SEARCH

SEARCH BY CITATION

Keywords:

  • rural;
  • breast cancer;
  • lung cancer;
  • cervical cancer;
  • anal cancer;
  • rectal cancer;
  • radiation therapy

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Note Added in Proof
  8. REFERENCES

BACKGROUND:

Rural populations have limited geographic access to radiation therapy. The current study examines whether rural patients with cancer are less likely than urban patients with cancer to receive recommended radiation therapy, and identifies factors influencing rural versus urban differences in radiation therapy receipt.

METHODS:

The current study included 14,692 rural and 107,834 urban patients with 5 cancer types and stages for which radiation therapy was recommended. The authors used 2000 to 2004 Surveillance, Epidemiology, and End Results (SEER) Limited-Use Data from 8 state-based (California, Connecticut, Hawaii, Iowa, Kentucky, Louisiana, New Mexico, and Utah) and 3 county-based (Atlanta, rural Georgia, and Seattle/Puget Sound) cancer registries. Adjusted radiation therapy receipt rates were calculated by rural versus urban residence overall, for different sociodemographic and cancer characteristics, and for different states based on logistic regression analyses using general estimating equation methods to account for patient clustering by county.

RESULTS:

Adjusted rates of radiation therapy receipt were lower for rural (62.1%) than urban (69.1%) patients with breast cancer (P ≤ .001). Among patients with breast cancer, radiation therapy receipt differed more by sociodemographic characteristics (eg, rural patients aged < 50 years had a 67.1% receipt rate, whereas those aged ≥ 80 years had a radiation therapy receipt rate of 29.1%) than rural versus urban residence. Adjusted rates of radiation therapy receipt were similar for rural and urban patients with other cancer types overall (66.1% vs 68.2%; difference not significant), although there were differences between urban and rural patients with regard to radiation therapy receipt for patients with stage IIIA nonsmall cell lung cancer (66.2% vs 60.7%; P ≤ .01).

CONCLUSIONS:

Sociodemographics, cancer types and stages, and state of residence appear to have a greater influence over receipt of radiation therapy than rural versus urban residence location, suggesting that factors such as social support, receipt of other cancer treatments, and regional practice patterns are important determinants of radiation therapy receipt. Cancer 2012. © 2012 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Note Added in Proof
  8. REFERENCES

Radiation therapy is recommended as primary or adjuvant therapy for many cancers to prevent local recurrence, improve survival, or palliate, depending on the cancer type and stage.1 Curative radiation therapy usually involves daily treatment for 3 to 9 weeks, which is a substantial commitment of time and resources. Radiation therapy requires regionalized, specialized treatment facilities for which geographic access can be limited. Indeed, nearly 40% of patients with colorectal cancer residing in small and isolated rural areas and 30% residing in large rural areas must travel > 50 miles to receive radiation oncology consultation. Geographic barriers may inhibit rural populations from receiving recommended radiation therapy.2

The literature examining radiation therapy use has focused on patients with early stage breast cancer, and demonstrated that increased distance from radiation treatment centers and decreased population density reduce the likelihood of obtaining recommended treatment involving adjuvant radiation therapy (lumpectomy and radiation vs mastectomy).3-7 Increased distance from radiation treatment centers also decreases the use of recommended adjuvant radiation treatment after breast-conserving surgery.3, 4, 8, 9 However, the majority of these studies were conducted in single states or a single medical center. In addition, none of these studies examined the association between receipt of radiation therapy among rural patients with breast cancer and county-level environmental characteristics such as level of rurality, poverty, education, and employment. Lastly, we found only 1 study comparing receipt of recommended radiation therapy for other, nonbreast cancers (in this case, lung cancer) between rural and urban patients.10

This research used national, population-based Surveillance, Epidemiology, and End Results (SEER) program data to fill these information gaps by comparing radiation therapy receipt for 5 cancer types (anus, breast, cervix, lung, and rectum) between rural and urban patients, and by examining whether county-level environmental factors are associated with the receipt of radiation therapy among rural patients. These findings can inform cancer centers, program planners, and policy makers as they launch cancer services and adapt cancer programs for rural areas.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Note Added in Proof
  8. REFERENCES

Data Sources

This research used 2000 to 2004 Surveillance, Epidemiology, and End Results (SEER) Limited-Use Data from 8 state-based (California, Connecticut, Hawaii, Iowa, Kentucky, Louisiana, New Mexico, and Utah) cancer registries and 3 county-based (Atlanta, rural Georgia, and Seattle/Puget Sound) cancer registries in 2 other states. These registries include 303 rural and 165 urban counties. Registries with urban counties only were excluded (Detroit, Michigan; and New Jersey). SEER data provide chart-abstracted information on cancer histology, type, stage, extent of disease, and initial treatment (eg, surgery, radiation); Federal Information Processing Standard (FIPS) county codes; and demographic characteristics (eg, age, sex, and race/ethnicity). Linked Area Resource File data provided radiation oncologist availability in each county and study year. The US Department of Agriculture 2004 Economic Research Service County Typology codes identify counties that were considered to have low education (≥ 25% of residents aged 25 years-64 years had neither a high school diploma nor a general education diploma in 2000), low employment (< 65% of residents aged 21 years-64 years were employed in 2000), and persistent poverty (≥ 20% of residents were considered to be poor across 4 censuses: 1970, 1980, 1990, and 2000).11

Study Population

We identified 159,133 patients aged ≥ 18 years at the time of diagnosis with 5 cancer types and stages for which the National Comprehensive Cancer Network (NCCN)-published clinical oncology practice guidelines recommend radiation therapy as the primary or adjuvant treatment: stages II or III anal cancer, ductal carcinoma in situ (DCIS) of the breast, T categories 1-4 breast cancer, stages IB2 to IVA cervical cancer, stages IA to IIIB small cell lung cancer, stages IIA or IIIB nonsmall cell lung cancer, or stages II or III rectal cancer.12 We excluded 1410 patients with atypical histologies, 23 patients with a cancer diagnosis noted on the death certificate or at autopsy only, and 29,726 patients with prior cancers other than nonmelanoma skin cancer. We identified 5448 patients with missing information regarding radiation treatment, 1062 of whom lived in rural counties and 4386 of whom lived in urban counties. Because proportionately more radiation therapy data were missing for rural patients, we conducted sensitivity analyses designating patients with missing data first as having received radiation and then as not having received radiation, and found no difference in our findings. On the basis of this analysis, we chose to exclude the 5448 patients for whom information regarding radiation treatment was missing. Our final sample size was 122,526 patients. Table 1 indicates the number of rural and urban patients by cancer type and stage. SEER TNM staging criteria changed in 2004. For data from 2000 through 2003, we used the extent of disease variables: tumor size, extension, and lymph node involvement to match 2004 staging criteria.

Table 1. Inclusion Criteria and Sample Sizes for Different Cancer Types and Stages
Cancer Type and StageCancer Site CodesHistology CodesPatient Inclusion CriteriaNo. of Rural PatientsNo. of Urban Patients
  1. Abbreviation: DCIS, ductal carcinoma in situ.

