined effects of age, breast density, and hormone replacement therapy use on the accuracy of screening mammography. Ann Intern Med  2003; 138: 168–75. Google Scholar (20) Greendale GA, Reboussin BA, Slone S, Wasilauskas C, Pike MC, Ursin G. Postmenopausal hormone therapy and change in mammographic density. J Natl Cancer Inst  2003; 95: 30–7. Google Scholar © The Author 2005. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org.

Journal

JNCI MonographsOxford University Press

Published: Nov 1, 2005

There are no references for this article.