SEARCH

SEARCH BY CITATION

Keywords:

  • health care disparities;
  • colorectal cancer;
  • National Cancer Institute-designated cancer centers;
  • socioeconomic factors;
  • access to care;
  • insurance coverage

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

BACKGROUND

National Cancer Institute (NCI) cancer centers provide high-quality care and are associated with better outcomes. However, racial and ethnic minority populations tend not to use these settings. The current study sought to understand what factors influence minority use of NCI cancer centers.

METHODS

A data set containing California Cancer Registry (CCR) data linked to patient discharge abstracts identified all patients with colorectal cancer (CRC) who were treated from 1996 through 2006. Multivariable models were generated to predict the use of NCI settings by race. Geographic proximity to an NCI center and patient sociodemographic and clinical characteristics were assessed.

RESULTS

Approximately 5% of all identified patients with CRC (n = 79,231) were treated in NCI settings. The median travel distance for treatment for all patients in all hospitals was ≤ 5 miles. A higher percentage of minorities lived near an NCI cancer center compared with whites. A baseline multivariable model predicting use showed a negative association between Hispanic ethnicity and NCI center use (odds ratio, 0.71; 95% confidence interval, 0.64-0.79). Asian/Pacific Islander patients were more likely to use NCI centers (odds ratio, 1.41; 95% confidence interval, 1.28-1.54). There was no difference in use noted among black patients. Increasing living distance from an NCI cancer center was found to be predictive of lower odds of use for all populations. Medicare and Medicaid insurance statuses were positively associated with NCI center use. Neighborhood-level education was found to be a more powerful predictor of NCI use than poverty or unemployment.

CONCLUSIONS

Select minority groups underuse NCI cancer centers for CRC treatment. Sociodemographic factors and proximity to NCI centers are important predictors of use. Interventions to address these factors may improve minority attendance to NCI cancer centers for care. Cancer 2014;120:399–407. © 2013 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Racial and ethnic disparities in colorectal cancer (CRC) outcomes have persisted over the last 20 years,[1-4] with a widening of the mortality gap more recently.[5, 6] The literature suggests that differences in the quality of care delivered in the hospitals in which minorities cluster for treatment may be important to cancer outcomes.[7-9] Many investigations have shown that minorities tend to use settings that provide a lower quality of care,[10] whereas others have clearly demonstrated minority distribution into hospitals associated with decreased cancer survival.[11, 12]

The National Cancer Institute (NCI) has recognized 67 cancer centers in the United States.[13] Patients treated in these centers have been shown to benefit from a decreased risk of postoperative mortality and improved long-term survival.[14, 15] Superior outcomes are believed to be due in part to higher compliance with evidence-based cancer care.[16, 17] To our knowledge, minority use of NCI centers has not been extensively studied; there is 1 previous investigation that evaluated use of NCI cancer centers in the Medicare population and found that urban African Americans were more likely than their white counterparts to use an NCI center for cancer care.[18] The authors also found that use of NCI centers decreased as travel time increased. The study was somewhat limited by the homogeneity of the study population (Medicare data limited the study to patients aged > 65 years despite the rising incidence of CRC in individuals aged < 50 years[19]) and could not evaluate the impact of insurance status on use. Finally, the study was underpowered to detect differences in any minority groups other than African Americans.

The purpose of the current study was to determine what factors influence minority use of NCI cancer centers using records from an all-payer, all-age, racially and ethnically diverse data set. We hypothesized that geographic accessibility, insurance status, and neighborhood socioeconomic status (SES) were important determinants of whether minorities use NCI centers for CRC care. Defining the impact of these factors will guide the development of actionable strategies to increase minority access to high-quality care, with the goal of addressing long-standing disparities in cancer treatment and outcomes.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Sources of Data

