Second Opinions From Medical Oncologists for Early-Stage Breast Cancer: Prevalence, Correlates, and Consequences | Breast Cancer | JAMA Oncology | JAMA Network
[Skip to Navigation]
Sign In
Figure 1.  Patient Flow Throughout the Study
Patient Flow Throughout the Study

Reasons for ineligibility include prior cancer diagnosis, Paget disease of nipple, disease stage of III or higher, residing outside Surveillance, Epidemiology, and End Results (SEER) registry region, unable to complete survey, and deceased. Reasons for exclusion include noninvasive (stage 0) disease and never having seen a medical oncologist.

Figure 2.  Multivariable Model of Characteristics Associated With Receipt of a Second Opinion
Multivariable Model of Characteristics Associated With Receipt of a Second Opinion

Squares indicate odds ratios; error bars, 95% CIs; and RS, recurrence score.

Original Investigation
March 2017

Second Opinions From Medical Oncologists for Early-Stage Breast Cancer: Prevalence, Correlates, and Consequences

Author Affiliations
  • 1Department of Medicine, Stanford University, Stanford, California
  • 2Departments of Health Research and Policy, Stanford University, Stanford, California
  • 3Department of Systems, Populations, and Leadership and Institute for Healthcare Policy and Innovation, University of Michigan School of Nursing, Ann Arbor
  • 4Department of Biostatistics, School of Public Health, University of Michigan, Ann Arbor
  • 5Department of Radiation Oncology, Center for Bioethics and Social Science in Medicine, University of Michigan, Ann Arbor
  • 6Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles
  • 7Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, Georgia
  • 8Department of Health Management and Policy, University of Michigan, Ann Arbor
  • 9Department of Internal Medicine, School of Public Health, Division of General Medicine, Univeristy of Michigan, Ann Arbor
JAMA Oncol. 2017;3(3):391-397. doi:10.1001/jamaoncol.2016.5652
Key Points

Question  As treatment decision making becomes more complex, is there an unmet need for second opinions from medical oncologists for the treatment of breast cancer?

Findings  In this survey of a contemporary diverse population sample of 1901 patients newly diagnosed with breast cancer, 168 patients (9.8%) received a second opinion, and 54 (3.2%) received chemotherapy from a second medical oncologist. Second opinions were not associated with overall patient satisfaction or receipt of chemotherapy.

Meaning  Use of second opinions from medical oncologists after diagnosis of breast cancer was low, but there was little evidence of unmet need.


Importance  Advances in the evaluation and treatment of breast cancer have made the clinical decision-making context much more complex. A second opinion from a medical oncologist may facilitate decision making for women with breast cancer, yet little is known about second opinion use.

Objective  To investigate the patterns and correlates of second opinion use and the effect on chemotherapy decisions.

Design, Setting, and Participants  A total of 1901 women newly diagnosed with stages 0 to II breast cancer between July 2013 and September 2014 (response rate, 71.0%) were accrued through 2 population-based Surveillance, Epidemiology, and End Results registries (Georgia and Los Angeles County, California) and surveyed about their experiences with medical oncologists, decision making, and chemotherapy use.

Main Outcomes and Measures  Factors associated with second opinion use were evaluated using logistic regression. Also assessed was the association between second opinion and chemotherapy use, adjusting for chemotherapy indication and propensity for receiving a second opinion. Multiple imputation and weighting were used to account for missing data.

Results  A total of 1901 patients with stage I to II breast cancer (mean [SD] age, 61.6 [11.0] years; 1071 [56.3%] non-Hispanic white) saw any medical oncologist. Analysis of multiply imputed, weighted data (mean n = 1866) showed that 168 (9.8%) (SE, 0.74%) received a second opinion and 54 (3.2%) (SE, 0.47%) received chemotherapy from the second oncologist. Satisfaction with chemotherapy decisions was high and did not differ between those who did (mean [SD], 4.3 [0.08] on a 1- to 5-point scale) or did not (4.4 [0.03]) obtain a second opinion (P = .29). Predictors of second opinion use included college education vs less education (odds ratio [OR], 1.85; 95% CI, 1.24-2.75), frequent use of internet-based support groups (OR, 2.15; 95% CI, 1.12-4.11), an intermediate result on the 21-gene recurrence score assay (OR, 1.85; 95% CI, 1.11-3.09), and a variant of uncertain significance on hereditary cancer genetic testing (OR, 3.24; 95% CI, 1.09-9.59). After controlling for patient and tumor characteristics, second opinion use was not associated with chemotherapy receipt (OR, 1.04; 95% CI, 0.71-1.52).

Conclusions and Relevance  Second opinion use was low (<10%) among patients with early-stage breast cancer, and high decision satisfaction regardless of second opinion use suggests little unmet demand. Along with educational level and use of internet support groups, uncertain results on genomic testing predicted second opinion use. Patient demand for second opinions may increase as more complex genomic tests are disseminated.


