Disparities in receipt of adjuvant radiation therapy after breast-conserving surgery among the cancer-reporting regions of California

Authors


Abstract

BACKGROUND:

Incidence and mortality of breast cancer vary according to demographic factors such as age, race/ethnicity, socioeconomic status (SES), and geographic region. This study assesses the variation of these factors in the use of adjuvant radiation therapy (RT) after breast-conserving surgery (BCS) among 8 regions of California.

METHODS:

The authors identified 85,574 cases of first primary female invasive breast cancer with complete data diagnosed between January 1, 2000 and December 31, 2007. Logistic regression was used to determine the association between race/ethnicity, age, SES, and receipt of RT after BCS within each of the regions of California. Odds ratios (ORs) and 95% confidence intervals (CIs) were computed.

RESULTS:

Age was a significant predictor of receipt of RT after BCS in all regions. In Los Angeles (LA), lower SES was associated with decreasing odds of RT. Racial disparities were evident only in LA, where black (OR, 0.85; 95% CI, 0.74-0.97) and Hispanic (OR, 0.86; 95% CI, 0.78-0.96) women were about 15% less likely to receive RT than white women.

CONCLUSIONS:

Racial disparities in the receipt of RT after BCS exist only in LA, where African American and Hispanic women are less likely to receive this form of adjuvant treatment. Lower SES was also associated with a reduced likelihood of receipt of RT in LA. Women age 70 years and older are less likely to receive RT after BCS in all regions of California. Cancer 2012. © 2011 American Cancer Society.

INTRODUCTION

Breast cancer is the most common cancer in women worldwide.1 In California, it is the most common cancer in women regardless of age or race/ethnicity.2, 3 Breast-conserving surgery (BCS) followed by radiation therapy (RT) is an established standard of care for early stage breast cancer,4, 5 with equivalent survival for invasive breast cancer patients when compared with mastectomy.6, 7

Racial and geographic disparities in breast cancer outcomes have been well documented,3, 8-12 and minority women with early stage breast cancer have been reported to have double the risk for failing to receive necessary adjuvant treatments when compared with white women.13 We have been investigating breast cancer in California with particular emphasis on subtypes, age, and race/ethnicity.14-17 In 2009, we presented preliminary results indicating that there were differences in the use of RT after BCS in the 8 regions of California defined by the California Cancer Registry.18 This study presents a more focused look at the use of RT after BCS in California.

MATERIALS AND METHODS

Case Identification

By using the population-based California Cancer Registry, we identified cases of first primary female invasive breast cancer (International Classification of Diseases for Oncology-3 sites C50.0-C50.9)19 diagnosed January 1, 2000 through December 31, 2007 and reported to the California Cancer Registry as of April, 2009. Cases are reported to the Cancer Surveillance Section of the California Department of Public Health from hospitals and any other facilities providing care or therapy to cancer patients residing in California.20 Cases identified outside of California, only at autopsy, or from death certificates were excluded.

The details of documentation of estrogen receptor (ER) progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) in breast cancer along with age and stage at diagnosis, tumor grade, tumor subtype, race/ethnicity, and socioeconomic status (SES) have been extensively described in our previous publications14-16, 21 and by the California Cancer Registry.20

RT

Radiation treatment was recorded in the cancer registry for all documented cancer-directed RT performed as the first course of treatment. The definition of a first course of RT includes all treatment as part of a planned first course, as part of a facility's standards of practice, before disease progression or treatment failure, or that begins within 1 year of diagnosis.20

Region of California

For administrative purposes, the California Cancer Registry has divided the entire state into the 10 regions (Fig. 1). For this study, we divided the regions into the following: 1) North, 2) Sacramento, 3) San Francisco Bay Area and Santa Clara (San Francisco), 4) Central, 5) Tri-County, 6) Desert Sierra, 7) Los Angeles (LA), and 8) San Diego and Orange.

Figure 1.

The 10 regions of the State of California as defined by the California Cancer Registry are shown.

Statistical Analysis

Frequencies and percentages were calculated for all variables within each region. Contingency tables were used to assess the distributions of characteristics within each of the regions. The distributions were analyzed and patterns examined with the chi-square test of independence and Pearson residuals.22 The association of race and region with RT after BCS was evaluated using logistic regression analysis. In our previous research, we computed only a main effects model, where all regions were entered into a single model with demographic and tumor characteristics.18

For the present analysis, a model for each region was fitted. This allowed us to readily determine whether the association between race/ethnicity and receipt of RT after BCS was different among the regions of California. Patient and tumor characteristics were all entered simultaneously. A variable was considered statistically significant if the Wald chi-square was P < .05. Odds ratios and 95% confidence intervals were computed. Analyses were performed using SAS 9.1 (SAS Institute, Cary, NC). The expected probabilities23 of RT after BCS were computed for a region by race for both low and high SES.

