Deimplementation of the Choosing Wisely Recommendations for Low-Value Breast Cancer Surgery

Abstract

IMPORTANCE

Overtreatment of early-stage breast cancer results in increased morbidity and cost without improving survival. Major surgical organizations participating in the Choosing Wisely campaign identified 4 breast cancer operations as low value: (1) axillary lymph node dissection for limited nodal disease in patients receiving lumpectomy and radiation, (2) re-excision for close but negative lumpectomy margins for invasive cancer, (3) contralateral prophylactic mastectomy in patients at average risk with unilateral cancer, and (4) sentinel lymph node biopsy in women 70 years or older with hormone receptor–positive cancer.

OBJECTIVE

To evaluate the extent to which these procedures have been deimplemented, determine the implications of decreased use, and recognize possible barriers and facilitators to deimplementation.

EVIDENCE REVIEW

A systematic review of published literature on use trends in breast surgery was performed in accordance with PRISMA guidelines. The Ovid, Embase, Cumulative Index to Nursing and Allied Health Literature, and Cochrane databases were searched for original research with relevance to the Choosing Wisely recommendations of interest. Eligible studies were examined for data about use, and any patient-level, clinician-level, or system-level factors associated with use.

FINDINGS

Concordant with recommendations, national rates of axillary lymph node dissection for patients with limited nodal disease have decreased by approximately 50% (from 44% in 2011 to 30% to 34% in 2012 and 25% to 28% in 2013), and national rates of lumpectomy margin re-excision have decreased by nearly 40% (from 16% to 34% before to 14% to 18% after publication of a consensus statement). Conversely, national rates of contralateral prophylactic mastectomy continue to rise each year, accounting for up to 30% of all mastectomies for breast cancer (range in all mastectomy cases: 2010–2012, 28%−30%; 1998, <2%), and rates of sentinel lymph node biopsy in women 70 years or older with low-risk breast cancer are persistently greater than 80% (range, 80%−88%). Factors associated with high rates of contralateral prophylactic mastectomy use are younger age, white race, increased socioeconomic status, and the availability of breast reconstruction; limited data exist on factors associated with high rates of sentinel lymph node biopsy in women 70 years or older. Successful deimplementation of axillary lymph node dissection and lumpectomy margin re-excision were associated with decreased costs and improved patient-centered outcomes.

CONCLUSIONS AND RELEVANCE

This review demonstrates variable deimplementation of 4 low-value surgical procedures in patients with breast cancer. Addressing specific patient-level, clinician-level, and system-level barriers to deimplementation is necessary to encourage shared decision-making and reduce overtreatment.

Low-value care represents a specific segment of health care waste associated with the delivery of services without a clinically meaningful benefit. The estimated cost associated with unnecessary medical tests, treatments, and procedures is more than $200 billion annually in the US.¹ In addition to the financial consequences, overtreatment can lead to significant physical, psychological, and social harm to individual patients.² Asaresult, a national spot-light has emerged on eliminating overtreatment and low-value care.

Choosing Wisely is a national campaign founded in 2012 by the American Board of Internal Medicine Foundation to address overtreatment.³ Seventeen surgical societies have participated in the campaign, generating more than 110 recommendations to avoid low-value care in surgery. Nearly all these recommendations target low-value perioperative care, such as tests or medications.⁴ A no-table exception is in the field of breast surgery, in which multiple operations were identified as low value. Based on data from randomized clinical trials and meta-analyses, the American College of Surgeons, the Society for Surgical Oncology (SSO), and the American Society for Breast Surgeons together identified 4 surgical targets for the deimplementation of low-value breast cancer care: (1) axillary lymph node dissection (ALND) for limited nodal disease in patients receiving lumpectomy and radiation, (2) re-excision for close but negative margins for invasive cancer, (3) contralateral prophylactic mastectomy (CPM) in patients at average risk with unilateral cancer, and (4) sentinel lymph node biopsy (SLNB) in women 70 years or older with hormone receptor–positive cancer.⁵

Despite substantial evidence supporting these recommendations and overall enthusiasm for reducing low-value cancer care, deimplementation rates and determinants are poorly understood.The objective of this study is to evaluate national trends in deimplementation across Choosing Wisely recommendations to identify potential targets for reducing the use of low-value procedures. Additional aims of this study are to determine the implications of decreased use and recognize possible barriers and facilitators to deimplementation.

Methods

Study Overview

We conducted a systematic review using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) schema. The study protocol was registered with PROSPERO prior to initiating the review (CRD42019127077).

Data Sources and Searches

We performed literature searches in the Ovid, Embase, Cumulative Index to Nursing and Allied Health Literature (CINAHL), and Cochrane databases in March 2019. Each search combined Medical Subject (MeSH) headings with keyword variations to identify studies on breast surgery recommendations and practice patterns. Search terms and a selection flowchart are provided in the eAppendix in the Supplement. No restrictions were applied to the searches. After excluding duplicates, articles were combined into a single file for review in Endnote X8 (Clarivate Analytics).

Study Selection

We screened titles for relevance to the 4 Choosing Wisely recommendations. Studies were limited to original research in English evaluating1ormoreoftherecommendationsintheUS.Becauseour study aimed to assess deimplementation rates, studies were excluded if they contained no data regarding procedure use trends. One reviewer (T.W.) assessed abstracts to ensure alignment with inclusion and exclusion criteria. Two reviewers (T.W. and A.B.) then independently reviewed the full-text articles to verify eligibility. After discussing discrepancies and reaching consensus, the final set of eligible articles for data abstraction was identified. We performed a hand search of the references from eligible articles to identify any additional relevant studies.

Data Extraction and Analysis

We captured detailed data from all studies, including study design, patient cohort, main variables, results, and conclusions. Data were extracted independently and in duplicate, and a third methodologist (L.A.D.) resolved disagreements. We searched the data for results and conclusions on procedure use rates, determinants of deimplementation, and implications of deimplementation. For each study, the quality of the evidence was rated from 1 to 5, based on the Oxford Centre for Evidence-Based Medicine rating system.