Advances in the evaluation and treatment of breast cancer have made the clinical decision-making context much more complex.1,2 Options for all modalities of treatment, including surgery, drug therapy, radiation, and reconstruction, have markedly expanded, as have preventive options for women at high genetic risk for second cancers. This is particularly true for decisions about systemic therapies because patients now must consider choices about 3 different medication categories: endocrine, chemotherapy, and biologic. Examples include whether to take tamoxifen or an aromatase inhibitor, with or without ovarian suppression, and for how long3-6; whether to take chemotherapy, with or without anthracyclines,7-9 and before or after surgery10,11; and whether to take a new biologic agent, such as pertuzumab.12,13 Moreover, diagnostic algorithms that guide treatment recommendations have become increasingly technical as genomic analyses, including germline genetic testing, are integrated into routine care.14-18 This complicated decision context can quickly overwhelm a patient seeking to understand her new diagnosis and choose a comprehensive care plan. Furthermore, most patients have only recently met the specialist physicians who are now in charge of their cancer care. Thus, at the same time when she must deliberate between treatment options, a patient must also appraise the quality of one or more therapeutic relationships. These simultaneous demands may especially burden patients with limited educational, social, or financial resources.19,20

Second opinions can facilitate treatment decision making and should be encouraged when patients are uncertain about their options or lack confidence in the treatment decision process. Given the increasing complexity of treatment decision making, second opinions may be an increasingly important opportunity for patients to gain confidence in their physicians and the proposed management plan. It is possible that a second opinion may indicate poor communication or care coordination if, for instance, there are socioeconomic gradients in use, evidence of discordance in communication or decision making, or differential use of indicated treatments in patients who do vs do not obtain second opinions.

However, little is known about how patients are referred to a medical oncologist after diagnosis, and, surprisingly, virtually nothing is known about the patterns and correlates of second opinions in community practice or the implications for quality of care. Also unknown are the characteristics of the patient-oncologist encounter, whether related to the patient, physician, or clinical situation, that prompt patients to seek a second opinion. Understanding these aspects of treatment decision making is necessary to inform interventions that can improve breast cancer care delivery and outcomes. We examined the patterns and correlates of second medical oncology opinions and patients’ perspectives on chemotherapy decision making and communication with oncologists in a large, diverse, contemporary population-based sample of patients newly diagnosed with breast cancer.

Study Sample

We selected from the iCanCare study women aged 20 to 79 years diagnosed with stages 0 to II breast cancer who were reported to the Surveillance, Epidemiology, and End Results (SEER) registries of Georgia and Los Angeles County, California. Eligible patients were identified approximately 2 months after surgery via pathology reports from definitive surgical procedures (those intended to remove the entire tumor with clear margins). To ensure a relatively homogeneous sample of patients with early-stage disease, patients with stages III to IV metastatic disease, tumors larger than 5 cm, or more than 3 involved lymph nodes were excluded. Black, Asian, and Hispanic women were oversampled in Los Angeles as previously described.21 Patients were selected between July 2013 and September 2014. This study was approved by the University of Michigan Institutional Review Board and received a waiver of documentation of informed consent. All data were deidentified before research use.

Questionnaire Design and Content

Questionnaire content was developed using a conceptual framework, research questions, and hypotheses. We developed measures by drawing from the literature and our prior research.15,22 We used standard techniques to assess content validity, including systematic review by design experts, cognitive pretesting with patients, and pilot studies in relevant populations.

Data Collection

Surveys were mailed approximately 2 months after surgery. To encourage response, we provided a $20 cash incentive and used a modified Dillman method,23 including reminders to nonrespondents. All materials were in English. We added Spanish-translated materials for all women with surnames that suggested Hispanic ethnicity.21 Each SEER registry provided limited SEER data (stripped of all identifiers) for participants to the University of Michigan: these data were then merged to survey data under institutional review board approval from partnering universities and the public health departments of Georgia and California.


Patients provided information about chemotherapy decisions, including how strongly the oncologist recommended chemotherapy on a 1- to 5-point scale (1, very strongly; 2, weakly; 3, left it up to me; 4, weakly against it; and 5, very strongly against it), whether they saw a second oncologist (the question was worded as “Did you see a second medical oncologist for an opinion about chemotherapy?” [yes or no]), and, if so, whether that second oncologist administered chemotherapy (yes or no). Patients reported their satisfaction with their amount of involvement and information about chemotherapy decisions (on a 1- to 5-point scale, with 1 indicating not enough; 3, just right; and 5, too much) and the chemotherapy decision itself (on a 1- to 5-point scale, with 1 indicating not at all satisfied; 2, a little; 3, somewhat; 4, quite; and 5, totally). Patients rated their decision-making preferences on a 1- to 5-point scale (1 indicating not at all true; 2, a little; 3, somewhat; 4, quite; and 5, very) as follows: “preferred to be told what to do,” “wanted my doctor to tell me,” or “wanted to make my own decisions.” Patients rated oncologists on a 1- to 5-point scale (1, not at all true; 2, a little; 3, somewhat; 4, quite; and 5, very) according to the Health Care Climate Questionnaire,24 which measures perceived physician support of patient autonomy with questions as follows: “provided me with choices,” “understood how I saw things,” “expressed confidence in my decision making,” “listened to how I would like to be treated,” “encouraged me to ask questions,” and “tried to understand how I saw things.”

