The proportion of patients undergoing SLNB is reported using the composite definition of both Surveillance, Epidemiology, and End Results (SEER) registrar coding and Medicare procedure claims (A), a definition derived only from Medicare claims (B), and a definition derived only from SEER registrar coding (C). Limit lines indicate 95% CIs. There was a significant difference in the use of SLNB for whites vs blacks regardless of the definition (all P < .001).
Cumulative incidence of lymphedema, defined by the presence of a diagnosis code indicating lymphedema, was calculated using the Kaplan-Meier method and stratified by race and type of axillary surgery. Differences across the strata were statistically significant (P < .001). ALND indicates axillary lymph node dissection; SLNB, sentinel lymph node biopsy.
eTable 1. Cohort selection criteria
eTable 2. Claims codes used to define type of axillary surgery, lymphedema, and other covariates
Black DM, Jiang J, Kuerer HM, Buchholz TA, Smith BD. Racial Disparities in Adoption of Axillary Sentinel Lymph Node Biopsy and Lymphedema Risk in Women With Breast Cancer. JAMA Surg. 2014;149(8):788–796. doi:10.1001/jamasurg.2014.23
Racial disparities exist in many aspects of breast cancer care. Sentinel lymph node biopsy (SLNB) was developed to replace axillary lymph node dissection (ALND) for staging early breast cancer to minimize complications. Racial disparities in the use of SLNB remain incompletely characterized, and their effect on lymphedema risk is not known.
To determine racial differences in SLNB use among patients with pathologically node-negative breast cancer during the period when SLNB became the preferred method for axillary staging as well as whether such differences affect lymphedema risk.
Design, Setting, and Participants
A retrospective study was conducted using the Surveillance, Epidemiology, and End Results–Medicare–linked database from 2002 through 2007 to identify cases of incident, nonmetastatic, pathologically node-negative breast cancer in women aged 66 years or older.
Main Outcomes and Measures
Sentinel lymph node biopsy use and 5-year cumulative incidence of lymphedema by race.
Of 31 274 women identified, 1767 (5.6%) were black, 27 856 (89.1%) were white, and 1651 (5.3%) were of other or unknown race. Sentinel lymph node biopsy was performed in 73.7% of white patients and 62.4% of black patients (P < .001). The use of SLNB increased by year for both black and white patients (P < .001); however, a fixed disparity of approximately 12 percentage points in SLNB use persisted through 2007. In adjusted analysis, black patients were significantly less likely than white patients to undergo SLNB (odds ratio, 0.67; 95% CI, 0.60-0.75; P < .001). Overall, the 5-year cumulative lymphedema risk was 8.2% in whites and 12.3% in blacks (hazard ratio [HR], 1.43; 95% CI, 1.23-1.67; P < .001). When stratified by type of axillary surgery, 5-year lymphedema risk was 6.8% in whites undergoing SLNB (HR, 1 [reference]), 8.8% in blacks undergoing SLNB (HR, 1.28; 95% CI, 1.02-1.60; P = .03), 12.2% in whites undergoing ALND (1.79; 1.63-1.96; P < .001), and 18.0% in blacks undergoing ALND (2.76; 2.25-3.39; P < .001).
Conclusions and Relevance
Although SLNB use increased in both black and white patients with pathologically node-negative breast cancer from 2002 through 2007, the rates of SLNB remained lower in black than white patients during this entire period by approximately 12 percentage points. This racial disparity in SLNB use contributed to racial disparities in lymphedema risk. Improvements in the dissemination of new techniques are needed to avoid disparities in breast cancer care and patient outcomes, particularly in disadvantaged groups.
Sentinel lymph node biopsy (SLNB) was developed to prevent major complications associated with axillary lymph node dissection (ALND) for surgical staging of clinically node-negative breast cancer. Several multicenter studies and randomized trials1- 5 have demonstrated that SLNB is an acceptable alternative for axillary staging and is associated with decreased complications and no significant differences in regional recurrence risk or survival compared with ALND. In 1998, the American Society of Breast Surgeons6 released its first guidelines for surgeons to appropriately incorporate the SLNB technique into their practice. In the National Comprehensive Cancer Network (NCCN) Clinical Practice Guidelines in Oncology (NCCN Guidelines) for Breast Cancer, version 1.2002, the NCCN panel and the American Society of Breast Surgeons recommended SLNB as an alternative to ALND in selected patients with clinically node-negative, clinical stage I breast cancer.7,8 Sentinel lymph node biopsy was then endorsed by other national and international consensus groups including the St. Gallen International Consensus Panel in 20039 and the American Society of Clinical Oncology in 2005.10 In the NCCN Guidelines for Breast Cancer, version 1.2007,11 the NCCN panel endorsed SLNB as the preferred axillary staging procedure in patients with clinically node-negative breast cancer.
Previous literature12- 14 has indicated that racial disparities in the use of SLNB were apparent early in the course of its development. However, the extent to which such disparities persisted once SLNB was fully embraced as the preferred method of axillary staging is unknown. Furthermore, the effect of disparities in SLNB use on patient outcomes, specifically the risk of lymphedema, has not previously been evaluated.
Accordingly, the purpose of this study was to determine (1) the pattern of adoption of SLNB as it became the preferred surgical staging procedure for node-negative breast cancer, (2) whether racial disparities existed in adoption of SLNB, and (3) the effect of racial disparities in SLNB use on the risk of lymphedema.
Using the Surveillance, Epidemiology, and End Results (SEER)–Medicare–linked database, we identified 51 063 women aged 66 years or older with local or regional invasive breast cancer diagnosed from January 1, 2002, to December 31, 2007, that was treated with lumpectomy or mastectomy. The women included had continuous fee-for-service insurance coverage (not health maintenance organization) spanning a window of 12 months before and after diagnosis. Of these, we selected an analytic cohort comprising 31 274 patients with pathologically node-negative disease and a documented axillary surgical procedure (eTable 1 in the Supplement]). This study was granted exempt status by The University of Texas MD Anderson Cancer Center Institutional Review Board.
The primary outcome was the type of axillary surgery performed by race. Axillary surgery was defined as SLNB (with or without completion ALND) or ALND alone. Patients were considered to have undergone SLNB if they had a Medicare claim for lymphatic mapping or a SEER code for SLNB (eTable 2 in the Supplement). Axillary lymph node dissection was similarly defined using claims data supplemented by SEER coding. Given concerns regarding the accuracy of SEER coding with respect to classification of nodal surgery,15 confirmatory analyses using only SEER coding or only Medicare coding to classify the type of axillary surgery were also conducted.
The secondary outcome was the 5-year cumulative incidence of lymphedema occurring at any time after diagnosis. Patients were considered to have lymphedema if they had at least 1 diagnosis code indicating postsurgical lymphedema.
Baseline patient covariates included age, race (white, black, and other/unknown), comorbidity, and year of diagnosis. Because black and white racial designations in Medicare data are highly accurate, our analyses focused primarily on these 2 groups.16 Comorbidity was determined using the Klabunde et al17 modification of the Charlson comorbidity score (0, no comorbidity; 1, mild comorbidity; and ≥2, moderate to severe comorbidity) using Medicare claims generated during the 12 months preceding the diagnosis. Tumor covariates included size, grade, and estrogen receptor status. Treatment covariates included type of breast surgery (lumpectomy vs mastectomy) determined using SEER data and Medicare claims, with the most extensive breast surgery within 6 months of diagnosis considered the definitive surgery, and receipt of radiotherapy determined using both SEER data and Medicare claims (eTable 2 in the Supplement). Sociodemographic covariates included Medicaid coverage, SEER registry, and area-level covariates such as median household income, median educational level in the patient’s census tract (or zip code if census tract data were unavailable), and density of general surgeons at the county level determined from the Area Resource File. Area-level covariates were divided into quartiles.
Unadjusted associations of SLNB with race and other covariates were evaluated using the χ2 test and Wilcoxon rank sum test as appropriate. The Cochran-Armitage test was used to evaluate trends over time. Multivariate logistic regression identified independent predictors of SLNB receipt. Covariates were selected on the basis of clinical significance or unadjusted association with the outcome at P ≤ .20, with iterative model refinement to minimize colinearity and remove nonsignificant covariates. Unknown covariate values were treated as separate strata using dummy variables. Given colinearities of black race with markers of socioeconomic deprivation (eg, Medicaid enrollment, median income, and educational level), models were created with and without these covariates to evaluate the contribution of these socioeconomic factors to any measured race effect. Goodness of fit was evaluated using the Hosmer-Lemeshow test.
The cumulative incidence of lymphedema after diagnosis was calculated using the Kaplan-Meier method, with patients censored at the time of loss of fee-for-service Medicare Part A or B coverage, end of the study period (December 31, 2009), or death. Unadjusted differences in lymphedema risk were evaluated using the log-rank test. Proportional hazards regression identified covariates associated with lymphedema risk, with covariate selection as indicated above. Two models were created. The first model sought to evaluate the association of race with lymphedema with risk adjustment for all relevant clinical pathologic factors. The second model used a 4-level dummy variable to account for each stratum of race and axillary surgery, thus enabling determination of the effects of the combinations of these 2 variables independently on lymphedema risk. For both models, the proportional hazards assumption was confirmed by inspection of the log(−log [survival]) curves.
All statistical tests were 2-sided. Analysis was conducted using SAS statistical software, version 9.3 (SAS Institute Inc).
The baseline characteristics of the 31 274 patients with pathologically node-negative breast cancer are summarized in Table 1. The median age was 74 years (interquartile range [IQR], 70-79), 5.6% of the patients were black, 61.8% of the patients underwent lumpectomy, and 72.6% underwent SLNB with or without completion ALND. Black patients were more likely than white patients to have moderate to severe comorbidity (24.7% vs 11.3%; P < .001). Black patients had larger tumors compared with white patients (median tumor size, 1.5 cm [IQR, 1.0-2.5] for blacks and 1.3 cm [IQR, 0.9-2.0] for whites; P < .001).
A total of 62.8% of white patients and 53.0% of black patients underwent lumpectomy (P < .001). Among patients with tumors 2 cm or smaller, 69.0% of white patients and 61.7% of black patients underwent lumpectomy (P < .001). Among patients with tumor size from 2.1 through 5.0 cm, 44.3% of white patients and 39.4% of black patients underwent lumpectomy (P = .04).
Among patients who underwent SLNB, the median number of axillary nodes removed was 2 (IQR, 1-5). Among patients who underwent ALND, the median number of axillary nodes removed was 11 (IQR, 7-15) (P < .001).
A total of 62.4% of black patients (1102 of 1767) underwent SLNB, compared with 73.7% (20 541 of 27 856) of white patients and 65.0% (1073 of 1651) of patients with other/unknown race/ethnicity (P < .001). The use of SLNB increased steadily from 2002 through 2007 for both races (P < .001) (Figure 1A). However, the absolute difference in the percentage use of SLNB between white and black patients remained statistically significant for each year analyzed, and there was no evidence to suggest that this disparity narrowed over the period covered by the study. Specifically, from 2002 to 2007, the rate of SLNB use increased from 45.4% to 73.1% in black patients and from 58.1% to 85.4% in white patients, for an absolute difference in the rate of SLNB use between the races of 12.3 percentage points for patients who received the diagnosis in 2007 (P < .001).
In confirmatory analyses, for patients whose axillary surgery was classified solely by Medicare claims (23 530), 70.7% of whites (15 694 of 22 184) and 59.9% of blacks (806 of 1346) were coded as undergoing SLNB (P < .001). For patients whose axillary surgery was classified solely by SEER coding (27 962), 68.7% of whites (18 082 of 26 314) and 57.3% of blacks (945 of 1648) were coded as undergoing SLNB (P < .001). Time trends in adoption of SLNB by race were nearly identical for these 2 subgroups (Figure 1B and C).
Rates of SLNB use differed geographically; the rates were highest in Seattle, Washington (88.8%), and lowest in Louisiana (52.5%) (P < .001) (Table 1). Patients who underwent ALND were more likely to have Medicaid coverage (1452 of 8558 [17.0%]) than were patients who underwent SLNB (1989 of 22 716 [8.8%]) (P < .001). Patients with the highest median area-level educational attainment had a higher rate of SLNB (83.7%) compared with those with the lowest median area-level educational attainment (61.2%) (P < .001). Higher median area-level income and higher surgeon density were also associated with greater likelihood of SLNB (both P < .001).
In adjusted analysis, black race was associated with lower odds of SLNB compared with white race (odds ratio [OR], 0.67; 95% CI, 0.60-0.75; P < .001 [Hosmer-Lemeshow, P = .17]). The addition of Medicaid coverage into the model slightly decreased the effect size of race (black vs white: OR, 0.75; 95% CI, 0.67-0.84; P < .001 [Hosmer-Lemeshow, P = .20]). Although other sociodemographic covariates were significant in unadjusted analyses, they resulted in a poor model fit when evaluated in a multivariate context and thus were not included. A later year of diagnosis was associated with increased odds of SLNB (2007 vs 2002: OR, 4.49; 95% CI, 4.06-4.97; P < .001). Mastectomy, compared with lumpectomy, was associated with lower odds of SLNB (OR, 0.19; 95% CI, 0.18-0.20; P < .001) (Table 2). Patients older than 80 years were also less likely to have SLNB.
Because the type of breast surgery was an independent predictor of SLNB use, a stratified analysis was performed to determine whether the disparity in use of SLNB between black and white women persisted regardless of the type of breast surgery. The odds of SLNB receipt were lower for blacks than whites both among patients undergoing lumpectomy (OR, 0.68; 95% CI, 0.57-0.81; P < .001) and among patients undergoing mastectomy (0.66; 0.57-0.77; P < .001).
Overall, the 5-year cumulative incidence of a diagnosis code for lymphedema was 8.2% in whites and 12.3% in blacks (hazard ratio [HR], 1.43; 95% CI, 1.23-1.67; P < .001). Other factors associated with increased lymphedema risk included ALND (HR, 1.76; 95% CI, 1.62-1.93; P < .001), obesity (HR, 1.72; 95% CI, 1.40-2.12; P < .001), receipt of radiotherapy (HR, 1.39; 95% CI, 1.22-1.59; P < .001), and mastectomy (HR, 1.60; 95% CI, 1.39-1.83; P < .001) (Table 3). When both race and type of axillary surgery were taken into account, the 5-year cumulative incidence of lymphedema was 6.8% in whites undergoing SLNB (HR, 1 [reference]), 8.8% in blacks undergoing SLNB (HR, 1.28; 95% CI, 1.02-1.60; P = .03), 12.2% in whites undergoing ALND (HR, 1.79; 95% CI, 1.63-1.96; P < .001), and 18.0% in blacks undergoing ALND (HR, 2.76; 95% CI, 2.25-3.39; P < .001) (Figure 2).
In this study of 31 274 patients with pathologically node-negative breast cancer identified in the SEER-Medicare–linked database, SLNB use increased in all patients from 2002 to 2007 as SLNB transitioned from an acceptable alternative to ALND to the preferred method of axillary staging. However, throughout this period, black patients had an absolute 12% lower likelihood of SLNB receipt compared with white patients. This disparity was consistent throughout all of the years of the study and indicated that adoption of SLNB in blacks lagged approximately 2 to 3 years behind adoption of SLNB in whites. With adjustment for clinical pathologic factors, the odds of SLNB remained one-third lower in blacks than in whites. Furthermore, our analysis illustrates the racial disparities in outcomes that result from this racial disparity in cancer treatment. Specifically, 18.0% of black patients undergoing ALND had a diagnosis code for lymphedema within 5 years of cancer diagnosis compared with 12.2% of white patients undergoing ALND. In contrast, when black patients underwent SLNB, the risk of lymphedema within 5 years of diagnosis was considerably lower (8.8%) and was similar to the risk in white patients who underwent SLNB (6.8%).
We found that several socioeconomic and geographic factors were associated with lower SLNB use including receipt of Medicaid, residence in an area with a lower median educational level or income, and residence in an area with lower surgeon density; these findings are consistent with those of prior studies. Keating et al18 showed that black patients with breast cancer were less likely to be treated at specialty hospitals, more likely to be treated at high-volume hospitals, and less likely to receive definitive primary breast cancer treatment. Using the 2000-2002 SEER-Medicare–linked database, Carpenter et al12 evaluated how the new SLNB technique was disseminated in various types of hospital settings. They found that patients receiving care at hospitals affiliated with National Cancer Institute cooperative groups and the American College of Surgeons Oncology Group were more likely to undergo SLNB and that black patients were half as likely to undergo SLNB (OR, 0.54). The small patient cohort limited analysis, but there was a trend toward a higher use of SLNB if black patients were treated at cooperative group–affiliated hospitals. Using the National Cancer Database, Chen et al13 evaluated trends in the use of SLNB as consensus statements were being published by national organizations from 1998 to 2005. Black and Hispanic patients had a lower receipt of SLNB, and receipt of SLNB was dependent on socioeconomic status and insurance status. The disparity in SLNB use between whites and minorities was more prominent in 2005 than in 1998, and SLNB use was more strongly associated with socioeconomic and insurance status in 2005 than in 1998.
We believe that our study is the first to show that the lower rate of SLNB use among black patients has a negative effect on patient outcomes. Specifically, black patients undergoing ALND not only experienced a higher risk of lymphedema compared with black patients undergoing SLNB but also experienced a higher risk of lymphedema compared with white patients undergoing ALND. Increased use of ALND among blacks is particularly problematic, as this population, potentially because of higher prevalence of obesity or other comorbid diseases, is particularly sensitive to the deleterious effects of ALND on lymphedema risk. Lymphedema can be a chronic complication significantly affecting breast cancer patients’ quality of life and places a substantial economic burden on the health care system. Lymphedema rates vary in the literature,19- 24 ranging from 3% to 30% depending on the extent of axillary surgery and adjuvant treatment. Morbidity results from the National Surgical Adjuvant Breast, Bowel Project B-32 trial25 showed that, at both short term and 3 years, patients who underwent ALND had worse outcomes in terms of arm mobility, lymphedema, and paresthesias compared with patients who underwent SLNB. At 36 months, 14% of the patients in the ALND group and 8% of those in the SLNB group had a difference in arm volume greater than 10%. Using the MarketScan Health and Productivity Management database with 2-year follow-up, Shih et al26 found that breast cancer–related lymphedema resulted in an adjusted increase in medical costs of more than $14 000. In that study, ALND was associated with a 6-fold increase in lymphedema, and patients with lymphedema had double the rate of lymphangitis or cellulitis that was observed in patients without lymphedema.
There are several limitations of this study. The SEER-Medicare–linked database does not include information on patients’ clinical presentation. In some cases, the decision to perform ALND may have been made because the patient presented with a palpable lymph node. This may have increased the ALND rate in black patients because they are known27- 29 to more often present with advanced disease. Another limitation is that tumor registries have differing abilities to accurately capture the type of axillary surgery. There is concern regarding undercoding of SLNB by SEER tumor registrars.15 To address this concern, we used a combination of SEER data and Medicare billing claims to determine receipt of SLNB vs ALND. The racial disparity in rates of SLNB remained stable when we analyzed just SEER data or just Medicare data, suggesting the validity of our study findings. Furthermore, the fact that our measured associations between ALND and increased risk of lymphedema are consistent with effect sizes from randomized trials provides additional internal validation of our surgical classification and outcomes.
Given that the present study focused on elderly patients, the usefulness of any type of axillary surgery in this population may be questioned. A primary purpose of SLNB or ALND is to identify patients with lymph node involvement who may benefit from adjuvant chemotherapy. However, many elderly patients may choose to forego chemotherapy even if they were found to have node-positive disease because of comorbid illness, limited life expectancy, or the possibly attenuated benefits of chemotherapy in older women. As a result, this purpose of axillary surgery may not be relevant to certain elderly women. Nevertheless, for many of this population, axillary surgery may still be favored, not only to inform the prognosis and selection of adjuvant systemic therapy but also to optimize regional control of the disease in the axilla30 and guide decisions regarding radiotherapy indications and target volumes.
This study demonstrates that racial disparities in SLNB use persisted even as SLNB became the preferred method of axillary staging and that these disparities in surgical treatment contributed to disparities in lymphedema risk. These findings emphasize that not all newly developed techniques in breast cancer care are made available in a timely fashion to all eligible patients. As new techniques continue to be developed, focused educational interventions must be developed to ensure that these techniques reach historically disadvantaged patients to avoid disparities in care. More contemporary data will be needed to determine whether this disparity still exists in black patients and other at-risk minorities.
Accepted for Publication: October 29, 2013.
Corresponding Author: Dalliah M. Black, MD, Department of Surgical Oncology, The University of Texas MD Anderson Cancer Center, 1400 Pressler St, Unit 1484, Houston, TX 77030 (firstname.lastname@example.org).
Published Online: June 18, 2014. doi:10.1001/jamasurg.2014.23.
Author Contributions: Drs Black and Smith had full access to all of 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: Black, Buchholz, Smith.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Black, Jiang, Smith.
Critical revision of the manuscript for important intellectual content: Black, Kuerer, Buchholz, Smith.
Statistical analysis: Jiang.
Administrative, technical, or material support: Black, Buchholz, Smith.
Study supervision: Black, Buchholz.
Conflict of Interest Disclosures: None reported.
Additional Contributions: We acknowledge the efforts of the Applied Research Program, National Cancer Institute; the Office of Research, Development and Information, Centers for Medicare and Medicaid Services; Information Management Services Inc; and the Surveillance, Epidemiology, and End Results (SEER) program tumor registries in the creation of the SEER-Medicare–linked database. Stephanie Deming, ELS, of MD Anderson Scientific Publications, provided editorial assistance. No financial compensation was given.