Annals of Surgical Oncology 15(10):2644–2652
DOI: 10.1245/s10434-008-0053-5
Disparities in Urban and Rural Mastectomy Populations The Effects of Patient- and County-Level Factors on Likelihood of Receipt of Mastectomy Lisa K. Jacobs, MD,1,2 Katherine A. Kelley,1 Gedge D. Rosson, MD,1,3 Meagan E. Detrani,1 and David C. Chang, PhD, MPH, MBA1
1 Department of Surgery, The Johns Hopkins University School of Medicine, Baltimore, MD, USA Division of Surgical Oncology, The Johns Hopkins University School of Medicine, Osler 624, 600 North Wolfe Street, Baltimore, MD 21287, USA 3 Division of Plastic Surgery, The Johns Hopkins University School of Medicine, Baltimore, MD, USA
2
Background: Using the 2006 Surveillance, Epidemiology, and End Results (SEER) database and the 2004 Area Resource File (ARF), the likelihood of mastectomy for stages I–III breast cancer patients in urban versus rural populations are examined. County and patient level data are evaluated for impact on receipt of mastectomy. Patient variables included age, stage, race, and marital status, and community variables are income, employment, and radiation facility staff density. The likelihood of mastectomy in urban and rural patients, and the impact of the different variables on that procedure, is reported. Methods: This retrospective analysis of a combined dataset from the 2006 SEER database and the 2004 ARF linked using the federal information processing standard (FIPS) state county variable evaluates patient and county variables with multivariate regression. Results: From 1992 to 2003, 137,303 patients were identified in the SEER database. The rural population (county population of <20,000) comprised 9.58% of the overall population. On bivariate analysis, the likelihood of mastectomy was significantly higher among rural patients (59.90% versus 44.92%, P < 0.001). Multivariate analysis demonstrated that rural residency is an independent factor affecting receipt of mastectomy (odds ratio [OR] 1.58, 95% confidence interval [CI] 1.26–1.97). The likelihood that a patient received a mastectomy was impacted by the significant patient factors of stage at diagnosis, race, and marital status, and significant community factors were employment, education level, and density of radiation technologists. Conclusion: An increased likelihood of mastectomy for rural patients with stages I–III breast cancer is shown with analysis of patient and community factors that may play a role. Key Words: Mastectomy—Breast cancer—Rural—Urban—Disparity.
Evidence accumulated over the past 20 years has established the equivalent survival of partial mastectomy with radiation therapy and mastectomy for management of breast cancer.1–3 Despite this evidence, there are still large differences in mastectomy rates across the USA.4,5 Regional variation in the
utilization of mastectomy over breast preservation is well documented in the literature.6,7 Studies of statewide breast cancer populations have shown an increase in mastectomy rates in elderly rural residents diagnosed with breast cancer.8–11 Some of the regional variation in utilization of breast preservation has been linked to availability of radiation oncology services, with some studies finding a statistically significant impact.12–14 This study analyzes the likelihood of an urban versus a rural patient to undergo mastectomy, including the access to radiation therapy
Published online July 29, 2008. Address correspondence and reprint requests to: Lisa K. Jacobs, MD; E-mail:
[email protected] Published by Springer Science+Business Media, LLC 2008 The Society of Surgical Oncology, Inc.
2644
DISPARITY IN URBAN AND RURAL MASTECTOMY
as a variable. We hypothesize that, in addition to the availability of radiation oncology services, other community factors play a role in the decision to pursue mastectomy in the rural population. Much of the variation in theraphy has been attributed to patient factors such as age,15–17 race,13,18–22 socioeconomic status, income level,18 and education level.15,18,22,23 More recent studies have investigated the impact of marital status on the treatment options regarding breast cancer.15 In addition, an employed patient may have difficulty committing time to receive radiation therapy and may have difficulty maintaining work-related relationships.24,25 Each of these variables are included in this analysis. The primary objective of this manuscript is to determine the impact of community factors and their relationship to patient factors on mastectomy choice in women with breast cancer.