Patients provided information on the following: race/ethnicity, insurance, educational level, travel time to the nearest hospital, comorbidities, marital status, employment, and household income. Patients reported on whether they received germline genetic testing for the breast cancer 1 (BRCA1) (OMIM 113705) and breast cancer 2 (BRCA2) (OMIM 600185) genes (BRCA1/2) and/or other genes (yes or no) and results (positive, negative, or variant of uncertain significance [VUS]). Patients reported whether they received 21-gene recurrence score (RS) testing (yes or no) and results (low, intermediate, or high). Patients reported on use of internet-based support groups (1- to 5-point scale: 1, almost never; 2, rarely; 3, sometimes; 4, often; and 5, almost always). The SEER registries provided age (years), cancer stage (I, II), cancer grade (1-3), and biomarkers, including expression of estrogen receptor (ER), progesterone receptor (PR), and the erb-b2 receptor tyrosine kinase 2 gene/human epidermal growth factor receptor 2 gene (ERBB2/HER2) (OMIM 164870).

We constructed a measure of chemotherapy indication according to the guidelines of the National Comprehensive Cancer Network (eTables 1 and 2 in the Supplement).25 Patients were categorized as having a high chemotherapy indication if they had a tumor larger than 1 cm and/or involved lymph nodes and also had ER- and PR-negative and/or ERBB2-positive disease. They were categorized as having a low chemotherapy indication if they had all of the following: age of 50 years or older, postmenopausal status, and stage I, grade 1, ER- and/or PR-positive, ERBB2-negative disease. All others were categorized as having an intermediate chemotherapy indication.

Statistical Analysis

Survey design and nonresponse weights were created to compensate for the differential probability of selecting patients by race, disease stage, and SEER site and to adjust for potential bias attributable to survey nonresponse. The weights were normalized to equal the observed sample size. Unless otherwise noted, all analyses were weighted so that statistical inferences are representative of our target population.26

Multiple Imputation

To account for item nonresponse and missing data, we multiply imputed data using a sequential regression multiple imputation framework.26 We generated 5 independently imputed data sets and then computed inferential statistics that combined estimates across the data sets.27


We described the unadjusted association of second opinion receipt with patient and tumor characteristics and patient appraisal of care yielded by observed unweighted data. A total of 436 patients (22.9%) had 1 or more missing values. We then multiply imputed data to which we applied inclusion and exclusion criteria to select an analytic sample (mean sample size, 1866 patients). We constructed a multivariable weighted logistic regression model to examine the association between the probability of second opinion receipt and SEER site, age at survey, race/ethnicity, comorbidities, educational level, employment, insurance, household income, marital status, travel time to nearest hospital, germline genetic testing receipt, 21-gene RS testing receipt, chemotherapy indication, internet-based support group use, and treatment decision-making preferences. We estimated the effect of second opinion receipt on the likelihood of chemotherapy receipt using an inverse probability of treatment weighting28,29 approach, adopted to address confounding. For each patient, we estimated the propensity of receiving a second opinion. Weighting each patient by the inverse propensity of her second opinion receipt, we created a synthetic sample in which second opinion receipt is independent of patient characteristics. After examining the properties of the weights, we estimated the mean effect of second opinion receipt on the probability of chemotherapy receipt. In a separate model using the F test for multiply imputed data, we tested for the presence of a joint effect of second opinion receipt and its interaction with chemotherapy indication. Unless otherwise noted, results were generated using multiply imputed weighted data. All analyses were conducted using SAS statistical software, version 9.4 (SAS Institute Inc). P < .05 was considered statistically significant (2-sided joint Wald test). Reported results were generated using multiply imputed, weighted data.

Patient Characteristics

We selected 3880 women diagnosed with early-stage breast cancer (mean [SD] age, 61.6 [11.0] years), of whom 3631 (93.6%) were eligible for the study. The survey response rate was 71.0% (N = 2578). We excluded 677 patients from this analysis (497 with noninvasive disease and 164 who never saw a medical oncologist) (Figure 1). eTable 3 in the Supplement indicates that the 1901 remaining patients were racially and socioeconomically diverse: 1071 (56.3%) were non-Hispanic white, 306 (16.1%) were non-Hispanic black, 328 (17.3%) were Hispanic, and 141 (7.4%) were Asian. For 1160 (61.0%), high school was the highest educational level (eTable 3 in the Supplement). A total of 1194 (62.8%) had stage I disease, 518 (27.2%) had grade 1 disease, and 471 (24.8%) had grade 3 disease. A total of 1597 (84.0%) had ER- and PR-positive tumors; 211 (11.1%) had ERBB2-positive tumors (eTable 1 in the Supplement). A total of 610 (32.1%) reported germline genetic testing, and 716 (37.7%) reported 21-gene RS testing.