Results

Fifty-seven articles were included in this systematic review. Four-teen articles^6–19 reported on the recommendation to avoid ALND for limited nodal disease (Table 1), 10 articles^20–29 discussed the recommendation to avoid re-excision for close but negative margins for invasive cancer (Table 2), 27 articles^30–35 evaluated the recommendation to avoid CPM in patients with unilateral cancer (Table 3), and 6articles^30–34studied the recommendation to avoid SLNB in women 70 years or older with hormone receptor–positive, invasive breast cancer (Table 4). The determinants associated with persistent procedure use for each recommendation are summarized in Table 5.

Table 1.

Findings on Avoiding Routinely Excising All the Lymph Nodes Under the Arm in Patients Undergoing a Lumpectomy for Breast Cancer

Source	Study period	Data source	Level of evidence	Cohort	Excluded patients	ALND before vs after Z00011, No./total No. (%)
Caudle et al,6 2012	2009–2011	Institutional	3	Patients with clinical stage T1–2N0M0 invasive breast cancer who received BCT	Patients who received chemotherapy or converted to mastectomy	53 of 62 Patients (85%) vs 10 of 42 patients (24%) who were SLN positive. In the first 6 mo after discussion of the Z0011 results, 7 of 25 patients (28%) who were SLN positive underwent ALND vs 3 of 17 (18%) in the subsequent 6 mo
Fillion et al,7 2016	2007–2013	Institutional	3	Included patients with clinical stage T1–2N0M0 invasive breast cancer, ≤2 positive nodes, and who received BCS and radiation	Patients who received neoadjuvant systemic therapy, underwent bilateral procedures, or required reconstructive procedures	52 of 67 Patients (78%) vs 7 of 34 patients (21%) who were SLN positive; 75% of patients who were SLN positive in 2010, 26% in 2011, and 16% in 2012 and 2013
Gabriel et al,8 2015	2007–2014	Institutional	3	All patients who received AlNd. Being Z0011 eligible was defined as patients with clinical stage T1-T2N0 invasive breast cancer and ≤2 positive nodes who received BCS and radiation	Z0011-ineligible defined as patients who received neoadjuvant chemotherapy followed by breast conservation or were found to have extranodal extension of the SLNB	Overall rate of ALND: 5.37 cases/mo before Z0011 (279 of 804 cases) vs 5.62 cases/mo after Z0011 (191 of 558 cases). The compliance rate after Z0011 was approximately 90% (53 of 59 cases); ALND rates did not change
Jorns et al,9 2016	18 mo prior to and after Z0011 study publication	Institutional	3	Patients who underwent SLNB, had a preoperative core biopsy diagnosis of ductal carcinoma in situ or invasive carcinoma, and were clinically node negative	None	17 of 24 Patients (70.8%) vs 13 of 34 patients (38.2%) who were SLN positive
Le et al,10 2016	2009–2013	Institutional	3	Patients with ≤T2 invasive breast cancer, ≤2 positive nodes, and no gross extranodal disease who received BCT	Patients who received neoadjuvant systemic therapy	34 of 45 Patients (75.6%) vs 1 of 46 patients (2.2%) who were SLN positive
Robinson et al,11 2014	2003–2013	Institutional	3	Patients with T1-T2 invasive breast cancer with a clinically negative axilla who received BCT	None	Completion in 84% vs 63% of patients who were SLN positive; numerators and denominators not reported
Subhedar et al,12 2015	2006–2013	Institutional	3	All patients who received BCT and SLNB	Patients who had a SLNB but no nodes identified on pathologic examination	379 of 2372 Patients (16.0%) vs 68 of 2841 patients (2.4%)
Weiss et al,13 2017	2015–2016	Institutional	3	Patients with clinical T1–2N0 invasive breast cancer, tumor ≤5 cm, ≤2 positive nodes, who received BCS and SLNB	Patients who received neoadjuvant chemotherapy or endocrine therapy, converted to mastectomy, or had nodal metastasis on fine-needle aspiration	Of patients who were SLN positive, 4 of 48 patients (8%) in 2015–2016 vs 24% (numerator and denominator not reported) in the first year after publication of the trial data
Wright et al,14 2015	28 mo prior to and 28 mo after Z0011 study publication	Institutional	3	Patients with invasive breast cancer, tumor ≤5 cm, clinically negative axilla, and ≤2 positive nodes who received BCS and radiation	Patients who received neoadjuvant therapy, had only isolated tumor cells identified in the SLN, or underwent preoperative axillary ultrasonographic examination with fine-needle aspiration	92 of 110 Patients (84%) vs 29 of 70 patients (41%) who were SLN positive; in 14-mo increments, the rate decreased from 90% (53 patients) to 77% (39 patients) to 51% (20 patients) to 29% (9 patients) in those meeting Z0011 criteria
Howard et al,15 2016	2004–2013	NCDB and SEER	3	Patients with tumor ≤5 cm, invasive breast cancer, and ≤2 positive nodes who received BCS and radiation	Patients who received neoadjuvant chemotherapy (NCDB)	57.4% to 64.3% of Patients who were SLN positive before Z0011 (NCDB) vs 29.6% in 2012 (NCDB) and 24.7% in 2013 (SEER); numerators and denominators not reported
Mann et al,16 2017	2012–2013	NCDB	3	Patients with clinical T1-T2N0M0 invasive breast cancer and ≤2 positive nodes who received BCS and radiation	Patients who received neoadjuvant systemic therapy or had matted nodes or gross extranodal disease	34.2% of Patients who were SLN positive in 2012 vs 27.9% in 2013; numerators and denominators not reported
Nocera et al,17 2018	2007–2014	ACS-NSQIP	3	Patients with invasive breast cancer	None	3787 of 10 710 Patients (35.4%) diagnosed with IBC in 2007 vs 4576 of 21 166 patients (21.6%) in 2014
Rosenberger et al,18 2019	2004–2014	NCDB	3	Patients with clinical stage I, II, or III invasive breast cancer who received BCS or mastectomy	Patients with an unknown number of lymph nodes retrieved	32% of Patients who received BCS or mastectomy for invasive breast cancer in 2004 vs 16% in 2014; numerators and denominators not reported
Yao et al,19 2015	1998–2011	NCDB	3	Patients with clinical stage T1–2N0M0 invasive breast cancer, negative resection margins, and ≤2 positive nodes who received BCS and radiation	Patients who received neoadjuvant chemotherapy	In patients who were SLN positive, the proportion of patients receiving SLNB only increased from 6.1% in 1998, 23.0% in 2009, and 56.0% in 2011; numerators and denominators not reported

Abbreviations: ACS-NSQIP, American College of Surgeons National Surgical Quality Improvement Program; ALND, axillary lymph node dissection; BCS, breast-conserving surgery; BCT, breast-conserving treatment; IBC, invasive breast cancer; NCDB, National Cancer Data Base; SEER, Surveillance, Epidemiology, and End Results; SLN, sentinel lymph nodes.