METHODS AND MATERIALS After obtaining approval from the Johns Hopkins Institutional Review Board, a retrospective analysis of a combined dataset that includes variables from the 2006 Surveillance, Epidemiology, and End Results (SEER) database and the Area Resource File (ARF) from 2004 was completed. The two databases are linked using the FIPS state county variable. The FIPS county codes were established by the National Bureau of Standards, US Department of Commerce in 1968.26 The SEER database of 2006 is maintained by the National Cancer Institute (NCI). It collects cancer incidence and survival rates from population based cancer registries covering the regions Alaska, Arizona, California, Connecticut, Detroit, Atlanta, Rural Georgia, Hawaii, Iowa, Kentucky, Louisiana, New Jersey, New Mexico, Seattle, and Utah.27 The Area Resource File (ARF) is a collection of data from more than 50 sources, including the American Medical Association, the American Hospital Association, the US Census Bureau, Centers for Medicare & Medicaid Services, the Bureau of Labor Statistics, and the National Center for Health Statistics, and includes over 6,000 variables concerning counties in the USA.28 Selection criteria for the study included stage and extent of disease. Female patients with stages I, II, or III breast cancer, as defined by SEER and modified by the American Joint Committee on Cancer (AJCC) staging, 3rd edition are included. Stage I cancer is defined as being confined to the breast tissue and fat, including the nipple and areola with a tumor size
2645
under 2 cm. As defined by the SEER database, stage II cancer is defined as having invaded the subcutaneous tissue, infiltrated the skin of the primary breast, and infiltrated local dermal lymphatic ducts with a tumor size between 2 and 5 cm. Stage III cancer is defined as the invasion or fixation to the pectoral fascia or underlying tissue with a tumor size greater than 5 cm. The extent of disease stage is established by size of tumor, extension, lymph node involvement, and the number of nodes examined and those that were positive. Literature supporting breast preservation as opposed to mastectomy was published in the 1980s, therefore data for this study is collected after 1992 to allow for the implementation of these treatment recommendations. Exclusion criteria included: T4 tumors and Paget’s disease, because the standard treatment recommendation is mastectomy;29 metastatic disease, because surgical management in this population is palliative; and prior cancer, because of the possibility of prior radiation therapy. Excluded patients were identified using the ‘‘extent of disease’’ variable in SEER. The outcome of interest is the receipt of mastectomy, defined by the ‘‘surgery to primary site’’ variable in SEER and includes the following surgeries: subcutaneous mastectomy, total mastectomy, modified radical mastectomy, radical mastectomy, and extended radical mastectomy. Surgeries categorized as breast preservation were nipple resection, lumpectomy, excisional biopsy, re-excision biopsy, wedge resection, quadrantectomy, and segmental mastectomy. The primary independent variable is rural/ urban status of the county of residence for each patient, as defined by two separate criteria, the rural–urban continuum code and the urban influence code. The rural–urban continuum codes are from the Economic Research Service (ERS) of the US Department of Agriculture. The codes form a classification scheme that distinguishes metropolitan counties by the population size of their metropolitan area and nonmetropolitan counties by degree of urbanization and adjacency to a metropolitan area or nonmetropolitan areas. Metropolitan counties are distinguished by population size of the metropolitan statistical area of which they are a part. Nonmetropolitan counties are classified according to the aggregate size of their urban population.30 The urban influence codes are from the US Department of Agriculture’s Economic Research Service (ERS) website http://www.ers.usda.gov/Data/ UrbanInfluenceCodes/. The urban influence codes divide the 3,141 counties, county equivalents, and the independent cities in the USA into 12 groups Ann. Surg. Oncol. Vol. 15, No. 10, 2008
