Racial and socioeconomic disparities in adjuvant chemotherapy for older women with lymph node-positive, operable breast cancer

Authors

  • Alessia Bhargava,

    1. Department of Economics, Yale University, New Haven, Connecticut
    Search for more papers by this author
    • Alessia Bhargava is an economics major student expecting a BA in economics from Yale University in 2010 and participated in the summer research project at University of Texas School of Public Health.

  • Xianglin L. Du MD, PhD

    Corresponding author
    1. School of Public Health, University of Texas Health Science Center at Houston, Texas
    • University of Texas School of Public Health, RAS-E631, 1200 Herman Pressler Drive, Houston, TX 77030
    Search for more papers by this author
    • Fax: (713) 500 9264


Abstract

BACKGROUND:

Consistent with findings from clinical trials, a recent population-based study indicated that adjuvant chemotherapy for lymph node-positive, operable breast cancer is effective at improving survival in older women, specifically those ages 65 years to 69 years; however, to the authors' knowledge, no conclusion has been reached about the relative benefit of chemotherapy for women aged ≥70 years, probably because of small number of patients. However, little is known about racial and socioeconomic disparities in adjuvant chemotherapy for breast cancer among older women.

METHODS:

This study included 14,177 white women and 1277 black women aged ≥65 years who were diagnosed with operable breast cancer (stage II-IIIA) and positive lymph nodes between 1991 and 2002. These women were identified from the Surveillance, Epidemiology, and End Results and Medicare-linked database. Multivariate logistic regression was used to compute the odds ratios of receiving chemotherapy among black women compared with white women, and the causal step approach was used to test whether census tract-level poverty mediated racial disparities.

RESULTS:

Interaction terms analyses indicated that regressions should be stratified by age group. In the group ages 65 years to 69 years, the adjusted odds ratio of receiving chemotherapy were lower for black women than for white women (odds ratio, 0.85; 95% confidence interval, 0.57-0.97). Poverty mediated the association between chemotherapy and race in this age group. No racial or socioeconomic disparities were observed among women aged ≥70 years.

CONCLUSIONS:

This study documented racial disparities in adjuvant chemotherapy that were mediated by poverty in women ages 65 years to 69 years, an age group for which there is clear evidence for the efficacy of chemotherapy, but no disparities were observed among women aged ≥70 years. The authors concluded that it is important to work toward reducing treatment disparities among older women. Cancer 2009. © 2009 American Cancer Society.

A disproportionate number of breast cancer deaths occur in racial and ethnic minorities.1, 2 In particular, African-American women have greater breast cancer recurrence rates and higher mortality rates than white women.3-5 The causes for these disparities include a more advanced stage of disease at diagnosis in African-American women, racial differences in tumor characteristics and comorbid conditions,4 and racial differences in treatment.1-2, 5-8 Several studies have indicated that adjusting for tumor characteristics and treatment eliminated or greatly attenuated racial differences in survival2, 9, 10; however, others observed a persistent disadvantage in survival for African-American women.11 In addition, several studies have investigated racial differences in breast cancer treatment after controlling for disease severity, prognostic factors, and comorbidities.3, 5-8 The findings about racial differences in adjuvant therapy in women with early breast cancer are inconsistent, depending on the type of adjuvant therapy and the stage at diagnosis.12-15 Several studies have indicated that, even when African-American women received chemotherapy, they received suboptimal treatment.16-19

Older women are affected disproportionately by breast cancer. Greater than 50% of breast cancer deaths occur among women aged ≥65 years, and the median age at diagnosis is 61 years.20 However, older women with breast cancer are more likely to be under-treated. Increasing age is associated with not receiving effective cancer therapies, even after controlling for comorbidity and prognostic factors.13, 21-23 Recent research using a population-based, observational cohort24 has demonstrated that adjuvant chemotherapy for lymph node-positive, operable breast cancer is effective at improving survival in older women, specifically those ages 65 years to 69 years, although no conclusion has been reached about the relative benefit of chemotherapy for women aged ≥70 years, probably because of the small number of patients. These results are consistent with findings from clinical trials.25 However, very little is known about racial disparities in adjuvant chemotherapy for breast cancer treatment among older women. Several studies have reported breast cancer chemotherapy in older women,6, 13, 21 but few have investigated racial disparities.12, 22, 26 Mandelblatt and colleagues12, 26 observed that, among Medicare beneficiaries aged ≥67 years with local breast carcinoma, African Americans were less likely to receive radiotherapy after breast-conservation surgery. In an article that focused on the relation between chemotherapy and age among Medicare breast cancer patients for all stages, Du and Goodwin22 noticed no racial differences in the use of chemotherapy. A recent article also indicated that there were no racial differences in the receipt of chemotherapy among breast cancer patients ages 66 years to 85 years.27