Table 2.

Findings on Avoiding Routinely Reoperating on Patients With Invasive Cancer if the Cancer is Close to the Edge of the Excised Lumpectomy Tissue

Source	Study period	Data source	Level of evidence	Cohort	Excluded individuals	Re-excision BCS or conversion to mastectomy before vs after guideline publication, No./total No. (%)
Bhutiani et al,20 2018	1998–2016	Institutional	3	Patients with stage I or II invasive breast cancer who received BCT	Patients who received neoadjuvant chemotherapy or were diagnosed by excisional biopsy	46 of 126 Patients (37%) vs 10 of 111 patients (9%)
Chung et al,21 2015	2011–2014	Institutiona1	3	Patients with invasive breast cancer who received BCS and radiation	None	115 of 597 Patients (19%) vs 32 of 248 patients (13%)
Heelen-Gladden et al,22 2017	2010–2016	Institutional	3	Patients with invasive breast cancer with or without ductal carcinoma in situ who received BCS	None	42 of 395 Patients (10.6%) vs 29 of 286 patients (10.1%)
Patten et al,23 2017	2011–2015	Institutional	3	Patients with stage I or II invasive breast cancer who received BCS and radiation	None	82 of 402 Patients (20.4%) vs 90 of 552 patients (16.3%)
Rosenberger et al,24 2016	2013–2014	Institutional	3	Patients with stage I or II invasive breast cancer who received BCS and radiation	Patients who received neoadjuvant chemotherapy, had bilateral invasive breast cancers, or had a personal history of breast cancer	108 of 504 Patients (21.4%) vs 106 of 701 patients (15.1%)
Van den Bruele et al,25 2018	2009–2017	Institutional	3	Patients with invasive breast carcinoma who received BCS	Patients who were diagnosed on excisional biopsy	23 of 116 Patients (20%) vs 13 of 85 patients (15%), which decreased further to 10 of 85 patients (12%) with addition of intraoperative specimen ink
Kaczmarski et al,26 2019	2012–2018	Medicare	3	Patients with breast cancer who received BCT	Patients without a diagnosis of breast cancer within 6 mo before or after initial surgery, with a history of surgery for breast cancer, who were cared for by multiple surgeons, or who were enrolled in Medicare parts A and B for fewer than 6 mo before BCT	23 363 of 105 838 Patients (22.1%) vs 31 949 of 185 227 patients (17.2%)
Kantor et al,27 2019	2004–2015	NCDB	3	Patients with stage I or II invasive breast cancer who received BCS	Patients who received neoadjuvant therapy, were diagnosed on excisional biopsy, or had positive margins at definitive operation	16.1% of Patients in 2004, 18.7% of patients in 2011, 17.2% of patients in 2013, and 14% of patients in 2015 underwent re-excision BCS, unilateral mastectomy, or bilateral mastectomy; numerators and denominators not reported
Morrow et al,28 2017	2013–2015	SEER	3	Patients with stage I or II invasive breast cancer who received BCS	Patients with a history of breast cancer or bilateral disease	34% vs 18%; Numerators and denominators not reported
Schulman et al,29 2016	2013–2015	ASBrS Mastery of Breast Surgery Database	3	Patients with invasive breast cancer who received BCS	Patients diagnosed on excisional biopsy	2482 of 13 297 Patients (20.2%) vs 1851 of 12 805 patients (16.5%)

Abbreviations: ASBrS, American Society for Breast Surgeons; BCS, breast-conserving surgery; BCT, breast-conserving treatment; NCDB, National Cancer Data Base; SEER, Surveillance, Epidemiology, and End Results.

Table 3.

Findings on Avoiding Routinely Performing a Double Mastectomy in Patients Who Have a Single Breast With Cancer

Article	Study period	Data source	Level of evidence	Cohort	Excluded individuals	Patients with CPM, No./total No.(%)
Chung et al,36 2012	1995–2008	Institutional	3	Patients with unilateral stage 0 to III breast cancer who received a mastectomy	Patients with a previous mastectomy	8 of 93 Patients (19.4%) with mastectomies in 1995–1999 vs 86 of 133 patients (64.7%) in 2005–2008
Damle et al,37 2010	2002–2009	Institutional	3	Patients with unilateral stage 0 to IV breast cancer who received surgical treatment	Patients who had had a mastectomy for a prior breast cancer	2 of 31 Patients (6%) with mastectomies in 2002 vs 25 of 68 patients (37%) in 2009
Hoskin et al,38 2016	2009–2014	Institutional	3	Patients with unilateral stage 0 to III breast cancer who received surgical treatment	Patients with a prior history of breast cancer	48% of Mastectomies in 2009 vs 58% in 2014; numerators and denominators not reported
Jerome-D’Emilia et al,39 2019	2004–2014	Institutional	3	Patients ages 40 to 80 y old with unilateral stage 0 to III breast cancer who received surgical treatment	Patients with a prior history of breast cancer, inflammatory breast cancer, unstaged tumors, or multifocal disease and those diagnosed in a nursing home, on autopsy, or death certificate	5 of 113 Patients (4.4%) with all procedures (12.2% [5 of 41] of mastectomies) in 2004 vs 54 of 190 patients (28.4%) with all procedures (69.2% [54 of 78] with mastectomies) in 2014
Jones et al,40 2009	1998–2007	Institutional	3	Patients with unilateral stage 0 to IV breast cancer who received a mastectomy	None	12 of 184 Patients (6.5%) with mastectomies in 1999 to 28 of 174 patients (16.1%) in 2007
King et al,41 2011	1997–2005	Institutional	3	Patients with unilateral stage 0 to III breast cancer who received a mastectomy	Patients with a prior history of breast cancer or metastatic disease or who had received neoadjuvant chemotherapy	From 6.7% of 225 mastectomies in 1997 to 24.2% of 491 mastectomies in 2005
Pinell-White et al,42 2015	2002–2012	Institutional	3	Patients with unilateral stage 0 to IV breast cancer who received a mastectomy	Patients who had a previous mastectomy	<25% of Mastectomies in 2002 vs 66% in 2012; numerators and denominators not reported
Sabel et al,43 2016	2000–2012	Institutional	3	Patients with unilateral stage I, II, or III breast cancer who received surgical treatment	Patients with a prior history of breast cancer	4% of All procedures in 2000 vs 19% in 2011; numerators and denominators not reported
Stucky et al,44 2010	2000–2008	Institutional	3	Patients with unilateral invasive breast cancer who received surgical treatment	None	0% of All procedures in 2000 vs 20% in 2008; numerators and denominators not reported
Suggs et al,45 2017	2006–2015	Institutional	3	Patients with unilateral stage T0–2N0–1M0 breast cancer who received surgical treatment	None	2.7% of All procedures (9.3% of mastectomies) in 2006–2010 vs 12.3% of all procedures (33.6% of mastectomies) in 2011–2015; numerators and denominators not reported
Yi et al,46 2010	2000–2006	Institutional	3	Patients with unilateral stage 0 to III breast cancer who received surgical treatment	None	Significant increase from 6.8% of all procedures (15.5% of mastectomies) in 2000 to 14.0% (29.8% of mastectomies) in 2006; numerators and denominators not reported
Clerville et al,47 2018	2005–2013	Single-state cancer registry	3	Patients with stage 0 to IV breast cancer who received surgical treatment	Patients diagnosed on autopsy or death certificate	From 4.21% of all procedures in 2005 to 9.63% in 2013; numerators and denominators not reported
Jerome-D’Emilia et al,48 2019	2004–2014	Single-state cancer registry	3	Patients aged 40 to 80 y with unilateral stage 0 to III breast cancer who received surgical treatment	Patients with a prior history of breast cancer, inflammatory breast cancer, unstaged tumors, or multifocal disease and those diagnosed in a nursing home, on autopsy, or death certificate	171 of 4600 Patients (3.7%) with all procedures (171 of 1527 [11.2%] with mastectomies) in 2004 vs 529 of 4891 patients (10.8%) with all procedures (529 of 1607 [32.9%] with mastectomies) in 2014
Kurian et al,49 2014	1998–2011	Single-state cancer registry	3	Patients with unilateral stage 0 to III breast cancer who received mastectomy or BCS with radiation	Patients diagnosed on autopsy or a death certificate, tumor size >5 cm or unknown size, a microscopic or diffuse tumor, Paget disease, inflammatory carcinoma, or unknown lymph node involvement	From 2.0% of all procedures in 1998 to 12.3% in 2011; numerators and denominators not reported
McLaughlin et al,50 2009	1995–2005	Single-state cancer registry	3	Patients with breast cancer who received surgical treatment	None	1.9% of All procedures in 1995–1996 (5.6% of mastectomies) to 4.2% (14.1% of mastectomies); numerators and denominators not reported
Yakoub et al,51 2015	1996–2009	Single-state cancer registry	3	Patients with unilateral invasive breast cancer who received surgical treatment	None	36 of 1640 Patients (2%) with all procedures in 1996 vs 163 of 2034 patients (8%) in 2009
Agarwal et al,52 2015	2000–2010	SEER	3	Patients with unilateral stage I to III breast cancer who received a mastectomy	Patients with recurrent breast cancer	1024 of 13 296 Patients (7.7%) with mastectomies in 2000 vs 4358 of 15 401 (28.3%) in 2010
Albornoz et al,53 2015	1998–2011	NCDB	3	Patients with unilateral stage 0 to II breast cancer who received surgical treatment	None	6862 of 125 789 Patients (5.4%) with all procedures (6862 of 45 345 [15.1%] with mastectomies) in 2005 vs 17 090 of 144 016 patients (11.8%) with all procedures (17 090 of 58 018 [29.5%] with mastectomies) in 2011
Cemal et al,54 2012	1998–2008	Nationwide Inpatient Sample Database	3	Patients with breast cancer or at increased risk of breast cancer who received a mastectomy; patients undergoing bilateral mastectomies with a cancer diagnosis were assumed to have CPMs if they had unilateral cancer	None	From 3.9% of mastectomies in 1998 to 20.7% in 2008; numerators and denominators not reported
Grimmer et al,55 2015	1998–2011	NCDB	3	Patients with unilateral stage 0 to II breast cancer	Patients who received neoadjuvant therapy or were diagnosed on excisional biopsy	From 1.9% of all breast cancer patients in 1998 to 10.2% in 2011; numerators and denominators not reported
Kummerow et al,56 2014	1998–2011	NCDB	3	Patients with unilateral stage T0–2N0–2M0 breast cancer who received surgical treatment	Patients who were diagnosed but not treated at the reporting hospital	From 5.4% of mastectomies in 1998 to 29.7% in 2011; numerators and denominators not reported
Nash et al,57 2017	2004–2012	North American Association of Central Cancer Registries Database	3	Patients with unilateral invasive breast cancer who received surgical treatment	Patients with distant disease	20 037 of 391 393 Patients (5.1%) with all procedures (20 037 of 158 516 [12.6%] with mastectomies) in 2004–2006 vs 49 592 of 416 030 patients (11.9%) with all procedures (49 592 of 174 917 [28.4%] with mastectomies) in 2010–2012
Pesce et al,58 2014	2003–2010	NCDB	3	Patients with stage 0 to II breast cancer who received surgical treatment	Patients who received neoadjuvant treatment or were diagnosed on excisional biopsy	From 4.1% of all procedures (12.7% of mastectomies) in 2003 to 9.7% of all procedures (28.9% of mastectomies) in 2010; numerators and denominators not reported
Tuttle et al,59 2007	1998–2003	SEER	3	Included patients with unilateral stage I to III breast cancer who received surgical treatment	Patients who received a radical or extended radical mastectomy; had unstaged tumors, multifocal disease, grade IV tumors; or were diagnosed in a nursing home, on autopsy, or death certificate	From 1.8% of all procedures (4.2% of mastectomies) in 1998 to 4.5% of all procedures (11.0% of mastectomies) in 2003; numerators and denominators not reported
Vaz-Luis et al,60 2016	2000–2009	National Comprehensive Cancer Network	2	Patients with unilateral stage I breast cancer who received surgical treatment	Patients who received neoadjuvant treatment, had unknown hormone receptor and human epidermal receptor 2 status, or were not treated with definitive surgery	From 13% of mastectomies in 2000 to 30% in 2009; numerators and denominators not reported
Wong et al,61 2017	1998–2012	SEER	3	Patients with unilateral stage I to III breast cancer who received surgical treatment	Patients who had previous primary malignant conditions, underwent radical or extended radical mastectomy, or had an unknown vital status or cause of death	From 3.9% of all procedures in 2002 to 12.7% in 2012; numerators and denominators not reported
Yao et al,62 2010	1998–2007	NCDB	3	Patients with unilateral stage 0 to III breast cancer who received surgical treatment	Patients with multiple primary breast cancers diagnosed within a year of one another	395 of 105 864 Patients (0.4%) with all procedures in 1998 to 5835 of 123 551 patients (4.7%) in 2007

Abbreviations: BCS, breast-conserving surgery; BCT, breast-conserving treatment; CPM, contralateral prophylactic mastectomy; NCDB, National Cancer Data Base; SEER, Surveillance, Epidemiology, and End Results.

Table 4.

Findings on Avoiding Routine Use of Sentinel Node Biopsy in Clinically Node Negative Women 70 Years or Older With Hormone Receptor–Positive Invasive Breast Cancer

Source	Study period	Data source	Level of evidence	Cohort	Excluded individuals	Patients receiving axillary surgery, No./total No. (%)
Ferrigni et al,30 2019	2005–2014	Institutional	3	Patients ≥80 y old with stage I, II, or III invasive breast cancer	Patients who had a previous diagnosis of ductal carcinoma in situ or invasive breast cancer	94/124 (76%)
Mamtani et al,31 2016	2001–2010	Institutional	3	Patients 80–89 y old who were referred for surgical evaluation of stage I or II invasive breast cancer	Patients who received neoadjuvant chemotherapy, had bilateral breast cancers, or had a previous diagnosis of breast cancer	81/178 (46.3%)
Chagpar et al,32 2017	2004–2013	NCDB	3	Patients 70–90 y old with clinically lymph node-negative, hormone receptor-positive invasive breast cancers	None	154 504/193 728 (79.8%)
Chagpar et al,33 2017	2004–2012	NCDB and SEER	3	Patients ≥70 y old with clinically lymph node-negative, hormone receptor-positive breast cancer	Patients with tumor size ≥20 cm	NCDB: 126 638/157 942 (80.2%); over time, the proportion of patients who did not have lymph node evaluation decreased; in 2004, 29.2% had no lymph node evaluation compared with 18.7% in 2012. SEER: 79.8%
Dominici et al,34 2017	2012–2013	NCDB	3	Patients >65 y old with stage I or II invasive breast cancer without clinically node-positive disease who received BCS or mastectomy	Patients with non-Medicaid or non-Medicare government insurance, missing/unknown tumor size or node status, preoperative radiation, or no surgery within 12 mo of diagnosis	35 679/40 851 Patients (87.3%) >70 y old and 8920/12 308 patients (72.5%) >80 y old. For patients ≥70 y old with clinically node-negative, stage I or II hormone receptor-positive disease, 88.0% underwent axillary surgery
Kantor et al,35 2016	2004–2012	NCDB	3	Patients ≥80 y old with estrogen receptor-positive breast cancer with 2 or less lifetime cancer diagnoses	Patients who had in situ disease, metastatic disease, or received adjuvant or neoadjuvant chemotherapy	65 250/95 357 (69%)

Abbreviations: BCS, breast-conserving surgery; BCT, breast-conserving treatment; NCDB, National Cancer Data Base; SEER, Surveillance, Epidemiology, and End Results.

Table 5.

Determinants Associated With Increased Procedure Use

Procedure	Factor
	Patient	Tumor	Clinician	Facility
Axillary lymph node dissection	Younger age, black race, and Hispanic ethnicity	Larger size, lobular histology, higher grade, estrogen receptor/progesterone receptor negative, larger volume of metastasis, and >1 positive node	Surgeon-level variation	Low breast cancer volume, nonacademic facility, and geographic region
Margin re-excision	Younger age, black race, rural area, and private insurance	Close margin for ductal carcinoma in situ, higher-stage disease, lobular histology, and human epidermal growth factor receptor 2 subtype	Surgeon-level variability, <100% breast surgery practice, and low volume of patients with breast cancer	Site not approved by the Society for Surgical Oncology and geographic region
Contralateral prophylactic mastectomy	Younger age, white race, socioeconomic status, marital status, family history, and preoperative magnetic resonance imaging	Smaller tumors; lower-grade tumors, lower stage tumors, negative nodes, lobular histology	None	Teaching hospitals, hospitals with increased surgical volume, and availability of breast reconstruction
Sentinel lymph node biopsy in patients ≥70 y	Younger age, fewer comorbidities, and private insurance	Mastectomy receipt	None	Increased surgical volume

Avoiding Routine Excision of All Lymph Nodes Beneath the Arm in Patients Having Lumpectomy for Breast Cancer

The recommendation to avoid routine excision of all axillary lymph nodes in patients undergoing lumpectomy is based on the American College of Surgeons Oncology Group Z0011 trial,⁶³ which found that ALND does not improve survival or recurrence rates in patients with small tumors with 1 or 2 positive lymph nodes who receive breast-conserving surgery and radiation. The trial results were first presented in 2010 and published in early 2011.

Procedure Use Rates

All but 1 study⁸ reported a significant decrease in ALND rates over time.^6–19 Nine studies^6–14 evaluated single-institution data and demonstrated great variation in ALND use. Overall, 71% to 85% of eligible patients with positive sentinel lymph nodes (SLNs) received an ALND before Z0011, compared with 2% to 63% after. The most significant reduction in use by a single institution resulted in a decrease in ALND rates for patients who had positive SLNs and were Z0011 eligible from 34 of 45 patients (75.6%) in the years 2009 to 2011 to 1 of 46 patients (2.2%) in 2011 to 2013.¹⁰ By comparison, 2 centers observed persistent ALND rates of greater than 38%.^9,11 Continued deimplementation for more than 5 years after the Z0011 trial was noted; at one institution, the ALND rate for eligible patients decreased from 8 of 48 (24%) in 2011 to 8% (numerator and denominator not reported) in 2015 to 2016.¹³ Five studies^15–19 evaluated national trends and found that ALND rates for patients eligible for Z0011 ranged from 57% to 77% prior to the Z0011 trial compared with 25% to 44% afterwards. Mirroring institutional trends, these rates decreased over several years; national rates of ALND were 44% in 2011, 30% to 34% in 2012, and 25% to 28% in 2013.^15,16,19

Determinants of Deimplementation

Several studies reported patient-level, tumor-level, and facility-level factors associated with ALND use. Patient demographics associated with increased use of ALND included younger age^10,16,18,19 and black or Hispanic race.^16,18,19 Patients eligible for Z0011 with high-risk tumor features, including lobular or mixed histology,^6,14,19 larger tumor size,^6,10,18 higher grade,^14,18,19 larger volume of nodal metastasis,^7,19 presence of extranodal extension,^7,13,14 negative hormone receptor status,^10,14,19 and more than 1 positive SLN^13,16,19 were more likely to receive ALND. Treatment at low-volume and nonacademic or research hospitals was associated with increased use of ALND,^16,18 and facilities located in New England had lower rates of ALND.^16,19 One study¹⁴ noted significant variation within a single health system in surgeon-level adoption of Z0011 results.

Implications of Deimplementation

Several studies reported a significant decrease in frozen-section analysis of SLNs, leading to decreased operative time and cost.^6,7,9,13,14 There also appears to be potential outcomes on adjuvant therapy, with 1 study⁶ showing an increase in the delivery of high-tangent breast radiation fields to target the axilla in patients with positive SLNs who omit ALND and another study⁷ showing a decrease in the use of adjuvant chemotherapy after Z0011. In terms of patient morbidity, Howard et al¹⁵ estimated that approximately 1700 women diagnosed with breast cancer in the US in 2012 did not undergo ALND as a result of the Z0011 trial. Based on typical complication rates, this reduction translated to an estimated 87 fewer wound infections, 139 fewer axillary seromas, 520 fewer cases of axillary paresthesia, and 225 fewer lymphedema cases nationwide.¹⁵ Finally, from an education and training perspective, multiple studies noted that the Z0011 trial decreased resident participation in ALNDs.^8,17,18

Avoiding Routine Reoperations on Patients With Invasive Cancer if the Cancer Is Close to the Edge of the Excised Lumpectomy Tissue

This recommendation is based on the 2014 Society of Surgical Oncology and American Society for Radiation Oncology (SSO/ASTRO) consensus statement endorsing a margin termed no tumor on ink for patients with stage I and II invasive cancer undergoing lumpectomy with whole-breast irradiation.⁶⁴ This guideline states that if the tumor extends close to the edge of the specimen but not to the inked margin, re-excision is not necessary.

Procedure Use Rates

All but 1 study²² noted a decrease in reoperation rates (including re-excision and conversion to mastectomy) after 2014.^20–29 Six institutional studies demonstrated overall decreased reoperation rates after lumpectomy from 11% to 37% prior to release of the consensus statement to 9% to 16% afterwards.^20–25 Two studies^20,21 evaluated reoperation rates in patients who had no tumor on ink and found a decrease from 22% (23 of 103 patients) to 5% (5 of 105 patients) and from 30% (35 of 115 patients) to 22% (7 of 32 patients) after the SSO/ASTRO recommendations. Four studies^26–29 evaluated national trends, demonstrating overall reoperation rates after lumpectomy significantly decreased from 16% to 34% before to 14% to 18% after publication of the consensus statement.

Determinants of Deimplementation

Four studies^21–24 found that patients with invasive breast cancer and concurrent noninvasive disease (ductal carcinoma in situ [DCIS]) were likely to experience re-excision for a close margin for the DCIS component. While the presence of DCIS at the margin was a factor significantly associated with re-excision, 1 study²⁴ did note a trend toward fewer re-excisions after the release of recommendations for cases in which the margin was negative for invasive cancer but close for DCIS. Although there were limited findings on tumor factors associated with increased reoperation, 1 study²⁷ found that patients with higher-stage disease, lobular histology, and human epidermal growth factor receptor 2-positive subtype had the highest rates of reoperation. Few patient factors were consistently found to contribute toward unnecessary re-excisions, although isolated studies noted increased rates of re-excisions in patients of black race and from rural areas,²⁶ with private insurance,²⁷ residing in the North-east region of the US,²⁶ and at younger ages.^23,26,27 Physician-level and facility-level factors associated with increased reoperation rates included a breast surgery practice that was less than 100%, practice at a non-SSO–approved teaching site, and low patient volume and breast-conserving surgery volume.^26,28,29 Two studies noted that the recommendation did directly affect clinician practice patterns, resulting in decreased national surgeon-level reoperation variability after 2014.^26,29

Implications of Deimplementation

Three articles^20,23,24 commented on improved patient-centered outcomes in the post–SSO/ASTRO guideline era, with decreased or rare conversion to mastectomy after lumpectomy and improved breast satisfaction scores. Morrow et al²⁸ found an increase in the absolute rate of lumpectomy and decline in the rate of unilateral and bilateral mastectomies (from 34% to 18%), suggesting that decreasing unnecessary surgery after initial lumpectomy may reduce the number of women who opt for bilateral mastectomies to treat small, unilateral breast cancers. From a financial perspective, analyses estimate a 17% to 25% cost savings by reducing reoperations; a national 3% reduction in re-excision rates would save more than $116 million dollars annually.^20,25

Avoiding Routinely Performing a Double Mastectomy in Patients Who Have a Single Breast With Cancer

This recommendation has been supported in various forms since the SSO position statement on CPM in 1993.⁶⁵ The recent 2016 recommendation by American Society for Breast Surgeons is the most strongly worded version to date.⁶⁶