RESULTS

Table 1 shows the total population of women in each of the regions of California stratified by race. LA was the most populous region, with almost 5 million women. The Tri-County Region, just northwest of LA, and the North region had the fewest women. White women made up 40% or more of the female population of all regions except for LA. Seventy-five percent of the female population in the North region was comprised of white women. Hispanics made up >44% of the Central, Desert Sierra, and LA regions' female population.

Table 1. Total Population of Women Within the Regions of California in 2008 (N=18,368,644)
RegionWhite, No. (%)African American, No. (%)Hispanic, No. (%)Asian/Pacific Islander, No. (%)Other, No. (%)Total Women, No. (%)
San Francisco1,507,543 (44.89)230,889 (6.88)783,497 (23.33)822,420 (24.49)13,854 (0.41)3,358,203 (18.28)
Central685,342 (42.69)68,964 (4.30)741,967 (46.22)92,374 (5.75)16,721 (1.04)1,605,368 (8.74)
Sacramento1,011,705 (56.67)144,908 (8.12)384,791 (21.56)226,907 (12.71)16,839 (0.94)1,785,150 (9.72)
Tri-County420,671 (57.80)12,880 (1.77)245,434 (33.72)44,472 (6.11)4380 (0.60)727,837 (3.96)
Desert Sierra845,588 (40.84)158,410 (7.65)921,567 (44.51)128,736 (6.22)16,079 (0.78)2,070,380 (11.27)
North582,048 (75.38)11,881 (1.54)129,579 (16.78)28,696 (3.72)19,949 (2.58)772,153 (4.20)
San Diego1,506,221 (49.01)101,209 (3.29)1,014,907 (33.02)435,010 (14.15)16,022 (0.52)3,073,369 (16.73)
Los Angeles1,463,763 (29.42)471,467 (9.47)2,327,808 (46.78)696,434 (14.00)16,712 (0.34)4,976,184 (27.09)
Total8,022,8811,200,6086,549,5502,475,049120,55618,368,644 (100.0)

We identified 147,078 cases of incident primary invasive female breast cancer. There were 61,504 patients missing data on at least 1 variable of interest (Table 2). The proportion of women with missing data in each of the regions was similar to the distribution of women in each of the regions (Tables 1 and 2). LA had the highest percentage of patients with at least 1 variable with missing data (27.1%). The Tri-County region had the lowest percentage of missing data (4.7%). There were 51,872 cases of BCS that were candidates for analysis.

Table 2. Number and Percentage of Patients With Data Missing for at Least 1 Variable Within Each Regiona
Race/ EthnicitySan Francisco, No. (%)Central, No. (%)Sacramento, No. (%)Tri-County, No. (%)Desert Sierra, No. (%)North, No. (%)San Diego, No. (%)Los Angeles, No. (%)Total
  • a

    Missing information regarding radiation therapy is unknown.

Non-Hispanic white7651 (65.75)2900 (71.84)4218 (75.48)2343 (81.02)4974 (69.92)3546 (89.59)6686 (69.71)7979 (47.82)40,297
Non-Hispanic black738 (6.34)138 (3.42)326 (5.83)36 (1.24)485 (6.82)24 (0.61)248 (2.59)2109 (12.64)4104
Hispanic1135 (9.75)700 (17.34)442 (7.91)353 (12.21)1263 (17.75)200 (5.05)1419 (14.80)3981 (23.86)9493
Asian/Pacific Islander1764 (15.16)119 (2.95)438 (7.84)112 (3.87)274 (3.85)69 (1.74)905 (9.44)2320 (13.90)6001
Other349 (3.00)180 (4.46)164 (2.93)48 (1.66)118 (1.66)119 (3.01)333 (3.47)298 (1.79)1609
Total11,637 (18.92)4037 (6.56)5588 (9.09)2892 (4.70)7114 (11.57)3958 (6.44)9591 (15.59)16,687 (27.13)61,504

Table 3 shows that the San Francisco, San Diego, and LA regions had the most cases of BCS. San Francisco had the highest percentage of cases that received RT (81%) and was used as the reference category in subsequent analyses. LA had the lowest (59%). Table 3 also shows that these 2 regions had the largest proportion of women younger than 46 years. The Central, Desert Sierra, and LA regions had a similar percentage of Hispanic cases (21%). LA had the highest percentage of African American cases (10%). The Central region had the highest percentage of people in SES 1 (29%).