After obtaining Institutional Review Board approval from the State of California Committee for the Protection of Human Subjects and Stanford University, we analyzed a large state, all-payer, administrative data set comprised of a linkage between the California Cancer Registry (CCR) and the California Office of Statewide Health Planning and Development Patient Discharge Database (OSHPD-PDD). The CCR is a statewide database containing demographic, clinical, and SES data for patients treated in the state for any cancer, excluding nonmelanoma skin cancers. By legislative mandate, all providers treating patients with a primary diagnosis of cancer are required to report clinical encounters, regardless of their nature (radiotherapy, chemotherapy, or surgery), to the registry. The CCR is recognized as one of the most comprehensive and complete cancer registries in the country.[20] Less than 3% of the data regarding race are missing. Demographic variables contained in the CCR include age, sex, and race/ethnicity. SES variables contained in the CCR include measures of median income, rate of poverty, rates of unemployment, and rates of college education at the census-defined block group level. The OSHPD-PDD is an all-payer database containing records for every discharge from a general, acute, nonfederal facility in California. Each record contains International Classification of Diseases, Ninth Revision, Clinical Modification (ICD9-CM) codes indicating the primary diagnosis for the index admission and coding for the primary procedure performed during the hospitalization. The OSHPD-PDD also contains the patient's insurance and a unique hospital identification number indicating where care was delivered. Records from the OSHPD-PDD were linked to the cancer database by the CCR staff using a probabilistic linkage algorithm based on the patient's sex, day and month of birth, and social security number. The matching variables were stripped from the data before disclosure to the investigators.

Study Population

International Classification of Diseases, Ninth Revision, Clinical Modification codes were used to identify patients with a primary diagnosis of CRC (153.xx, 154.0, and 154.1) and concomitant coding for colorectal surgical procedures (45.7-8, 48.31-2, and 48.5-6). Because the outcome of interest relates to the location of hospital-based care, we only included those patients for whom an inpatient discharge was recorded. Patients enrolled in California's largest health maintenance organization system were excluded because enrollment, by definition, dictates the location of care.

Patient Characteristics and Hospital Classification

Patient demographic characteristics (race, sex, and age) were based on classification in the CCR data. Age was treated as a categorical variable in 10-year increments starting after age 44. Neighborhood characteristics were based on census block group data reflecting the percentage unemployed, percentage with a college education, and the percentage of residents living at < 200% of the federal poverty level. These variables were summarized to describe the neighborhood characteristics. NCI hospital classification was based on official recognition and funding support through a Cancer Center Support Grant from the NCI.

Geocoding and Defining Nearby Hospitals

Geocoding was performed using geographical information systems software (ArcGIS 10; Esri Inc, Redlands, Calif). Hospital locations were geocoded based on street addresses. Patient locations were geocoded into geographic coordinates based on their home zip code (or zip code plus; 4 when available) using zip code geometric centroids. This method has been validated in previous studies.[21, 22] All geocoded data were projected to the North America Equidistant Conical Projection coordinate system. The median straight-line distance between patients and their treatment hospital was calculated for all patients across the state. Previous empiric comparisons have shown that straight-line distance is highly correlated with road distance and travel time.[23-25] A hospital was considered to be “nearby” and theoretically available to a patient if it fell within a radius less than or equal to the calculated median travel distance for the state.

Statistical Analysis

Chi-square analysis was used to compare the use of NCI cancer centers between racial/ethnic groups. A baseline multivariable logistic regression model adjusted for race, sex, age, year of diagnosis, and stage of disease estimated the odds of using an NCI center (model 1). Subsequent models assessed the independent association of each factor thought to impact the use of NCI cancer centers and their effect on the coefficients associated with different minority group use. Model 2 assessed travel distance, model 3 assessed insurance type, and model 4 assessed the impact of SES factors. Model 5 was built to distinguish the relative effects of those factors found to have both an independent association with use of an NCI center and significant effects on odds of minority use. Distance was entered into the regression models as a continuous variable, increasing by increments of 5 miles. SES categories were run as continuous variables to reflect increasing levels of unemployment, college education, and poverty, after log transformation of the data. All tests of significance were 2-tailed. Odds estimates were considered to be statistically significant when they were not equal to 1, the 95% confidence interval (95% CI) excluded 1, and the P value was ≤ .05.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

