Surgeon specialization and use of sentinel lymph node biopsy for breast cancer

Abstract

IMPORTANCE

Sentinel lymph node biopsy (SLNB) is the standard of care for axillary staging in clinically node-negative breast cancer patients. It is not known whether SLNB rates differ by surgeon expertise. If surgeons with less breast cancer expertise are less likely to offer SLNB to clinically node-negative patients, this practice pattern could lead to unnecessary axillary lymph node dissections (ALND) and lymphedema.

OBJECTIVE

To explore potential measures of surgical expertise (including a novel objective specialization measure – percentage of a surgeon’s operations devoted to breast cancer determined from claims) on the use of SLNB for invasive breast cancer.

DESIGN

Population-based prospective cohort study. Patient, tumor, treatment and surgeon characteristics were examined.

SETTING

California, Florida, Illinois

PARTICIPANTS

Elderly (65+ years) women identified from Medicare claims as having had incident invasive breast cancer surgery in 2003.

MAIN OUTCOME MEASURES

Type of axillary surgery performed.

RESULTS

Of the 1,703 women treated by 863 different surgeons, 56% underwent an initial SLNB, 37% initial ALND and 6% no axillary surgery. The median annual surgeon Medicare volume of breast cancer cases was 6 (range: 1.5–57); the median surgeon percentage of breast cancer cases was 4.6% (range: 0.7%–100%). After multivariable adjustment of patient and surgeon factors, women operated on by surgeons with higher volumes and percentages of breast cancer cases had a higher likelihood of undergoing SLNB. Specifically, women were most likely to undergo SLNB if operated on by high volume surgeons (regardless of percentage) or by lower volume surgeons with a high percentage of cases devoted to breast cancer. In addition, membership in the American Society of Breast Surgeons (OR 1.98, CI 1.51–2.60) and Society of Surgical Oncology (OR 1.59, CI 1.09–2.30) were independent predictors of women undergoing an initial SLNB.

CONCLUSIONS AND RELEVANCE

Patients treated by surgeons with more experience and focus in breast cancer were significantly more likely to undergo SLNB, highlighting the importance of receiving initial treatment by specialized providers. Factors relating to specialization in a particular area, including our novel surgeon percentage measure, require further investigation as potential indicators of quality of care.

INTRODUCTION

Over the last two decades, sentinel lymph node biopsy (SLNB) has been the single most important advance in the primary surgical treatment of clinically node-negative breast cancer. Compared to axillary lymph node dissection (ALND), which involves the removal of most axillary lymph nodes, SLNB involves the removal of only a few lymph nodes (average 2–3 nodes).^1–5 Several large, randomized controlled trials demonstrate that SLNB is associated with similar disease-free survival when compared to ALND^1,4–9 as well as a reduced likelihood of developing lymphedema and other arm/shoulder morbidity.^{1,4–6,10–17} For these reasons, SLNB is now the standard of care for axillary staging in clinically node-negative patients^18–20 and is considered a surgical quality measure by certain organizations.^21–23

Many women with breast cancer in the US are cared for by surgeons with low volumes of breast cancer cases.^17,24–26 However, it is not known whether patients of low volume surgeons are as likely to undergo SLNB. If surgeons with less breast cancer expertise are less likely to offer SLNB to clinically node-negative patients, this practice pattern could lead to unnecessary ALNDs and cases of lymphedema.

In this report, we evaluate the relationship between surgeon characteristics and axillary surgery for invasive breast cancer by exploring several potential measures of surgeon expertise. One measure, surgeon volume of breast cancer cases, has been used as a measure of technical proficiency with regard to margin status and re-excision rates after breast-conserving surgery, use of axillary surgery, and adequacy of ALND.^27–30 For this study, we also developed a novel objective measure of surgical specialization, the percentage of a surgeon’s operative cases that is devoted to breast cancer, as determined from claims data. Prior studies have evaluated the relationship of self-reported percentages of a surgeon’s practice devoted to breast surgery to the type of breast surgery performed, adequacy of ALND, and patient satisfaction.^29–32 Other measures of expertise investigated included the number of years in practice and membership in surgical oncology and breast-specific societies. We explored these five measures of surgical expertise and receipt of initial axillary surgery in a population-based cohort of older breast cancer survivors.