2646
L. K. JACOBS ET AL.
based on population and commuting data from the 2000 Census of Population in the case of metropolitan counties, and adjacency to metropolitan area in the case of nonmetropolitan counties.30 In this analysis a code of 6 is used for each system which defined a rural community as areas with population of 20,000 or below; this is an appropriate level because: firstly, it categorizes 9.61% of the study population as rural, which is similar to urban rural differences as calculated by the Federal Highway Administration, which sites that 20.78% of the US population is rural; secondly, 62.5% of the counties in the dataset were rural, which effectively separates the two groups based on geographic criteria. After setting this criterion, post hoc analysis demonstrated a significant increase in mastectomy in the levels above 6, indicating that this was an appropriate cut point. Other patient level data extracted from the SEER are age at diagnosis, calendar year at diagnosis, race, marital status, and stage of cancer. Age at diagnosis was grouped into 5-year intervals. The calendar year of diagnosis was analyzed to determine if there are any variations in treatment from year to year. Race was defined as White, Black, and an ‘‘other’’ group that includes all other races. Hispanics are not included as a selection because most are categorized as White. Marital status is broken into four groups: married, single, divorced or separated, and widowed. Community-level variables extracted from the year 2000 ARF include median household income, education level, number of people employed, and radiation oncology and radiation technology staff availability. Median household income is defined as follows: low income falls below the 30th percentile of the USA, middle income falls between the 30th and 50th percentile, and high income level lies above the 50th percentile. The education level of the community was the percentage of the population with a highschool diploma or more. The percentage of people employed was determined per county by using the values of the census population. For the purpose of this study, those who worked from home, unpaid volunteers for specific organizations, institutionalized population and people on active duty in the US Armed Forces, and civilians 16 years and older who did not work or did not have a job during the reference week and are looking for a job during the four last weeks were considered unemployed. Radiation oncologist and technologist density were based upon mean values within the county. Sensitivity analysis was done to determine whether the method of calculating radiation oncology staff Ann. Surg. Oncol. Vol. 15, No. 10, 2008
availability would affect the results. Each density was calculated in two ways, first by number of radiation oncologist or technologists by population density and second by land area. This analysis was undertaken because of reports in the literature indicating that availability of radiation therapy significantly alters the decision to pursue mastectomy or lumpectomy with radiation therapy. We hypothesized that radiation technologist density may have more impact than radiation oncologist density because the technologist would be able to conduct treatment in more rural areas with oversight from the radiation oncologist. We evaluated the availability of radiation oncology services by calculating the density of radiation oncologist and the density of radiation technologists. Multivariate statistical analysis was carried out to determine the likelihood of mastectomy for a patient against rural/urban residency, age, race, marital status, income, stage, employment, education, and transportation of the study population. A t-test was done in order to determine the significance of these values and these proportions were compared using a two-sided z-test. Using a logistic regression model, the odds ratio and P-value were determined. All calculations were performed using Stata SE, version 9.2.
RESULTS Using the SEER database, 137,303 patients met the inclusion and exclusion criteria. Table 1 demonstrates the characteristics of the overall population, with 46.38% undergoing mastectomy and 9.58% residing in a rural area. A summary of the covariates analyzed for both patient level and community level data are displayed in Table 1. Figure 1 demonstrates a decline in mastectomy rates for the overall population over the time period studied. Rural Versus Urban Analysis Patient-level data from the SEER database linked to the ARF for the defined urban versus rural populations is depicted in Table 2. There was a significant difference in the percentage of mastectomies between urban and rural populations: 44.92% of urban women received a mastectomy, while 59.90% of rural women received a mastectomy (P < 0.001). There was no significant difference in age and stage between urban and rural populations. There were, however, significant differences between the populations in race (with more Blacks residing in the urban population) and marital status (with more single,
2647
DISPARITY IN URBAN AND RURAL MASTECTOMY
TABLE 1. Characteristics of the study population Covariate Patient-level data (People) N (People) Age Race White Black Other Mastectomy Rural Stage I II III Marital status Married Single Separated/divorced Widowed County-level data N Income (US $/ year) <30,000 30,000–50,000 50,000 Percentage employed Percentage with high-school education or more Density of radiation oncologists – per land area (number of providers per 1,000,000 sq. miles) Density of radiation technologists – per land area (number of providers per 1,000,000 sq. miles)
Mean or percentage 137,303 60.05 84.42% 7.86% 7.72% 46.38% 9.58% 53.22% 42.48% 4.30% 59.82% 11.22% 10.51% 18.45% 200 1.68% 11.45% 86.87% 46.55% 54.02% 5.39 12.08
separated, and divorced in the urban population), Analysis of the community level data as depicted in Table 2 demonstrated no difference in the education level of the urban and rural populations. Significant differences between the populations were identified in community income level, employment level, and density of radiation facility staff, all of which were higher in the urban population. Mastectomy Versus Other Surgical Treatments Analysis Analysis of the entire population was completed to determine which covariates play a role in the receipt of mastectomy versus breast preservation. Bivariate analysis of covariates is displayed in Table 3 and demonstrates in patient-level data that there were no significant differences in age and race distribution. Of the women who did not undergo a mastectomy, 7.26% lived in a rural location, and of those women who did receive a mastectomy 12.54% resided in a rural population. There were significant differences based on stage at diagnosis, with more women with stage III disease undergoing mastectomy. Interestingly, analysis of marital status only demonstrated a difference in the widowed population.