Table 3. Demographic and Tumor Characteristics of First Primary Invasive Female Breast Cancer Cases With Complete Data Within the 8 Regions of the California Cancer Registry, 2000-2007 (N=85,574)
CharacteristicSan Francisco, n=18,455, No. (%)Central, n=6588, No. (%)Sacramento, n=9653, No. (%)Tri-County, n=3584, No. (%)Desert Sierra, n=6540, No. (%)North, n=3865, No. (%)San Diego, n=16,346, No. (%)Los Angeles, n=20,543, No. (%)Total
  1. Abbreviations: AJCC, American Joint Committee on Cancer; BCS, breast-conserving surgery; SES, socioeconomic status.

Age at diagnosis         
 <46 years3208 (17.4)1084 (16.5)1454 (15.1)499 (13.9)1131 (17.3)455 (11.8)2754 (16.8)3646 (17.7)14,231
 46-69 years10,910 (59.1)3816 (57.9)5685 (58.9)2100 (58.6)3844 (58.8)2312 (59.8)9274 (56.7)11,718 (57.0)49,659
 70+ years4337 (23.5)1688 (25.6)2514 (26.0)985 (27.5)1565 (23.9)1098 (28.4)4318 (26.4)5179 (25.2)21,684
ER/PR/HER2 Phenotype         
 +/+/+1768 (9.6)527 (8.0)1076 (11.1)275 (7.7)763 (11.7)486 (12.6)1718 (10.5)2292 (11.2)8905
 +/+/−10,825 (58.7)3560 (54.0)5309 (55.0)2086 (58.2)3255 (49.8)2110 (54.6)9373 (57.3)10,823 (52.7)47,341
 +/−/+544 (2.9)218 (3.3)333 (3.4)103 (2.9)230 (3.5)171 (4.4)523 (3.2)694 (3.4)2816
 +/−/−1692 (9.2)720 (10.9)867 (9.0)359 (10.0)583 (8.9)376 (9.7)1522 (9.3)1925 (9.4)8044
 −/+/+73 (0.4)25 (0.4)32 (0.3)11 (0.3)25 (0.4)21 (0.5)71 (0.4)120 (0.6)378
 −/+/−148 (0.8)54 (0.8)82 (0.8)32 (0.9)47 (0.7)27 (0.7)187 (1.1)203 (1.0)780
 −/−/+1201 (6.5)476 (7.2)663 (6.9)251 (7.0)571 (8.7)262 (6.8)983 (6.0)1584 (7.7)5991
 −/−/−2204 (11.9)1008 (15.3)1291 (13.4)467 (13.0)1066 (16.3)412 (10.7)1969 (12.0)2902 (14.1)11,319
Race/ethnicity         
 Non-Hispanic white12,037 (65.2)4702 (71.4)7527 (78.0)2910 (81.2)4453 (68.1)3540 (91.6)12,103 (74.0)11,625 (56.6)58,897
 Non-Hispanic black1123 (6.1)233 (3.5)555 (5.7)43 (1.2)449 (6.9)25 (0.6)386 (2.4)1953 (9.5)4767
 Hispanic2038 (11.0)1426 (21.6)844 (8.7)483 (13.5)1339 (20.5)235 (6.1)2316 (14.2)4318 (21.0)12,999
 Asian/Pacific Islander3257 (17.6)227 (3.4)727 (7.5)148 (4.1)299 (4.6)65 (1.7)1541 (9.4)2647 (12.9)8911
AJCC stage at diagnosis
 I9346 (50.6)3048 (46.3)4680 (48.5)1836 (51.2)2840 (43.4)2000 (51.7)7638 (46.7)8910 (43.4)40,298
 II7135 (38.7)2622 (39.8)3876 (40.2)1346 (37.6)2706 (41.4)1425 (36.9)6748 (41.3)8773 (42.7)34,631
 III1646 (8.9)777 (11.8)903 (9.4)345 (9.6)839 (12.8)361 (9.3)1667 (10.2)2468 (12.0)9006
 IV328 (1.8)141 (2.1)194 (2.0)57 (1.6)155 (2.4)79 (2.0)293 (1.8)392 (1.9)1639
Grade         
 I4620 (25.0)1437 (21.8)2362 (24.5)843 (23.5)1265 (19.3)1027 (26.6)3845 (23.5)3711 (18.1)19,110
 II8011 (43.4)2525 (38.3)4152 (43.0)1532 (42.7)2615 (40.0)1634 (42.3)7007 (42.9)8337 (40.6)35,813
 III5551 (30.1)2506 (38.0)3003 (31.1)1126 (31.4)2477 (37.9)1080 (27.90)5186 (31.7)8145 (39.6)29,074
 IV273 (1.5)120 (1.8)136 (1.4)83 (2.3)183 (2.8)124 (3.2)308 (1.9)350 (1.7)1577
Tumor size         
 ≤2 cm12,087 (65.5)3965 (60.2)6130 (63.5)2357 (65.8)3746 (57.3)2559 (66.2)10,131 (62.0)11,717 (57.0)52,692
 2.1-5 cm5234 (28.4)2122 (32.2)2988 (31.0)1022 (28.5)2269 (34.7)1085 (28.1)5108 (31.2)7085 (34.5)26,913
 >5 cm1134 (6.1)501 (7.6)535 (5.5)205 (5.7)525 (8.0)221 (5.7)1107 (6.8)1741 (8.5)5969
SES         
 1, low453 (2.5)1916 (29.1)845 (8.8)117 (3.3)1104 (16.9)360 (9.3)996 (6.1)3090 (15.0)8881
 21279 (6.9)1770 (26.9)1783 (18.5)377 (10.5)1683 (25.7)1138 (29.4)1957 (12.0)3466 (16.9)13,453
 32265 (12.3)1671 (25.4)3062 (31.7)765 (21.3)1656 (25.3)1177 (30.5)3006 (18.4)3964 (19.3)17,566
 44523 (24.5)1064 (16.2)2768 (28.7)1121 (31.3)1488 (22.8)757 (19.6)4450 (27.2)4574 (22.3)20,745
 5, high9935 (53.8)167 (2.5)1195 (12.4)1204 (33.6)609 (9.3)433 (11.2)5937 (36.3)5449 (26.5)24,929
Surgery         
 BCS12,110 (65.6)3697 (56.1)5236 (54.2)2362 (65.9)3513 (53.7)2421 (62.6)9664 (59.1)12,869 (62.6)51,872
 Mastectomy6345 (34.4)2891 (43.9)4417 (45.8)1222 (34.1)3027 (46.3)1444 (37.4)6682 (40.9)7674 (37.4)33,702
Radiation therapy after BCS
 Yes9806 (81.0)2681 (72.5)4134 (79.0)1520 (64.4)2680 (76.3)1667 (68.9)6968 (72.1)7575 (58.9)37,031
 No2304 (19.0)1016 (27.5)1102 (21.0)842 (35.6)833 (23.7)754 (31.1)2696 (27.9)5294 (41.1)14,841

The Central, Desert Sierra, and LA regions had the most cases diagnosed in stage III with grade III and IV tumors. The LA and Desert Sierra regions also reported the highest percent of tumors ≥5 cm (≥8%). ER/PR/HER2+ and ER/PR/HER2 tumors were more prevalent in the Central (23%), Desert Sierra (25%), and LA (22%) regions.

The logistic regression models for each separate region (Table 4) showed wide variation among the variables associated with receipt of RT after BCS. Table 4 indicates where a variable for a region had a statistically significant Wald chi-square, that is, all coefficients for a variable were not equal to zero for that region. Within all regions, age was a significant predictor of receipt of RT after BCS. All regions except for the Tri-County region showed that American Joint Committee on Cancer stage was associated with receipt of RT. Tumor phenotype was a significant factor only in the San Diego region. SES was a statistically significant predictor of RT in several regions. In LA, decreasing income (lower SES category) was associated with decreasing odds of RT. In the North region, the trend was the opposite. Women in the lowest SES category were 2.5× as likely as women in SES 5 to receive RT.

Table 4. ORs and 95% CIs for Variables Associated With RT After BCS in the Regions of California Defined by the California Cancer Registry
VariableSan Francisco, OR (95% CI)Central, OR (95% CI)Sacramento, OR (95% CI)Tri-County, OR (95% CI)Desert Sierra, OR (95% CI)North, OR (95% CI)San Diego, OR (95% CI)Los Angeles, OR (95% CI)
  • Abbreviations: AJCC, American Joint Committee on Cancer; CI, confidence interval; OR, odds ratio; SES, socioeconomic status.

  • a

    Indicates where Wald chi-square indicated that all coefficients for a variable were statistically significantly (P<.05) different from zero for the region.

Age at diagnosis        
 <46 years0.88 (0.76-1.00)a0.79 (0.63-0.98)a0.90 (0.73-1.10)a1.03 (0.79-1.36)a0.77 (0.61-0.97)a0.91 (0.68-1.22)a0.95 (0.83-1.09)a0.95 (0.86-1.06)a
 46-69 years [reference]1.00a1.00a1.00a1.00a1.00a1.00a1.00a1.00a
 70+ years0.52 (0.46-0.57)a0.55 (0.47-0.65)a0.51 (0.44-0.60)a0.67 (0.55-0.81)a0.67 (0.55-0.81)a0.68 (0.55-0.83)a0.57 (0.51-0.63)a0.83 (0.76-0.90)a
ER/PR/HER2 Phenotype        
 +/+/− [reference]1.001.001.001.001.001.001.00a1.00
 +/+/+0.93 (0.79-1.11)1.03 (0.77-1.38)0.82 (0.66-1.03)1.14 (0.82-1.60)0.96 (0.74-1.24)1.08 (0.81-1.45)0.83 (0.71-0.97)a0.92 (0.82-1.04)
 +/−/+0.69 (0.53-0.91)1.32 (0.79-2.19)1.40 (0.90-2.17)0.97 (0.55-1.70)1.14 (0.73-1.78)1.05 (0.65-1.68)0.82 (0.63-1.07)a0.99 (0.80-1.22)
 +/−/−0.91 (0.77-1.07)1.15 (0.90-1.47)1.06 (0.82-1.36)1.07 (0.80-1.43)1.37 (1.01-1.87)1.05 (0.77-1.41)1.11 (0.94-1.30)a0.97 (0.85-1.09)
 −/+/+0.82 (0.36-1.89)0.31 (0.06-1.60)0.64 (0.21-1.94)3.85 (0.48-31.09)2.14 (0.45-10.18)0.80 (0.26-2.46)1.41 (0.63-3.20)a0.84 (0.50-1.38)
 −/+/−0.77 (0.48-1.26)0.83 (0.34-2.02)1.75 (0.73-4.23)0.63 (0.24-1.65)1.56 (0.58-4.18)0.91 (0.33-2.48)0.80 (0.53-1.20)a0.98 (0.68-1.41)
 −/−/+0.97 (0.78-1.22)0.83 (0.58-1.19)0.78 (0.57-1.06)0.89 (0.60-1.33)1.04 (0.75-1.45)0.92 (0.60-1.40)0.79 (0.63-1.00)a0.87 (0.74-1.03)
 −/−/−0.81 (0.69-0.95)0.84 (0.66-1.07)0.95 (0.75-1.21)1.03 (0.76-1.38)1.19 (0.92-1.54)0.81 (0.58-1.12)0.83 (0.71-0.98)a0.90 (0.79-1.01)
Race/ethnicity        
 Non-Hispanic white [reference]1.001.001.001.001.001.001.001.00a
 Non-Hispanic black0.86 (0.71-1.04)0.73 (0.49-1.08)0.83 (0.63-1.10)0.61 (0.28-1.31)0.69 (0.51-0.93)1.01 (0.35-2.88)0.77 (0.57-1.05)0.85 (0.74-0.97)a
 Hispanic0.95 (0.81-1.11)0.93 (0.76-1.13)1.28 (0.98-1.67)1.37 (1.03-1.82)1.00 (0.81-1.25)0.93 (0.64-1.36)0.95 (0.82-1.10)0.86 (0.78-0.96)a
 Asian/Pacific Islander1.06 (0.93-1.22)0.96 (0.60-1.51)1.20 (0.88-1.63)1.24 (0.78-1.96)1.03 (0.67-1.58)1.21 (0.59-2.50)0.82 (0.69-0.98)0.93 (0.83-1.05)a
AJCC stage at diagnosis        
 I [reference]1.00a1.00a1.00a1.001.00a1.00a1.00a1.00a
 II0.55 (0.48-0.63)a0.86 (0.68-1.08)a0.78 (0.63-0.95)a0.99 (0.76-1.28)1.04 (0.81-1.34)a0.75 (0.58-0.97)a0.80 (0.70-0.91)a0.83 (0.75-0.93)a
 III0.57 (0.44-0.74)a0.60 (0.41-0.88)a0.41 (0.28-0.57)a0.87 (0.53-1.49)0.69 (0.47-1.02)a0.82 (0.50-1.35)a0.61 (0.47-0.78)a0.62 (0.51-0.76)a
 IV0.18 (0.13-0.27)a0.30 (0.15-0.61)a0.32 (0.19-0.56)a0.50 (0.21-1.22)0.36 (0.21-0.62)a0.37 (0.17-0.83)a0.24 (0.16-0.36)a0.41 (0.29-0.60)a
Grade        
 I [reference]1.001.001.001.001.00a1.00a1.001.00a
 II1.04 (0.92-1.17)0.90 (0.74-1.08)0.95 (0.80-1.12)0.99 (0.80-1.23)0.88 (0.71-1.10)a0.91 (0.73-1.13)a0.99 (0.88-1.11)0.98 (0.89-1.08)a
 III0.91 (0.78-1.05)1.15 (0.91-1.45)0.85 (0.68-1.06)0.86 (0.65-1.13)0.68 (0.53-0.88)a0.75 (0.57-0.98)a0.85 (0.74-0.99)0.84 (0.75-0.94)a
 IV0.84 (0.56-1.26)0.79 (0.42-1.16)0.89 (0.47-1.68)1.13 (0.56-2.26)0.77 (0.42-1.41)a0.23 (0.13-0.40)a1.05 (0.71-1.54)1.12 (0.82-1.52)a
Tumor size        
 ≤2 cm [reference]1.00a1.001.001.001.00a1.001.001.00a
 2.1-5 cm1.14 (0.99-1.32)a0.68 (0.53-0.87)a0.77 (0.62-0.95)0.78 (0.58-1.04)0.69 (0.53-0.90)a1.01 (0.76-1.35)0.89 (0.77-1.02,)0.89 (0.80-1.00)a
 >5 cm0.57 (0.41-0.81)a0.57 (0.32-1.01)a0.82 (0.45-1.51)0.64 (0.28-1.50)0.73 (0.42-1.27)a1.03 (0.43-2.47)0.97 (0.68-1.39)0.81 (0.63-1.06)a
SES        
 1, low0.72 (0.54-0.95)a1.01 (0.63-1.60)a0.76 (0.55-1.04)a1.46 (0.80-2.65)0.75 (0.54-1.05)2.54 (1.69-3.81)a1.07 (0.86-1.33)a0.64 (0.56-0.73)a
 20.78 (0.65-0.94)a1.09 (0.69-1.71)a0.62 (0.48-0.80)a0.80 (0.58-1.10)0.68 (0.50-0.93)2.44 (1.81-3.28)a0.86 (0.73-1.00)a0.74 (0.66-0.83)a
 30.85 (0.74-0.99)a0.99 (0.63-1.56)a0.77 (0.61-0.97)a0.91 (0.72-1.15)0.80 (0.59-1.08)2.19 (1.64-2.93)a0.85 (0.75-0.97)a0.79 (0.71-0.88)a
 40.96 (0.89-1.12)a1.53 (0.95-2.47)a0.92 (0.72-1.16)a0.85 (0.69-1.05)0.86 (0.63-1.17)1.46 (1.08-1.99)a0.96 (0.86-1.08)a0.87 (0.79-0.96)a
 5, high [reference]1.00a1.00a1.00a1.001.001.00a1.00a1.00a

The logistic regression models for each individual region showed that race was only a statistically significant predictor of RT in the LA region. Black and Hispanic women living in LA were about 15% less likely to receive RT than white women (Table 4). Race was not a statistically significant predictor of RT in any other region of California.

Figure 2 presents the expected probabilities computed from the individual model for LA for a 46- to 69-year-old woman with a stage 1, grade 1, ER+/PR+/HER2 tumor that was <2.0 cm. Black women in SES 1 had a 58% probability of RT after BCS, whereas white women in SES 5 had a probability of around 71%. For all races, women in SES 1 were approximately 10% less likely to receive RT than women of the same race in SES 5.

Figure 2.

Expected probabilities computed from the individual model for Los Angeles for a 46 to 69-year-old woman with a stage 1, grade 1, estrogen receptor-positive/progesterone receptor-positive/human epidermal growth factor receptor 2-negative tumor <2.0 cm are shown. SES, socioeconomic status.

DISCUSSION

Age, race/ethnicity, and SES are known to affect breast cancer incidence and mortality.3, 9, 24 Our previous research revealed that women with triple-negative breast cancer were significantly more likely to be African American or Hispanic and live in socioeconomically deprived areas,14, 15 and age was found to have an important role in determining the ER/PR/HER2 subtype.16, 17

Although racial disparities in breast cancer treatment and outcomes have been previously well documented,3, 25-27 the interplay between race/ethnicity and SES and their respective causative effect on breast cancer outcomes remain controversial.28 Furthermore, it remains difficult to completely unravel the intertwining aspects of these factors in addition to tumor biology and age.29 This dilemma is evident from studies that have shown comparable outcomes after adjustment for sociodemographic factors if patients have equal access to healthcare.30, 31 Others have shown that racial disparities persist even after controlling for SES and access to healthcare.32 Newman concluded from 2 meta-analyses that African American ethnicity is a significant and independent predictor of poor outcome from breast cancer, even after accounting for SES.33 A Southwest Oncology Group study concluded that after adjustment for SES, African American patients with breast cancer had worse adjusted survival, despite enrollment on phase 3 clinical trials with uniform stage, treatment, and follow-up.34 Several studies have revealed disparities because of race/ethnicity, older age, and geographic location in breast cancer care, including the receipt of RT after BCS.12, 13, 27, 35-43

The present study, however, is the first to document a disparity in the receipt of RT because of race/ethnicity and SES in a single region within the state of California, and an age disparity throughout the entire state. LA, the most populous of the California regions with the largest number and highest percentage of African American and Hispanic residents in this study, is the only region where the association of RT after BCS varied by race/ethnicity. Our analysis of the association of race and SES within each region, not done in our preliminary research,18 illustrates how conducting an analysis without consideration of interactions among variables can lead to results that are misleading.

It is not surprising that women 70 years and older are less likely to receive RT after BCS, because several studies have demonstrated that RT may be eliminated in selected, good-risk older patients.44-46 As the use of RT reduces mortality and local recurrence at 15-year follow-up, lowering of the threshold for RT is not advised in light of the ageing population and increased life expectancy.41, 47

We do not know why race and SES are associated with reduced receipt of RT in LA, but we may speculate on the reasons for this disparity. Lack of physician referral,48 type of hospital performing breast surgery,49 lack of health insurance, and distance from radiation treatment centers50, 51 have all been cited as influencing disparities in breast cancer treatment patterns. However, missing data cannot be overlooked as a potential reason for our results.

The limitations of population-based cancer registry investigations are well known.14, 15, 21 The exclusion of subjects without ER, PR, and HER2 results has been noted in other population-based registry studies.14, 15, 52-54 As pointed out by Krieger et al,55 studies on race/ethnicity that fail to account for missing data and socioeconomic status may yield inflated estimates of racial/ethnic disparities.

Cancer registries rely on data submitted from many sources, including hospitals, private clinics, and outpatient facilities, including physician offices and radiation oncology centers. In California, the regional registries report all incident cancer cases to the California Cancer Registry and have an obligation to include first courses of treatment for breast cancer. Because adjuvant RT is most often administered after a breast cancer patient is discharged from a hospital, the regional registry depends upon these outpatient facilities for accurate data submission. It is not known if all regional registries attempt to ascertain treatment data with equal vigor. Data from ambulatory care centers have been found to be less accurate than hospital-based care.56 A study comparing the completeness of hospital medical records with records from outpatient treatment facilities is warranted, but unfortunately is well beyond the scope of the present investigation.

Despite the shortcomings listed above, we believe our study is of value because of the large number of cases reported to the statewide cancer registry from an ethnically diverse population. The inclusion of SES further strengthens the findings.

We have highlighted the importance of addressing race/ethnicity, SES, and region when analyzing data from a population-based cancer registry, because all of these variables influence results. On the basis of the existing data from the California Cancer Registry, we conclude that 1) racial disparities in the receipt of RT after BCS exist only in LA, where African American and Hispanic women are less likely to receive this form of adjuvant treatment; 2) a clear disparity based on SES exists in LA; and 3) women 70 years and older are less likely to receive RT after BCS in all regions of California.

FUNDING SOURCES

This research was supported by a grant from the Sutter Institute for Medical Research.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary