The Contribution of Clinical Breast Examination to the Accuracy of Breast Screening

Abstract

Background

There is controversy about whether adding clinical breast examination (CBE) to mammography improves the accuracy of breast screening. We compared the accuracy of screening among centers that offered CBE in addition to mammography with that among centers that offered only mammography.

Methods

The cohort included 290 230 women aged 50–69 years who were screened at regional cancer centers or affiliated centers within the Ontario Breast Screening Program between January 1, 2002, and December 31, 2003, and were followed up for 12 months. The regional cancer centers offer screening mammography and CBE performed by a nurse. All affiliated centers provide mammography but not all provide CBE. Performance measures for 232 515 women who were screened by mammography and CBE at the nine regional cancer centers or 59 affiliated centers that provided CBE were compared with those for 57 715 women who were screened by mammography alone at 34 affiliated centers that did not provide CBE.

Results

Sensitivity of referrals was higher for women who were screened at regional cancer centers or affiliated centers that offered CBE in addition to mammography than for women screened at affiliated centers that did not offer CBE (initial screen: 94.9% and 94.6%, respectively, vs 88.6%; subsequent screen: 94.9% and 91.7%, respectively, vs 85.3%). Mammography sensitivity was similar between centers that offered CBE and those that did not. However, women without cancer who were screened at regional cancer centers or affiliated centers that offered CBE had a higher false-positive rate than women screened at affiliated centers that offered only mammography (initial screen: 12.5% and 12.4%, respectively, vs 7.4%; subsequent screen: 6.3% and 8.3%, respectively, vs 5.4%).

Conclusions

Women should be informed of the benefits and risks of having a CBE in addition to mammography for breast screening.

CONTEXT AND CAVEATS

Prior knowledge

Although clinical breast examination (CBE) can detect cancers that are missed by mammography, controversy exists about whether adding CBE to mammography improves the accuracy of breast screening.

Study design

A comparison of the accuracy of screening among centers within the Ontario Breast Screening Program that offered CBE performed by a nurse in addition to mammography with that among centers that offered only mammography.

Contribution

Cancer detection rates and sensitivity were higher, as were the abnormal call rates and false-positive rates, among centers that offered CBE in addition to mammography than among centers that offered only mammography.

Implications

The benefit in increased sensitivity from adding CBE to mammography needs to be weighed against the potential risks and costs of further follow-up due to false-positive results as well as the anxiety associated with additional diagnostic evaluations.

Limitations

Comparison of performance measures across center types in a nonrandomized study design is subject to confounding. The findings may not be generalizable to CBE use in clinical practice or for women outside of the age range of this study.

From the Editors

The Canadian Task Force on Preventive Health Care recommends that women aged 50–69 years undergo screening for breast cancer by mammography and clinical breast examination (CBE) every 1–2 years ( 1 ). Results of several randomized controlled trials have provided evidence that screening reduces the breast cancer mortality rate ( 2 , 3 ). Long-term follow-up of participants in these trials revealed a statistically significant reduction in mortality upon achieving 70% attendance among women in the target age group ( 4 , 5 ). Because most organized screening programs require a sufficient amount of time to reach a substantial proportion of the target population, initial evaluation of such programs includes evaluation of interim indicators of effectiveness, such as cancer detection rates and screening accuracy.

Measures of screening accuracy are particularly important interim indicators of the effectiveness of breast screening programs. Two such indicators, the cancer detection rate and screen sensitivity (which measures the proportion of women with cancer who had a positive screen), are used to determine whether screening is effective in detecting breast cancer in women. If the sensitivity is low, the test will miss many cancers, resulting in a high rate of false-negative results or interval cancers that contribute substantially to mortality in the screened population ( 6 , 7 ). Two other measures of screening accuracy, the referral rate and screen specificity, are related to the efficiency of the screening test and the need to perform as few procedures as possible on women who do not have cancer. If a screening modality produces a high proportion of false-positive results, then women may be inappropriately referred for further potentially invasive diagnostic tests. In addition, women who have a false-positive mammogram can experience anxiety, distress, and intrusive thoughts, which may persist for several months or years after the completion of assessment ( 8–10 ).

Although CBE can detect cancers that are missed by mammography ( 11 ), there is controversy on the value of adding CBE to mammography. A review of breast screening randomized clinical trials showed that mortality reduction in the four trials including CBE in addition to mammography was similar to those in trials including mammography only ( 3 ). A study of four Canadian provinces (including Ontario) that provide CBE found that CBE added to mammography had a small contribution to the early detection of breast cancer ( 12 ).

The accuracy of CBE can be affected by examiner experience and technique. A pooled analysis found that the accuracy of CBE varies according to factors related to the examiner, such as the duration of the examination, the examiner's experience, and the CBE technique used ( 11 ). The highest CBE sensitivity reported was achieved by the Canadian National Breast Screening Study-2, a randomized trial of women aged 50–59 years who were offered CBE alone or CBE and mammography, in which CBE was performed by specially trained health professionals who received a specific protocol for conducting breast examinations ( 13 , 14 ).

The Ontario Breast Screening Program (OBSP) is conducted under the auspices of Cancer Care Ontario and has operated since July 1, 1990, to deliver a population-based breast screening program. From 1990 to 1994, screening took place at nine regional cancer centers located across Ontario (including one in a mobile unit that targets communities in the northwestern region of the province). All nine regional cancer centers have dedicated breast screening centers that offer biennial screening consisting of two-view mammography and CBE performed by a nurse. Beginning in 1995, the OBSP began offering screening mammography and CBE at affiliated centers located within established mammography facilities in hospitals or independent health facilities. Centers that were affiliated with OBSP after April 1998 were encouraged but not required to offer CBE.

The OBSP provides a unique opportunity to evaluate the value of adding CBE to mammography alone for the detection of breast cancer. The purpose of this study was to compare the accuracy of referrals for further assessment made on the basis of CBE alone, mammography alone, and CBE and/or mammography among women who attended regional cancer centers or affiliated centers that provided mammography and CBE with the accuracy of referrals for further assessment among women who attended affiliated centers that offered mammography alone.

Study Population and Methods

Study Population

A cohort of women aged 50–69 years who were screened through the OBSP between January 1, 2002, and December 31, 2003, was identified from information routinely collected by an integrated client management system on all women screened in the OBSP and observed for up to 12 months after their last screening examination. Women who participate in the OBSP must be residents of Ontario, have no history of breast cancer or augmentation mammoplasty, and have no acute breast symptoms. The OBSP offers all eligible women biennial screening consisting of two-view mammography. Women who are considered to be at high risk of breast cancer because they have a family history of breast cancer or ovarian cancer or a personal history of ovarian cancer or benign breast conditions such as lobular carcinoma in situ, benign phyllodes tumor, radial scar, atypical ductal hyperplasia, or atypical lobular hyperplasia are screened annually. Women are also advised to be screened at 1-year intervals if their breast density is 75% or higher and/or based on their assessment results or if requested by the screening radiologist or physician. Of the 102 OBSP screening centers in operation during the study period, 68 (67%) offered CBE in addition to mammography. A complete description of the details of the operation of the OBSP has been published ( 15 , 16 ).

At OBSP centers that employ nurses, the primary role of the nurses is to perform CBE at both the initial and subsequent screens. The CBE is performed while the woman is in the upright, supine, and lateral oblique positions by using the clock method of examination and includes palpation of nodes in the axillae and the supra- and infra-clavicular areas as well as visual observation of the entire breast area. The CBE takes 8–10 minutes, on average, and generally occurs before mammography is performed.

There were 167 nurses in the OBSP during the study period. Nurses in the OBSP are awarded certification for the performance of CBE following an intense didactic training period during which nurses complete 40 CBEs under the direct supervision of a designated nurse and 40 unsupervised CBEs and the completion of a written examination. To maintain their OBSP certification, nurses are regularly assessed through an OBSP quality assurance program that consists of practice reviews of documentation of findings and CBE performance (given at intervals that depend on the nurse's years of experience) and by performing a minimum of 500 CBEs per year. Nurses are also evaluated annually with regard to their personal performance measures (ie, cancer detection rate, positive predictive value, sensitivity), which are expected to fall within 2 SDs of the program average. In addition to certification and ongoing evaluation of nurses, comprehensive standards of practice have been developed for the performance of CBE (visual inspection and palpation), documentation of findings, and criteria for referral.