Factors Associated With Receipt of Second Opinions

Multiple imputation (mean n = 1866 patients) yielded an estimated mean (SD) prevalence of second opinion receipt of 168 (9.8% [0.74%]). Figure 2 shows that factors significantly associated with second opinion receipt were a college education vs less (odds ratio [OR], 1.85; 95% CI, 1.24-2.75), a preference for making one’s own treatment decisions quite a bit of the time or always vs never or sometimes (OR, 1.15; 95% CI, 1.01-1.31), frequent use of internet-based support groups vs none (OR, 2.15; 95% CI, 1.12-4.11), an intermediate result on the 21-gene RS assay vs not tested (OR, 1.85; 95% CI, 1.11-3.09), and a VUS (OR, 3.24; 95% CI, 1.09-9.59) or negative result (OR, 1.58; 95% CI, 1.04-2.42) on germline genetic testing vs not tested. Odds of second opinion receipt were significantly lower in Georgia vs Los Angeles County (OR, 0.58; 95% CI, 0.39-0.87), but there were no interactions between site and other model covariates. No other factor, including comorbidities, employment, income, or race/ethnicity, was associated with second opinion receipt.

Receipt of Chemotherapy and Use of Second Opinions

Based on an analysis using multiply imputed, weighted data (average n = 1866), 823 patients (44.0% [SE, 1.2%]) reported chemotherapy receipt, with somewhat higher rates among patients who did (94 [52.1%]) vs did not (729 [43.2%]) receive a second opinion on univariate analysis (OR, 1.45; 95% CI, 1.07-1.97). However, several patient characteristics were correlated with second opinion and chemotherapy use, raising concern about confounding. To test whether second opinion use had a significant effect on chemotherapy use, we used an inverse probability of treatment weighting model to control for differences in the distribution of characteristics between patients who did vs did not receive a second opinion and thus reduce bias from confounding. We observed no significant interaction between second opinion receipt and chemotherapy indication on the probability of receiving chemotherapy (P = .45): high indication (OR, 0.42; 95% CI, 0.11-1.56), low indication (OR, 0.41; 95% CI, 0.06-2.77), and intermediate indication (OR, 1.28; 95% CI, 0.83-1.97). After controlling for patient and tumor characteristics, second opinion use was not associated with chemotherapy receipt (OR, 1.04; 95% CI, 0.71-1.52). Among patients who received a second opinion, 54 (3.2%) (SE, 0.47%) received chemotherapy from the second oncologist.

Patient Appraisal of Decision Making and Second Opinions

Satisfaction with chemotherapy decisions was high and did not differ between those who did (mean [SD], 4.3[.08] on a 1- to 5-point scale) or did not (4.4[.03]) obtain a second opinion (P = .29) (eTable 4 in the Supplement). Patients rated oncologists highly (mean score, 4.1 of 5) on the Health Care Climate Questionnaire, signifying perceived clinician support of patient autonomy (eTable 4 in the Supplement).


Second opinions may substantially affect a breast cancer treatment plan. Recent studies30-33 have focused on second opinions provided by one clinician to another: for example, indications that discordant reads in pathology and radiology reports change treatment in 10% to 25% of cases. Tumor boards, which enable clinicians to synthesize a combined multidisciplinary opinion, are associated with improved care quality.34-36 We previously reported on the role of second opinions in surgical decision making.37,38 However, despite this evidence of effect, little is known about the prevalence and consequences of second opinions that oncologists provide directly to patients.

In this large, diverse contemporary cohort, second opinion use was remarkably low: less than 10%, with less than 5% of all patients receiving chemotherapy from a second oncologist. There were regional differences, with second opinions less common in Georgia than Los Angeles County. Reassuringly, we did not observe racial/ethnic or socioeconomic gradients, patient-reported dissatisfaction with communication or decision making, or differential chemotherapy use by patients who did vs did not receive second opinions. We conclude that there is little evidence of unmet demand for second opinions and that their potential effect on chemotherapy decisions in community practice appears to be small.

Overall, our findings are encouraging with regard to the quality of breast cancer care. However, we identified key predictors of second opinion use that suggest opportunities for improvement. We observed a distinct profile of patients who were more likely to obtain second opinions. These patients were more often college educated, more frequently used internet-based support groups, and preferred to make their own treatment decisions. Such patients may desire greater engagement in their care and pursue second opinions for more information and support. This process may constitute an appropriate use of second opinions, yet interventions that enable the first oncologist to recognize and address these patients’ needs may also be desirable.