Multivariate Analysis Using multivariate analysis we determined that women in rural communities are 58–69% more likely to undergo mastectomy after controlling for all other variables, as demonstrated in Table 4. Patient-level covariates that demonstrated a statistically significant difference on multivariate analysis included: race, with Blacks having an odds ratio of 0.93 and other races having an odds ratio of 1.25; stage, with stage II having an odds ratio of 2.61 and stage III an odds ratio of 10.16; and marital status, with widowed women having an odds ratio of 1.10 and separated and divorced women an odds ratio of 0.94. Community-level covariates demonstrating a statistically significant difference on multivariate analysis were: employment level, with an odds ratio of 1.02; education level, with an odds ratio of 0.96; and the availability of radiation technologist when density was calculated by population. Adjusting the analysis for radiation oncology staff availability as a density per population found that rural individuals are 58% more likely to undergo a mastectomy than their urban counter parts; adjusting for radiation oncology density by area revealed a similar significant association. This also affected the impact of the percentage employed in the community, as depicted in Table 4. As opposed to other studies, the only patient-level covariate that had no significant difference on multivariate analysis was age. On the community level there were no significant differences in treatment with community income level.
DISCUSSION In 1898 Halsted first reported the use of radical mastectomy with a locoregional recurrence of 16%;31 from that moment it was considered the best treatment for breast cancer. However, in 1985 the New England Journal of Medicine published a randomized controlled trial demonstrating the equivalence of breast-conserving surgery with radiation therapy to mastectomy.1,2 With those results and the results of similar studies a change in practice occurred with a reduction in mastectomy rates over time.10 This study analyzed patients after 1992 and found that mastectomy rates continue to fall (Fig. 1). Yet, there are still a large percentage of women today who receive mastectomy. Previous studies have focused on patient-level data to identify factors that impact mastectomy rates, as noted earlier. The objective of this study was to identify the patient- and communityAnn. Surg. Oncol. Vol. 15, No. 10, 2008
2648
L. K. JACOBS ET AL.
1 0.9
Odds Ratio relative to year 1992
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
Year
FIG. 1. The decline in mastectomy in comparison with the year 1992. The odds ratios are depicted.
level factors and the interaction between those that would influence likelihood of mastectomy, focusing on the rural–urban comparison. This study supports other reports in the literature indicating an increased mastectomy rate in the rural population as compared with the urban population.8,10,32 We found that 59.90% of the rural population receives a mastectomy and 44.92% of the urban population receives mastectomy. Variables that have been identified to impact mastectomy rates in the literature include: availability of academic institutions,13 ease of treatment,17 cost,33 and lack of oncology consultations.6 The literature indicates that the access of the rural population to radiation oncology facilities significantly impacts the selection of mastectomy over breast preservation.10,12,14,16,34 To include that variable in this analysis we calculated the density of both radiation oncologists and radiation technologists in the rural and urban populations. The ARF allows calculation of the density of radiation oncologists in two ways. The first is based on population and the second on land area. Calculations using land area were not significant; however, calculations using population were, which broadly suggests that in rural areas access to services may be more of an issue than distance to travel for services. Rural populations had an average of 2.25 radiation oncologists and 2.57 radiation technologists per county, whereas the urban Ann. Surg. Oncol. Vol. 15, No. 10, 2008
population had 10.63 radiation oncologists and 27.93 radiation technologists per county. Therefore, it is possible that communities with large numbers of people, or urban centers, are more likely to undergo breast-conserving surgery because there are more radiation facilities available to them. A weakness of this study is the inability to link the location of radiation facilities directly to patient residency. Studies which have made this link have found distance to a radiation facility to have a significant impact on treatment decisions.10,12 In addition to the availability of radiation facilities, availability of screening facilities in urban and rural populations may result in a later stage at presentation.6,35,36 Higher stage does significantly increase the likelihood of mastectomy, as reported here and supported in the literature.37 However, in agreement with other studies,9 this study found no significant difference in stage at presentation for rural and urban populations. The literature demonstrates conflicting results on the impact of race on mastectomy rates.12,20 Some authors report lower likelihood of screening,38 higher stage at diagnosis,21 and differences in utilization of radiation therapy between the White and Black populations,16,18 which would likely impact mastectomy rates. This study identifies a difference in mastectomy rates between Whites, Blacks, and other races. Blacks were less likely to have mastectomy,