At OBSP centers that do not employ nurses, women do not receive a CBE. At these centers, a medical radiation technologist who has special training in mammography and is accredited with the Canadian Association of Radiologist's Mammography Accreditation Program (CAR-MAP) performs a visual inspection of both breasts with the woman in an upright position and documents any visual findings as well as physical findings identified by the woman herself. These findings are available to the radiologist when he or she reads the mammogram. All women who attend centers that do not have a nurse on-site are encouraged to visit their family physician to receive a CBE for routine screening regardless of the findings documented by the technologist.

At all OBSP centers, mammography consists of standard craniocaudal and mediolateral oblique views that are obtained for each woman using high-quality mammography machines and processors that are optimized for mammography by certified mammography technologists. Quality assurance on the equipment meets or exceeds that specified by CAR-MAP. Films are read once in batch by a radiologist who can refer to previous films when they are available. There were 290 radiologists in the OBSP during the study period. All OBSP radiologists are accredited under CAR-MAP and must meet or exceed the program's reading volume standards by reading a minimum of 480 mammograms per year.

Women may be referred for further assessment by either the radiologist or the nurse. Referral information is collected in a standardized method on the screening report and recorded as “normal” or “recommended clinical assessment by physician” by the nurse and as “normal/benign” or “needs additional evaluation by imaging and/or surgical consultation” by the radiologist. The mammogram results are not available to the nurse. Thus, the nurse makes an independent decision about whether the woman has an abnormality that requires further assessment based on standardized referral criteria for visual and palpable findings. The nurse informs the woman of any clinical abnormality at the time of the visit and refers the woman to her family physician for assessment. The radiologist makes his or her decision to refer the woman for diagnostic assessment regardless of the CBE findings recorded by the nurse and indicates his or her recommendations for assessment (eg, additional views, ultrasound, or a surgical consultation).

Between January 1, 2002, and December 31, 2003, the OBSP provided 343 711 screens to 301 362 women who were aged 50–69 years and had completed follow-up. We excluded 4604 women because they had a mammogram and CBE from three OBSP centers that no longer offered CBE, 6520 women because they only had a mammogram in centers that offer both mammography and CBE, and eight women because they attended an affiliated center that screened fewer than 10 eligible women. The final sample size for the analyses included 290 230 women. For women who were diagnosed with breast cancer (screen-detected cancer = 1733; interval cancer = 197), we included the last OBSP screen before the diagnosis; among the women without breast cancer (n = 288 300), for those who had more than one screen, we included the last OBSP screen during the study period. This study was approved by the Health Sciences 1 Research Ethics Board at the University of Toronto.

Definition of Screening Center Characteristics

OBSP screening services are delivered through nine regional cancer centers and 93 affiliated centers. The regional cancer centers provide administration, coordination, education, and training to the affiliated centers in their region. Affiliated centers provide screening mammography services through hospitals or independent health facilities. The annual volume of screens was calculated as the number of screens performed at the center between January 1, 2002, and December 31, 2003, divided by the number of years the center was in operation during the same time period. The total number of years of operation refers to the number of years the center operated as part of the OBSP from July 1, 1990, to December 31, 2003.

Definition of Risk Factors

Information on risk factors for breast cancer was based on self-reported data collected at the woman's most recent OBSP screening appointment during the study period through a personal interview with the nurse or technologist. Women who had at least one first-degree relative with breast cancer were classified as having a positive family history. Women were classified as current users of hormone therapy if they were using hormone therapy at the time of their last screen during the study period. Mammographic density was recorded by the radiologist as less than 75% or 75% or greater based on results of a previous study ( 17 ), which found that a breast density of 75% or higher was associated with a lower mammography sensitivity.

Definition of Screening Visit Characteristics

Age at last screen was defined as the age of the woman at her most recent OBSP screening examination from January 1, 2002, to December 31, 2003. The modality of referral for further assessment was classified as mammography (ie, referral based on an abnormal mammogram regardless of the CBE result), CBE (ie, referral based on an abnormal CBE regardless of mammogram result), or CBE and/or mammography (ie, referral based on an abnormal CBE and/or an abnormal mammogram). The most recent screen during the study period was defined as an initial screen for women who had only one OBSP screen during their lifetime or as a subsequent screen for women who had more than one OBSP screen during their lifetime. For women who were diagnosed with a screen-detected or an interval breast cancer, pathological confirmation was obtained from regional staff during the recall process or through record linkage with the Ontario Cancer Registry. Case ascertainment in the Ontario Cancer Registry was estimated in 1996 to be 98% complete for breast cancer ( 18 ).

Definition of Performance Measures

Performance measures for mammography were calculated for all women who had a two-view mammogram of both breasts that was interpreted by a radiologist. Performance measures for CBE were calculated for all women who had a complete CBE of both breasts performed by a nurse. Abnormal call rate was defined as the number of women who were referred for further testing based on an abnormal OBSP screen per 100 women screened. Cancer detection rate was defined as the number of women with a screen-detected ductal carcinoma in situ or invasive cancer per 1000 women screened. Sensitivity was defined as the proportion of women with breast cancer who had a positive screening test (the number of true-positive screens divided by the sum of the true-positive and false-negative screens). For mammography, a true-positive result (ie, a screen-detected cancer) included breast cancers that were diagnosed within 12 months after an abnormal mammogram. A false-negative result (ie, an interval cancer) included breast cancers that were diagnosed after a normal mammogram as well as screen-detected breast cancers that were detected by CBE alone. Conversely, for CBE, a true-positive result (ie, a screen-detected cancer) included breast cancers that were diagnosed within 12 months after a positive CBE. A false-negative result (ie, an interval cancer) included breast cancers that were diagnosed after a negative CBE as well as screen-detected cancers that were diagnosed by mammography alone.

False-positive rate was defined as the proportion of women without breast cancer who had a positive screening test (the number of false-positive screens divided by the sum of the true-negative and false-positive screens). For mammography, a true-negative result included women who did not have a breast cancer diagnosis within 12 months after a normal mammogram and a false-positive result included women who did not have a breast cancer diagnosis after an abnormal mammogram, irrespective of their CBE result. Conversely, for CBE, a true-negative result included women who did not have a breast cancer diagnosis within 12 months after a negative CBE and a false-positive result included women who did not have a breast cancer diagnosis after a positive CBE, irrespective of their mammogram result.

Statistical Analysis

Characteristics of the screening centers and women were compared by center types by using a two-sided Pearson χ ² test. Performance measures (ie, abnormal call rate, cancer detection rate, sensitivity, and false-positive rate) were calculated by type of screening center (regional cancer center with CBE, affiliated center with CBE, affiliated center without CBE) and by modality of referral, and 95% confidence intervals (CIs) were calculated for binomial proportions. Statistical analyses were conducted using SAS version 9.1 (SAS Institute, Cary, NC) and all P values are two-sided. A P value less than .05 was considered statistically significant.

Results

CBE was offered in addition to mammography at nine regional cancer centers and at 59 affiliated centers ( Table 1 ). Compared with the affiliated centers, the regional cancer centers had a higher average annual volume of screens (7939 vs 1560 and 1710 for affiliated centers with and without CBE, respectively) and had offered breast screening as part of the OBSP for a longer time (mean number of years of operation within OBSP: 13 vs 5 and 3 for affiliated centers with and without CBE, respectively). Most of the affiliated centers that offered CBE were located within a hospital, whereas those that provided mammography without CBE were more likely to be part of an independent health facility. The majority of the affiliated centers that offered mammography without CBE were located in the central east administrative region of Ontario.

Table 1

Comparison of facility characteristics among screening centers by type of center *

Characteristic	Regional cancer centers with CBE	Affiliated centers with CBE	Affiliated centers without CBE
No. of centers	9	59	34
Annual screening volume
Mean † (SD)	7939 (2941)	1560 (1318)	1710 (1952)
Range	5376–13 717	129–6567	70–10 035
No. of years in operation ‡
Mean (SD)	13 (1)	5 (2)	3 (2)
Range	11.5–13.5	0.2–9.0	0.2–5.6
Affiliated center type, No. (%)
Hospital	—	47 (79.7)	8 (23.5)
Independent health facility	—	12 (20.3)	26 (76.5)
Administrative region, No. (%)
Central East	2 (22.2)	5 (8.5)	27 (79.4)
Central West	1 (11.1)	10 (16.9)	0 (0.0)
Eastern	1 (11.1)	6 (10.2)	1 (2.9)
North East	1 (11.1)	9 (15.3)	2 (5.9)
North West	1 (11.1)	1 (1.7)	2 (5.9)
South	1 (11.1)	3 (5.1)	1 (2.9)
South East	1 (11.1)	8 (13.6)	1 (2.9)
South West	1 (11.1)	17 (28.8)	0 (0.0)

*

CBE = clinical breast examination; — = not applicable.

†

The annual volume of screens was averaged over the period between January 1, 2002, and December 31, 2003.