METHODS

Data Sources

The study sample consists of a population-based cohort of elderly breast cancer survivors who participated in a National Cancer Institute-sponsored survey study examining breast cancer care outcomes. Women between the ages of 65 and 89 were identified from Medicare claims as having had an incident breast cancer surgery in 2003, using our validated claims-based algorithm.³³ Details regarding study recruitment and assessments have been previously described.^17,34 Briefly, potential participants were contacted by mail in September 2005 and four subsequent annual structured telephone interviews were conducted. For this study, the cohort consists of 1,703 women with invasive breast cancer who completed all four waves of the survey, had complete information regarding pathologic nodal status, and had a surgeon identifiable by Medicare claims.

Tumor characteristics and stage was provided by state cancer registries.³⁵ Medicare claims information was collected from inpatient, outpatient and carrier Standard Analytical Files. Surgeon characteristics were obtained from the American Medical Association’s (AMA) Physicians Professional Database (Table 1).^36,37 Society membership was determined from online directories of the Society of Surgical Oncology (SSO)³⁸ and American Society of Breast Surgeons (ASBrS).³⁹ Surgeons were not contacted directly for this study.

Table 1.

Variable definitions and data source

Variable	Data source
Patient characteristics
Age	Medicare claims
State of residence	Medicare claims
California, Illinois or New York
Race	Patient survey
White, Black or African American, Asian, native
Hawaiian or other Pacific Islander, American
Indian/Alaska Native, or other
Body mass index (BMI) at time of surgerya	Patient survey
Comorbidityb	Medicare claims
Pathologic tumor characteristics
Tumor size	State tumor registryc
Lymph node status	State tumor registry
Number of lymph nodes removed	State tumor registry
Treatment variables
Type of breast surgery	Medicare claims
Breast-conserving surgery
Mastectomy
Type of axillary surgery	Medicare claims
None
Sentinel node biopsy (SLNB)
Axillary lymph node dissection (ALND)
Radiation therapy	Patient survey
Chemotherapy	Patient survey
Hormonal therapy within one year of surgery	Patient survey
Surgeon factors
Surgeon characteristics	American Medical Association
Gender (self-report)	Physicians Professional Databased
Medical school location (LCME primary source)
Year of medical school graduation (ACGME primary source)
Surgical specialty (self-report)
General surgery, surgical oncology or other
Academic affiliation (self-report)
American Society of Breast Surgeons (ASBrS) membership	Society directory
Society of Surgical Oncology (SSO) membership	Society directory
Surgeon volume of Medicare breast cancer cases	Medicare claims
Surgeon percentage of Medicare operations performed for breast cancer	Medicare claims

Abbreviations: ACGME, Accreditation Council for Graduate Medical Education; LCME, Liaison Committee on Medical Education

^aBody mass index was calculated by self-reported height and weight at the time of surgery.

^bComorbidity was determined from inpatient, outpatient and carrier Medicare claims for the year preceding the incident breast cancer diagnosis, based on the NCI Combined Comorbidity Index algorithm by Klabunde which is specific for breast cancer patients.⁶¹

^cThe three state tumor registries (Florida, Illinois, California) have achieved Gold-level certification by the North American Association of Central Cancer Registries.³⁵

^dThe American Medical Association’s Physician Masterfile was established in 1906 and currently includes information from more than 1.4 million physicians, residents and medical students in the US. It is considered the most accurate source of information on physician characteristics. Information regarding medical education, residency training or other professional certification data is verified from the primary source while other information (surgical specialty and academic affiliation) is self-reported. Physicians are contacted by the AMA periodically to update their Masterfile.

Variable Definitions

Table 1 summarizes the various patient, tumor, treatment and surgeon characteristics examined and the source of information. Medicare claims data were used to determine the type of axillary surgery performed.^40–42 Patients were classified into one of three groups: 1) no axillary surgery; 2) initial SLNB, which includes patients who underwent only SLNB and those who went on to completion ALND either during the same operation or subsequent surgery; 3) initial ALND. Procedure codes in 2003 for SLNB included: injection of the radioactive tracer or dye (Current Procedural Terminology [CPT] A9520, 38790, 38792) and removal of sentinel lymph nodes (CPT 38500, 38525). ALND procedural codes included codes that referred to only axillary surgery (CPT 38740, 38745) and those that included combined breast and axillary procedures (CPT 19162, 19200, 19220, 19240).

Surgeon Characteristics

Provider codes from Medicare claims of the surgeons operating on cohort patients were linked to the 2004 AMA’s Physicians Professional Database to obtain surgeon characteristics. Surgeons were considered to have an academic affiliation if they performed the majority of their breast operations in an Association of American Medical Colleges Council of Teaching Hospitals and Health systems hospital,⁴³ were employed by a medical school, or primarily practiced medical research or education. Medicare claims were used to determine annual Medicare surgeon volume of breast cancer cases, as previously described,^17,44 based on claims for all breast cancer cases treated in each state, not solely for cohort subjects.