Anus, stages II and IIIC210, C211, C212, C2188000.3, 8010.3, 8070.3, 8071.3, 8072.3, 8076.3, 8083.3, 8124.3With or without surgery96887
Breast, DCISC500-C506, C508, C5098230.2, 8453.2, 8500.2, 8501.2, 8503.2, 8504.2, 8507.2With lumpectomy146112,353
Breast, T categories 1-4C500-C506, C508, C5098000.3, 8001.3, 8140.3, 8211.3, 8230.3, 8255.3, 8310.3, 8323.3, 8440.3, 8450.3, 8453.3, 8470.3, 8480.3, 8481.3, 8490.3, 8500.3, 8501.3, 8502.3, 8503.3, 8504.3, 8508.3, 8510.3, 8512.3, 8513.3, 8514.3, 8520.3, 8521.3, 8522.2, 8523.3, 8524.3, 8525.3, 8540.3, 8541.3, 8543.3, 8550.3, 8560.3, 8562.3, 8570.3, 8571.3, 8572.3, 8573.2, 8573.3T1/T2 with lumpectomy, or with ≥4 positive lymph nodes and mastectomy897174,133
T3/T4 with either lumpectomy or mastectomy
Cervix, stages IB2-IVAC530, C531, C538, C5398001.3, 8010.3, 8070.3, 8071.3, 8076.3, 8140.3, 8144.3, 8255.3, 8261.3, 8262.3, 8263.3, 8323.3, 8384.3, 8481.3, 8482.3, 8570.3Stages IB2 and IIA without hysterectomy or pelvic exenteration3752367
All other stages with or without hysterectomy or pelvic exenteration
Small cell lung, stages IA-IIIBC341-343, C3488002.3, 8041.3, 8042.3, 8043.3, 8044.3, 8045.3Stages I and II without pneumonectomy9723541
Stage III with or without pneumonectomy
Nonsmall cell lung, stages IIIA and IIIBC341-343, C3488000.3, 8001.3, 8010.3, 8011.3, 8012.3, 8020.3, 8021.3, 8046.3, 8070.3, 8071.3, 8072.3, 8073.3, 8076.3, 8083.3, 8084.3, 8094.3, 8140.3, 8141.3, 8145.3, 8230.3, 8231.3, 8255.3, 8263.3, 8480.3, 8560.3, 8570.3, 8572.3Stage IIIA without pneumonectomy17398412
Stage IIIB without T4N0M0 and T4N1M0
Excluded if 1) separate tumor nodules within the same lobe, 2) malignant pleural effusion, 3) pleural tumor foci separate from direct pleural invasion, or 4) involvement of vertebra or neural foramina
Rectum, stages II and IIIC2098000.3, 8020.3, 8021.3, 8124.3, 8140.3, 8144.3, 8145.3, 8210.3, 8211.3, 8220.3, 8221.3, 8230.3, 8255.3, 8261.3, 8262.3, 8263.3, 8480.3, 8481.3, 8490.3, 8560.3With surgical resection10786141
Total14,692107,834

Study Variables

Outcome variable

The study's outcome variable was the receipt of any radiation therapy. The SEER data indicate whether radiation therapy was offered or received, and the radiation therapy method used in the first treatment course. Very few patients were offered but did not receive radiation therapy (1.2% of patients with breast cancer and 1.8% of patients with 1 of the other 4 cancer types). This occurred when the patient or the patient's guardian refused radiation therapy or when radiation therapy was recommended but it is unknown whether it was administered. Thus, these patients were designated as not having received radiation therapy.

Independent variables of interest

The primary independent study variable was the urban versus rural residence location of the cancer patients. Patient county of residence was classified as metropolitan (urban) or nonmetropolitan (rural) using Urban Influence Codes (UICs)13 linked to residence county FIPS codes. We subdivided nonmetropolitan counties into “adjacent rural” (geographically adjacent to a metropolitan area [UICs 3-7]), “remote micropolitan” (not adjacent to a metropolitan county and with a town/urban cluster of 10,000 to 49,999 residents [UIC 8]), and “small rural” (not adjacent to a metropolitan county and without a town of 10,000 residents [UICs 9-12]). For the final analyses, we aggregated cancer patients across the 3 rural county types because of their small sample sizes and because their adjusted radiation therapy rates were not statistically significantly different from one another.

Additional county-level environmental variables, which served as secondary independent variables, included county designation as having persistent poverty, low employment, or low education, and radiation oncologist availability in the diagnosis year.

Control variables

Control variables included patient sociodemographics (age, sex, marital status, and race/ethnicity). Residence state controlled for regional practice variation.

Statistical Analysis

Because the majority of study patients had breast cancer (n = 96,918) and there were limited numbers of patients with other cancers, we divided the study population into those with breast cancer and those with other, nonbreast cancers (called “other cancers”), and conducted separate analyses for these 2 groups. We calculated unadjusted rates of radiation therapy receipt by rural versus urban residence location overall, for patients with different sociodemographic and cancer characteristics, and by state. We used logistic regression analysis to examine the relationship between residence location (rural vs urban) and receipt of radiation therapy, controlling for sociodemographic characteristics, cancer type and stage, and state of residence.