‡

Total years of operation within the Ontario Breast Screening Program from July 1, 1990, to December 31, 2003.

Of the 290 230 women who were screened between January 1, 2002, and December 31, 2003, 232 515 had CBE in addition to mammography and 57 715 women had mammography alone ( Table 2 ). Compared with women who were screened at 34 affiliated centers that did not offer CBE, those screened at regional cancer centers or affiliated centers that offered CBE were statistically significantly more likely to have a first-degree relative with breast cancer ( P < .001), to be a current user of hormone therapy ( P < .001), and to have lower breast density ( P < .001). Of the women who were screened at regional cancer centers, approximately 46% were 60–69 years of age and approximately 82% had been screened previously. By contrast, of the women who attended affiliated centers that did not offer CBE, 38% were 60–69 years of age and approximately 45% had been screened previously compared with women screened at regional cancer centers. The age and previous screening history of women who attended affiliated centers that offered CBE were intermediate to those of women screened at regional cancer centers or affiliated centers without CBE.

Table 2

Comparison of risk factors and screening characteristics among women by type of center attended *

Characteristic	Regional cancer centers with CBE, No. (%)	Affiliated centers with CBE, No. (%)	Affiliated centers without CBE, No. (%)
No. of women	104 072	128 443	57 715
First-degree relative with breast cancer, † No. (%)
No	89 961 (86.4)	110 382 (85.9)	51 380 (89.0)
Yes	14 111 (13.6)	18 061 (14.1)	6335 (11.0)
Current use of hormone therapy, † No. (%)
No	73 241 (70.6)	89 319 (70.0)	43 517 (77.0)
Yes	30 468 (29.4)	38 373 (30.0)	12 988 (23.0)
Unknown	363	751	1210
Breast density, † No. (%)
<75%	99 520 (95.6)	119 306 (92.9)	51 287 (88.9)
≥75%	4552 (4.4)	9133 (7.1)	6428 (11.1)
Unknown	0	4	0
Age at last screen in y, † No. (%)
50–59	56 397 (54.2)	74 463 (58.0)	35 616 (61.7)
60–69	47 675 (45.8)	53 980 (42.0)	22 099 (38.3)
Type of screen, † No. (%)
Initial	19 130 (18.4)	45 396 (35.3)	31 944 (55.4)
Subsequent	84 942 (81.6)	83 047 (64.7)	25 771 (44.6)

*

CBE = clinical breast examination.

†

P < .001 (two-sided Pearson χ ² test) for regional cancer centers with CBE or affiliated centers with CBE vs affiliated centers without CBE.

The cancer detection rate and sensitivity for referrals based on CBE alone were higher at initial screens for women who were screened at regional cancer centers than for those who were screened at affiliated centers that offered CBE (cancer detection rate: 3.8 vs 2.4 per 1000; sensitivity: 47.4% vs 32.0%) but not for subsequent screens (cancer detection rate: 1.8 vs 1.5 per 1000; sensitivity: 26.7% vs 25.6%) ( Table 3 ). Compared with affiliated centers that offered CBE, regional cancer centers had lower abnormal call rates (2.2% vs 2.9%) and false-positive rates (2.0% vs. 2.7%) for CBE referrals on subsequent screens; however, the abnormal call rates (4.1% vs 4.1%) and false-positive rates (3.7% vs 3.8%) for regional cancer centers and affiliated centers with CBE were similar for women who had initial screens. For referrals based on mammography alone, compared with women who had an initial screen at affiliated centers that did not offer CBE, those who had an initial screen at regional cancer centers or affiliated centers that offered CBE had higher abnormal call rates (10.4% and 10.1%, respectively, vs 7.9%), cancer detection rates (7.3 and 6.8, respectively, vs 5.4 per 1000), and false-positive rates (9.7% and 9.5%, respectively, vs 7.4%) ( Table 3 ). This pattern of higher abnormal call rates, cancer detection rates, and false-positive rates was also observed for women who had a subsequent screen at affiliated centers that offered CBE. However, women who had a subsequent screen at a regional cancer center had lower abnormal call rates (5.2% vs 5.8%) and false-positive rates (4.6% vs 5.4%) and higher cancer detection rates (6.0 vs 4.1 per 1000) for mammography referrals than women who had a subsequent screen at an affiliated center that did not offer CBE. Mammography sensitivity was similar between centers that did and did not offer CBE.

Table 3

Performance measures for initial and subsequent screens and 95% confidence intervals by type of center and modality of referral within 12 months of follow-up *

	Modality of referral
Performance measure	CBE		Mammography		CBE and/or mammography
Initial screen	Subsequent screen	Initial screen	Subsequent screen	Initial screen	Subsequent screen
Abnormal call rate per 100 women (95% CI)
Regional cancer centers with CBE	4.1 (3.8 to 4.4)	2.2 (2.1 to 2.3)	10.4 (10.0 to 10.8)	5.2 (5.0 to 5.3)	13.1 (12.7 to 13.6)	6.9 (6.7 to 7.1)
Affiliated centers with CBE	4.1 (3.9 to 4.2)	2.9 (2.7 to 3.0)	10.1 (9.9 to 10.4)	6.6 (6.4 to 6.7)	13.1 (12.8 to 13.4)	8.8 (8.6 to 9.0)
Affiliated centers without CBE	—	—	7.9 (7.6 to 8.2)	5.8 (5.5 to 6.1)	7.9 (7.6 to 8.2)	5.8 (5.5 to 6.1)
Cancer detection rate per 1000 women (95% CI)
Regional cancer centers with CBE	3.8 (3.0 to 4.8)	1.8 (1.5 to 2.1)	7.3 (6.2 to 8.6)	6.0 (5.5 to 6.6)	7.6 (6.4 to 8.9)	6.3 (5.8 to 6.9)
Affiliated centers with CBE	2.4 (2.0 to 2.9)	1.5 (1.3 to 1.8)	6.8 (6.1 to 7.6)	5.0 (4.6 to 5.5)	7.2 (6.5 to 8.0)	5.4 (4.9 to 5.9)
Affiliated centers without CBE	—	—	5.4 (4.7 to 6.3)	4.1 (3.4 to 4.9)	5.4 (4.7 to 6.3)	4.1 (3.4 to 4.9)
Sensitivity, % (95% CI)
Regional cancer centers with CBE	47.4 (39.7 to 55.3)	26.7 (23.3 to 30.5)	90.4 (84.7 to 94.1)	90.1 (87.4 to 92.3)	94.9 (90.1 to 97.4)	94.9 (92.8 to 96.4)
Affiliated centers with CBE	32.0 (27.4 to 37.1)	25.6 (22.0 to 29.6)	89.0 (85.2 to 91.8)	85.9 (82.6 to 88.7)	94.6 (91.7 to 96.5)	91.7 (88.9 to 93.8)
Affiliated centers without CBE	—	—	88.6 (83.4 to 92.3)	85.3 (78.1 to 90.4)	88.6 (83.4 to 92.3)	85.3 (78.1 to 90.4)
False-positive rate, % (95% CI)
Regional cancer centers with CBE	3.7 (3.5 to 4.0)	2.0 (1.9 to 2.1)	9.7 (9.3 to 10.2)	4.6 (4.4 to 4.7)	12.5 (12.0 to 12.9)	6.3 (6.1 to 6.4)
Affiliated centers with CBE	3.8 (3.7 to 4.0)	2.7 (2.6 to 2.8)	9.5 (9.3 to 9.8)	6.1 (5.9 to 6.2)	12.4 (12.1 to 12.7)	8.3 (8.1 to 8.5)
Affiliated centers without CBE	—	—	7.4 (7.1 to 7.7)	5.4 (5.1 to 5.7)	7.4 (7.1 to 7.7)	5.4 (5.1 to 5.7)

*

CBE = clinical breast examination; CI = confidence interval; — = not applicable.

For referrals based on CBE and/or mammography, the abnormal call rates and false-positive rates for both initial and subsequent screens were higher for women who were screened at regional cancer centers or affiliated centers that offered CBE in addition to mammography than for women who were screened at affiliated centers that offered only mammography (abnormal call rate for initial screen: 13.1% and 13.1%, respectively, vs 7.9%; abnormal call rate for subsequent screen: 6.9% and 8.8%, respectively, vs 5.8%; false-positive rate for initial screen: 12.5% and 12.4%, respectively, vs 7.4%; false-positive rate for subsequent screen: 6.3% and 8.3%, respectively, vs 5.4%) ( Table 3 ). In addition, cancer detection rates and sensitivity were higher for women who were screened at centers that offered CBE than for women who were screened at affiliated centers that did not offer CBE (cancer detection rate for initial screen: 7.6 and 7.2 per 1000, respectively, vs 5.4 per 1000; cancer detection rate for subsequent screen: 6.3 and 5.4 per 1000, respectively, vs 4.1 per 1000; sensitivity for initial screen: 94.9% and 94.6%, respectively, vs 88.6%; sensitivity for subsequent screen: 94.9% and 91.7%, respectively, vs 85.3%).

