• colon adenocarcinoma;
  • disparities;
  • blacks;
  • quality of care;
  • diagnostic stage


  1. Top of page
  2. Abstract
  6. Acknowledgements


Several studies have attributed racial disparities in cancer incidence and mortality to variances in socioeconomic status and health insurance coverage. However, an Institute of Medicine report found that blacks received lower quality care than whites after controlling for health insurance, income, and disease severity.


To examine the effects of race on colorectal cancer outcomes within a single setting, the authors performed a retrospective cohort study that analyzed the cancer registry, billing, and medical records of 365 university hospital patients (175 blacks and 190 whites) diagnosed with stage II-IV colon cancer between 2000 and 2005. Racial differences in the quality (effectiveness and timeliness) of stage-specific colon cancer treatment (colectomy and chemotherapy) were examined after adjusting for socioeconomic status, health insurance coverage, sex, age, and marital status.


Blacks and whites had similar sociodemographic characteristics, tumor stage and site, quality of care, and health outcomes. Age and diagnostic stage were predictors of quality of care and mortality. Although few patients (5.8%) were uninsured, they were more likely to present at advanced stages (61.9% at stage IV) and die (76.2%) than privately insured and publicly insured patients (p = .002).


In a population without racial differences in socioeconomic status or insurance coverage, patients receive the same quality of care, regardless of racial distinction, and have similar health outcomes. Age, diagnostic stage, and health insurance coverage remained independently associated with mortality. Future studies of disparities in colon cancer treatment should examine sociocultural barriers to accessing appropriate care in various healthcare settings. Cancer 2010. © 2009 American Cancer Society.

Colorectal cancer is the third most common cancer. In 2008, an estimated 148,810 new cases of colorectal cancer were expected to be diagnosed, and 8.8% of all cancer deaths (49,960) were expected to occur because of colorectal cancer.1 Blacks are burdened with higher mortality rates in all cancer sites combined, including colorectal cancer. Colorectal cancer incidence rates are 1.2× higher in black males than white males and 1.3× higher in black females than white females; mortality rates are 1.4× higher in both black males and females than their white counterparts.2 Over the past 25 years, the black-white disparity in mortality has increased annually in males and females.

Some studies suggest racial disparities in colorectal cancer incidence and mortality may be attributed to tumor biology.3-7 However, before the mid 1970s, blacks did not experience disparities in colorectal cancer incidence or mortality. As treatment options for colorectal cancer advanced, disparities between blacks and whites grew substantially. Incidence rates increased sharply in black men between 1973 and 1980 before leveling off and remained level in black women, whereas incidence rates steadily decreased in white men and women.8 Similarly, colorectal cancer mortality rates among whites progressively decreased since the 1950s in women and the 1980s in men, while increasing before leveling off during the mid 1980s and early 1990s in black women and men, respectively.8 These statistics suggest that tumor biology alone may not fully explain the excess mortality seen among black patients. Racial disparities in colorectal cancer mortality may be more aptly explained by variances in treatment.

Several empirical studies have identified statistically significant differences between blacks and whites in colorectal cancer screening practices,9-12 whereas other studies have found no significant differences.13-15 Recent reports from the American Cancer Society suggest that gaps between blacks and whites in colorectal cancer screening are relatively small. Fecal occult blood testing rates were 19.2% and 19.4%, respectively, and colonoscopy/sigmoidoscopy rates were 44.3% and 47.0%, respectively.2

Most studies that examined the quality of stage-specific colorectal cancer treatment found notable black-white differences.16-26 Blacks generally did not receive surgery, chemotherapy, radiation therapy, or follow-up surveillance at the same rate as whites even when their disease stage was less advanced. Although racial disparities in treatment were consistently indicated; in many of these studies, the ratio of blacks to whites was generally low, and sociodemographic factors were dissimilar.16, 17, 19, 20-27 The majority of studies did not apply standardized quality of care measures and did not examine treatment delay. It is evident that further black-white comparison research on the quality of colon cancer treatment is needed.

The authors undertook this study to assess black-white differences in the quality of stage-specific colon cancer treatment in a patient population comparable in size and demographic characteristics, thereby adjusting for sociodemographic factors. The second objective was to determine whether differences in colon cancer treatment quality explained differences in mortality rates.


  1. Top of page
  2. Abstract
  6. Acknowledgements

A retrospective study was developed to examine the medical records of colon cancer patients who received care within the Emory University Hospital System between 2000 and 2005. The Emory University Hospital System includes Emory Hospital, Crawford Long Hospital, Winship Cancer Institute, and the Emory Clinics. After approval by the Emory University Institutional Review Board, algorithms were developed to query electronic medical records from linked data sources to identify patients using cancer registry International Classification of Diseases for Oncology codes and to determine diagnosis, treatment, and health outcomes.28 Fields from administration, hospital, and clinic information were used to identify distinguishing patient characteristics such as race, socioeconomic status (SES), health insurance coverage, sex, age, and marital status, as well as to assess the quality of care received within each domain of care including time to initiation of treatment.

Using Practice Guidelines to Define Quality of Care Measures

Quality of care measures (effective and timely treatment) were determined using the conceptual framework developed by the Institutes of Medicine to characterize healthcare quality as effective and timely.29 Effective care was defined as the receipt of appropriate recommended treatment based on stage distribution.29 Measures used to determine effective and timely colon cancer treatment were derived from the National Comprehensive Cancer Network (NCCN) Clinical Practice Guidelines in Oncology version 2.2006 and guidelines developed by the American Society of Clinical Oncology's Quality Oncology Practice Initiative, an oncologist-led, practice-based quality improvement program.30, 31 NCCN and Quality Oncology Practice Initiative guidelines for colon cancer include a pathologic workup (colonoscopy, pathology review, and marking of cancerous site) for all patients. For stage II-IV patients, a colectomy (surgical removal of the tumor and regional lymph nodes and nearby tissues or organs for patients with tumors that have spread through the wall of the colon) is the standard of care. Chemotherapy, including 5-fluorouracil, capecitabine, oxaliplatin, and irinotecan, is recommended for patients with stage III or IV colon cancer. Timeliness of appropriate care was examined with respect to receipt of a colectomy within 30 days of diagnosis in stage II-IV patients, and using the Quality Oncology Practice Initiative recommendation of chemotherapy within 120 days of diagnosis in stage III-IV patients.29

Detailed reviews of individual medical records were conducted to identify provided treatment, diagnostic reports, surgical procedures, and chemotherapy regimens. Fields from the cancer registry dataset and pathology notes were used to confirm diagnosis of colon cancer. Healthcare identifiers, procedure codes, and procedure dates were used to identify the occurrence and dates of treatment received. Pharmacy information was used to determine chemotherapy regimens and other medications ordered for patients. Medical record chart notes were reviewed for additional details regarding pathology reports, chemotherapy, and medications.

A total of 472 black (n = 210) and white (n = 262) patients identified by cancer registry records were diagnosed with colon cancer between 2000 and 2005. Patients with stage I (8, black/white ratio 3:5) and unknown stage (99, black/white ratio 1:2) were excluded. Patients with stage II-IV colon cancer (n = 365) met the inclusion criteria for this study. With 190 white subjects and 175 black subjects, this study was powered to detect a 15% difference in mortality from a baseline mortality rate for whites of 44% (2-sided alpha = .05, beta = .79).

Descriptive statistics were calculated for the demographic profile variables, which included race, sex, age, marital status, employment status, and health insurance coverage; the diagnostic variables, which included diagnostic stage and site; the treatment variables, which included effective and timely receipt of appropriate treatment; and mortality. Measures for treatment were calculated as the percentage receiving guideline-recommended treatment. The following demographic, diagnostic, treatment, and outcomes variables, respectively were analyzed: race, sex, age, marital status, employment status, and health insurance coverage; diagnostic stage and site; effective and timely receipt of a colectomy and adjuvant chemotherapy; and mortality. Racial comparisons of demographic variables were performed. Demographic variables were also compared by stage distribution, effective and timely treatment, and mortality rates.

Binary forward stepwise logistic regressions were performed to determine: 1) if there were racial differences in the effectiveness and timeliness of stage-specific treatment after adjusting for other sociodemographic factors (sex, age, employment status, marital status, and health insurance coverage); and 2) if differences in sociodemographic characteristics and quality of care explained differences in mortality rates. The Statistical Package for Social Sciences 15.0 (SPSS Inc., Chicago, Ill) was used to perform statistical analyses.


  1. Top of page
  2. Abstract
  6. Acknowledgements

The patient population consisted of 190 whites (52.1%) and 175 blacks (47.9%), of whom 188 were male (51.5%), and 177 were female (48.5%). Most patients were between the ages of 50 and 79 years (71.5%), married (59.5%), employed (49.0%), or retired (38.4%) and publicly insured (60.0%). There were no statistically significant racial differences in age, marital status, employment status, or health insurance coverage. As shown in Table 1, there were racial differences in sex, as women represented 54.3% of black patients but only 43.2% of white patients (P = .034). About 40% of patients presented with stage III disease (n = 155), and cancer was diagnosed most frequently in the sigmoid (24.7%), ascending (18.4%), and cecum (18.4%) sites of the colon.

Table 1. Racial Comparison of Demographic and Diagnostic Variables
VariablesRaceCumulative TotalP
  1. NOS indicates not otherwise specified.

Sex      .034
Age, y      .210
Marital status      .103
Employment status      .211
Health insurance coverage      .163
 Private insurance6031.66537.112534.2 
 Public insurance12264.29755.421960.0 
Diagnostic stage      .606
 Stage II5830.54827.410629.0 
 Stage III7640.07945.115542.5 
 Stage IV5629.54827.410428.5 
Diagnostic site      .328
 Hepatic flexure63.284.6143.8 
 Splenic flexure31.674.0102.7 
 Overlap lesion10.500.010.3 
Diagnostic site right/left distribution      .335
 Right side7841.18246.916043.8 
 Left side10253.78850.319052.1 

Demographic differences were found in stage distribution. Uninsured patients (n = 21, 5.8%) were more likely to present at advances stages (61.9% at stage IV) than privately insured patients (32.0% at stage IV) or publicly insured patients (23.3% at stage IV, P = .002). Patients <50 years of age (n = 34, 9.3%) were more likely to be diagnosed at later stages (47.1% at stage III and 41.2% at stage IV, P = .07) than patients in older age groups. No significant racial differences were found in diagnostic stage or site, including right and left side distribution (see Table 1).

Effective Treatment

Table 2 presents data on appropriate receipt of treatment by demographics. A colectomy, which is recommended for all stage II-IV patients, was performed on 86.8% of the study population. Chemotherapy, which is recommended for all stage III and IV patients, was administered in 54.1% of stage III and IV patients. No demographic variable was significantly associated with decreased rates of colectomy. Statistically significant associations were seen between receipt of chemotherapy and age, marital status, and employment status.

Table 2. Comparison of Demographic Variables and Receipt of Effective Treatment
VariablesColectomy Stages II-IVPChemotherapy Stages III-IVP
Race  .350  .871
 White16285.3 7254.5 
 Black15588.6 6853.5 
Sex  .480  .824
 Men16185.6 7153.4 
 Women15688.1 6954.8 
Age, y  .398  .000
 <502779.4 2583.3 
 50-6410387.3 5663.6 
 65-7912889.5 5052.6 
 ≥805984.3 919.6 
Marital status  .309  .010
 Single2683.9 1664.0 
 Married18685.7 9059.2 
 Widowed7286.7 1834.0 
 Separated/divorced3397.1 1655.2 
Employment status  .385  .019
 Employed15988.8 8162.8 
 Retired10886.4 3744.6 
 Unemployed/unknown5082.0 2246.8 
Health insurance coverage  .065  .057
 Private insurance10785.6 5762.0 
 Public insurance19589.0 7047.6 
 No insurance/unknown1571.4 1365.0 

Timely Treatment

Table 3 presents data on the receipt of timely stage-appropriate treatment (colectomy and chemotherapy) by demographics. The diagnosis date for 10.1% (n = 32) of patients followed their surgical date, and these patients were excluded from the analysis. Of the remaining 285 patients, 245 (86.0%) received a timely colectomy (within 30 days of diagnosis). The date of diagnosis followed the date of chemotherapy for 3.6% (n = 5) of patients with stage III-IV colorectal cancer, and these patients were excluded as well. Of the remaining 135 patients, 113 (96.4%) received timely chemotherapy (within 120 days of diagnosis). As seen in Table 4, there were statistically significant associations between effective and timely treatment variables and diagnostic stage. Stage of disease at diagnosis significantly influenced receipt of a colectomy and chemotherapy and the administration of chemotherapy within 120 days of diagnosis. There were no significant associations between diagnostic site and treatment variables Table 5).

Table 3. Comparison of Demographic Variables and Receipt of Timely Treatment
VariablesDiagnosis to Colectomy ≤30 DaysPDiagnosis to Chemotherapy ≤120 DaysP
Race  .611  .308
 White12787.0 5776.0 
 Black11884.9 5882.9 
Sex  .645  .388
 Men12686.9 6082.2 
 Women11985.0 5576.4 
Age, y  .393  .755
 <502492.3 2284.6 
 50-648588.5 4781.0 
 65-799081.8 4175.9 
 ≥804686.8 571.4 
Marital status  .515  .884
 Single2080.0 1381.3 
 Married13984.8 7680.0 
 Widowed6090.9 1372.2 
 Separated/divorced2686.7 1381.3 
Employment status  .150  .985
 Employed12888.9 6779.8 
 Retired8385.6 2978.4 
 Unemployed/unknown3477.3 1979.2 
Health insurance coverage  .188  .473
 Private insurance9089.1 4877.4 
 Public insurance14585.3 5578.6 
 No insurance/unknown1071.4 1292.3 
Table 4. Comparison of Effective/Timely Treatment Variables and Diagnostic Stage
VariablesDiagnostic StageP
Stage IIStage IIIStage IV
Colectomy      .000
 Not received1118.1905.82826.9 
Chemotherapy      .000
 Not received9388.27749.74240.4 
Diagnosis to colectomy      .449
 ≤30 days7286.711183.56289.9 
 ≥30 days1113.32216.5710.1 
Diagnosis to chemotherapy      .000
 ≤120 days220.06081.15386.9 
 ≥120 days880.01418.9813.1 
Table 5. Comparison of Effective/Timely Treatment Variables and Diagnostic Site
VariablesDiagnostic SiteP
Right SideLeft Side Stage IV
Colectomy    .451
 Not received217.32010.5 
Chemotherapy    .704
 Not received9559.410957.4 
Diagnosis to colectomy    .346
 ≤30 days10888.513284.6 
 ≥30 days1411.52415.4 
Diagnosis to chemotherapy    .103
 ≤120 days5461.35461.3 
 ≥120 days738.71438.7 

Nearly half of patients (45.5%) died during the study period (2000-2005). Univariate analyses uncovered statistically significant associations between mortality and diagnostic stage, sex, age, marital status, employment status, and health insurance coverage. Race was not statistically associated with mortality (Table 6). Increased mortality was observed among those who did not receive a colectomy (75.0% vs 41%; P = .000) and those who did not receive adjuvant chemotherapy for stage III disease (41.6% vs 25.6%; P = .036). No differences in mortality were observed based on the timeliness of colectomy or chemotherapy.

Table 6. Comparison of Demographic Variables and Mortality Rates
Race  .353
Sex  .027
Age, y  .000
Marital status  .046
Employment status  .003
Health insurance coverage  .002
 Private insurance4636.8 
 Public insurance10447.5 
 No insurance/unknown1676.2 
Diagnostic stage  .000
 Stage II2725.5 
 Stage III5233.5 
 Stage IV8783.7 
 Diagnostic site  .805
 Right side7245.0 
 Left side8343.7 

In multivariate logistic regression models shown in Table 7, demographic factors were not significant predictors of receipt of a colectomy in stage II-IV patients, although diagnostic stage was a predictor. Stage IV patients (odds ratio [OR], 0.314, 95% confidence interval [CI], 0.1-0.7) were less likely than stage II patients to receive a colectomy. Age was a significant predictor of receipt of chemotherapy in stage III-IV patients. Patients in the oldest age groups were less likely to receive chemotherapy (age 65-79 years: OR, 0.222; 95% CI, 0.1-0.6 and age ≥80 years: OR, 0.049; 95% CI, 0.0-0.2). Demographic factors were not significant predictors of timely receipt of a colectomy or chemotherapy, although diagnostic stage was a significant predictor of receipt of timely chemotherapy. Stage III (OR, 17.1; 95% CI, 3.3-89.7) and stage IV patients (OR, 26.5; 95% CI, 4.7-147.8) were more likely to receive timely chemotherapy than stage II patients. Although chemotherapy is not routinely recommended for stage II patients, chemotherapy may be considered in stage II patients with inadequately sampled nodes, T4 lesions, perforation, or poorly differentiated histology.31

Table 7. Multivariate Stepwise Logistic Regression Analysis
VariablesOR95% CIP
  1. OR indicates odds ratio; CI, confidence interval.

Model for receipt of colectomy (stage II-IV)    
 Stage II (reference)    
 Stage III1.878.7504.704.178
 Stage IV.314.147.672.003
Model for receipt of chemotherapy (stage III-IV)    
 Age <50 years (reference)    
 Age 50-64 years.350.1221.004.051
 Age 65-79 years.222.078.629.005
 Age ≥80 years.
Model for diagnosis to chemotherapy ≤120 days    
 Stage II (reference)    
 Stage III17.1433.27689.718.001
 Stage IV26.5004.752147.782.000
Model for mortality    
 Age <50 (reference)    
 Age 50-641.000.045.4021.000
 Age 65-793.
 Age ≥807.436.226.869.000
 Stage II (reference)    
 Stage III2.0081.1033.657.023
 Stage IV4.50611.11354.039.000
 Colectomy received (reference)    
 Colectomy not received3.1541.3717.257.007

The following variables were found to be significant predictors of mortality: age 65-79 years (OR, 3.29; 95% CI, 1.2-9.0), age ≥80 years (OR, 7.44; 95% CI, 2.5-22.2), and not receiving a colectomy (OR, 3.15; 95% CI, 1.4-7.3). Diagnostic site was not a predictor in logistic regression models. However, diagnostic site was a predictor of mortality in blacks when black/white subjects were analyzed separately (right side: OR, 0.401; 95% CI, 0.168-0.956; P = .039).


  1. Top of page
  2. Abstract
  6. Acknowledgements

The Institutes of Medicine report Unequal Treatment: Confronting Racial and Ethnic Disparities in Healthcare recognized that much of the racial disparity in colorectal cancer outcomes may be attributed to variances in the quality of care.32 Numerous federal reports on healthcare quality have identified systemic barriers to quality healthcare including bias, discrimination, and differential treatment; increasing gaps between blacks and whites in SES and health insurance coverage; and disproportionate access to medical services between blacks and whites.33-36 Although access to healthcare is essential to eliminating racial disparities in cancer, studies have found that considerable racial disparities exist in colorectal cancer care quality even when access to care is not a barrier.4, 9, 16-19, 37-39 The President's Cancer Panel (2001) reported that a gap exists between advances in cancer research and the delivery of affordable, accessible, and equitable care.36

Clinical trials and Veteran Administration studies have shown that when equitable care is provided to patients from different racial and ethnic backgrounds, treatment response, including disease recurrence, mortality rates, disease-free survival, and overall survival, is similar.27, 40-43 In this study, black-white comparisons with respect to sociodemographic characteristics, severity of disease, quality of treatment, and health outcomes were similar, which supports the premise that equal treatment yields equal outcomes. This study found that blacks were slightly more likely than whites to undergo a colectomy (88.6% vs 85.3%, respectively) and just as likely to receive chemotherapy. These findings differ from several studies that demonstrate significant disparities between blacks and whites in stage distribution, tumor location, grade, and histology4, 6, 18, 34, 42-45; quality of treatment (receipt of appropriate surgery and chemotherapy)16-19, 22, 23, 43; and mortality rates and survival.22, 24, 25, 45

Some studies attribute disparities in severity of disease, quality of care, and health outcomes to differences in SES and health insurance coverage.24-26, 44, 45 Two studies, both conducted in Michigan, found that SES and health insurance status explained later stage distribution (stage III-IV).44, 45 Notwithstanding, blacks were more likely than whites to die of colorectal cancer, even at earlier stages of the disease.4, 6, 18, 34, 44-47 Blacks were also more likely to present with cancer in the proximal and descending colon4, 6, 34, 48 and to present at younger ages.4, 6, 44, 45 In these studies, a greater proportion of blacks were in a lower economic strata and lacked adequate health insurance.

Although very few patients in this study were <50 years old (9.3%) or uninsured (5.8% compared with the national rate of 15.3%),49 these patients were more likely to be diagnosed at later stages. Recent studies assessing the association between health insurance coverage and diagnostic stage also found uninsured patients were more likely to be diagnosed at advanced stages and experience unfavorable outcomes.50, 51 Whereas US Census data indicate racial disparities in health insurance coverage (19.5% of blacks compared with 14.3% of whites without adequate health insurance) and median income ($32,876 and $53,910, respectively),49 black patients in this study were more likely than their white counterparts to be employed (52.0% and 46.3%, respectively) and privately insured (37.1% and 31.6%, respectively). These findings showing similarities in employment status and health insurance coverage may be reflective of the demographics of metropolitan Atlanta, where more blacks are represented in higher educational and economic strata compared with the national average.

This study provides a comprehensive comparison of black and white patients treated within a common setting, but clearly is limited in its scope by the retrospective nature of the study and missing data common to these clinical studies. Results indicated that the percentage of patients without health insurance in this study was significantly lower than the national average, and blacks were more likely to have private insurance and less likely to have public insurance compared with whites; therefore, study findings may not be generalized.

Socioeconomic data such as income and educational level were not captured in uniformly administrative, hospital, or clinical records. Diagnostic stage was not determined in nearly 1 of 3 of identified patients (n = 99). Treatment data were obtained from 2 data sources, a Healthcare Data Warehouse and Power Chart (an electronic medical record). There were inconsistencies across data sources, although in most instances information was identical.

Just over half of patients (54%) diagnosed with stage III and IV colon cancer reportedly received chemotherapy. It would be highly unusual in clinical practice for patients with stage III and IV colon cancer not to have undergone chemotherapy. The authors suspect that numerous patients who underwent surgery in the Emory University Hospital System may have received chemotherapy at a clinic outside of the hospital system, and this was not captured by university data systems. Lastly, there were treatment advances in therapy and supportive care during the study period that could have affected outcomes for patients diagnosed in later years.


The Institutes of Medicine devised a definition for healthcare quality—patient-centered care that meets clinical performance standards consistent with current knowledge and makes desired outcomes more likely. Donabedian asserted that healthcare quality involved the setting in which care occurred, the processes of care, and the outcomes of care, and he established 7 pillars of quality: efficacy, efficiency, optimality, acceptability, legitimacy, equity, and cost.52 The Institutes of Medicine adopted his views in the development of its conceptual framework that characterized healthcare quality as effective, safe, timely, and patient centered.53

Several factors such as SES and lack of adequate health insurance impact the quality of healthcare provided to black colon cancer patients. Disparities between blacks and whites resulting from the intersection of race and SES may be more apparent in healthcare systems that serve greater percentages of low-income and uninsured or underinsured patient populations than those serving more affluent and insured patients. Future studies of disparities in colon cancer treatment should use standardized quality of care measures, including timeliness of care, and examine sociocultural barriers to accessing quality healthcare in public and private settings. Disparities in colorectal cancer and many other diseases could be drastically reduced or eliminated if quality healthcare was made affordable, accessible, and equitable.


  1. Top of page
  2. Abstract
  6. Acknowledgements

We thank the following institutions for direction and support: Morehouse School of Medicine, Grady Health System, Emory University, and Clark Atlanta University.


  1. Top of page
  2. Abstract
  6. Acknowledgements

Supported (in part) by grant #5U48DP000049-03 from the Centers for Disease Control and Prevention; Dr. Flowers' Georgia Cancer Coalition Distinguished Scientist Award and Amos Medical Faculty Development Award; and the Georgia Center for Health Equality, National Institutes of Health/National Center on Minority Health and Health Disparities, 5P60MD000525-04 (principal investigator: Otis W. Brawley, Maryland).


  1. Top of page
  2. Abstract
  6. Acknowledgements