Along with patient demographics and preferences, we identified a clinical predictor of second opinion use: uncertain results of genomic tests. Patients who reported having a VUS on germline genetic testing were 3 times more likely to obtain a second medical oncology opinion. These unclassified results may confuse patients; moreover, one study39 found that few (<15%) physicians who order BRCA1/2 testing understand how to manage a VUS. Oncologists confronted by VUSs may struggle to explain them to patients’ satisfaction, prompting patients to seek another oncologist who can. Although VUS rates are low (2%-5%) when BRCA1/2 are the only genes sequenced, they increase 10-fold (35%) with use of the multiple-gene panels that are rapidly emerging into breast cancer care.40-44 Furthermore, VUS rates are significantly higher in racial/ethnic minorities than non-Hispanic whites.45,46 Thus, the demand for second opinions may increase with dissemination of more comprehensive genetic testing and, to a greater extent, among vulnerable populations. This occurrence raises concern about future access disparities and emphasizes the need to follow trends in second opinion use over time. Studies are urgently needed to improve the interpretation of VUSs and physicians’ ability to manage them.

Patients with intermediate results on the 21-gene RS assay were 2 times more likely than untested patients to receive a second opinion. Although the clinical utility of low and high RS is well established,14,47-49 the appropriate management of intermediate RS remains unknown pending results of clinical trials.14 A recent study50 of oncologists reported low “genomic confidence,” namely, the ability to use genomic testing results effectively for patient care. Some patients may perceive their oncologists’ low confidence about treatment recommendations in the setting of uncertain germline or tumor genomic results and seek greater confidence through a second opinion. This finding underscores the need for educational interventions that help oncologists’ knowledge and competence to keep pace with the rapid expansion of precision medicine technology.

Strengths and Limitations

Aspects of this study warrant comment. Its strengths include a large, racially/ethnically diverse, contemporary sample of patients with breast cancer enrolled from 2 population-based cancer registries; specific measures of patients’ clinical decision making; and a high response rate. Furthermore, weighting and multiple imputation techniques were used to account for potential bias attributable to missing data and to ensure that results were representative of the overall population. Its limitations include restriction to 2 geographic areas (Georgia and Los Angeles County); thus, results may not apply fully to all US patients with breast cancer. Furthermore, the sample was selected for earlier cancer stages (stages I-II) and had generally favorable tumor biology. The patterns, correlates, and outcomes of second opinion use may be different in patients whose stage or tumor biology renders them at higher risk for metastatic recurrence. We have not yet validated patients’ reports of genetic testing or oncologists’ perspectives on second opinions and other aspects of treatment. There has been insufficient follow-up time to ascertain long-term outcomes of cancer recurrence and survival. Nonetheless, this study offers a novel and clinically relevant view of breast cancer treatment decision making.


In an era of concern about the cost and value of cancer care, guidelines advise that we choose wisely before ordering diagnostic tests.51 However, there are no guidelines as to whether a second opinion (with costs similar to those of diagnostic tests) is potentially valuable or merely redundant. Given the subjective and personal nature of the therapeutic encounter, second opinions may sometimes be necessary to address a poor fit between patient and physician. We were encouraged to find high endorsement of perceived autonomy supportiveness of medical oncologists, with few patients (<10%) seeking a second opinion and little evidence of an unmet need. Our results indicate that a patient’s preference for greater engagement is one factor contributing to second opinion use, and uncertain results of diagnostic testing are another. As treatment options proliferate and molecular diagnostic tests expand, physicians may face increasing pressure to enable patients’ preferences about treatment decision making and to navigate the increasingly murky landscape of genomic testing. These tasks demand effective physician-patient communication, and developing interventions to enhance the quality of such communication is a high priority.

Back to top
Article Information

Corresponding Author: Allison W. Kurian, MD, MSc, Department of Medicine, Stanford University, HRP Redwood Building, Room T254A, 150 Governor’s Ln, Stanford, CA 94305-5405 (

Accepted for Publication: October 17, 2016.

Published Online: December 29, 2016. doi:10.1001/jamaoncol.2016.5652

Author Contributions: Dr Kurian and Ms Bondarenko had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Kurian, Bondarenko, Li, Ward, Katz.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Kurian, Friese, Bondarenko, Katz.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Bondarenko, Li.

Obtained funding: Katz.

Administrative, technical, or material support: Hamilton, Ward, Katz.

Study supervision: Ward, Hamilton, Katz.

Conflict of Interest Disclosures: None reported.

Funding/Support: Research reported in this publication was supported by grant P01CA163233 from the National Cancer Institute to the University of Michigan. The collection of cancer incidence data used in this study was supported by the California Department of Public Health pursuant to California Health and Safety Code Section 103885; cooperative agreement 5NU58DP003862-04/DP003862 from the Centers for Disease Control and Prevention’s National Program of Cancer Registries, and contract HHSN261201000140C awarded to the Cancer Prevention Institute of California, contract HHSN261201000035C to the University of Southern California, and contract HHSN261201000034C to the Public Health Institute from the National Cancer Institute’s Surveillance, Epidemiology, and End Results program. The collection of cancer incidence data in Georgia was supported by contract HHSN261201300015I and task order HHSN26100006 from the National Cancer Institute and by cooperative agreement 5NU58DP003875-04-00 from the Centers for Disease Control and Prevention.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The ideas and opinions expressed herein are those of the authors, and endorsement by the state of California, US Department of Public Health, the National Cancer Institute, and the Centers for Disease Control and Prevention or their contractors and subcontractors is not intended nor should be inferred.

Meeting Presentation: Preliminary results were presented in part at the Annual Meeting of the American Society of Clinical Oncology; May 30, 2015; Chicago, Illinois.

Additional Contributions: We acknowledge the outstanding work of our project staff (Mackenzie Crawford, MPH, and Kiyana Perrino, MPH, from the Georgia Cancer Registry; Jennifer Zelaya, Pamela Lee, Maria Gaeta, Virginia Parker, BA, and Renee Bickerstaff-Magee from the University of Southern California; Rebecca Morrison, MPH, Rachel Tocco, MA, Alexandra Jeanpierre, MPH, Stefanie Goodell, BS, Rose Juhasz, PhD, Paul Abrahamse, MA, and Kent Griffith from the University of Michigan). These people were compensated for their contributions. We acknowledge with gratitude our survey respondents.

Berry  DA, Cronin  KA, Plevritis  SK,  et al; Cancer Intervention and Surveillance Modeling Network (CISNET) Collaborators.  Effect of screening and adjuvant therapy on mortality from breast cancer.  N Engl J Med. 2005;353(17):1784-1792.PubMedGoogle ScholarCrossref
Munoz  D, Near  AM, van Ravesteyn  NT,  et al.  Effects of screening and systemic adjuvant therapy on ER-specific US breast cancer mortality.  J Natl Cancer Inst. 2014;106(11):dju289.PubMedGoogle ScholarCrossref
Davies  C, Pan  H, Godwin  J,  et al; Adjuvant Tamoxifen: Longer Against Shorter (ATLAS) Collaborative Group.  Long-term effects of continuing adjuvant tamoxifen to 10 years versus stopping at 5 years after diagnosis of oestrogen receptor-positive breast cancer: ATLAS, a randomised trial.  Lancet. 2013;381(9869):805-816.PubMedGoogle ScholarCrossref
Francis  PA, Regan  MM, Fleming  GF,  et al; SOFT Investigators; International Breast Cancer Study Group.  Adjuvant ovarian suppression in premenopausal breast cancer.  N Engl J Med. 2015;372(5):436-446.PubMedGoogle ScholarCrossref
Goss  PE, Ingle  JN, Pritchard  KI,  et al.  Extending aromatase-inhibitor adjuvant therapy to 10 years.  N Engl J Med. 2016;375(3):209-219.PubMedGoogle ScholarCrossref
Pagani  O, Regan  MM, Walley  BA,  et al; TEXT and SOFT Investigators; International Breast Cancer Study Group.  Adjuvant exemestane with ovarian suppression in premenopausal breast cancer.  N Engl J Med. 2014;371(2):107-118.PubMedGoogle ScholarCrossref
Blum  JL, Flynn  PJ, Yothers  G,  et al. Interim joint analysis of the ABC (Anthracyclines in Early Breast Cancer) phase III trials (USOR 06-090, NSABP B-46I/USOR 07132, NSABP B-49 [NRG Oncology]) comparing docetaxel + cyclophosphamide (TC) v anthracycline/taxane-based chemotherapy regimens (TaxAC) in women with high-risk, HER2-negative breast cancer. Paper presented at: American Society of Clinical Oncology Annual Meeting; June 4, 2016; Chicago, IL.
Giordano  SH, Lin  YL, Kuo  YF, Hortobagyi  GN, Goodwin  JS.  Decline in the use of anthracyclines for breast cancer.  J Clin Oncol. 2012;30(18):2232-2239.PubMedGoogle ScholarCrossref
Jones  S, Holmes  FA, O’Shaughnessy  J,  et al.  Docetaxel with cyclophosphamide is associated with an overall survival benefit compared with doxorubicin and cyclophosphamide: 7-year follow-up of US Oncology Research Trial 9735.  J Clin Oncol. 2009;27(8):1177-1183.PubMedGoogle ScholarCrossref
DeMichele  A, Yee  D, Berry  DA,  et al.  The neoadjuvant model is still the future for drug development in breast cancer.  Clin Cancer Res. 2015;21(13):2911-2915.PubMedGoogle ScholarCrossref
Kaufmann  M, von Minckwitz  G, Mamounas  EP,  et al.  Recommendations from an international consensus conference on the current status and future of neoadjuvant systemic therapy in primary breast cancer.  Ann Surg Oncol. 2012;19(5):1508-1516.PubMedGoogle ScholarCrossref
Schneeweiss  A, Chia  S, Hickish  T,  et al.  Pertuzumab plus trastuzumab in combination with standard neoadjuvant anthracycline-containing and anthracycline-free chemotherapy regimens in patients with HER2-positive early breast cancer: a randomized phase II cardiac safety study (TRYPHAENA).  Ann Oncol. 2013;24(9):2278-2284.PubMedGoogle ScholarCrossref