The novel objective measure (surgeon percentage of practice devoted to breast cancer cases) was created. For each surgeon, the annual number of patients who underwent an initial breast cancer surgery was divided by the number of patients who underwent a general surgery operation performed that same year. General surgery operations were defined by CPT codes 10021 –69990 and all patient refined diagnosis related groups (APR-DRGs) surgical “P” codes.^41,45 This specialization measure was based on Medicare claims for all surgeries performed in each state, not solely for cohort subjects.

Statistical Analysis

The conceptual framework was to model 1) the decision to perform axillary surgery; and if done, 2) which type of axillary surgery (initial SLNB or ALND) was performed. We hypothesized that surgeon characteristics would affect the type of axillary surgery performed in a fashion incremental to clinically important, preoperative factors. Variables included in these analyses were determined a priori: five surgeon characteristics (surgeon percentage, surgeon volume, membership in ASBrS or SSO, and years since medical school graduation), three patient characteristics (age, BMI, comorbidities), and geographic location. Tumor characteristics (clinical stage, hormone receptor and HER2-neu status) could not be included as this information was not consistently recorded in the state tumor registries.

The two axillary surgery outcomes were analyzed by separate multiple logistic regression modeling. The first model included the entire cohort and examined the independent effects of various surgeon and patient characteristics on the likelihood of undergoing axillary surgery (initial SLNB or ALND vs. no axillary surgery). In the subset of women who underwent axillary surgery, the second model examined the relationship between these same characteristics and the likelihood of undergoing initial SLNB vs. ALND.

Data analyses were conducted using SAS statistical software (Version 9.3, SAS Institute; Cary, NC). This study was approved by the Centers for Medicare and Medicaid Services Privacy Board, our institution’s and each tumor registry’s Institutional Review Board.

RESULTS

The mean age of the 1,703 women at the time of surgery was 72.9 (SD 5.6) years; 91% were white. At the time of surgery, 36% of women were underweight or normal weight (BMI <25), 31% overweight (BMI 25 – <30), 22% obese (BMI≥30) and 11% unknown. The majority of women (62%) were healthy with an NCI combined comorbidity score of 0; 32% had a score > 0, and 6% were unknown. Most women had early stage disease: 70% had T1 tumors (tumor size < 2 cm); 19% had T2 tumors (2–5 cm) and 3% had tumors > 5 cm in size; 9% had no reported tumor size. Twenty-two percent of women had lymph node involvement. Almost two-thirds (62%) underwent breast-conserving surgery; 38% underwent mastectomy. No axillary surgery was performed in 6%, while 56% underwent an initial SLNB and 37% initial ALND. The mean number of lymph nodes removed was 7.1 [SD 6.7; range: 0 – 44; interquartile range: 9 (2 – 11)]. Women undergoing ALND had an average of 9.2 (SD 6.8) nodes removed, compared to 3.0 (SD 3.2) for those undergoing SLNB (p<0.001). Two-thirds of the cohort received radiation therapy, 23% chemotherapy, and 68% hormonal therapy.

Surgeon characteristics

Table 2 details the characteristics of the 863 surgeons who operated on these 1,703 women. The majority were male, graduated from a U.S. medical school, and were general surgeons. Small minorities were surgical oncologists, had an academic affiliation, or were ASBrS or SSO members. Figure 1 shows the distribution of surgeon volume and surgeon percentage. The median annual Medicare volume was 6 cases (range: 1.5 to 57; interquartile range: 7.5 (3.0 – 10.5). Overall, 41% of surgeons performed fewer than 6 incident Medicare breast cancer cases annually, which equates to fewer than 15 cases in women of all ages.

Table 2.

Characteristics of 863 surgeons, by surgeon percentage measure

Surgeon characteristics	Total surgeon cohort (n = 863)	Surgeon percent ≥ 10% (n = 181)	Surgeon percent < 10% (n = 682)	p-value
Gendera				<0.0001
Male	739 (86%)	100 (56%)	639 (94%)
Female	120 (14%)	80 (44%)	40 (6%)
Medical school traininga				0.010
U.S.	702 (81%)	159 (88%)	543 (80%)
Non-U.S.	157 (18%)	21 (11%)	136 (20%)
Years since medical school graduationa				0.38
More than 30 (1973 and earlier)	277 (32%)	61 (34%)	201 (30%)
20 – 30 (1974 – 1984)	320 (37%)	68 (38%)	252 (37%)
Less than 20 (since 1985)	262 (30%)	51 (28%)	226 (33%)
Surgeon specialty				<0.0001
General surgery	781 (90%)	157 (87%)	24 (92%)
Surgical oncology	21 (2%)	14 (8%)	7 (1%)
Other	61 (7%)	10 (6%)	51 (7%)
Academic surgeon	88 (10%)	45 (25%)	43 (6%)	<0.0001
ASBrS memberb	165 (19%)	69 (39%)	96 (14%)	<0.0001
SSO memberb	84 (10%)	50 (28%)	34 (5%)	<0.0001
Mean annual Medicare surgeon volume of breast cancer cases (SD)	8.5 (7.3)	19.7 (12.3)	7.5 (5.7)	<0.0001
0 – <6	351 (41%)	27 (15%)	324 (48%)
6 – <12	308 (36%)	55 (30%)	253 (37%)
12 or more	204 (24%)	99 (55%)	105 (15%)

Abbreviations: ASBrS, American Society of Breast Surgeons; SD, standard deviation; SSO, Society of Surgical Oncology; U.S., United States.

^aExcludes 4 surgeons with missing information

^bExcludes 8 surgeons with missing information

Figure 1.

Distribution of surgeon volume (Figure 1A) and surgeon percentage (Figure 1B) among the 863 surgeons. Surgeon volume or surgeon percentage is displayed on the x-axis; the percentage of surgeons in each category is displayed on the y-axis. In Figure 1B, two surgeons are excluded who had a surgeon percentage > 60%. In each Figure, the percent of surgeons in each bar is represented by the number above the bar. Bars marked with an “*” represent percentage values less than 1%. In Figure 1B, the sum of all bar percents (24%, 30%, 17% and 8%) between the 0 and 10% x-axis hatch mark shows that 79% of surgeons devote less than 10% of their operative cases to breast cancer. The remaining 21% of surgeons (sum of all bar percents after the surgeon percentage 10% x-axis hatch mark) devote 10% or more of their operative cases to breast cancer.

Among all Medicare operations performed by the 863 surgeons during one year, the median percentage of breast cancer cases was 4.6% (range: 0.7% to 100%; interquartile range: 6.2% [2.6% to 8.8%]). Twenty-one percent of surgeons (n=181) devoted 10% or more of their operations to breast cancer (Table 2, Figure 1b) and performed more than twice as many breast cancer operations annually as surgeons devoting less than 10% of their operations to breast cancer (19.7 vs. 7.5; Table 2). These surgeons were more likely to be female, a surgical oncologist, members of the ASBrS and SSO, and have an academic affiliation. Only 8% and 4% of surgeons devoted more than 20% and 30% of their operations to breast cancer, respectively.

Independent predictors of undergoing axillary surgery

Of the entire cohort, 94% underwent either SLNB or ALND and 6% (n=108) underwent no axillary surgery. In the multiple logistic regression model, when adjusting for surgeon and patient characteristics, the only independent predictor of undergoing any axillary surgery (SLNB or ALND) was patient age. Compared to women aged 65–69 years, women 80 years and older were significantly less likely to undergo any axillary surgery (OR 0.21, 95% CI 0.12 – 0.36). Patient BMI, comorbidity and all five surgeon characteristics were not predictors of a woman’s likelihood of undergoing axillary surgery.

Independent predictors of undergoing initial SLNB

Among the 1,585 women who underwent axillary surgery, 60% underwent initial SLNB and 40% ALND. Women operated on by surgeons who were members of ASBrS or SSO were more likely to undergo SLNB (Table 3). Since there was no interaction between ASBrS and SSO membership (p = 0.92), the effects of society membership are additive. Women undergoing an operation by surgeons who were members of both ASBrS and SSO had a 3.14 odds ratio of undergoing SLNB (95% CI 2.02 – 4.87).

Table 3.

Logistic regression model predicting the performance of initial SLNB vs. ALND in 1,585 women who underwent axillary surgery^a

Variables	Odds Ratio	95% CI	p-value
Surgeon characteristics
Percent of Medicare operations performed annually for breast cancer	Due to the presence of an interaction between surgeon volume and percentage of breast cancer cases, the effects are shown in Figure 2.		<0.0001b
Annual Medicare surgeon volume of breast cancer cases
Interaction between surgeon volume and surgeon percent
ASBrS member			<0.0001
No	1.00
Yes	1.98	1.51 – 2.60
SSO member			0.02
No	1.00
Yes	1.59	1.09 – 2.30
Years since medical school graduation			0.22
More than 30	1.00
20 – 30	1.08	0.84 – 1.39
Less than 20	1.27	0.96 – 1.68
Patient characteristics
Age at time of surgery			0.0009
65 – 69	1.00
70 – 74	0.70	0.54 – 0.92
75 – 79	0.87	0.65 – 1.17
80+	0.50	0.35 – 0.72

Adjusted for comorbidity, BMI, geographic location

Abbreviations: ALND, axillary lymph node dissection; ASBrS, American Society of Breast Surgeons; SLNB, sentinel lymph node biopsy; SSO, Society of Surgical Oncology.

^aExcludes 10 patients with missing information

^bThe presence of an interaction renders the p-values for each individual variable not interpretable. The reported p-value is for a three-way degree of freedom test.

Women operated on by surgeons who had higher breast cancer volumes and higher percentages of breast cancer cases were also more likely to undergo SLNB. These two effects, however, were not additive as there is an interaction between these two variables (p<0.0001). Figure 2 displays the relationship between surgeon volume, surgeon percentage and the odds ratio of undergoing SLNB vs. ALND. All odds ratios are in reference to the baseline odds for a surgeon who performs six breast cancer cases annually and devotes 3% of all surgeries to breast cancer cases (baseline surgeon). For a low volume surgeon (annual volume of 6 Medicare cases), the likelihood of a patient undergoing SLNB increases significantly with increasing surgeon percentage of breast cancer cases. Compared to the baseline surgeon, the odds ratio of a patient undergoing an initial SLNB ranges from 1.05 (95% CI 1.02– 1.09) if her surgeon’s percentage of breast cancer operations is 5% to 1.96 (95% CI 1.03 – 3.05) if surgeon percentage is 30%. As surgeon volume increases, the incremental effect of surgeon percentage on the odds ratio of receiving an SLNB diminishes. If a surgeon performs 12 Medicare breast cancer cases annually, the odds ratio of a patient receiving an SLNB ranges from 1.40 (95% CI 1.25 – 1.56) to 2.09 (95% CI 1.44 – 3.05) as her surgeon’s percentage rises from 5% to 30%. At a high surgeon volume of 24 annual Medicare cases (equates to about 60 annual cases of all ages), the effect of surgeon proportion is negligible.

Figure 2.

Graph of annual Medicare surgeon volume (x-axis) and the odds ratio of undergoing sentinel node biopsy (SLNB) vs. axillary lymph node dissection (ALND; y-axis) relative to the baseline odds of undergoing SLNB for a surgeon who performs six annual incident Medicare breast cancer and devotes 3% of all surgeries to breast cancer cases. Various surgeon percentage values are represented by the colored boxes (red, 5%; green, 10%; light blue, 20%; dark blue, 30%). For each box, the black line represents the odds ratio estimate; the box represents the limits of the 50% confidence interval; the whiskers (dotted lines) represent the 95% confidence intervals.

Similarly, the incremental effect of surgeon volume on surgeon percent is greatest at the lower surgeon percentages (5% and 10%; red and green bars). As surgeon percentage increases, the effect of surgeon volume becomes minimal. When surgeon percentage is 30% (dark blue bars; only 4% of the surgeons in this cohort), the incremental effect of surgeon volume is negligible. With increasingly higher surgeon volume and surgeon percentage values, the confidence intervals widen significantly; caution must be used in interpreting the model results at the high end of surgeon volume and surgeon percentage values.

DISCUSSION

In this population-based cohort of over 1,700 older invasive breast cancer survivors treated by 863 predominantly community-based general surgeons, 56% underwent an initial SLNB procedure while 37% underwent an initial ALND and 6% underwent no axillary surgery. The surgeon characteristics evaluated in this study were measurable factors which we believed would reflect experience with breast cancer (case volume, years since medical school graduation) or degree of focus or specialization in breast cancer (percentage of total surgeries performed for breast cancer, ASBrS and SSO membership). None of these five surgeon factors were predictive of a woman’s likelihood of undergoing axillary surgery. However, if a woman underwent axillary surgery, surgeon percentage of breast cancer surgeries, case volume, and membership in either ASBrS or SSO were associated with the likelihood of a woman undergoing an SLNB, controlling for patient age, BMI, and comorbidity.

Specialization has been found to be beneficial in other areas of health care delivery (e.g., critical care intensivists^46,47). Our study results would support some degree of specialization as benefitting breast cancer patients. Surgeons who dedicate a larger percentage of their practice to breast cancer or are members of breast and surgical oncology societies may have a greater motivation to become facile with treatment advances, such as SLNB. However, it would be difficult from a health policy perspective to base provider recommendations on professional society membership as it is a surgeon’s choice to join professional societies and membership requirements vary by society. Therefore, we favor the objective measure of percentage of a surgeon’s practice that is devoted to breast cancer (specialization construct), in addition to surgeon volume of breast cancer cases (experience construct), as a potential surgical quality marker.

Both higher surgeon volume and higher surgeon percentage of breast cancer cases were associated with a higher likelihood of a patient undergoing an initial SLNB. However, this relationship between surgeon volume and surgeon percentage is somewhat complex (Figure 2). As one variable increases (e.g. surgeon volume), the effect of the other variable (e.g. surgeon percentage) decreases. The incremental effect of higher surgeon percentage on a high volume surgeon’s likelihood of performing an SLNB is marginal but the incremental effect of higher surgeon percentage on a low volume surgeon’s likelihood of performing SLNB is substantial. This finding could partially explain why some low volume surgeons have good outcomes, if they are high percentage surgeons. Similarly, the incremental effect of surgeon volume on a high percentage surgeon’s likelihood of performing an SLNB is minor but the incremental effect of surgeon volume on a low percentage surgeon’s likelihood of performing SLNB is significant. We believe that this novel surgeon percentage measure further refines the surgeon volume measure and vice versa and warrants further investigation as a potential quality marker for surgical outcomes.

This study has several limitations. First, several tumor characteristics are typically known preoperatively and affect a surgeon’s decision to recommend axillary surgery. Information regarding the clinical tumor stage prior to surgery, hormone receptor and HER2-neu status were not routinely recorded in the 2003 state tumor registries; therefore, we could not conclusively determine which patients were appropriate candidates for SLNB. However, these factors are unlikely to vary systematically by surgeon characteristics. Second, misclassification of the outcome is possible since this axillary surgery variable was claims-based and defined by billing codes available in 2003. However, Medicare claims are generally accurate for breast cancer surgical billing codes.^48,49

Lastly, as this study involves Medicare patients treated in 2003, a period when SLNB was considered an option to ALND for women with early stage breast cancer^50,51 but was already widely adopted into clinical practice,^52–54 our findings may not be generalizable to younger and more contemporary populations. Surprisingly, only 6% of women in this elderly cohort did not undergo any axillary surgery and if axillary surgery was performed, older women were more likely to receive ALND rather than SLNB. Studies in more current cohorts of elderly women should be performed as these findings are in contrast to the National Comprehensive Cancer Network (NCCN) guidelines which consider both SLNB and ALND optional in elderly patients.^19,55,56 However, the fact that all patients were Medicare beneficiaries is a strength as we could identify surgeons based on claims information, calculate surgeon volume and percentage of breast cancer cases, and perform our analyses of various surgeon characteristics by linkage with other sources. An observational study of this magnitude affords us an opportunity to evaluate practice patterns in the community, where the majority of breast cancer treatment occurs.

Several previously reported measures of surgeon specialization in breast cancer have been based on subjective measures: local reputation, opinion, and surgeon self-reported proportion of their total practice devoted to breast cancer-related procedures.^{29–32,57,58} We believe that our novel objective measure, computed from health care claims, is likely to be a more reliable and sensitive surgeon-specific measure than these subjective measures and warrants further evaluation as a potential indicator of other aspects of breast cancer quality of care. Finally, this surgeon percentage/specialization measure may help refine studies that include surgeon volume in evaluating outcomes.

In summary, among this population-based, large cohort of older invasive breast cancer survivors, women operated on by surgeons who perform more breast cancer cases, focus their practice more on breast cancer, or are members of breast and surgical oncology societies were more likely to undergo an initial SLNB. Since women who undergo SLNB are at lower risk of developing lymphedema than those who undergo ALND, the importance of receiving initial treatment at centers that have experienced SLNB teams¹⁹ and surgeons who specialize in breast cancer must be emphasized to decrease lymphedema rates and its associated complications which significantly impact the quality of life and medical costs of survivors.^13,16,59,60 Future studies addressing provider specialization are needed to determine whether this novel objective surgeon percentage measure is an indicator of improved outcomes and quality of cancer and non-cancer care.