A total of 95,771 patients with CRC were identified. Patients enrolled in the large health maintenance organization (15,987 patients [16.7%]) were excluded. Another 147 records were excluded because they could not be geocoded, and 406 records were excluded because they were coded as American Indian/Alaska Native, for which the cell sizes were too small for reliable estimates. Ultimately, 79,231 records were retained for analysis. There were 387 treating hospitals including 9 NCI-designated or comprehensive cancer centers treating patients with CRC during the study period. Slightly fewer than 5% of patients with CRC were treated in NCI settings. The median travel distance to obtain any care for CRC across the state was 4.7 miles. This calculated median travel distance was used to define “nearby” hospitals for the remainder of the analysis.

NCI Center Availability and Use by Race/Ethnicity

Table 1 shows the sociodemographic and clinical characteristics of the patients living “nearby” an NCI cancer center, (defined as a radius less than the median travel distance for the state (≤ 5 miles)), and those who did not. A slightly higher percentage of patients aged ≥ 75 years and those with rectal cancer lived within the median travel distance to an NCI center. Higher percentages of patients with public insurance (Medicare or Medicaid) or no insurance lived close to an NCI setting compared with those with private insurance. Of those living nearby an NCI cancer center, comparison of neighborhood SES characteristics revealed that there was a slightly lower median percentage unemployment, a higher median percentage college education, and a slightly higher poverty rate compared with those living > 5 miles away. Of those who lived nearby, a higher percentage were racial/ethnic minorities. Although only 10% of the nearby population was white, 26% were Asian/Pacific Islander (API), 14% were Hispanic, and 12% were African American.

Table 1. Distribution of Sociodemographic and Clinical Characteristics by Geographic Proximity to an NCI Cancer Center for Colorectal Cancer: California, 1996 to 2006
CharacteristicTotalLive ≤5 Miles From NCI No. (%)Live >5 Miles From NCI No. (%)
  1. Abbreviations: % College education, percentage of the census block with a college education; % Poverty, percentage of those living below the 200% federal poverty level; % Unemployment, percentage of the neighborhood unemployed; API, Asian/Pacific Islander; IQR, interquartile range; NCI, National Cancer Institute.

Total79,2319816 (12.4)69,415 (87.6)
Age, y   
0-443310425 (12.8)2885 (87.2)
45-547745860 (11.1)6885 (88.9)
55-6412,9291419 (11.0)11,510 (89.0)
65-7421,4882546 (11.9)18,942 (88.1)
75-8424,0243161 (13.2)20,863 (86.8)
≥8597351405 (14.4)8330 (85.6)
Sex   
Male39,1004737 (12.1)34,363 (87.9)
Female40,1315079 (12.7)35,052 (87.3)
Race   
White57,7645970 (10.3)51,794 (89.7)
Black4260512 (12.0)3748 (88.0)
Hispanic93811301 (13.9)8080 (86.1)
API78262033 (26.0)5793 (74.0)
Cancer type   
Colon62,5107463 (11.9)55,047 (88.1)
Rectum16,7212353 (14.1)14,368 (85.9)
Stage of disease (AJCC consolidated staging)   
I16,5632046 (12.4)14,517 (87.6)
II25,0573148 (12.6)21,909 (87.4)
III21,8842671 (12.2)19,213 (87.8)
IV11,2351385 (12.3)9850 (87.7)
Unknown2786348 (12.5)2438 (87.5)
Insurance status   
Private28,5863367 (11.8)25,219 (88.2)
Medicare38,4015024 (13.1)33,377 (86.9)
Medicaid4016646 (16.1)3370 (83.9)
Uninsured1190187 (15.7)1003 (84.3)
Neighborhood characteristics (median, IQR)   
% Unemployment3.2 (1.8-5.2)3.0 (1.6-4.9)3.2 (1.8-5.2)
% College education33.0 (19.4-50.1)43.6 (22.8-60.8)32.1 (19.1-48.4)
% Poverty23.1 (12.3-39.8)24.2 (14.0-41.8)22.9 (12.1-39.6)

Figure 1 shows the percentage of patients who lived nearby and were treated in an NCI setting by race/ethnicity. A higher percentage of African American patients living nearby used NCI centers compared with whites living nearby (15.0% vs 12.4%), but the difference was not statistically significant (P = .09). Hispanics (7.6%; P < .001) and API populations (8.2%; P < .001) used NCI settings in significantly lower percentages than white patients, despite higher proportions of these populations living nearby.

image

Figure 1. National Cancer Institute (NCI) cancer center use rates by those living nearby (Colorectal cancers, California, 1996-2006) demonstrate the distribution of each racial/ethnic group living within the median travel distance (< 5 miles) of an NCI-designated or comprehensive cancer center.

Download figure to PowerPoint

Factors Influencing Minority Use of NCI Settings

Table 2 depicts the results of multivariable models predicting the odds of using a NCI cancer center setting by race/ethnicity with adjustment for demographic and clinical characteristics (baseline model). Similar to the unadjusted comparisons, there was no difference in the odds of use between black and white patients. While Hispanic ethnicity was associated with 29% lower odds (odds ratio [OR], 0.71; 95% CI, 0.64-0.79), API ethnicity was associated with a 41% increased odds of using an NCI center (OR, 1.41; 95% CI, 1.30-1.54) compared with whites. To evaluate the effect of travel distance to an NCI center on racial/ethnic use, the baseline model was adjusted by a continuous variable representing living distance from an NCI center. This is shown in model 2. Distance from home to an NCI setting, in increasing increments of 5 miles, was found to be independently associated with a slightly decreased odds of treatment in an NCI cancer center (OR, 0.93; 95% CI, 0.93-0.94). When the baseline model was adjusted for increasing distance, the negative association between Hispanic ethnicity and NCI use worsened slightly (OR, 0.68; 95% CI, 0.61-0.76). There was a similarly negative effect on odds of use by API individuals (OR, 1.12; 95% CI, 1.02-1.23) and a strong and significantly negative effect on predicted use by African American patients (OR, 0.81; 95% CI, 0.70-0.93).

Table 2. Multivariable Models Predicting Use of an NCI Cancer Center for Treatment, With Adjustment for Proximity to an NCI Cancer Center for Colorectal Cancer (California, 1996 to 2006)
 Baseline ModelModel 2: Baseline Model + Distance From Home to NCI
Patient CharacteristicsOR95% CIPOR95% CIP
  1. Abbreviations: 95% CI, 95% confidence interval; API, Asian/Pacific Islander, NCI, National Cancer Institute; OR, odds ratio.

  2. All models were also adjusted for year of diagnosis.

  3. a

    - indicates reference category.

  4. b

    P <.05.

  5. c

    P <.001.

  6. d

    Entered into the model continuously increasing.

Age, y      
0-44-a  -  
45-540.810.71-0.93b0.820.72-0.94b
55-640.600.53-0.69c0.610.53-0.69c
65-740.390.34-0.45c0.400.35-0.45c
75-840.320.28-0.37c0.320.28-0.36c
≥850.230.19-0.27c0.220.19-0.26c
Sex      
Male-  -  
Female0.880.83-0.94c0.880.82-0.94b
Race      
White-  -  
Black0.960.84-1.10 0.810.70-0.93b
Hispanic0.710.64-0.79c0.680.61-0.76c
API1.411.30-1.54c1.121.02-1.23b
Stage of disease      
I-  -  
II0.840.77-0.92c0.840.77-0.92c
III0.900.82-0.99b0.900.82-0.98b
IV1.050.95-1.17 1.050.95-1.16 
Distance to NCI-designated cancer center (+5 mi)d---0.930.93-0.94c

To assess the effect of select social factors on the use of NCI centers, the baseline model was adjusted for insurance status (model 3) and neighborhood SES characteristics (model 4). These results are shown in Table 3. Although both Medicare (OR, 1.52; 95% CI, 1.39-1.66) and Medicaid (OR, 1.31; 95% CI, 1.15-1.50) insurance were found to be independent predictors of use of NCI centers, having no insurance (OR, 0.65; 95% CI, 0.49-0.86) was negatively correlated with use. Adjustment of the baseline model by type of insurance did not change the correlations between race/ethnicity and use of an NCI center previously noted. In contrast, neighborhood characteristics (increasing percentage unemployment, percentage with a college education, and percentage of residents living at < 200% of the federal poverty level) were each found to be independently associated with NCI center use. Although increasing unemployment and poverty rates were associated with approximately 4% (P = .03) and 12% (P < .001) lower odds of NCI cancer center use, respectively, patients from neighborhoods with increasing percentages of college-educated residents had a 42% increased odds of use (P < .0001). Adjustment of the baseline model for neighborhood SES characteristics (model 4) neutralized the negative association between Hispanic ethnicity and use of an NCI cancer center (OR, 1.05; 95% CI, 0.93-1.18). There was also a significant increase in the odds of NCI center use by black patients compared with their white counterparts (OR, 1.39; 95% CI, 1.20-1.60). There was a 5% increase found in the odds of use by API patients after adjusting for neighborhood characteristics.

Table 3. Multivariable Models Predicting Use of an NCI Cancer Center for Treatment by Race, With Adjustment for Insurance Status and Neighborhood Factors for Colorectal Cancer: California, 1996 to 2006
 Model 3: Insurance StatusModel 4: Neighborhood Characteristics
Patient CharacteristicsOR95% CIPOR95% CIP
  1. Abbreviations: % College education, percentage of the census block with a college education; % Poverty, percentage of those living below the 200% federal poverty level; % Unemployment, percentage of the neighborhood unemployed; 95% CI, 95% confidence interval; API, Asian/Pacific Islander, NCI, National Cancer Institute; OR, odds ratio.

  2. All models were adjusted for age, sex, race, stage of disease, year of diagnosis, and insurance status (model 3) and neighborhood characteristics (model 4).

  3. a

    - indicates reference category.

  4. b

    P <.001.

  5. c

    P <.05.

  6. d

    Entered into model continuously increasing.

Race      
White- a  -  
Black0.940.82–1.08 1.391.20–1.60b
Hispanic0.710.63–0.78b1.050.93–1.18c
API1.401.28–1.54b1.451.31–1.60 
Insurance type      
Private-     
Medicare1.521.39–1.66b---
Medicaid1.311.15–1.50b   
No insurance0.650.49–0.86c   
Neighborhood factors (continuous)      
% College educationd---1.421.35–1.50b
% Unemploymentd   0.960.92–0.99c
% Povertyd   0.880.83–0.93b

To clarify the relative importance of the 2 factors that were found to be independently associated with NCI center use and that had strong, opposing effects on estimates of minority use, we compared the effects of proximity and neighborhood college education. To do so, we built a fifth model to predict NCI cancer center use with adjustment for travel distance and subsequently for neighborhood college education. In this model (data not shown) there was an independent and significant association noted between increasing distance to an NCI center (OR, 0.94; 95% CI, 0.94-0.95) and college education (OR, 1.47; 95% CI, 1.41-1.53). Figure 2 shows that adjusting the baseline model for these 2 opposing forces resulted in significant changes for all minority groups. There was a 14% net increase in the odds of NCI center use by black patients. For Hispanic patients, there was a net 24% increase in the odds of use and the previously negative association between Hispanic ethnicity and use was neutralized. There was a 20% decrease in the odds of NCI cancer center use by API patients when accounting for both factors; however, the overall positive correlation between API ethnicity and use of an NCI cancer center persisted.

image

Figure 2. Changes in the odds of use of National Cancer Institute cancer centers with multivariable model adjustment for distance from an NCI setting and neighborhood college education level stratified by race/ethnicity for colorectal cancers (California, 1996-2006) are shown. The figure compares the effect of travel distance to an NCI cancer center and neighborhood education level on the odds of using an NCI cancer center among each racial and ethnic group.

Download figure to PowerPoint

Given the potential for interactions between 1) race/ethnicity and insurance status and 2) race/ethnicity and SES measures, interaction terms using these factors were run in the various models. Although there were some correlations noted between select interaction terms, there was no qualitative differences observed in the previously reported associations, and therefore these data are not shown.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

The current study investigates the factors associated with minority use of NCI cancer centers for the treatment of CRC in California. The results demonstrate that fewer than 5% of all patients with CRC in the state of California use NCI centers. The majority of patients travel relatively short distances (≤5 miles) for CRC treatment anywhere and although higher percentages of minority groups live near an NCI center, the likelihood of use was lower among select groups. In particular, Hispanic patients were less likely to use NCI cancer centers. This negative association persisted despite adjustment for insurance type, neighborhood-level unemployment, and poverty. Living farther away from an NCI setting predicted a lower odds of use for all populations. Residence in a neighborhood with a higher percentage of college graduates equalized the odds of NCI cancer center use for Hispanic and black patients when compared with their white counterparts. There was a small increase in the already positive association between API ethnicity and use of an NCI cancer center.

When comparing the effect of proximity to an NCI setting against the effect of neighborhood education level, we found differences in the impact for each racial/ethnic group. Simultaneous adjustment for the 2 factors resulted in significant increases in the odds of use by Hispanic and black patients. These positive changes reflect a net stronger impact of neighborhood education (positive predictor) compared with the negative effect of increasing living distance from an NCI setting. By contrast, the association between API patients and use of NCI cancer centers was decreased by 20% after adjusting for these 2 factors, which suggests that distance from an NCI center may have a stronger effect on use among these subgroups.

To the best of our knowledge, the current study is the first to simultaneously assess the effect of proximity, insurance type, and neighborhood SES to explain patterns of NCI cancer center use in an all-payer racially and ethnically diverse populations. The use of an all-payer data set is important because nearly one-third of patients with CRC are diagnosed before the age of 65 years, and the disparity in the incidence of the disease between black and white patients is greatest in this younger age group.[26] Moreover, we included a broader distribution of patients across the SES spectrum and found that select markers of increasing SES predict an increased odds of use of an NCI cancer center. Thus, these findings are robust and potentially more generalizable than those of previous studies.

The results of the current study are in keeping with those of others who have shown a correlation between black race and increased use of NCI centers for cancer care.[19] The results are also consistent with studies that have shown that patients are unwilling to travel significant distances for care regardless of potential gains in outcome associated with the use of high-quality settings.[27] The results of the current study advance our understanding of predictors of location of care by differentiating the importance of various SES factors on use. Patients living in neighborhoods dense with college graduates have higher odds of using NCI cancer centers. The findings suggest the importance of education, independent of neighborhood-level poverty and unemployment or travel distance.

Racially and ethnically based differences in the use of NCI cancer centers as observed in the current study may be driven by number of factors. First, minorities are less likely to be referred to high-quality settings for surgical care.[28, 29] There is evidence to suggest that providers who care for minority and low-income patients face challenges in obtaining access to specialists, imaging, and nonemergency hospital admissions for their patients.[30] Furthermore, in counties with high levels of racial segregation, there appear to be fewer options for surgical facilities and surgeons[10, 31] and limited availability of specialists such as gastroenterologists and radiation oncologists.[32]

Alternatively, minority subgroups may demonstrate certain preferences that determine the location of care. Patients have generally shown a predilection for the use of nearby hospitals regardless of the quality of care.[27] Our finding of a 5-mile median travel distance is consistent with this statement. In addition, patients may prefer certain hospitals at which the staff shares social, cultural, or linguistic backgrounds that are similar to their own. Finally, mistrust of the health care system in select minority populations may strongly influence the types of hospitals they chose for care.[33, 34] Regardless of the explanation, the results of the current study are important because NCI cancer centers are known to be associated with better outcomes in comparison with other care settings. Identifying points of intervention to increase minority treatment in these settings may positively impact cancer outcome disparities.

Limitations of the Current Study

The current study is limited by the use of cross-sectional data. Although we were able to demonstrate strong associations between race/ethnicity and use of an NCI cancer center, we were not able to prove causation. In diseases such as CRC, in which elective surgery is central to care, both patient preference and provider referral patterns may play an important role in determining the location of care. These 2 factors cannot be distinguished with the current approach. Nonetheless, the results have defined potential target populations for intervention: minorities living in neighborhoods with low educational attainment. Second, the population in the current study was limited to patients diagnosed and treated for CRC. We chose CRC as a proxy for all NCI cancer center use because it is a high-incident cancer,[35] affects both men and women nearly equally, and is often treated in a wide variety of hospital settings. The resultant variation in the quality of CRC care[36, 37] and survival[7] is believed to contribute to disparities in cancer outcomes.[37] Future studies should repeat the analysis using other cancers to validate the distribution of patients across both NCI and non-NCI settings. Finally, we used SES measures at the census block group. Although neighborhood SES measures may mask individual-level SES, previous studies have shown that using block group data may still capture neighborhood contextual effects on individual-level health.[38-40] Moreover, the current data set, through the inclusion of block group data, provides to the best of our knowledge the most granular SES data currently available in an administrative cancer data set. By comparison, national data sets such as the Surveillance, Epidemiology, and End Results program rely on census tract data, which, similar to zip code and county-level data, are less sensitive to SES gradients within the geographic unit of measurement.[41] Thus, census block group data provide a significant advantage in reflecting patient behavior over previous studies that have used census tract,[41, 42] zip code,[8, 43] or county-level data.

CONCLUSIONS

The results of the current study have important implications. These findings suggest that although there is variation between minority groups with respect to the use of NCI cancer centers for the treatment of CRC, Hispanic and black populations share common barriers to use. In particular, increasing neighborhood education level appears to have a positive impact on the odds of use. Because this may relate to a lack of information regarding differences in hospital quality and the potential impact on outcomes related to the use of NCI centers, targeted community outreach and education efforts may serve to increase minority and overall use of these centers. However, these efforts are unlikely to change the inherent characteristics of the neighborhood. Thus, the results of the current study should also serve as encouragement for primary care providers to refer minority patients to NCI settings. Increasing minority use of NCI centers may be positively impacted by the upcoming implementation of the Patient Protection and Affordable Care Act and its core of universal insurance coverage and value-based purchasing. Financial incentives for high-quality care may increase attendance by Medicare and newly covered Medicaid beneficiaries, many of whom are likely to be racial and ethnic minorities. Improving use in minority groups, either by influencing patient preference or provider referral patterns, could enhance access to higher-quality care and may positively impact long-standing cancer disparities.

FUNDING SUPPORT

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Dr. Rhoads and Dr. Huang were supported by the National Institutes of Health National Institute on Minority Health and Health Disparities Loan Repayment Program. Dr. Rhoads' work on this project was also supported by a Harold Amos Medical Faculty Development Award from the Robert Wood Johnson Foundation and a grant from the National Cancer Institute (1R21CA161786-01A).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES