Racial differences in tumor stage and survival for colorectal cancer in an insured population†
This research was made possible by the cooperation of the participating Cancer Research Network members. We are particularly grateful to Donald J. Bachman, MS; Jennifer Ellis, MSPH; Hassan Fouayzi, MS; Rick Krajenta; Liyan Liu, MSPH; Janis Yao, MS; and Fagen Xie, PhD, whose collective work in building the Cancer Research Network's Virtual Data Warehouse made this research project possible.
Abstract
BACKGROUND.
Despite declining death rates from colorectal cancer (CRC), racial disparities have continued to increase. In this study, the authors examined disparities in a racially diverse group of insured patients.
METHODS.
This study was conducted among patients who were diagnosed with CRC from 1993 to 1998, when they were enrolled in integrated healthcare systems. Patients were identified from tumor registries and were linked to information in administrative databases. The sample was restricted to non-Hispanic whites (n = 10,585), non-Hispanic blacks (n = 1479), Hispanics (n = 985), and Asians/Pacific Islanders (n = 909). Differences in tumor stage and survival were analyzed by using polytomous and Cox regression models, respectively.
RESULTS.
In multivariable regression analyses, blacks were more likely than whites to have distant or unstaged tumors. In Cox models that were adjusted for nonmutable factors, blacks had a higher risk of death from CRC (hazard ratio [HR], 1.17; 95% confidence interval [95% CI], 1.06–1.30). Hispanics had a risk of death similar to whites (HR, 1.04; 95% CI, 0.92–1.18), whereas Asians/Pacific Islanders had a lower risk of death from CRC (HR, 0.89; 95% CI, 0.78–1.02). Adjustment for tumor stage decreased the HR to 1.11 for blacks, and the addition of receipt of surgical therapy to the model decreased the HR further to 1.06. The HR among Hispanics and Asians/Pacific Islanders was stable to adjustment for tumor stage and surgical therapy.
CONCLUSIONS.
The relation between race and survival from CRC was complex and appeared to be related to differences in tumor stage and therapy received, even in insured populations. Targeted interventions to improve the use of effective screening and treatment among vulnerable populations may be needed to eliminate disparities in CRC. Cancer 2007;109:612–620. © 2006 American Cancer Society.
Colorectal cancers are the third most common cancers among men and women in the United States and are projected to cause 55,170 deaths in 2006.1 The majority of colorectal cancers arise in adenomatous polyps through a well-defined, multistage process over several years.2, 3 Therefore, timely identification and removal of precursor lesions potentially can prevent colorectal cancer,4 and early stage tumors are curable.1 Despite declining incidence and mortality from colorectal cancer since the early 1980s, racial and ethnic disparities have continued to widen, particularly between blacks and whites.5-8 Previous studies in the general population established that blacks with colorectal cancer were significantly more likely to be diagnosed at an advance stage and to have lower survival rates compared with whites.6, 9, 10 In contrast, Asians/Pacific Islanders and, in some reports, Hispanics have lower incidence and mortality from colorectal cancer than whites.5, 6
Although the exact mechanisms for racial disparities in colorectal cancer are not clear, the results of several studies suggest that inequalities in access to health care related to colorectal cancer prevention and treatment, including lack of health insurance or a regular source of care, may be important factors.11-13 In general, individuals who are enrolled in health insurance plans are able to select a regular source of primary care. Many health plans provide colorectal cancer screening and treatment as covered benefits and have systems to encourage the use of cancer-prevention services.14 Consequently, individuals who are enrolled in health plans are more likely to report the use of cancer screening15-18 and to be diagnosed with cancers at an earlier stage than individuals who are not enrolled in health plans.19 However, the extent to which various racial groups derive the benefits of earlier cancer detection and/or potentially lower case-fatality rates from colorectal cancer through membership in health plans is unknown. The objective of this study was to examine racial differences in tumor stage and survival rates among patients who were diagnosed with and treated for colorectal cancer in integrated health care systems.
MATERIALS AND METHODS
Setting
For this study, we used data from 6 geographically diverse, integrated health care systems that are affiliated with the Cancer Research Network (the Network). The details of the Network and the data sources used for this study have been described previously.20, 21 Funded by the National Cancer Institute under a cooperative agreement, the Network is a consortium of 14 research centers, which are affiliated with integrated health care systems that provide the continuum of cancer care for patients of all ages. The health care systems that participated in this study (Henry Ford Health System, Michigan; Group Health, Washington; and 4 Kaiser Permanente Regions—Colorado, Northwest [Oregon and southern Washington], and Northern and Southern California) are predominantly staff-model health care delivery systems. Each of the participating health care systems has a tumor registry that collects information in a manner consistent with the Surveillance, Epidemiology, and End Results (SEER) Program. The Institutional Review Boards of all participating institutions approved this study.
Patients
The study cohort was comprised of patients aged ≥20 years who were diagnosed with incident, invasive colorectal cancer between January 1, 1993 and December 31, 1998 while they were enrolled in 1 of the 6 health care systems. We identified eligible patients who had carcinoma of the colon/rectum from tumor registries by using SEER site-of-origin codes (C18.0, C18.2–C18.9, C19.9, and C20.9) and morphology codes derived from the International Classification of Diseases for Oncology, 2nd edition. We restricted our analyses to the 13,958 patients who were identified as non-Hispanic white (white), non-Hispanic black (black), Hispanic, or Asian/Pacific Islander.
Data Sources and Study Parameters
The outcomes of interest were tumor stage at the time of diagnosis and survival after cancer diagnosis. Information on tumor stage was obtained from tumor registries and was categorized according to the SEER staging scheme as local (localized to the primary tumor site), regional (regional spread), distant (distant metastases), or unstaged (unknown tumor stage).22 Survival information was obtained from tumor registries and was confirmed with information from administrative databases. We obtained survival information on all patients up to the date of disenrollment from the health plan, the date of death, or the last date of follow-up, December 31, 2003. Cause of death information was incomplete for 15% of the decedents, and previous studies have documented that cancer-related deaths often are misclassified.23 Therefore, for our survival endpoint, we examined both colorectal cancer-specific (n = 4024) and all-cause (n = 6799) death-rate ratios.
Information on demographic and tumor characteristics and on the type of initial cancer-related therapy received (surgery, chemotherapy, or radiation therapy) was extracted from tumor registries. Tumor location was categorized as the right colon (relative to the splenic flexure), the left colon (the descending and sigmoid colon), or the rectum (rectum and rectosigmoid). Tumor grade was categorized as low-grade (well-differentiated or moderately differentiated), high-grade (poorly differentiated or undifferentiated), or unknown. Data on the length of health plan enrollment (categorized as <1 year, from 1 to 9 years, from 5 to 9 years, or ≥10 years) and comorbid medical conditions were obtained through linkage of patient records with electronic administrative databases at each of the health care systems. We derived a modified Charlson comorbidity index24 (categorized as 0, 1, and ≥2) by using electronic information from hospitalizations during the 2-year period prior to diagnosis.
Statistical Analyses
We examined differences in the distribution of demographic and clinical characteristics of patients by using the chi-square test or an analysis of variance, as appropriate. We also examined differences in tumor location by using polytomous logistic regression models.25
We used bivariate and multivariable, polytomous, logistic regression models to examine racial and ethnic differences in tumor stage. We used local tumors as the reference category and obtained the estimated odds ratio (OR) for regional, distant, and unstaged tumors among blacks, Hispanics and Asians/Pacific Islanders compared with whites while controlling for potential confounding factors. Covariates in our multivariable analyses included age, gender, diagnosis year, tumor location, comorbidity index, study site, and length of health plan enrollment prior to cancer diagnosis.
We performed analyses of racial and ethnic differences in survival rates among the cohort by using Kaplan-Meier survival analyses and Cox proportional hazards models. Survival was censored at the date of disenrollment from the health plan (n = 1817), at the date of death, or at the last date of follow-up, whichever occurred first. Our Cox proportional hazards model development strategy was focused on identifying potential explanations for racial and ethnic differences in survival. Therefore, our multivariable Cox modeling commenced with a basic model comprised of race and ethnicity and nonmutable covariates (ie, age, gender, diagnosis year, and study site). Then, additional variables were added to this model, 1 at a time, beginning with an indicator of comorbid disease burden followed by tumor stage and receipt of colorectal cancer surgery. We also performed additional analyses stratified according to tumor stage and location. We restricted our survival analyses to colorectal cancer-specific death-rate ratios, because analyses of all-cause death-rate ratios did not change our findings. We examined the assumption of proportional hazards for race and ethnicity by plotting the log of the negative log of the survival functions against the log of time for each racial and ethnic group; the plotted lines were reasonably parallel. The analyses for this study were generated by using SAS (version 9.1 for Windows; SAS Institute Inc., Cary, NC) and Stata 8.2 for Windows (StataCorp, College Station, TX).
RESULTS
Of the 13,958 eligible patients with incident, invasive colorectal cancer who were diagnosed between 1993 and 1998 at the participating study sites, 10,585 patients (75.8%) were white, 1479 patients (10.6%) were black, 985 patients (7.1%) were Hispanic, and 909 patients (6.5%) were Asians/Pacific Islanders. The characteristics of the study population according to race and ethnicity are shown in Table 1. White patients were older than patients in the other racial or ethnic groups. The majority of patients had been enrolled with their respective health plans for ≥5 years prior to the date of diagnosis and had a low comorbid disease burden. Black patients were more likely to have tumors diagnosed at distant stage or to have tumors that were unstaged at the time of diagnosis. Black patients had slightly higher comorbidity, and both Hispanic and Asian/Pacific Islander patients had lower comorbidity than white patients. A lower proportion of black patients underwent surgical therapy compared with white patients. A higher proportion of Hispanic and Asian/Pacific Islander patients underwent surgery or received chemotherapy compared with the proportion of white patients.
| Characteristics | % Whites | % Blacks | % Hispanics | % Asians/ Pacific Islanders |
|---|---|---|---|---|
| Total no. of patients | 10,585 | 1479 | 985 | 909 |
| Sex | ||||
| Women | 47 | 49.3 | 39.3 | 46.6 |
| Age, y | ||||
| Mean ± SD, y | 68.3 ± 12.0 | 63.7 ± 12.3 | 62.1 ± 12.4 | 61.8 ± 12.2 |
| 20–50 | 7 | 12.6 | 15.7 | 17.1 |
| 50–64 | 27.6 | 38.8 | 40.3 | 39.8 |
| 65–79 | 47.5 | 39.2 | 36.8 | 36.5 |
| ≥80 | 17.9 | 9.5 | 7.2 | 6.6 |
| Length of enrollment prior to diagnosis, y | ||||
| <1 | 7.9 | 8.3 | 11.5 | 10.1 |
| 1–4 | 17.5 | 16.6 | 24.2 | 22.6 |
| 5–9 | 18.8 | 15.7 | 22.6 | 17.7 |
| ≥10 | 55.9 | 59.4 | 41.7 | 49.6 |
| Comorbidity index* | ||||
| Mean ± SD | 0.20 ± 0.65 | 0.23 ± 0.73 | 0.13 ± 0.48 | 0.11 ± 0.47 |
| 0 | 86.8 | 86.1 | 90 | 92.6 |
| 1 | 7 | 7.6 | 6.1 | 4.4 |
| ≥2 | 4.6 | 5.7 | 2.5 | 2.6 |
| Unknown | 1.6 | 0.6 | 1.4 | 0.3 |
| Tumor location | ||||
| Right colon | 42.9 | 46 | 34.4 | 30.8 |
| Left colon | 29.1 | 27.5 | 28.8 | 34.4 |
| Rectum | 28 | 26.5 | 36.8 | 34.8 |
| Stage | ||||
| Local | 38.8 | 37.3 | 40 | 40.2 |
| Regional | 41.1 | 37.8 | 41.8 | 41.3 |
| Distant | 17.9 | 20.8 | 15.8 | 17.6 |
| Unstaged | 2.2 | 4.1 | 2.3 | 1 |
| Grade | ||||
| Low | 71.2 | 70.2 | 70.9 | 67.8 |
| High | 16.9 | 13.3 | 14.9 | 15.8 |
| Unknown | 11.9 | 16.6 | 14.2 | 16.4 |
| Therapies | ||||
| Surgery | 91.1 | 85.1 | 93.4 | 93.3 |
| Chemotherapy | 33.6 | 34.3 | 42.5 | 44.3 |
| Radiation | 9.9 | 7.7 | 13.6 | 11.2 |
- * Modified Charlson comorbidity index (see Deyo et al., 199224).
The majority of tumors were located distal to the splenic flexure (see Table 1). In unadjusted, polytomous, logistic regression analyses, blacks (OR, 1.17; 95% confidence interval [95% CI], 1.01–1.36) were more likely to be diagnosed with tumors located in the right colon compared with whites. Both Hispanics (OR, 0.64; 95% CI, 0.55–0.76) and Asians/Pacific Islanders (OR, 0.58; 95% CI, 0.49–0.70) were less likely to have right colon tumors compared with whites.
We performed polytomous regression analyses of the association of race and ethnicity with a diagnosis of regional, distant, or unstaged tumors relative to tumors that were diagnosed at local stage. In unadjusted analyses, blacks were more likely than whites to be diagnosed with distant tumors (OR, 1.21; 95% CI, 1.04–1.41) or unstaged tumors (OR, 1.96; 95% CI, 1.46–2.63). Adjustment for age, gender, diagnosis year, tumor location, comorbidity index, study site, and length of enrollment prior to diagnosis did not alter these associations significantly (see Table 2). In unadjusted analyses, Hispanics were similar to whites in their patterns of diagnoses of regional tumors (OR, 0.99; 95% CI, 0.83–1.13), distant tumors (OR, 0.86; 95% CI, 0.71–1.04), and unstaged tumors (OR, 1.03; 95% CI, 0.67–1.61). With adjustment for the selected covariates, the estimated OR was unchanged for distant tumors (OR, 0.91; 95% CI, 0.75–1.12) but increased to 1.20 for regional tumors and to 1.42 for unstaged tumors, and the estimate for regional tumors was statistically significant. In unadjusted analyses, Asians/Pacific Islanders were similar to whites in their patterns of diagnoses of regional tumors (OR, 0.97; 95% CI, 0.83–1.13) and distant tumors (OR, 0.95; 95% CI, 0.78–1.15), but they were significantly less likely to have unstaged tumors (OR, 0.44; 95% CI, 0.22–0.86). These findings were stable to adjustment for the selected covariates (see Table 2). However, the associations were not statistically significant in the adjusted analyses.
| Tumor stage | Whites | Blacks | Hispanics | Asians/Pacific Islanders | ||||
|---|---|---|---|---|---|---|---|---|
| No. | OR† | No. | Adjusted OR (95% CI)† | No. | Adjusted OR (95% CI)† | No. | Adjusted OR (95% CI)† | |
| Local | 4104 | — | 551 | — | 394 | — | 365 | — |
| Regional | 4354 | 1.00 (Ref) | 559 | 1.02 (0.90–1.17) | 412 | 1.20 (1.03–1.40) | 375 | 1.12 (0.96–1.31) |
| Distant | 1895 | 1.00 (Ref) | 308 | 1.16 (0.99–1.36) | 156 | 0.91 (0.75–1.12) | 160 | 0.99 (0.81–1.21) |
| Unstaged | 232 | 1.00 (Ref) | 61 | 1.94 (1.41–2.68) | 23 | 1.42 (0.90–2.24) | 9 | 0.61 (0.31–1.21) |
- Numbers shown are the point estimates of the ORs and the 95% CI.
- * The reference tumor stage was local disease, and the reference group was whites. Covariates in the adjusted models included tumor location, gender, age, comorbidity, and duration of enrollment prior to cancer diagnosis, year of diagnosis, and study site.
- † Numbers shown are the point estimates of the ORs and the 95% CI.
In Kaplan Meier analyses, the survival-rate ratios among the racial and ethnic groups diverged in the first year of follow-up after diagnosis, and black patients had the lowest survival rate (see Fig. 1). The estimated 5-year survival rates were 70% for whites, 66% for blacks, 69% for Hispanics, and 73% for Asians/Pacific Islanders (see Fig. 1). We used Cox proportional hazards models to examine the degree to which modifiable factors explained racial and ethnic differences in survival rates. In unadjusted analyses, blacks had a higher risk of death (hazard ratio [HR], 1.13; 95% CI, 1.03–1.25) compared with whites (Table 3). With adjustment for age, gender, diagnosis year, and study site (nonmutable covariates), the HR for death was 1.17 (95% CI, 1.06–1.30). Adding an indicator for comorbid disease burden at the time of diagnosis to the regression model did not change the HR. Further adjustment for tumor stage decreased the HR of death to 1.11 for blacks; excluding comorbidity from this model did not change the HR. Adding an indicator for the receipt of surgical therapy further decreased the HR of death to 1.06 (95% CI, 0.96–1.17); including additional covariates in the regression model did not change this HR. In unadjusted Cox model analyses, Hispanics had a risk of death (HR, 0.99; 95% CI, 0.88–1.12) similar to that of whites. This finding was robust to inclusion of the other covariates to the model. In unadjusted Cox model analyses, Asians/Pacific Islanders had a lower risk of death (HR, 0.85; 95% CI, 0.75–0.97) from colorectal cancer than whites. This finding was robust to the addition of other covariates to the model, but the difference was not statistically significant in the adjusted analyses (see Table 3).
Colorectal cancer survival-rate ratios by race/ethnicity for patients who were diagnosed from 1993 to 1998.
| Model components* | Whites (Ref) | Blacks | Hispanics | Asians/Pacific Islanders | |||
|---|---|---|---|---|---|---|---|
| HR | 95% CI | HR | 95% CI | HR | 95% CI | ||
| 1. Unadjusted model (Model 1) | 1.00 | 1.13 | 1.03–1.25 | 0.99 | 0.88–1.12 | 0.85 | 0.75–0.97 |
| 2. Model 1 + basic predictors† | 1.00 | 1.17 | 1.06–1.30 | 1.04 | 0.92–1.18 | 0.89 | 0.78–1.02 |
| 3. Model 2 + comorbidity | 1.00 | 1.15 | 1.04–1.27 | 1.04 | 0.92–1.17 | 0.90 | 0.79–1.03 |
| 4. Model 3 + stage at diagnosis | 1.00 | 1.11 | 1.00–1.23 | 1.09 | 0.96–1.23 | 0.94 | 0.82–1.07 |
| 5. Model 4 + receipt of surgical therapy | 1.00 | 1.06 | 0.96–1.17 | 1.08 | 0.95–1.22 | 0.93 | 0.81–1.06 |
| 6. Full model‡ | 1.00 | 1.06 | 0.96–1.17 | 1.08 | 0.95–1.22 | 0.93 | 0.81–1.06 |
- Ref indicates reference group; HR, hazard ratio; 95% CI, 95% confidence interval.
- * The models are numbered sequentially with the unadjusted model designated Model 1 and the full model designated Model 6.
- † The basic model was adjusted for age at diagnosis, gender, year of diagnosis, and study site.
- ‡ The full model was adjusted for age, gender, year of diagnosis, study site, stage, cancer-related therapies, tumor location, comorbidity, enrollment before cancer diagnosis, and tumor grade.
To explore further the potential explanations for differences in survival between blacks and whites, we performed additional Cox proportional hazards models analyses on strata of tumor stage and location. In analyses that adjusted for the nonmutable variables, we observed that blacks who were diagnosed with local-stage disease (HR, 1.03; 95% CI, 0.74–1.43), or regional-stage disease (HR, 1.08; 95% CI, 0.91–1.29) had survival rates that were similar to those among whites. Blacks who were diagnosed with distant-stage disease had a higher risk of death (HR, 1.18; 95% CI, 1.03–1.36), and adjustment for comorbidity in the model did not change this finding (HR, 1.17; 95% CI, 1.02–1.35). Adding an indicator for the receipt of surgical therapy and chemotherapy decreased the HR to 1.04 (95% CI, 0.91–1.20). In analyses stratified on tumor location that controlled for the nonmutable factors, blacks with right colon cancers were similar to whites in their risk of death (HR, 1.07; 95% CI, 0.93–1.24). Blacks who were diagnosed with left colon cancers had a higher risk of death (HR, 1.27; 95% CI, 1.04–1.55); this HR was robust to adjustment for comorbidity in the model (HR, 1.23; 95% CI, 1.01–1.50). Further adjustment for tumor stage in the model decreased the HR to 1.14 (95% CI, 0.94–1.40); this HR was stable to the addition of indicators of surgical therapy to the model (HR, 1.12; 95% CI, 0.91–1.36). In analyses on patients who were diagnosed with rectal cancer, blacks had a higher risk of death (HR, 1.19; 95% CI, 0.98–1.45), and adjustment for comorbidity and tumor stage in the regression model did not change the finding (HR, 1.16; 95% CI, 0.95–1.42). Adjustment for the receipt of surgical therapy reduced the HR to 1.04 (95% CI, 0.85–1.27).
DISCUSSION
Our findings suggest that, among patients with colorectal cancer who are diagnosed and treated in integrated health care systems, blacks have a higher risk of dying, whereas Asians/Pacific Islanders have better survival rates compared with whites. There was no statistically significant difference in survival between Hispanics and whites. A substantial portion of the survival disadvantage observed in our study among blacks was explained by differences in tumor stage and the receipt of surgical therapy. In contrast to Hispanics and Asians/Pacific Islanders, blacks were more likely to be diagnosed with tumors at distant stage and to have unstaged tumors, and they were less likely to undergo surgery compared with whites.
Several previous studies in the general population have reported racial differences in survival from colorectal cancer.5-8 The racial differences observed in our study are more modest than those from previous studies conducted in settings in which the provision of cancer care services is less uniform than in integrated health care systems. Roetzheim et al. examined racial disparities among 9548 patients who were diagnosed with colorectal cancer in 1994 using data from the Florida Cancer Data System.26 In their analyses that adjusted for a variety of sociodemographic and clinical factors, including health insurance, socioeconomic indicators, tumor location, and comorbidity, blacks had a 22% higher risk of death than whites (HR, 1.22; 95% CI, 1.05–1.42). Even after further adjustment for tumor stage and treatment indicators in their analyses, blacks had an 18% increased risk of death, a difference that may be attributed in part to unmeasured variations in cancer care in the general population.
Our study was conducted in a population for whom risk-based screening27 and standard therapy for colorectal cancer were available as covered benefits. Despite the availability of these services, in our study, we observed racial differences, albeit more modest than in the general population, in survival among patients who were diagnosed with colorectal cancer between 1993 and 1998. There are several possible explanations for the survival differences observed in our study.
Racial disparities in colorectal cancer survival often are attributed to differences in tumor stage that result from differential use of cancer screening services.10, 12 In our study, we observed significant differences in tumor stage, especially between blacks and whites. Because colorectal carcinogenesis progresses through a slow, multistage process,2 the magnitude of racial differences for regional and distant tumors likely reflect delays in diagnosis, specifically, elapsed time from malignant transformation to cancer detection. Among blacks, the strength of the association with tumor stage increased monotonically from regional through distant to unstaged tumors. The likely explanation is that blacks are less likely to be diagnosed through screening, as documented in other settings.28-30 In addition, the higher proportion of tumors located on the right colon among blacks also suggests that they either may have a higher incidence of right colon cancers or are less likely to use screening tests that adequately detect right colon tumors. Colonoscopy, a procedure that can visualize the entire colon directly, was recommended for routine colorectal cancer screening beginning in 1997.27 In previous studies, a greater delay was reported in the adoption of new technology among some minority populations.31, 32 Thus, slower adoption of colonoscopy for colorectal cancer screening among blacks during the study period may have been a potential contributing factor to the differences in tumor location and stage observed in our study. In addition, the significantly increased risk of unstaged cancers suggests that blacks also were less likely to have had complete diagnostic testing around the time of their cancer diagnosis.
Previous studies also suggested that racial differences in the receipt of effective treatments may contribute to differences in cancer survival.13 We observed that blacks were less likely to receive cancer-related therapies, whereas Hispanics and Asians/Pacific Islanders were more likely to receive these therapies than whites. These differences were corroborated by our multivariable analyses. Among blacks, adjusting for cancer-related therapies in the Cox models reduced the survival differences, particularly in the subgroup of patients with advanced or rectal tumors. These findings are consistent with some previous reports that, when both blacks and whites received similar treatment, such as in the Veterans Affairs (VA) health care system33, 34 or among patients enrolled in clinical trials,35-37 there were either no differences or only attenuated disparities in survival from colorectal cancer.
In an analysis of 3176 patients who were diagnosed with colorectal cancer in 1989 in the VA health care system, Dominitz et al. reported that blacks had a 13% higher risk of death (HR, 1.13; 95% CI, 1.01–1.28) than whites.33 Their study did not provide information on tumor stage.33 In a similar analysis of 46,044 patients with colorectal cancer who were diagnosed between 1987 and 1998 in the VA health care system, an even smaller difference was reported in the risk of death between blacks and whites (adjusted HR, 1.04; 95% CI, 1.01–1.08).34 Previous studies have indicated that Medicare beneficiaries enrolled in health maintenance organization plans have better survival than patients enrolled in fee-for-service plans.38, 39 However, in a study that was restricted to patients who were enrolled in both Medicaid and Medicare at the time of their colorectal cancer diagnosis, no racial difference in survival was observed.40 Dignam et al. reported no significant racial difference in cancer-specific survival among patients with colon cancer (HR, 1.08; 95% CI, 0.94–1.25) or rectal cancer (HR, 1.25; 95% CI, 0.94–1.66) who were enrolled in the National Surgical Adjuvant Breast and Bowel Project trials, in which treatment regimens are controlled more rigorously.35, 36 Analyses of patients enrolled in the Phase III Intergroup Trial 0089, which assessed the efficacy of adjuvant chemotherapy (1988–1992), also indicated that there was no racial difference in survival.37 Those findings, combined with our results, suggest that differences in colorectal cancer survival could be reduced or eliminated if differences in the receipt of colorectal cancer screening and therapy were removed. Within the setting of integrated health care systems, coordinated implementation of standardized protocols for monitoring the receipt of cancer-related services specifically targeted at vulnerable populations may reduce these disparities.41
The extent to which comorbid disease burden explains racial differences in survival from colorectal cancer is not well known.12 In our analyses, adjustment for comorbidity did not change the observed racial differences. This finding suggests that racial differences in survival from colorectal cancer are not because of differential comorbidity. However, the differences in comorbid disease burden between blacks and whites in this study were relatively small, and indicators of comorbidity were not updated after cancer diagnosis.
Strengths and Limitations
This study was conducted in a population with health insurance, providing added insights into disparities. Together, the health care delivery systems that were included in this study provided care to approximately 7 million patients in 1998. This, along with the geographically diverse regions of the country from which the study population was drawn, enhances the generalizability of our findings. The majority of patients in the study in all racial groups had been enrolled with their health plans for ≥5 years prior to their cancer diagnosis. Therefore, most patients had the opportunity to have a regular source of health care and to be screened. However, there are other factors that may affect access to cancer care that were not available for this study.
The results of some previous studies suggest that racial disparities in colorectal cancer may be caused in part by differences in socioeconomic status.42 Although the health care systems that were included in this study serve predominantly working class populations, there are likely to be socioeconomic differences by race. The health care systems offer a variety of health insurance plans with differing out-of-pocket expenses that may vary by race and affect the receipt of screening or treatment services. Furthermore, the burden of time taken out of work for patients and caregivers for cancer screening, diagnosis, and treatment may have a disproportionate impact on vulnerable populations. There also may be racial differences in the patient-physician communication, including recommendation of cancer-related services, that may affect decisions on the use of screening and treatment services.43-46 In addition, blacks are more likely than whites to report mistrust of the medical care system, to be less satisfied with their care, and to refuse therapy.47 In general, even when insurance coverage is similar, minorities have greater difficulties gaining access to the medical system.48, 49 The impact of these potential barriers and other socioeconomic factors on colorectal cancer disparities in insured populations is not understood well and requires further study.
Outcomes for patients with early colorectal cancer are related to the aggressiveness of surgical therapy.50 Although we were able to determine the receipt of initial cancer-directed therapy after diagnosis, the treatment data available in the tumor registries did not include information on the details of specific types of surgery, radiation or chemotherapy agents used, or whether a complete course of therapy was received.
Conclusions
The relation between race and survival for colorectal cancer is complex. Despite the availability of health care insurance, we observed racial differences in survival from colorectal cancer among patients who received care from integrated health care systems. Our findings suggest that these disparities may be caused by racial differences in the receipt of cancer prevention, detection, and treatment services. To understand whether these disparities persist in the colonoscopy era, these findings need to be updated with more recent data as they become available. Additional studies that include measures of the process of care, including physician recommendation, the extent of evaluation of abnormal colorectal cancer screening, and measures of socioeconomic status in settings of similar health insurance coverage, would provide added insights into colorectal cancer disparities. Given the potential to prevent or cure colorectal cancer, more attention needs to be devoted to assuring that vulnerable populations receive equitable care. Targeted interventions to improve the use of colorectal cancer screening and treatment among vulnerable minority populations may be needed to eliminate disparities in colorectal cancer.