The pattern of an increased sensitivity and false-positive rate in centers that offered CBE compared with centers that did not was also observed after adjusting for characteristics of the women, the facilities, and the providers (data not shown). Among all centers that offered CBE, the cancer detection rate for mammography referrals was calculated to be 5.9 per 1000 and for CBE and/or mammography referrals 6.3 per 1000, and the false-positive rate for mammography referrals was 6.5% and for CBE and/or mammography referrals 8.7% (data not shown). Therefore, with CBE, an additional 0.4 cancers are detected per 1000 women screened with an increase of 2.2 percentage points in the false-positive rate. For 10 000 women screened, there would be an additional four cancers detected and of the 9937 women without cancer (based on 63 cancers detected per 10 000 women), there would be an additional 219 false-positive screens. Therefore, for each additional cancer detected by CBE per 10 000 women, there would be 55 additional false-positive screens.

Discussion

Overall, this study revealed that cancer detection rates and sensitivity were higher, as were the abnormal call rates and false-positive rates, among centers that offered CBE in addition to mammography than among centers that offered only mammography. These findings are consistent with results from previous breast screening randomized controlled trials ( 19–23 ) and demonstration projects ( 24 , 25 ). Several studies have examined the effectiveness of CBE in community-based populations by comparing referrals based on CBE and mammography with those based on CBE alone or mammography alone among women who received both CBE and mammography ( 26–29 ). However, this study differed from the previous studies in that we examined the accuracy of CBE in addition to mammography by comparing women who were screened by both modalities with women who were screened by mammography alone; in addition, in this study, CBE and mammography were performed according to standardized protocols. In general, we found that standardized CBE provided by highly trained nurses within a population-based screening program resulted in a slightly higher CBE accuracy than what was previously reported ( 26–29 ). This study design also allowed us to evaluate the influence of CBE results on the accuracy of mammography, and we found no improvement in sensitivity for mammography referrals among centers that offered CBE compared with centers without CBE.

CBE sensitivity is typically lower, and specificity is typically higher, in clinical settings than in trials ( 26–30 ). A pooled analysis of breast screening trials and observational studies found that CBE had an overall sensitivity of 54% and specificity of 94% ( 11 ). This pattern was also found in this study, where CBE sensitivity was lower than that in trials (ie, it ranged from 32% to 47% for initial screens and from 26% to 27% for subsequent screens) and CBE specificity was higher (ie, false-positive rates ranged from 3.7% to 3.8% for initial screens and from 2.0% to 2.7% for subsequent screens, resulting in specificities of 96% for initial screens and of 97% to 98% for subsequent screens).

Both the regional cancer centers and the affiliated centers in this study had higher cancer detection rates and sensitivities for CBE referrals than those previously found in other community-based programs and practices ( 26–29 ). The higher detection rates could reflect differences in certain patient characteristics; for example, the OBSP targets screening to asymptomatic women who are 50–69 years of age. Most of the studies that reported lower CBE sensitivities included women outside this age range, symptomatic women, and/or women who were diagnosed with relatively aggressive breast cancers. In addition, the women in these previous studies were screened in a variety of practice settings that did not use standardized protocols for conducting CBE unlike women who attended OBSP centers, who received standardized CBEs that lasted 8–10 minutes from nurses who had extensive training in CBE. Our findings also suggested a greater detection rate for CBE referrals than an earlier Canadian study that included data from OBSP in addition to three other provincial screening programs ( 12 ). One explanation for this finding may be that the nurses who performed CBE in this study as part of the OBSP had more years of experience, given that the OBSP started approximately 5 years earlier than the other Canadian screening programs that provide CBE by a nurse.

Even though referrals occur within the OBSP if either the nurse or the radiologist recommends follow-up, it is important to examine the effect of information from the CBE on the referral patterns of the radiologist. An earlier study among OBSP women in which radiologists had access to the nurses’ CBE recommendation found that the radiologists’ referral rates changed in the same direction as the referral of the nurses, but the change was statistically significant only when the nurse did not refer the woman for further assessment ( 31 ). In this study, we found that radiologists’ abnormal call rates for initial screens were also likely to have been influenced by nurses’ referrals based on CBE because they were higher for all centers that employed nurses and there were higher radiologist referral rates for subsequent screens in affiliated centers that employed nurses than for those that did not employ nurses. However, the nurses’ CBE recommendation did not increase the referral rate for radiologists employed at regional cancer centers because regional cancer centers had lower referral rates on subsequent screens compared with affiliated centers without CBE. This finding suggests that the influence of nurse referral on the radiologist may be dependent on the experience of the radiologist and the volume of screens performed at the center because the nurses’ recommendations were less influential for radiologists at regional cancer centers. A pan-Canadian study found that the positive predictive value of mammography, which is influenced by the abnormal interpretation rates, increased with increasing volume of screening examinations interpreted by the radiologist ( 32 ). Irrespective of the influence of nurse referrals on radiologist referrals, the previous OBSP study concluded that CBE information had no effect on the accuracy of cancer detection by mammography ( 31 ). We also found that the sensitivity for mammography referrals was similar between regional cancer centers or affiliated centers that offer CBE and affiliated centers that do not offer CBE.

This study has several strengths. In particular, we used routine follow-up data collected on a cohort of women, all of whom participated in the same screening program. Ascertainment of false-negative cancers was comparable among OBSP centers because all cancers that are diagnosed in the interval following a negative CBE or mammogram are tracked consistently by active follow-up of the program or through linkage with the Ontario Cancer Registry.

This study has four limitations. First, there is the possibility of workup bias because a positive CBE is less likely than a positive mammogram to receive intensive follow-up. In the OBSP, women who are referred on the basis of CBE alone were statistically significantly more likely than women referred on the basis of mammography to have an unknown final diagnosis ( 33 ). Although workup bias of a positive CBE may have led to slightly underestimated accuracy measures, it is unlikely that the patterns of distributions observed would have been appreciably altered. Second, we did not evaluate other benefits of employing nurses within screening programs. In addition to providing CBE, nurses play several roles in the OBSP, including recruitment, education, and follow-up. A future study will examine these additional roles of nurses and how they are associated with compliance to screening recommendations.

Third, comparison of performance measures across center types in this study is subject to confounding; comparisons among centers within any one type (ie, regional cancer centers, affiliated centers with CBE, or affiliated centers without CBE) may be more informative because the covariates for each referral modality would be similar. However, our main finding—that the sensitivity and false-positive rates were greater in centers that offered CBE than in centers that did not—concurs with our findings that both the sensitivity and false-positive rates for referrals based on CBE and/or mammography were greater than the sensitivity and false-positive rates for mammography referrals within regional cancer centers or affiliated centers that offered CBE.

Finally, the findings of this study may have limited generalizability to other populations. The CBEs in this study were performed by highly trained nurses within a population-based screening program and thus the results may not be generalizable to CBE use in clinical practice. In addition, because the study included women aged 50–69 years, the results of this study should not be interpreted as supporting CBE use for women outside of this age range.

Overall, we found higher breast cancer detection rates and sensitivities for CBE referral than those previously found in other community-based studies, which suggests that the accuracy of CBE can be improved in screening programs that offer high-quality CBEs by specially trained nurses. Although the addition of CBE to mammography improved the sensitivity of referral for women who received both modalities, the sensitivity of mammography referral was not improved and women screened by both modalities were at a higher risk of a false-positive test than women screened by mammography alone. Therefore, the benefit in increased sensitivity from the addition of CBE to mammography must be weighed against potential risks and costs of further follow-up due to false-positive results as well as the anxiety associated with additional diagnostic evaluations. Among centers that offered both CBE and mammography, for each additional cancer detected by CBE per 10 000 women screened, there were an additional 55 false-positive screens. Women should be informed of the risks and benefits of having a CBE in addition to mammography for breast screening.

Funding

Research project grant (Fall 2003 competition) from the Canadian Breast Cancer Foundation–Ontario Region (awarded to A.M.C.).

The study sponsor had no role in the design of the study; the collection, analysis, or interpretation of the data; the writing of the manuscript; or the decision to submit the manuscript for publication.

The authors thank the Ontario Breast Screening Program, a program of Cancer Care Ontario, for use of its data for this study.