2649
DISPARITY IN URBAN AND RURAL MASTECTOMY
TABLE 2. Rural versus urban populations and their characteristics comparatively. All P values £0.001
Patient-level data N (People) Mastectomy Age (years) 18–24 25–29 30–34 35–39 40–44 45–49 50–54 55–60 60–64 65–69 70–74 75–79 80–84 85–89 90 and over Race White Black Other Stage I II III Marital status Married Single Separated/divorced Widowed County-level data N Income (US $/ year) <30,000 30,000–50,000 >50,000 Percentage employed Percentage with high-school education or more Density of radiation oncologist – per land area (number of providers per 1,000,000 sq. miles) Density of radiation technologist – per land area (number of providers per 1,000,000 sq. miles)
Urban population
Rural population
124,143 44.92%
13,160 59.90%
0.06% 0.47% 1.76% 4.33% 8.37% 11.84% 12.57% 11.72% 10.64% 10.41% 10.17% 8.71% 5.52% 2.56% 0.87%
0.08% 0.33% 1.21% 2.98% 6.45% 8.68% 10.55% 10.33% 11.06% 12.39% 12.40% 11.15% 7.25% 3.82% 1.32%
83.70% 8.66% 7.64%
91.24% 0.33% 8.43%
53.10% 42.53% 4.37%
54.35% 41.98% 3.67%
59.33% 11.74% 10.85% 18.09%
64.43% 6.35% 7.38% 21.85%
75
125
0.68% 5.52% 93.80% 48.47% 53.83%
11.17% 67.35% 21.48% 45.39% 54.14%
10.63
2.25
27.93
2.57
while other races were more likely. These results partially agree with Schreon, who found that White and Black women with cancer are treated similarly while other races had significantly higher rates of mastectomy,10 and with Gelber who found that Japanese and Philippinos are more likely to receive mastectomy.39 Additionally, the difference in mastectomy likelihood between Black and White patients as observed in this study may be explained by the fact that our analysis adjusted for rural/urban residency, and the distribution of racial groups between rural and urban areas are not the same. The racial
TABLE 3. Mastectomy outcome populations and their characteristics comparatively
n Rural Urban Age (years) 18–24 25–29 30–34 35–39 40–44 45–49 50–54 55–60 60–64 65–69 70–74 75–79 80–84 85–89 90 and over Race White Black Other Stage I II III Marital status Married Single Separated/divorced Widowed
No mastectomy
Mastectomy
66,156 7.26% 92.74%
57,220 12.54% 87.46%
0.06% 0.41% 1.48% 3.90% 8.19% 12.07% 13.70% 12.50% 11.26% 10.82% 10.09% 8.01% 4.60% 2.11% 0.80%
0.07% 0.51% 1.95% 4.60% 8.10% 10.89% 10.84% 10.35% 9.84% 10.69% 11.08% 10.09% 6.81% 3.18% 0.99%
85.04% 7.59% 7.37%
84.30% 7.83% 7.87%
64.69% 33.94% 1.36%
40.38% 51.91% 7.71%
61.34% 11.21% 10.95% 16.50%
58.30% 10.83% 9.68% 21.20%
difference reported in the literature may, therefore, be a reflection of the differential distribution of racial groups in rural versus urban areas. Many studies have been published demonstrating a direct correlation between age and mastectomy rates.10,11,16,17,40 Our analysis of age demonstrated minimal significance on the likelihood of mastectomy. However, we did show a significant difference based on marital status, with widowed women being 10% more likely to receive a mastectomy and separated or divorced women being 6% less likely to receive a mastectomy. It is possible that the age differences previously identified are primarily affected by the likelihood of being widowed in the older population. This is supported by literature indicating that women who are married are somewhat more likely to prefer breast-conserving surgery than unmarried, single, divorced or widowed women.17 The influence of interpersonal relationships has been studied by Stafford who found that a husband, significant other or the possibility of a future partner influences treatment decisions,7 and by Clifford, who found that 64% of women reported their husbands were very upset about their mastectomy.40 Ann. Surg. Oncol. Vol. 15, No. 10, 2008
2650
L. K. JACOBS ET AL.
TABLE 4. Multivariate analysis of mastectomy outcome (n.s.= not significant) Adjusting by density per population
Rural (versus urban) Age (years): 18–24 25–29 30–34 35–39 40–44 45–49 50–54 55–60 60–64 65–69 70–74 75–79 80–84 85–89 90 and over Race White Black Other Stage I II III Marital status Married Single Separated/divorced Widowed County-level factors Income (US $/ year) <30,000 30,000– 50,000 >50,000 Percentage employed Percentage with high-school education or more Density of radiation oncologist – per land area (number of providers per 1,000,000 sq. miles) Density of radiation technologists – per land area (number of providers per 1,000,000 sq. miles)
Odds ratio
95% confidence interval
1.58
1.26–1.97
1.03 1.14 1.11 1.01 0.95 0.89 0.97 1.03 1.17 1.30 1.51 1.74 1.66 1.21
Odds ratio
95% confidence interval
0.00
1.69
1.38–2.08
0.00
0.60–1.76 0.66–1.98 0.64–1.92 0.59–1.71 0.56–1.62 0.53–1.49 0.58–1.63 0.62–1.73 0.69–1.97 0.77–2.21 0.89–2.56 1.01–2.99 0.98–2.80 0.74–1.96
n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s n.s. n.s. n.s. 0.045 n.s. n.s.