Given the recent findings in support of the effectiveness of chemotherapy among older women ages 65 years to 69 years with early (stage II and IIIA) breast cancer24 and the persistent disadvantage in breast cancer-specific mortality among older African-American women with early stage breast cancer compared with white women,2 it is important to investigate further the racial disparities in adjuvant chemotherapy among older women with early stage breast cancer. In the current study, we investigated racial disparities in the receipt adjuvant chemotherapy by patients with lymph node-positive, operable breast cancer in a much larger and recently updated, population-based cohort of women aged ≥65 years from 1991 through 2002 who had up to 16 years of follow-up. We studied this population because adjuvant chemotherapy is strongly recommended based on guidelines for patients with lymph node-positive, operable breast cancer ages 65 years to 69 years.28, 29 The guidelines for chemotherapy in women aged >70 years are less clear.28, 29 Furthermore, we investigated whether racial disparities were mediated by socioeconomic status (SES). We used the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) cancer registry data linked to Medicare data in the 16 areas that account for >25% of the total US population. Because of the large sample size, we were able to investigate racial differences in the receipt of adjuvant chemotherapy among 4 subgroups of older women ages 65 years to 69 years, 70 years to 74 years, 75 years to 79 years, and ≥80 years.

MATERIALS AND METHODS

Data Sources

For the current study we used the SEER-Medicare linked database, which is supported by the National Cancer Institute and covers the 16 SEER areas. Because of the small number of patients, we combined rural Georgia with Atlanta and combined 4 registries in California (San Francisco, Los Angeles, San Jose, and the rest of California), making the 16 areas into 12 geographic areas. The Medicare program provides payment for >97% of medical care for individuals aged ≥65 years.

Study Population and Sample Size Needed

The study population consisted of 15,598 African-American and white women aged ≥65 years who were diagnosed with operable breast cancer (stage II or IIIA) and who also had positive lymph nodes during the period from 1991 to 2002. These women were identified from the 12 SEER areas. Women in the study population had full coverage with both Medicare Parts A and B and were not enrolled with Health Maintenance Organizations during the year of diagnosis. The study population did not include women who had other primary cancers. Women who had missing SES variables were excluded; thus, 15,454 women (14,177 white women and 1277 African-American women) were included in the final analyses. On the basis of a sample-size equation for comparing 2 proportions,30 395 women would be needed in each racial group to have 80% power to detect a prevalence difference of 6% in receiving chemotherapy with a P value < .05.

Study Variables

Chemotherapy

Confirmation that chemotherapy was received within 6 months of breast cancer diagnosis was obtained from Medicare claims. The identification of chemotherapy through Medicare claims has been described previously described and reportedly is reliable.31

Comorbidity Index

Comorbidity was ascertained from Medicare claims by identifying diagnoses made or procedures undergone 1 year before and 1 month after the diagnosis of breast cancer, including heart disease, diabetes, and dementia but excluding any malignant tumor. Comorbid diseases within 1 month of diagnosis were included because they are likely to influence the choice of chemotherapy (as discussed in previous studies2, 5).

Socioeconomic Status

SES was measured by poverty rate at the census tract level from the 1990 Census and the 2000 Census available in the SEER-Medicare linked data. Data from the 1990 Census were linked to women who were diagnosed during 1991 through 1999, and data from the 2000 Census were linked to women who were diagnosed in 2000 or later. The census tract-level poverty rates, defined as the percentage of individuals living below the federal poverty line, were categorized into quartiles.

Other Characteristics

Patient and tumor characteristics, such as age, race, marital status, tumor stage, tumor size, tumor grade, hormone receptor status, surgery and radiation, geographic area (12 SEER areas), and year of diagnosis, were available from the SEER data. Age was categorized into 4 subgroups (ages 65-69 years, 70-74 years, 75-79 years, and ≥80 years).

Analyses

Differences in the distribution of baseline characteristics between African-American women and white women were tested using the chi-square statistic. Differences in the receipt of chemotherapy by race stratified by age subgroup also were tested using the chi-square statistic. The odds ratios (OR) of receiving chemotherapy for African-American women compared with white women were generated from multivariate logistic regression analyses. Three models with increasing numbers of covariates were estimated. The first model included only race and age. The second model was adjusted further for tumor biologic factors and treatment (tumor stage, tumor size, tumor grade, hormone receptor status, surgery, and radiation) and other patient characteristics (marital status, comorbidity index, geographic area, and year of diagnosis). The third model investigated whether poverty mediated the relation between race and chemotherapy by adding the poverty variable to all covariates from the second model. Mediation was tested using the causal step approach.32, 33 The test of whether poverty mediated the relation of race to chemotherapy included 3 steps: 1) a regression of chemotherapy adjusted for race, 2) a regression of chemotherapy adjusted for race and poverty, and 3) a regression of race adjusted for poverty. The analysis indicated that poverty mediated the association of race on chemotherapy if, in the first regression, race was statistically significant; if, in the second regression, race was not significant and poverty was significant; and if, in the third regression, poverty was significant.

We tested whether the associations between chemotherapy and race varied by age group by examining interactions between race and age. The statistical significance of interaction terms that were added to the logistic regression of chemotherapy adjusted for race and age groups indicated that separate regressions by age group were appropriate. To investigate racial disparities in chemotherapy within the different age groups (ages 65-69 years, 70-74 years, 75-79 years, and ≥80 years), all multivariate logistic regression analyses were repeated for stratifications by age group. Parameter estimates were obtained using the maximum-likelihood method of estimation, and ORs with corresponding 95% confidence limits and P values were reported. All tests of significance were 2 sided, and P values <.05 were considered statistically significant. All analyses were completed using Stata/IC 10 software (Stata Corp., College Station, Tex).34

RESULTS

Table 1 compares the use of chemotherapy, tumor biologic factors, and other patient characteristics between white women and black women. When tumor, treatment, and patient characteristics were not controlled for, white women and black women had similar rates of chemotherapy. Racial differences were significant for all other characteristics except year of diagnosis. Compared with white women, black women were more likely to be younger, less likely to be married, and more likely to live in census tracts with higher poverty rates. They were more likely have stage IIIA (vs stage II) disease; they had larger tumors, more positive lymph nodes, and less favorable tumor grade; they were less likely to have positive hormone receptor status and to undergo breast-conserving surgery with radiation; and they had higher comorbidity scores.

Table 1. Comparison of Characteristics of White Versus Black Older Women With Lymph Node-positive, Operable Breast Cancer
CharacteristicAllWhiteBlackP*
No.%No.%No.%
  • BCS indicates breast-conserving surgery; SEER, Surveillance, Epidemiology, and End Results; SES, socioeconomic status.

  • *

    Chi-square test.

  • No group with <5 patients was reported, as required by the data-user agreement.

Total15454100141771001277100 
Chemotherapy
 No chemotherapy965662.5884562.481163.5 
 Chemotherapy579837.5533237.646636.5.43
Age group, y
 65-69376924.434072436228.3 
 70-74384124.935392530223.6 
 75-79363823.5332923.530924.2 
 ≥80420627.2390227.530423.8<.01
Marital status
 Married636841.2604942.731925 
 Unmarried860755.7770154.390670.9 
 Unknown4793.14273524.1<.01
Tumor stage
 II1382689.51272089.7110686.6 
 IIIA162810.5145710.317113.4<.01
Tumor size, cm
 <18015.27535.3483.8 
 1-<2381124.7359125.322017.2 
 2-<3477630.9441231.136428.5 
 3-<4257916.7233916.524018.8 
 ≥4303419.6267018.836428.5 
 Unknown4532.94122.9413.2<.01
No. of positive lymph nodes
 1491231.845423237029 
 2-3325821.129842127421.5 
 4-9267417.3241917.125520 
 10-5115169.813819.713510.6 
 Unknown but positive309420285120.124319.03
Tumor grade
 Well differentiated163510.6155010.9856.7 
 Moderately differentiated577737.4537037.940731.9 
 Poorly differentiated577437.4518836.658645.9 
 Unknown226814.7206914.619915.6<.01
Hormone receptor status
 Positive1077769.7100597171856.2 
 Negative213813.8187613.226220.5 
 Unknown253916.4224215.829723.3<.01
Surgery and radiation
 BCS only233815.1210614.923218.2 
 BCS with radiation340122317722.422417.5 
 Mastectomy alone600938.9551938.949038.4 
 Mastectomy with radiation370624337523.833125.9<.01
Comorbidity score
 0995364.4929065.566351.9 
 1345622.431262233025.8 
 ≥2204513.2176112.428422.2<.01
SEER area
 Connecticut179411.6173012.2645.0 
 Detroit222514.4180612.741932.8 
 Hawaii590.4570.4<50.2 
 Iowa178811.6176712.5211.6 
 New Mexico4933.24863.470.5 
 Seattle12778.312608.9171.3 
 Utah6754.46744.8<50.1 
 Atlanta/Rural Georgia7705.06014.216913.2 
 Kentucky5723.75363.8362.8 
 Louisiana5243.43962.812810 
 New Jersey11067.210027.11048.1 
 California417127386227.230924.2<.01
Year of diagnosis
 19918935.88435.9503.9 
 199210486.89556.7937.3 
 199310046.59266.5786.1 
 199410136.69336.6806.3 
 199510006.59266.5745.8 
 19968955.88105.7856.7 
 19979776.38986.3796.2 
 19989796.38906.3897.0 
 199910556.89616.8947.4 
 2000207613.4188313.319315.1 
 2001224114.5206114.518014.1 
 2002227314.7209114.718214.3.11
SES quartile
 First (<3.9%; high)385825379326.8655.1 
 Second (3.9%-7.4%)386225375826.51048.1 
 Third (7.4%-13.5%)386925368226.018714.6 
 Fourth (>13.5%; low)386525294420.892172.1<.01

To investigate racial differences in the receipt of chemotherapy by age, Table 2 compares chemotherapy in white women and black women within age groups. Black women were less likely to receive chemotherapy in the group ages 65 years to 69 years, but there were no racial differences in the older age groups. To further investigate the relation between race and age in chemotherapy, we estimated a logistic regression of chemotherapy based on race and age group. The ORs for race (0.85; 95% confidence interval [CI], 0.57-0.97; P = .02) were statistically significant. Next, we investigated interactions between race and age by adding interaction terms to the regression. Interaction terms were significant (P = .03; chi-square test of joint significance), indicating that the associations between chemotherapy and race varied by age group. Therefore, we estimated logistic regressions stratified by age group for chemotherapy adjusted for race while controlling for tumor, treatment, and patient characteristics (Table 3).

Table 2. Comparison of White and Black Older Women With Lymph Node-positive, Operable Breast Cancer Who Did and Did Not Receive Adjuvant Chemotherapy, by Age Group
Age Group/RaceTotal No. of PatientsReceived ChemotherapyP*
No. of Patients%
  • *

    Chi-square test.

Ages 65-69 y
 White3407224465.9 
 Black36220456.4<.01
 Total3769244865.0 
Ages 70-74 y
 White3539166247.0.40
 Black30213444.4 
 Total3841179646.8 
Ages 74-79 y
 White3329102730.9.75
 Black3099229.8 
 Total3638111930.8 
Aged ≥80 y
 White390239910.2.38
 Black3043611.8 
 Total420643510.3 
Table 3. Multivariate Analysis of Receiving Adjuvant Chemotherapy in Blacks Compared With Whites Stratified by Age and Adjusted for Tumor Characteristics, Patient Characteristics, and Socioeconomic Status
VariableAges 65-69 y, n = 3769Ages 70-74 y, n = 3841Ages 75-79 y, n = 3638Aged ≥80 y, n = 4206
OR (95% CI)POR (95% CI)POR (95% CI)POR (95% CI)P
  • OR indicates odds ratio; CI, confidence interval; BCS, breast-conserving surgery; SES, socioeconomic status.

  • *

    Model 1 was adjusted for marital status, tumor stage, tumor size, number of positive lymph nodes, tumor grade, hormone receptor status, surgery and radiation, and comorbidity scores plus Surveillance, Epidemiology, and End Results area and year of diagnosis (not shown). Model 2 also was adjusted for SES (poverty) in addition to the factors in Model 1.

  • The P value shown is for a chi-square trend test of significance for SES (census-tract level poverty) of quartiles 2, 3, and 4.

Model 1*
 Race
  WhiteReference Reference Reference Reference 
  Black0.68 (0.52-0.90).010.75 (0.56-1.00).050.82 (0.60-1.12).210.93 (0.61-1.42).75
Model 2*
 Race
  WhiteReference Reference Reference Reference 
  Black0.78 (0.58-1.04).090.84 (0.62-1.15).280.79 (0.57-1.09).160.93 (0.59-1.44).73
 Marital status
  MarriedReference Reference Reference Reference 
  Unmarried0.88 (0.75-1.03).120.79 (0.68-0.91)<.010.86 (0.73-1.02).080.65 (0.51-0.83)<.01
  Unknown0.64 (0.41-1.00).050.47 (0.28-0.77)<.011.26 (0.76-2.09).380.63 (0.33-1.21).17
 Tumor stage
  IIReference Reference Reference Reference 
  IIIA1.41 (1.02-1.96).040.87 (0.65-1.17).371.12 (0.84-1.49).430.95 (0.66-1.37).79
 Tumor size, cm
  <1Reference Reference Reference Reference 
  1-<21.35 (1.00-1.81).051.04 (0.77-1.41).790.82 (0.54-1.22).320.79 (0.43-1.46).45
  2-<31.49 (1.10-2.03).011.05 (0.77-1.43).771.01 (0.67-1.51).980.58 (0.32-1.07).08
  3-<41.46 (1.03-2.07).031.19 (0.85-1.68).310.97 (0.63-1.49).890.65 (0.35-1.22).18
  ≥41.48 (1.02-2.13).041.22 (0.85-1.75).281.06 (0.69-1.64).790.67 (0.36-1.27).22
  Unknown1.66 (1.03-2.68).041.84 (1.12-3.01).021.33 (0.74-2.38).341.01 (0.44-2.32).98
 No. of positive lymph nodes
  1Reference Reference Reference Reference 
  2-31.92 (1.57-2.34)<.011.97 (1.63-2.40)<.011.73 (1.39-2.16)<.010.93 (0.67-1.30).69
  4-92.65 (2.09-3.36)<.012.88 (2.31-3.58)<.012.84 (2.24-3.61)<.011.71 (1.25-2.33)<.01
  10-513.99 (2.89-5.50)<.013.70 (2.79-4.91)<.013.50 (2.65-4.64)<.012.36 (1.63-3.43)<.01
  Unknown but positive0.31 (0.21-0.45)<.010.28 (0.19-0.41)<.010.27 (0.18-0.39)<.010.33 (0.21-0.50)<.01
 Tumor grade
  Well differentiatedReference Reference Reference Reference 
  Moderately differentiated0.91 (0.70-1.17).451.33 (1.02-1.70).031.30 (0.97-1.75).081.06 (0.68-1.65).80
  Poorly differentiated1.19 (0.91-1.57).201.87 (1.44-2.42)<.011.74 (1.29-2.36)<.011.41 (0.91-2.21).13
  Unknown0.90 (0.66-1.23).501.53 (1.13-2.07).011.39 (0.97-1.98).070.87 (0.51-1.47).60
 Hormone receptor status
  PositiveReference Reference Reference Reference 
  Negative3.00 (2.31-3.91)<.013.70 (2.90-4.72)<.013.45 (2.72-4.38)<.013.77 (2.89-4.92)<.01
  Unknown1.10 (0.89-1.38).331.14 (0.92-1.40).230.91 (0.72-1.15).431.25 (0.91-1.71).16
 Surgery and radiation
  BCS aloneReference Reference Reference Reference 
  BCS with radiation0.72 (0.54-0.94).020.57 (0.43-0.75)<.010.91 (0.67-1.25).571.33 (0.87-2.04).19
  Mastectomy alone0.75 (0.57-0.99).040.48 (0.37-0.63)<.010.69 (0.51-0.93).010.95 (0.65-1.38).78
  Mastectomy with radiation1.31 (0.97-1.78).080.94 (0.71-1.26).701.27 (0.92-1.73).142.15 (1.44-3.21)<.01
 Comorbidity score
  0Reference Reference Reference Reference 
  10.77 (0.63-0.94).010.72 (0.60-0.87)<.010.82 (0.68-0.99).040.87 (0.67-1.13).29
  ≥20.31 (0.24-0.41)<.010.44 (0.34-0.57)<.010.42 (0.33-0.55)<.010.63 (0.45-0.88).01
 SES quartile
  First [<3.9%; high]Reference Reference Reference Reference 
  Second [3.9%-7.4%]0.82 (0.65-1.02).080.99 (0.80-1.21).891.15 (0.92-1.45).220.94 (0.69-1.29).69
  Third [7.4%-13.5%]0.76 (0.60-0.96).020.98 (0.79-1.22).851.08 (0.84-1.38).560.91 (0.65-1.26).56
  Fourth [>13.5%; low]0.69 (0.54-0.88)<.010.79 (0.62-1.00).051.15 (0.88-1.49).300.97 (0.69-1.37).88
  P for SES .02 .16 .61 .94

Two models are presented in Table 3. Model-1 presents the OR of receiving chemotherapy for black women compared with white women adjusted for patient and tumor characteristics except for SES. Model 2 presents the OR of receiving chemotherapy for black women compared with white women by adding SES to the model in addition to other factors from Model 1. Model 2 also includes the effects of other factors on the receipt of chemotherapy. After adjusting for patient and tumor characteristics, black women were significantly less likely than white women to receive chemotherapy in the group ages 65 years to 69 years (OR, 0.68; 95% CI, 0.52-0.90; P = .01), and black women were marginally less likely to receive chemotherapy in the group ages 70 years to 74 years (OR, 0.75; 95% CI, 0.56-1.00; P = .05). There were no racial differences in chemotherapy the group ages 75 years to 79 years or in the group aged ≥80 years.

However, after also adjusting for SES (poverty variable) in the model, racial disparity in the receipt of chemotherapy no longer was significant in patients ages 65 years to 69 years, indicating that poverty may explain the racial differences in chemotherapy for this age group and is a candidate mediator between chemotherapy and race. By using the causal step approach, we observed that poverty indeed mediated the correlation of race with chemotherapy, because: 1) race was significant in the regression of chemotherapy adjusted for race (Model 1 in Table 3); 2) race was not significant and poverty was significant in the regression of chemotherapy adjusted for both race and poverty (Model 2 in Table 3); and 3) poverty was significant in the regression of race adjusted for poverty (P < .05; chi-square test of significance for poverty quartiles). In Table 3, among patients ages 70 years to 74 years, race no longer was marginally significant after adding poverty to the model, and poverty was not significant. In the groups ages 75 years to79 years and aged ≥80 years, neither race nor poverty was significant. In addition, racial disparity in the receipt of chemotherapy did not differ significantly in the early period (1991-1997) or in the later period (1998-2002). In both periods, the ORs by race were not significant in patients ages 65 years to 69 years (OR, 0.85 [95% CI, 0.54-1.32] vs 0.71 [95% CI, 0.47-1.06]) and patients aged ≥70 years (OR, 0.84 [95% CI, 0.60-1.17] vs 0.99 [95% CI, 0.78-1.24]), although the magnitude differed slightly with overlapping confidence intervals. However, this stratified analysis by period had smaller numbers of patients, which made the estimates relatively unstable and made the confidence intervals wider. Although the sample size for comparison between blacks and whites on the receipt of chemotherapy was sufficient overall (395 women for each group of blacks and whites were required), subgroups analyses had smaller numbers and less power to detect differences. Furthermore, we repeated all analyses with the population of 15,598 women, which included women who had missing SES variables, and the magnitude and direction of the findings were similar.

Model 2 in Table 3 also presents the effect of other factors on the receipt of chemotherapy by age group. The higher likelihood of receiving chemotherapy generally was associated with married women, higher tumor stage, more positive lymph nodes, poor tumor grade, negative hormone receptor status, and lower comorbidity scores among women ages 65 years to 69 years. In the older age groups, the general patterns were similar; however, the results were less stable, because the numbers of patients who received chemotherapy among older women (particularly those aged >80 years) were small.

DISCUSSION

In women who had lymph node-positive, operable breast cancer, we observed racial differences in adjuvant chemotherapy in women ages 65 years to 69 years, marginal differences in women aged 70 years to 74 and no differences in women aged over 74. Given the strong evidence about the effectiveness of adjuvant chemotherapy and current treatment guidelines recommending adjuvant chemotherapy in this population of women aged 65 years to 69, our findings of racial differences in this age group are of concern. It is interesting that racial disparities narrowed with age and that no racial differences were found in women aged 75 and over for which the evidence of efficacy and the treatment guidelines are less clear.25, 29

Some previous studies failed to identify racial differences in the receipt of adjuvant chemotherapy among older women.12, 22, 27 However, none of those studies specifically investigated adjuvant chemotherapy in the group ages 65 years to 69 years. Analyzing the receipt of adjuvant chemotherapy among all women aged ≥65 years can be misleading, because evidence of efficacy and treatment guidelines varies by age. Our results indicate that there are interactions between race and age in the receipt of adjuvant chemotherapy. It is well known that, as women age, they are less likely to receive adjuvant chemotherapy.13, 21-23 In fact, evidence of efficacy and treatment guidelines are less clear for women aged ≥70 years, because there is a lack of women in that age group who participate in clinical trials.25, 29 However, the data from the current study indicate that, although the receipt of adjuvant chemotherapy decreased with age in this large nationwide, population-based cohort of women with lymph node-positive, operable breast cancer, a substantial portion of older women in the population did receive chemotherapy (47% of women ages 70-74 years, 30% of women ages 75-79 years, and 10% of women aged ≥80 years). Therefore, there is a need to improve enrollment of breast cancer patients aged ≥70 years in clinical trials investigating the effectiveness of adjuvant chemotherapy to develop clearer guidelines for older women.

The relation between race and treatment may not be explored fully without understanding the role of SES.11, 35 Because of the strong correlation between minority status and SES, race often is a surrogate measure for SES in health studies. This leads some investigators to conclude inappropriately that there are racial differences in outcomes.2 We observed that racial differences in the receipt of adjuvant chemotherapy in the group ages 65 years to 69 years were mediated by census tract poverty rates. These results demonstrate that socioeconomic factors like poverty that are associated with race may be responsible for the disparities between African-American women and white women in the receipt of adjuvant chemotherapy for this age group. Even in our population of Medicare-insured women, poverty can influence treatment decisions through access and cost of care because of the difficulty poorer women may have in paying out-of-pocket costs and copayments. Quality of care also may be worse in poorer neighborhoods. Poverty has been associated with limited access to care, limited treatment options, and difficulty with treatment compliance.36 Women living in poorer neighborhoods also may be more likely to have poorer health, more comorbid conditions, worse employment, and different health beliefs. In addition, married women are more likely to receive chemotherapy, possibly because of higher motivation for chemotherapy and more family support, which includes transportation for visiting outpatient clinics to receive chemotherapy every few weeks for several months. All of these factors may have an impact on the receipt of adjuvant chemotherapy.15

Limitations of the current study include the availability of SES indicators only at the census tract level. Although living in disadvantaged neighborhood is an independent predictor of healthcare behavior, the lack of individual-level SES measures may introduce residual confounding. Lack of information on provider characteristics, patient and physician treatment preferences, and hospitals or sites of care is another limitation. In addition, there was no information on whether the quality of claims for chemotherapy in Medicare data varies by blacks and whites, which may have affected racial differences in the receipt of chemotherapy. However, this was not supported by our previous study, which demonstrated that the quality of Medicare claims for radiation therapy did not vary significantly between blacks and whites.37 Furthermore, although sample size for comparing 2 proportions of chemotherapy between blacks and whites were sufficient at 80% power with P < .05 overall or even in stratified analysis by ages 65 years to 69 years and ≥70 years, the sample size became smaller in further stratifications by ages 70 years to 74 years, 75 years to 79 years, and ≥80 years. Therefore, the estimates from these subgroups may be interpreted with caution.

In conclusion, the current study documents racial disparities in the receipt of adjuvant chemotherapy that were mediated by census tract-level poverty in women ages 65 years to 69 years with lymph node-positive, operable breast cancer, a group for which there is clear evidence for the efficacy of chemotherapy. There were no racial or socioeconomic disparities in receiving chemotherapy among women aged ≥70 years, for whom there is no clear evidence of the efficacy of chemotherapy. Disparities in treatment may explain some of the racial disparities in breast cancer outcomes.2 It is important to work toward reducing these treatment disparities, especially in light of evidence that there is no difference in response to adjuvant chemotherapy between African-American women and white women.38, 39

Acknowledgements

We acknowledge the efforts of the National Cancer Institute; Center for Medicare and Medicaid Services; Information Management Services, Inc.; and the Surveillance, Epidemiology, and End Results Program tumor registries in the creation of this database. The interpretation and reporting of these data are the sole responsibilities of the authors.

We thank Kavita Sail for providing some literature review and references and Luisa Franzini for helpful comments on the early drafts.

Conflict of Interest Disclosures

Supported by a grant from the Agency for Healthcare Research and Quality (R01-HS016743).

Ancillary