Gianni  L, Pienkowski  T, Im  YH,  et al.  Efficacy and safety of neoadjuvant pertuzumab and trastuzumab in women with locally advanced, inflammatory, or early HER2-positive breast cancer (NeoSphere): a randomised multicentre, open-label, phase 2 trial.  Lancet Oncol. 2012;13(1):25-32.PubMedGoogle ScholarCrossref
Sparano  JA, Gray  RJ, Makower  DF,  et al.  Prospective validation of a 21-gene expression assay in breast cancer.  N Engl J Med. 2015;373(21):2005-2014.PubMedGoogle ScholarCrossref
Jagsi  R, Griffith  KA, Kurian  AW,  et al.  Concerns about cancer risk and experiences with genetic testing in a diverse population of patients with breast cancer.  J Clin Oncol. 2015;33(14):1584-1591.PubMedGoogle ScholarCrossref
Jagsi  R, Kurian  AW, Griffith  KA,  et al. Genetic testing decisions of breast cancer patients: results from the iCanCare study. Paper presented at: American Society of Clinical Oncology Annual Meeting; May 30, 2015; Chicago, IL.
Dinan  MA, Mi  X, Reed  SD, Hirsch  BR, Lyman  GH, Curtis  LH.  Initial trends in the use of the 21-gene recurrence score assay for patients with breast cancer in the Medicare population.  JAMA Oncol. 2015;1(2):158-166.PubMedGoogle ScholarCrossref
Hassett  MJ, Silver  SM, Hughes  ME,  et al.  Adoption of gene expression profile testing and association with use of chemotherapy among women with breast cancer.  J Clin Oncol. 2012;30(18):2218-2226.PubMedGoogle ScholarCrossref
Livaudais  JC, Franco  R, Fei  K, Bickell  NA.  Breast cancer treatment decision-making: are we asking too much of patients?  J Gen Intern Med. 2013;28(5):630-636.PubMedGoogle ScholarCrossref
Martinez  KA, Kurian  AW, Hawley  ST, Jagsi  R.  How can we best respect patient autonomy in breast cancer treatment decisions?  Breast Cancer Manag. 2015;4(1):53-64.PubMedGoogle ScholarCrossref
Hamilton  AS, Hofer  TP, Hawley  ST,  et al.  Latinas and breast cancer outcomes: population-based sampling, ethnic identity, and acculturation assessment.  Cancer Epidemiol Biomarkers Prev. 2009;18(7):2022-2029.PubMedGoogle ScholarCrossref
Hawley  ST, Jagsi  R, Morrow  M,  et al.  Social and clinical determinants of contralateral prophylactic mastectomy.  JAMA Surg. 2014;149(6):582-589.PubMedGoogle ScholarCrossref
Dillman  DASJ, Christian  LM.  Internet, Mail, and Mixed-Mode Surveys: The Tailored Design Method. 3rd ed. Hoboken, NJ: John Wiley & Sons; 2009.
Williams  GC, Grow  VM, Freedman  ZR, Ryan  RM, Deci  EL.  Motivational predictors of weight loss and weight-loss maintenance.  J Pers Soc Psychol. 1996;70(1):115-126.PubMedGoogle ScholarCrossref
Gradishar  WJ, Anderson  BO, Balassanian  R,  et al.  Invasive breast cancer version 1.2016, NCCN clinical practice guidelines in oncology.  J Natl Compr Canc Netw. 2016;14(3):324-354.PubMedGoogle Scholar
Kish  L.  Survey Sampling. New York, NY: John Wiley & Sons; 1965.
Rubin  DB.  Multiple Imputation for Nonresponse in Surveys. Hoboken, NY: John Wiley & Sons; 1987.
Lunceford  JK, Davidian  M.  Stratification and weighting via the propensity score in estimation of causal treatment effects: a comparative study.  Stat Med. 2004;23(19):2937-2960.PubMedGoogle ScholarCrossref
Joffe  MM, Have  TRT, Feldman  HI, Kimmel  SE.  Model selection, confounder control, and marginal structural models: review and new applications.  Am Stat. 2004;58(4):272-279.Google ScholarCrossref
Elmore  JG, Longton  GM, Carney  PA,  et al.  Diagnostic concordance among pathologists interpreting breast biopsy specimens.  JAMA. 2015;313(11):1122-1132.PubMedGoogle ScholarCrossref
Geller  BM, Nelson  HD, Carney  PA,  et al.  Second opinion in breast pathology: policy, practice and perception.  J Clin Pathol. 2014;67(11):955-960.PubMedGoogle ScholarCrossref
Khazai  L, Middleton  LP, Goktepe  N, Liu  BT, Sahin  AA.  Breast pathology second review identifies clinically significant discrepancies in over 10% of patients.  J Surg Oncol. 2015;111(2):192-197.PubMedGoogle ScholarCrossref
Spivey  TL, Carlson  KA, Janssen  I, Witt  TR, Jokich  P, Madrigrano  A.  Breast imaging second opinions impact surgical management.  Ann Surg Oncol. 2015;22(7):2359-2364.PubMedGoogle ScholarCrossref
Kesson  EM, Allardice  GM, George  WD, Burns  HJ, Morrison  DS.  Effects of multidisciplinary team working on breast cancer survival: retrospective, comparative, interventional cohort study of 13 722 women.  BMJ. 2012;344:e2718.PubMedGoogle ScholarCrossref
Farrugia  DJ, Fischer  TD, Delitto  D, Spiguel  LR, Shaw  CM.  Improved breast cancer care quality metrics after implementation of a standardized tumor board documentation template.  J Oncol Pract. 2015;11(5):421-423.PubMedGoogle ScholarCrossref
Newman  EA, Guest  AB, Helvie  MA,  et al.  Changes in surgical management resulting from case review at a breast cancer multidisciplinary tumor board.  Cancer. 2006;107(10):2346-2351.PubMedGoogle ScholarCrossref
Katz  SJ, Hofer  TP, Hawley  S,  et al.  Patterns and correlates of patient referral to surgeons for treatment of breast cancer.  J Clin Oncol. 2007;25(3):271-276.PubMedGoogle ScholarCrossref
Morrow  M, Jagsi  R, Alderman  AK,  et al.  Surgeon recommendations and receipt of mastectomy for treatment of breast cancer.  JAMA. 2009;302(14):1551-1556.PubMedGoogle ScholarCrossref
Pal  T, Cragun  D, Lewis  C,  et al.  A statewide survey of practitioners to assess knowledge and clinical practices regarding hereditary breast and ovarian cancer.  Genet Test Mol Biomarkers. 2013;17(5):367-375.PubMedGoogle ScholarCrossref
Kurian  AW, Ford  JM.  Multigene panel testing in oncology practice: how should we respond?  JAMA Oncol. 2015;1(3):277-278.PubMedGoogle ScholarCrossref
Desmond  A, Kurian  AW, Gabree  M,  et al.  Clinical actionability of multigene panel testing for hereditary breast and ovarian cancer risk assessment.  JAMA Oncol. 2015;1(7):943-951.PubMedGoogle ScholarCrossref
Kurian  AW, Hare  EE, Mills  MA,  et al.  Clinical evaluation of a multiple-gene sequencing panel for hereditary cancer risk assessment.  J Clin Oncol. 2014;32(19):2001-2009.PubMedGoogle ScholarCrossref
Kurian  AW, Antoniou  AC, Domchek  SM.  Refining breast cancer risk stratification: additional genes, additional information.  Am Soc Clin Oncol Educ Book. 2016;35:44-56.PubMedGoogle ScholarCrossref
Swisher  EM.  Usefulness of multigene testing: catching the train that’s left the station.  JAMA Oncol. 2015;1(7):951-952.PubMedGoogle ScholarCrossref
Hall  MJ, Reid  JE, Burbidge  LA,  et al.  BRCA1 and BRCA2 mutations in women of different ethnicities undergoing testing for hereditary breast-ovarian cancer.  Cancer. 2009;115(10):2222-2233.PubMedGoogle ScholarCrossref
Maxwell  KN, Wubbenhorst  B, D’Andrea  K,  et al.  Prevalence of mutations in a panel of breast cancer susceptibility genes in BRCA1/2-negative patients with early-onset breast cancer.  Genet Med. 2015;17(8):630-638.PubMedGoogle ScholarCrossref
Goldstein  LJ, Gray  R, Badve  S,  et al.  Prognostic utility of the 21-gene assay in hormone receptor-positive operable breast cancer compared with classical clinicopathologic features.  J Clin Oncol. 2008;26(25):4063-4071.PubMedGoogle ScholarCrossref
Habel  LA, Shak  S, Jacobs  MK,  et al.  A population-based study of tumor gene expression and risk of breast cancer death among lymph node-negative patients.  Breast Cancer Res. 2006;8(3):R25.PubMedGoogle ScholarCrossref
Paik  S, Tang  G, Shak  S,  et al.  Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor–positive breast cancer.  J Clin Oncol. 2006;24(23):3726-3734.PubMedGoogle ScholarCrossref
Gray  SW, Hicks-Courant  K, Cronin  A, Rollins  BJ, Weeks  JC.  Physicians’ attitudes about multiplex tumor genomic testing.  J Clin Oncol. 2014;32(13):1317-1323.PubMedGoogle ScholarCrossref
Schnipper  LE, Lyman  GH, Blayney  DW,  et al.  American Society of Clinical Oncology 2013 top five list in oncology.  J Clin Oncol. 2013;31(34):4362-4370.PubMedGoogle ScholarCrossref