Procedure Use Rates

All 27 studies^36–62 demonstrated rates of CPM rising each year. Eleven studies^36–46 evaluated trends at a single institution and noted wide variability in CPM use, and in all studies, the highest CPM rate was observed during the last year of the study period. From 2011 to 2014, the CPM rate ranged from 34% to 69% of all mastectomy cases and from 12% to 28% of all breast cancer procedures.^{38,39,42,43,45} Five articles^47–51 evaluated data from single-state cancer registries and demonstrated similar trends in CPM rates, with increases from approximately 2% of all breast cancer procedures in 1995 through 1998 to 10% to 12% in 2011 to 2014. Eleven articles^52–62 evaluated national trends in CPM. Nationally, CPM rates from 2010 to 2012 ranged from 10% to 13% of all breast cancer procedures and from 28% to 30% of all mastectomy cases, an increase from less than 2% of all cases in 1998.^52–62

Determinants of Deimplementation

Overall, studies were congruent in identifying factors associated with higher CPM rates across institutional, state, and national levels. Nearly every article^{37–46,48,49,51–55,57–59,61} identified younger age as a factor associated with CPM. Additional patient factors associated with CPM included white race,^{37,39,41–43,46,48–55,57,59,61,62} increased socioeconomic status or private insurance,^{39,48–52,54,55,57,62} marital status,^38,45,46 and a family history of breast cancer.^{36,37,41,42,44,46}

Patients with smaller, lower-grade tumors and decreased stage or node-negative disease were more likely to undergo CPM compared with those with unilateral mastectomy.38,40–42,48,52,57,59 Nine studies38,^{39,46,49,53,57,59,61,62} found increased CPM rates in patients with lobular histology, although 3 studies^42–44 did not find any association between histology and CPM receipt. Chemotherapy receipt was associated with increased CPM rates.^38,48,51 The outcome of tumor hormone status^{39–41,44,46,49,55,59} and multifocal or multicentric disease on CPM rates were inconsistent across studies, with some^44,46 finding that these factors increased CPM rates while others^38,41 found that they decreased or had no association with CPM rates.

For facility-level factors, women treated at teaching hospitals and hospitals with increased surgical volume and patients undergoing preoperative magnetic resonance imaging were more likely to undergo CPM.^{36,41,42,44–46,51,54,55,60,62} Breast reconstruction was a also a factor significantly associated with CPM, and women who chose immediate breast reconstruction were more than twice as likely to undergo CPM than those who did not, regardless of any other patient-level or tumor-level factors.^{36–39,41–44,46,48,52,53}

Avoiding Routine Use of SLNB in Women 70 Years or Older Who Have Clinically Node-Negative Hormone Receptor–Positive Invasive Breast Cancer

This recommendation was released in 2016 based on the Cancer and Leukemia Group B C9343 study.⁶⁷ Initial results in 2003 and updated results in 2013 demonstrated no difference in overall survival in women who received axillary evaluation compared with those who did not.⁶⁷

Procedure Use Rates

Six studies were included in this systematic review; 2 were single-institution studies^30,31 and 4 evaluated national data.^32–35 All studies noted high rates of axillary surgery in elderly individuals, ranging from 80% to 88% in patients 70 years or older and 46% to 76% in women 80 years or older.^30–35 The only study that provides data on trends over time noted that the rate of lymph node evaluation in patients aged 70 years or older with clinically node negative, hormone receptor–positive breast cancer increased from 71% in 2004 to 81% in 2012.³³

Determinants of Deimplementation

There are limited data on factors influencing SLNB use in elderly patients. One study³⁴ found higher odds of receiving axillary surgery in patients who were younger (odds ratios [ORs]: ages 71–75 years, 6.04 [95% CI, 0.57–0.71]; 76–80 years, 0.33 [95% CI, 0.30–0.37]; >80 years, 0.08 [95% CI, 0.07–0.08]; 65–70 years, reference), had lower levels of comorbidity (ORs: patients who had a comorbidity score of 1, 0.84 [95% CI, 0.78–0.91]; patients with a score of 2, 0.61 [95% CI, 0.53–0.79]; patients with a score of 0, reference), had private insurance (ORs: patients who had Medicaid, 0.65 [95% CI, 0.72–0.93]; Medicare, 0.81 [95% CI, 0.74–0.90]; private insurance, reference), underwent mastectomy (ORs: patients who received mastectomy with no reconstruction, 3.33 [95% CI, 3.04–3.65]; mastectomy with reconstruction, 2.68 [95% CI, 2.07–3.46]; patients who underwent BCS, reference), and were treated at high-volume centers (OR: patients treated at low-volume centers, 0.82 [95% CI, 0.72–0.93]; patients treated at high-volume centers, reference). Several studies investigated the correlation between SLNB receipt and receipt of adjuvant therapy. Dominici et al³⁴ noted that patients older than 65 years who received axillary surgery had approximately 2-folds odds of receiving chemotherapy (ORs: patients who received ALND, 2.07 [95% CI, 1.10–3.90]; SLNB, 2.29 [95% CI, 1.25–4.20]; no axillary surgery, reference) and/or radiation (ORs: patients who received ALND, 1.65 [95% CI, 1.33–2.06]; SLNB, 1.95 [95% CI, 1.61–2.37]; no axillary surgery, reference). Similarly, Chagpar et al³² found that in patients aged 70 to 90 years old, lymph node evaluation was an independent factor that correlated with receipt of chemotherapy (OR, 2.23 [95% CI, 2.06–2.50]) and radiation therapy (OR, 3.98 [95% CI, 3.83–4.14]), although it is unclear in these cases whether the decision to use SLNB is driving the subsequent use of adjuvant therapy or the desire to potentially use chemotherapy is driving the use of SLNB.

Discussion

This systematic review demonstrates that 4 Choosing Wisely breast cancer procedures have been variably deimplemented. Given that all 4 recommendations are based on high-quality evidence demonstrating a reduction in morbidity without adversely affecting survival, these data raise critical questions regarding why surgeons have stopped performing some procedures but not others and the determinants and implications of successful deimplementation.

Possible Reasons Surgeons Have Stopped Doing Some Procedures

This review confirms that ALND rates for patients eligible for Z0011 have substantially decreased. The Z0011 trial data appear to have been well disseminated among breast surgeons. In a 2011 survey of American Society for Breast Surgeons members, 97% of surgeons indicated familiarity with the trial and 57% stated they would infrequently or never perform ALND on a patient eligible for Z0011.⁶⁸ Willingness to avoid ALND may be influenced by known risks of lymphedema, range of motion restriction, numbness, and chronic pain. Patients also likely play a role, because women are reluctant to undergo ALND unless they can be assured of substantial improvements in local cancer control and long-term survival.⁶⁹ However, while ALND rates for eligible patients have been reduced, this review demonstrates that a subset of patients with higher-risk tumor features on the spectrum on Z0011 eligibility continue to undergo this procedure. The lack of complete deimplementation for eligible patients may reflect clinician concerns about the underrepresentation of high-risk patient groups in the Z0011 trial.

Similarly, this review indicates that there has been significant deimplementation of reoperation after lumpectomy, with current national re-excision rates less than 20%. This confirms findings from a meta-analysis that showed a decrease in reoperation rates from 22% to 14% after 2014 and has been validated by a time-trend model.⁷⁰ The recommendation for establishing a negative margin under the standard of no tumor on ink also appears to be well disseminated among clinicians, with 1 survey⁷¹ demonstrating that 92% of surgeons were familiar with the recommendation and that more than 95% of clinicians would appropriately avoid re-excision according to these guidelines. This study did note discordance about whether patients with triple-negative tumors or extensive DCIS extending close to the margins should undergo re-excision, consistent with the findings of our review that DCIS at the margin was a factor significantly associated with re-excision. This suggests that surgeons remain uncertain about the applicability of the recommendation to these patients.

Possible Reasons Surgeons Have Not Stopped Doing Some Procedures

This review demonstrates that CPM rates in patients with unilateral breast cancer continue to rise despite substantial work dedicated to understanding the lack of deimplementation with a specific focus on patient decision-making. The consensus of this work is that increasing use of CPM is driven by patients, who are influenced by fear of recurrent breast cancer, peace of mind, desire for breast symmetry, and perceived improved survival.⁷² The strong influence of patient factors is reflected in this review, in that determinants such a sage, race, socio economic status, receipt of breast reconstruction, and family history largely explain variation in CPM use. Although patients with an extensive family history may potentially benefit from CPM, CPM has not been shown to improve outcomes in cases where a patient has a single relative who developed breast cancer.⁶⁵ Despite the emphasis on patient preference, evidence does suggest a role for the clinician in influencing decision-making on CPM. A survey of more than 1500 women demonstrated that patients reporting a surge on recommendation against CPM had lower rates of CPM compared with patients who received no recommendation either for or against CPM.⁷³ Overall, this supports the possibility that improved patient-clinician relationships and risk communication could reduce CPM.

Although the results of the Cancer and Leukemia Group B C9343 trial⁶⁷were instrumental in informing there commendations to avoid SLNB and radiation therapy in older women with hormone receptor– positive cancer who undergo lumpectomy and adjuvant hormone therapy, neither has been substantially deimplemented.^67,74 Few data exist on why clinicians continue to perform SLNB in older patients at high rates, but some evidence suggests that clinicians may disagree with the recommendation or use may be driven by uncertainty about whether the biopsy results would influence adjuvant chemotherapy decisions that could theoretically impact survival. A positive SLNB finding significantly correlates with receipt of adjuvant therapy, with lower overall survival for patients who do not receive a SLNB.^32,33 However, there is no difference in breast cancer–specific survival, suggesting that clinicians are offering SLNB and adjuvant treatments to heal their patients.³³ While these practices may reflect a desire on the part of the clinical team to offer aggressive curative intent therapies, there is little evidence that this reflects the preferences of these patients. For example, investigators at Johns Hopkins used a geriatric assessment tools to stratify patients into low-risk and high-risk groups to determine whether to recommend SLNB. More than 90% of patients with low functional status opted to forgo SLNB. Even among patients with good health status, more than 60% opted against SLNB.⁷⁵ Similar trends were identified with radiation therapy, suggesting that elderly patients in shared decision-making models may actually prefer a less aggressive treatment strategy.

For all recommendations, the barriers and facilitators to deimplementation can be broadly categorized to patient-level, tumor-level, and facility-level factors. Common themes emerged, with increased use of procedures among patients of younger age, increased socioeconomic status, or private insurance and patients with perceived higher-risk tumor characteristics. While important, the outcome of hospital volume and surgeon-level variability was not consistent among guidelines; for example, larger-volume hospitals have lower rates of ALND and lumpectomy margin re-excision but higher rates of CPM and SLNB in older patients. These findings emphasize the need to address both patient and clinician motivations for the use of low-value procedures.

Limitations

There are several limitations to this systematic review. There were variations in the quality of the included studies, and all but 1 study⁶⁰ were retrospective cohort studies; however, methodologic rigor was acceptable to support the conclusions noted in this review. Although the recommendation to avoid routine SLNB in women 70 years or older was based on older trial data, the guideline was released in 2016. As a result, there was limited information on trends in SLNB use or determinants of deimplementation. Additionally, the studies in this review did not systematically evaluate the determinants of deimplementation and broadly reported factors associated with procedure use. Thus, there could be other barriers or facilitators to deimplementation that are not captured here, such as the influence of multidisciplinary teams or potential lack of dissemination to clinicians of the recommendation to avoid SLNB.

Conclusions

In summary, this review demonstrates that 4 recommendations supported by the Choosing Wisely campaign have been variably effective in the deimplementation of low-value surgical care in patients with breast cancer. These findings provide some insight into why 2 recommendations have had greater success in deimplementation compared with the others. Further work should address specific clinician and patient concerns through shared decision-making to decrease the use of low-value care, medical costs, and patient morbidity.

Key Points.

Question

Has low-value breast cancer surgery been successfully deimplemented?

Findings

Rates of axillary lymph node dissection in patients with limited nodal disease and re-excision of close but negative margins after lumpectomy have significantly decreased over time, reflecting changes in practice concordant with current evidence and recommendations; conversely, rates of contralateral prophylactic mastectomy in patients at average risk with unilateral cancer and sentinel lymph node biopsy in patients 70 years or older with hormone receptor–positive cancer remain elevated, despite recommendations against their routine use. This review identifies some of the patient-level, tumor-level, and facility-level factors that contribute to this differential deimplementation.

Meaning

Efforts toward the reduction of low-value breast cancer surgery have been variably influential; areas of incomplete deimplementation can provide insight into potential barriers at guideline, patient, clinician, and system levels contributing to the persistent use of some procedures.