We conducted logistic regression analysis among rural patients only to determine whether key county-level variables (level of rurality [adjacent rural/remote micropolitan/small rural], radiation oncologist availability, and environmental economic and social characteristics [persistent poverty, low education, and low employment]) influenced whether patients received radiation therapy. We tested for and found statistically significant interactions between the county-level environmental economic and social characteristics and both the level of rurality and radiation oncologist availability. For this reason, we stratified this analysis by these environmental economic and social characteristics. The 3 county-level environmental economic and social characteristics were found to be highly correlated and produced similar regression analysis findings. We chose the best-fitting model based on the lowest −2 log-likelihood score.

We applied general estimating equation methods to all the logistic regression models to account for patient clustering by county.14 Because we tested multiple comparisons, only statistical differences at a P level of ≤ .01 were considered to be statistically significant. From the logistic models, we calculated adjusted rural and adjusted urban rates of radiation therapy receipt overall and for each of the different sociodemographic characteristics, cancer types and stages, and states of residence. We also calculated adjusted rates of radiation therapy receipt for rural patients in different county types. We tested for statistically significant differences between the adjusted rates using z-scores. Because state of residence was found to be strongly associated with receipt of radiation therapy, and rural patients were largely found in only a few states, we report adjusted findings only.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Note Added in Proof
  8. REFERENCES

Of the 122,526 cancer patients meeting study criteria, 107,834 were urban and 14,692 were rural residents (Table 2). Just over 10% of patients with breast cancer and 16.6% of patients with other cancers were rural residents. The mean age was 59.9 years for breast cancer patients and 64.6 years for patients with other cancers (P ≤ .001). We excluded male breast cancer patients, for whom treatment recommendations differ. Approximately one-half of the patients with other cancers were female. Approximately three-quarters of both cancer groups were white non-Hispanic, with the remainder relatively evenly divided between African American, Asian American, and Hispanic/Latino patients. Just greater than one-half of patients in both the breast and other cancers groups were married or partnered; approximately one-quarter were single, separated, or divorced; and the remainder were widowed. Approximately half of all patients were from California, which includes 4 SEER registries. Of those patients with other cancers, greater than one-half had small cell or nonsmall cell lung cancer, and greater than one-quarter had rectal cancer.

Table 2. Patient Sociodemographic and Cancer Characteristics, and State of Residence by Breast and Other Cancer Types
 Breast CancerOther Cancers
CharacteristicsUrbanb (N = 86,486)Rural (N = 10,432)Totalc (N = 96,918)Urbanb (N = 21,348)Rural (N = 4260)Total (N = 25,608)
  • Abbreviations: AI/AN, American Indian/Alaska Native; DCIS, ductal carcinoma in situ.

  • a

    Missing values for race: breast cancer, 450; other cancers, 36. Missing values for marital status: breast cancer, 2746; other cancers, 853.

  • b

    Symbols adjacent to “urban” columns indicate significance of urban versus rural differences within the group of patients with breast cancer or the group of patients with other cancers.

  • c

    Symbols adjacent to “total” breast cancer column indicate significance of overall differences between the group of patients with breast cancer and the group of patients with other cancers.

  • d

    P ≤ .001.

  • e

    P ≤ .01.

Age, y      
<5025.5d22.725.2d14.5d11.714.1
50-6438.138.038.131.933.232.1
65-6910.111.010.213.715.614.1
70-749.09.69.114.115.214.3
75-798.38.58.312.913.113.0
≥809.010.29.113.011.112.5
Female100.0100.0100.050.6d46.249.9
Racea      
White, non-Hispanic74.1d87.275.5d71.5d87.274.1
Black, non-Hispanic7.94.77.69.96.59.4
AI/AN, non-Hispanic0.30.70.30.40.90.5
Asian, non-Hispanic7.83.87.48.33.07.4
Latino/Hispanic9.83.59.19.82.38.5
Other non-Hispanic0.10.00.10.10.00.1
Marital statusa      
Married/partnered59.8d64.160.2d54.3d59.455.1
Single/separated/divorced24.717.723.927.221.726.3
Widowed15.618.215.818.519.018.6
State      
California57.7d13.552.9d53.7d10.946.6
Connecticut8.16.07.97.43.76.8
Georgia5.61.55.15.01.74.5
Hawaii2.36.72.72.04.32.4
Iowa3.123.45.33.819.66.4
Kentucky3.819.15.47.029.010.7
Louisiana6.412.87.19.718.711.2
New Mexico2.17.42.71.95.82.5
Utah2.82.82.81.71.61.7
Washington8.26.88.07.74.77.2
Cancer site and stage      
Anus, stage II   2.7d1.42.5
Anus, stage III   1.4d0.81.3
Cervix, stage I   0.50.40.5
Cervix, stage II   3.93.53.9
Cervix, stage III   6.1d4.05.8
Cervix, stage IV   0.60.80.6
Nonsmall cell lung, stage IIIA   21.522.621.7
Nonsmall cell lung, stage IIIB   17.918.318.0
Small cell lung, stage I   1.82.31.9
Small cell lung, stage II   0.7d1.20.8
Small cell lung, stage III   14.1d19.315.0
Rectum, stage II   12.6e11.212.4
Rectum, stage III   16.1d14.115.8
Breast, T157.656.457.5   
Breast, T220.921.320.9   
Breast, T34.95.25.0   
Breast, T42.3d3.12.4   
Breast, DCIS14.314.014.3   
Level of rurality      
Urban100.0 89.2d100.0 83.4
Adjacent rural 57.76.2 57.79.6
Remote micropolitan 26.82.9 22.63.8
Remote rural 15.51.7 19.73.3
Radiation oncologist in residence county94.6d25.887.2d92.6d21.080.7
Living in county with:      
Persistent poverty2.0d17.93.7d2.7d27.66.8
Low employment5.4d27.07.7d6.8d36.611.8
Low education21.420.821.3d21.8d33.023.6

Overall, patients with other cancers were significantly more likely to live in rural counties and counties with persistent poverty, lower employment, and less education than patients with breast cancer (Table 2). Greater than one-half of rural patients with breast and other cancers lived in counties adjacent to metropolitan counties. Patients with other cancers were less likely to live in a county with a radiation oncologist than those with breast cancer. Only 25.8% of rural patients with breast cancer and 21.0% of rural patients with other cancers had a radiation oncologist in their counties (P ≤ .001).

In the adjusted analysis, a lower percentage of rural (62.1%) compared with urban (69.1%) patients with breast cancer received radiation therapy (P ≤ .001). These differences between rural and urban patients were consistent across patient age, race, and marital status (Fig. 1). Notably, there were no significant differences with regard to radiation therapy receipt between the rural racial and ethnic groups (P = .25). For the African American and Hispanic/Latino patient groups, the absence of a significant rural versus urban difference was not because rural populations reached a high urban radiation therapy receipt rate; rather, urban populations had a low rate of radiation therapy receipt that paralleled that of rural patients. African American and Hispanic/Latino patients had the lowest rates of radiation therapy receipt of all racial/ethnic groups.

thumbnail image

Figure 1. Adjusted rates of radiation therapy receipt are shown for patients with breast cancer by rural versus urban residence and patient characteristics. Rates were adjusted for age, race, marital status, state of residence, and cancer site and stage. † indicates P ≤ .001; ‡, P ≤ .01.

Download figure to PowerPoint

For patients with breast cancer, differences in radiation therapy receipt by sociodemographic characteristics were greater than by rural versus urban residence location (Fig. 1). The oldest and widowed rural patients had the lowest rates among patients in different sociodemographic groups; rural residents in these groups had some of the very lowest rates of radiation therapy receipt. There was striking regional variation in the receipt of radiation therapy, with a nearly 30% absolute difference in radiation therapy receipt noted between rural patients with breast cancer in Connecticut and Georgia compared with Washington. There were similar differences in these rates noted between urban patients with breast cancer in Connecticut and Washington. Some states had more dramatic rural versus urban differences (Georgia and Hawaii) than others (Iowa, Utah, and Washington). Rates of radiation therapy receipt were significantly lower for rural compared with urban patients with T1 and T2 breast cancer, but not for those with DCIS, T3, or T4 tumors. Patients with T3 and T4 breast cancer had the lowest rates of radiation therapy receipt.

Among patients with other, nonbreast cancer, rural (66.1%) and urban (68.2%) patients had similar rates of radiation therapy receipt (P = .18). This lack of a rural versus urban difference was consistent across most sociodemographic groups, cancer types and stages, and states of residence (Fig. 2). Similar to patients with breast cancer, the oldest patients with other cancers had the lowest rates of radiation therapy receipt, and rural residents in these groups had the very lowest rates. Rural populations who were less likely than their urban counterparts to receive radiation therapy included only those aged 65 years to 69 years, non-Hispanic whites, and patients with stage IIIA nonsmall cell lung cancer. Unlike for breast cancer, there was substantial consistency in radiation therapy receipt across states.

thumbnail image

Figure 2. Adjusted rates of radiation therapy receipt are shown for patients with other cancers by rural versus urban residence and patient characteristics. Rates were adjusted for age, race, marital status, state of residence, and cancer site and stage. † indicates P ≤ .001; ‡, P ≤ .01.

Download figure to PowerPoint

County-based environmental characteristics influenced radiation therapy receipt among rural patients with breast cancer but not other cancers (Fig. 3). For patients with breast cancer living in low-employment counties only, having a radiation oncologist located within the county was associated with the increased receipt of radiation therapy. For both patients with breast cancer and those with other cancers, rural counties with some of the highest radiation therapy rates were remote micropolitan counties without low employment, whereas the lowest rates of radiation therapy receipt were noted among residents of remote micropolitan counties with low employment. With one exception, there were no differences in radiation therapy rates noted between patients living in adjacent rural, remote micropolitan, and small rural counties for either those with breast cancer or other cancers.

thumbnail image

Figure 3. Adjusted rates of radiation therapy receipt are shown for rural cancer patients by county characteristics. Rates were adjusted for age, race, marital status, state of residence, and cancer site and stage. † indicates P ≤ .001; ‡, P ≤ .01.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Note Added in Proof
  8. REFERENCES

The findings of the current study are consistent with literature demonstrating that rural patients with breast cancer, especially those living farther from radiation therapy facilities, are less likely than their urban counterparts to receive guideline-recommended radiation therapy.3, 15 This was not true for patients with other cancers. We found few published studies from the United States examining radiation therapy receipt for patients with cancers other than those of the breast. One lung cancer study demonstrated similar rural versus urban radiation therapy receipt rates, but did not take into account cancer stage and type and receipt of other treatments.10 We hypothesized that the similar radiation therapy receipt rate noted between patients with other cancers in rural and urban residence locations in the current study might be explained by the heterogeneity of the group of patients with other cancers, but the lack of a rural versus urban difference in radiation therapy receipt was generally consistent across cancer types in this group. Additional studies examining rural versus urban differences in receipt of radiation therapy, ideally including large numbers of patients with different types of cancers, are needed to confirm this study's results.

The finding that, for rural patients with breast cancer, radiation oncologist availability was associated with receipt of radiation therapy only in low-employment counties suggests that access to local services may be more important in socioeconomically disadvantaged locations. The higher rate of radiation therapy receipt in remote micropolitan counties without low employment compared with remote micropolitan counties with low employment deserves further exploration. Rural counties with larger population centers remote from urban centers may be more self-contained, and those with more individuals experiencing socioeconomic distress may be less able to facilitate access to specialty medical care for economically disadvantaged patients.

Among patients with breast cancer, sociodemographic factors were found to be associated with a much larger absolute variation in radiation therapy receipt than residence location. As in other studies,16-18 the oldest patients had the lowest rates of radiation therapy receipt. There was a nearly 30-percentage point difference in the rates between breast cancer patients aged ≥ 80 years and those aged ≤ 50 years. The finding that widowed rural patients with breast cancer had the lowest radiation treatment rates compared with rural married and single/separated/divorced patients suggests that loss of social support may strongly influence a patient's choice to seek complete therapy.

Variation in radiation therapy receipt rates between patients with other types of cancer and stages of disease appears to be associated with the indication for and effectiveness of this therapy. The highest rates have been noted among patients with anal and cervical cancers, which use radiation therapy as primary treatment and for which treatment is quite effective.19, 20 Patients with lung cancer, for which survival and radiation effectiveness are much lower, had lower radiation therapy rates.21-23

Patients with T3 and T4 breast cancer who underwent lumpectomy had the lowest rates of radiation therapy receipt, a finding we are unable to explain. The dramatic state-based differences in radiation therapy receipt suggest that health professionals may vary with regard to the benefit they attribute to radiation therapy for patients with breast cancer, and that medical norms regarding radiation therapy for breast cancer vary by region. Punglia et al24 demonstrated variation by institutions participating in the NCCN with regard to the use of guideline-recommended radiation therapy after mastectomy. Katz et al25 found that breast cancer patients at different sites reported differential rates of surgeon recommendation for radiation therapy. There is published literature questioning whether radiation therapy is needed for all patients with breast cancer.26, 27 Given these findings, professional discretion is likely used in recommending radiation therapy to breast cancer patients.

There were few differences noted with regard to receipt of radiation therapy between rural and urban patients with other cancers. The presence of a radiation oncologist in rural counties influenced radiation therapy receipt only in those counties with lower socioeconomic resources. As in other studies, these findings demonstrate the importance of financial resources in obtaining recommended medical services.28, 29

SEER cancer registry data have several limitations. First, the geographic identifiers are at the county level, prohibiting the calculation of distances between cancer patients' residences and radiation therapy services. Thus, although we have reported radiation therapy receipt in rural locations with different population sizes and levels of isolation from urban counties, we were unable to examine the degree to which distance from a radiation treatment center mediates our findings. Other studies have demonstrated that distance to radiation therapy centers is associated with radiation therapy receipt after breast-conserving surgery.3, 30, 31

Information concerning other factors that could influence the receipt of radiation therapy are not available in SEER, such as individual measures of socioeconomic status (eg, education, income), insurance status, comorbidity, and a direct measure of social support. Fortunately, SEER data include age and marital status, which are associated with comorbidity and social support, respectively. Our analysis used general estimating equation methods to account for factors such as clinical practice patterns that might be clustered by county. This hierarchical analysis will not change the rate estimates of radiation therapy receipt, but will increase the standard errors around the estimates. Because rural versus urban status is measured at the county level, this analysis may lead to an overly conservative estimate of the statistical significance of rural versus urban differences. Lastly, it is difficult to study cancer care in rural areas because of the relatively small number of patients, even in the powerful SEER population-based data source. To mitigate this limitation, we aggregated cancers other than those of the breast to gain an adequate sample size for study. We acknowledge, however, that variation in the indication for and effectiveness of radiation therapy among patients with different cancers makes this aggregation less desirable.

To the best of our knowledge, the current study is the first to examine radiation therapy receipt for multiple cancer types in rural areas. We found that rural residence can influence radiation therapy receipt, but in selected cancers only, and that sociodemographic factors, cancer types and stage, and state of residence have even greater influence over cancer patients' receipt of radiation therapy. These factors are powerful for urban as well as rural patients with cancer. This illustrates the need for further qualitative research to determine whether patients fully understand and health care providers effectively communicate the benefits and risks of radiation therapy so that patients can make fully informed decisions about its use, regardless of where they live.

Note Added in Proof

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Note Added in Proof
  8. REFERENCES

FUNDING SUPPORT

This University of Washington Rural Health Research Center study was supported by the Office of Rural Health Policy, Health Resources and Services Administration, US Department of Health and Human Services (U1CRHQ3712-04).

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Note Added in Proof
  8. REFERENCES