1.05 1.16 1.13 1.02 0.97 0.90 0.99 1.05 1.18 1.33 1.53 1.76 1.69 1.22
0.61–1.81 0.67–2.01 0.65–1.95 0.60–1.74 0.57–1.65 0.53–1.52 0.59–1.66 0.62–1.77 0.70–2.01 0.78–2.26 0.90–2.61 1.02–3.03 0.99–2.86 0.75–1.99
n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s n.s. n.s. n.s. n.s. n.s. n.s.
0.93 1.25
0.84–1.02 1.07–1.45
n.s. 0.004
0.92 1.27
0.84–1.01 1.09–1.49
0.042 0.002
2.61 10.16
2.53–2.70 8.25–12.52
0.00 0.00
2.61 10.15
2.54–2.70 8.21–12.53
0.00 0.00
1.03 0.94 1.10
0.97–1.09 0.89–1.00 1.06–1.14
n.s. 0.046 0.00
1.03 0.94 1.10
0.98–1.07 0.89–1.00 1.06–1.14
n.s. 0.016 0.00
1.05 0.86 1.02 0.96 1.00
0.68–1.60 0.55–1.35 1.01–1.04 0.95–0.97 0.99–1.00
n.s. n.s. 0.008 0.00 n.s.
1.25 0.98 1.02 0.96 1.01
0.84–1.87 0.63–1.52 1.00–1.04 0.95–0.97 0.95–1.07
n.s. n.s. 0.028 0.00 n.s.
1.00
0.99–1.00
0.014
0.99
0.97–1.02
n.s.
Other variables identified in the literature that impact mastectomy rates include education level,18,21,22 employment level, and income level of the patient. These variables have shown conflicting results, with some studies showing that higher education level results in a lower mastectomy rate8 and others showing a higher mastectomy rate with higher education level.15 The impact of employment on mastectomy rate has been interpreted in a variety of ways. Some report that employed women are apprehensive about returning to work after a mastectomy25 and others demonstrate that unemployed individuals present at later stages and are therefore more likely to receive mastectomy.24 Similar to employment, income levels may play a role in mastectomy by affecting access to screening and the stage at diagnosis.22 Conflicting results of the impact of income on Ann. Surg. Oncol. Vol. 15, No. 10, 2008
Adjusting by density per area
P - value
P - value
mastectomy rates report an impact due to cost of treatment33 while others found no differences based on income.18 All of these measures examined in the literature are at the patient level, and none have examined these variables at the community level as we have done in our analysis. We found a significant impact of education level of the community, with a decrease in the likelihood of a patient to undergo mastectomy in a community with a higher education level. We found an influence of employment, but not income level, of the community on the mastectomy rates. It is possible that, if patient-level data were to be examined for these variables, the results would be different. Although there are interesting findings in this study there are some weaknesses that must be recognized. First, the SEER population is more affluent, more educated, less rural, and lower in unemployment than
2651
DISPARITY IN URBAN AND RURAL MASTECTOMY
the US population as a whole.41 While the SEER database is not inclusive of the entire USA in that it covers specific regions, it is the most comprehensive database available for public use. Second, the SEER provides important patient-level data, but it does not include many variables that have a potential impact on receipt of mastectomy. We attempted to analyze some of those factors by linking it to communitylevel data in the ARF. While this does give some insight into the influence of those factors it must be recognized that those are community-level factors. We believe that this is an important analysis because we have previously compared the impact of patientand community-level data regarding the utilization of immediate breast reconstruction and found that community-level factors had more of an impact in some categories than patient-level factors.42 While this may not provide definitive results it does provide a basis for further study. Lastly, our definition of rural and urban populations may not be the same for others who would attempt this analysis. We defined counties under 20,000 people as rural; however, this definition is contradictory to the census definition of rural, which is a population under 2,500.43 We believe that the selected criteria is a valid because it defines 62.5% of the counties as rural and the remainder as urban, providing a good separation of the communities, and it places 10% of the people in the rural category, which is consistent with the US population. A significant difference in mastectomy rates has been identified between rural and urban populations. Patient-level variables that play a role in that difference are stage, race, and marital status. Community level factors that played a role were employment and education level of the community and density of radiation technologists. While these variables are significant in our analysis, the increased likelihood of mastectomy in the rural population persists despite controlling for both patient- and community-level factors.
CONCLUSION Significant differences between likelihood to undergo mastectomy in rural and urban populations are identified. This difference is partially attributable to the availability of radiation facilities, specifically, the number of radiation technologists per population density. In addition, age did not impact the likelihood of mastectomy, but marital status did. Despite controlling for both patient- and community-level factors, the disparity between urban and rural breast
cancer care persisted. Future work is needed to make breast-conserving surgery a viable option for rural patients needing radiation therapy.
REFERENCES 1. Early Breast Cancer Trialists’ Collaborative Group. Effects of radiotherapy and surgery in early breast cancer. An overview of the randomized trials. N Engl J Med 1995; 333:1444–55. 2. Fisher B, Anderson S, Bryant J, et al. Twenty-year follow-up of a comparing total mastectomy, lumpectomy, and lumpectomy plus irradiation for the treatment of invasive breast cancer. N Engl J Med 2000; 347:1233–41. 3. Veronesi U, Cascinelli N, Mariani L, et al. Twenty-year followup of a randomized study comparing breast-conserving surgery with radical mastectomy for early breast cancer. N Engl J Med 2002; 347:1227–32. 4. Farrow DC, Hunt WC, Samet JM. Geographic variation in the treatment of localized breast cancer. N Engl J Med 1992; 326:1097–101. 5. Farrow DC, Samet JM, Hunt WC. Regional variation in survival following the diagnosis of cancer. J Clin Epidemiol 1996; 49:843–7. 6. Howe HL, Johnson TP, Lehnherr M, et al. Patterns of breast cancer treatment: a comparison of a rural population with an urban population and a community clinical oncology program sample. Cancer Control 1995; 2:113–20. 7. Howe HL, Katterhagen JG, Yates J, et al. Urban-rural differences in the management of breast cancer. Cancer Causes Control 1992; 3:533–9. 8. Gold HT, Dick AW. Variations in treatment for ductal carcinoma in situ in elderly women. Med Care 2004; 42:267–75. 9. Mitchell KJ, Fritschi L, Reid A, et al. Rural-urban differences in the presentation, management and survival of breast cancer in western Australia. Breast 2006; 15:769–76. 10. Schroen AT, Brenin DR, Kelly MD, et al. Impact of patient distance to radiation therapy on mastectomy use in early-stage breast cancer patients. J Clin Oncol 2005; 23:7074–80. 11. Tropman SE, Ricketts TC, Paskett E, et al. Rural breast cancer treatment: evidence from the reaching communities for cancer care (REACH) project. Breast Cancer Res Treat 1999; 56:59–66. 12. Meden T, St John-Larkin C, Hermes D, et al. MSJAMA. Relationship between travel distance and utilization of breast cancer treatment in rural northern Michigan. JAMA 2002; 287:111. 13. Nattinger AB, Gottlieb MS, Veum J, et al. Geographic variation in the use of breast-conserving treatment for breast cancer. N Engl J Med 1992; 326:1102–7. 14. Nattinger AB, Kneusel RT, Hoffmann RG, et al. Relationship of distance from a radiotherapy facility and initial breast cancer treatment. J Natl Cancer Inst 2001; 93:1344–6. 15. Fontana V, Castro T, Polynice A. Preferences of healthy inner city women and the surgical treatment of early stage breast cancer. Am Surg 2007; 73:215–21. 16. Schootman M, Aft R. Rural-urban differences in radiation therapy for ductal carcinoma in-situ of the breast. Breast Cancer Res Treat 2001; 68:117–25. 17. Stafford D, Szczys R, Becker R, et al. How breast cancer treatment decisions are made by women in North Dakota. Am J Surg 1998; 176:515–9. 18. Desch CE, Penberthy L, Newschaffer CJ, et al. Factors that determine the treatment for local and regional prostate cancer. Med Care 1996; 34:152–62. 19. Johantgen ME, Coffey RM, Harris DR, et al. Treating earlystage breast cancer: Hospital characteristics associated with breast-conserving surgery. Am J Public Health 1995; 85:1432–4.
Ann. Surg. Oncol. Vol. 15, No. 10, 2008
2652
L. K. JACOBS ET AL.
20. Gelber RP, McCarthy EP, Davis JW, et al. Ethnic disparities in breast cancer management among Asian Americans and Pacific Islanders. Ann Surg Oncol 2006; 13:977–84. 21. Lannin DR, Mathews HF, Mitchell J, et al. Influence of socioeconomic and cultural factors on racial differences in latestage presentation of breast cancer. JAMA 1998; 279:1801–7. 22. Liff JM, Chow WH, Greenberg RS. Rural-urban differences in stage at diagnosis. Possible relationship to cancer screening. Cancer 1991; 67:1454–9. 23. Fagerlin A, Lakhani I, Lantz PM, et al. An informed decision? Breast cancer patients and their knowledge about treatment. Patient Educ Couns 2006; 64:303–12. 24. Bradley CJ, Neumark D, Luo Z, et al. Employment and cancer: findings from a longitudinal study of breast and prostate cancer survivors. Cancer Invest 2007; 25:47–54. 25. Maunsell E, Brisson C, Dubois L, et al. Work problems after breast cancer: an exploratory qualitative study. Psychooncology 1999; 8:467–73. 26. Area Resource File (ARF) 2004 Release. National Center for Health Workforce Analysis, Bureau of Health Professions, Health Resources and Services Administration, Department of Health and Human Services. 27. US National Institutes of Health. Surveillance epidemiology and end results: Providing information on cancer statistics to help reduce the burden of this disease on the US population. Available: http://www.seer.cancer.gov/ [accessed June 25, 2007]. 28. Cultice J (2000) Area resource file: National county-level health resource information database. Available: http://www.arfsys. com/ [accessed June 25, 2007]. 29. National Cancer Institute. Available: http://www.cancer.gov/ [accessed June 25, 2007]. 30. Ries L, Eisner M, Kosary C, et al. (2005) SEER Cancer Statistics Review, 1975–2002. Bethesda, MD: National Cancer Institute. 31. Osborne MP. William Stewart Halsted: his life and contributions to surgery. Lancet Oncol 2007; 8:256–65. 32. Beaulieu JE, Massey CS, Tucker TC, et al. Rural-urban variation in breast-conserving surgery in Kentucky. J Ky Med Assoc 2003; 101:455–9.
Ann. Surg. Oncol. Vol. 15, No. 10, 2008
33. Kim P. Cost of cancer care: the patient perspective. J Clin Oncol 2007; 25:228–32. 34. Athas WF, Adams-Cameron M, Hunt WC, et al. Travel distance to radiation therapy and receipt of radiotherapy following breast-conserving surgery. J Natl Cancer Inst 2000; 92:269–71. 35. Schootman M, Fuortes LJ. Breast and cervical carcinoma: the correlation of activity limitations and rurality with screening, disease incidence, and mortality. Cancer 1999; 86:1087–94. 36. Liff JM, Sung JF, Chow WH, et al. Does increased detection account for the rising incidence of breast cancer? Am J Public Health 1991; 81:462–5. 37. Answini GA, Woodard WL, Norton HJ, et al. Breast conservation: trends in a major southern metropolitan area compared with surrounding rural counties. Am Surg 2001; 67:994–8. 38. Bao Y, Fox SA, Escarce JJ. Socioeconomic and racial/ethnic differences in the discussion of cancer screening: ‘‘between-’’ versus ‘‘within-’’ physician differences. Health Serv Res 2007; 42:950–70. 39. Gelber RP, McCarthy EP, Davis JW, et al. Ethnic disparities in breast cancer management among Asian Americans and Pacific islanders. Ann Surg Oncol 2006; 13:977–84. 40. Clifford E, Clifford M, Georgiade NG. Breast reconstruction following mastectomy: II. Marital characteristics of patients seeking the procedure. Ann Plast Surg 1980; 5:344–6. 41. Nattinger AB, McAuliffe TL, Schapira MM. Generalizability of the surveillance, epidemiology, and end results registry population: factors relevant to epidemiologic and health care research. J Clin Epidemiol 1997; 50:939–45. 42. Rosson GD, Singh NK, Ahuja N, et al. Multilevel analysis of the impact of community versus patient factors on access to immediate breast reconstruction following mastectomy in the state of Maryland. Arch Surg (in press). 43. United States Census Bureau (2007) Available: http://www.census. gov. [accessed on June 25, 2007].