Residual Treatment Disparities After Oncology Referral for Rectal Cancer

Transcription

1 JNCI Journal of the National Cancer Institute Advance Access published May 13, 2008 ARTICLE Residual Treatment Disparities After Oncology Referral for Rectal Cancer Arden M. Morris, Kevin G. Billingsley, Awori J. Hayanga, Barbara Matthews, Laura-Mae Baldwin, John D. Birkmeyer Background Black patients with rectal cancer are considerably less likely than white patients to receive adjuvant therapy. We examined the hypothesis that the lower treatment rate for blacks is due to underreferral to medical and radiation oncologists. Methods We used Surveillance, Epidemiology, and End Results Medicare data to identify elderly ( 66 years of age) patients who had been hospitalized for resection of stage II or III rectal cancer (n = 2716). We used 2 tests to examine associations between race and 1) consultation with an oncologist and 2) receipt of adjuvant therapy. We then used logistic regression to analyze the influence of sociodemographic and clinical characteristics (age at diagnosis, sex, marital status, median income and education in area of residence, comorbidity, and cancer stage) on black white differences in the receipt of adjuvant therapy. All statistical tests were two-sided. Results Conclusion There was no statistically significant difference between the 134 black patients and the 2582 white patients in the frequency of consultation with a medical oncologist (73.1% for blacks vs 74.9% for whites, difference = 1.8%, 95% confidence interval [CI] = 5.9% to 9.5%, P =.64 ) or radiation oncologist (56.7% vs 64.8%, difference = 8.1%, 95% CI = 0.5% to 16.7%, P =.06 ), but blacks were less likely than whites to consult with both a medical oncologist and a radiation oncologist (49.2% vs 58.8%, difference = 9.6%, 95% CI = 0.9% to 18.2%, P =.03 ). Among patients who saw an oncologist, black patients were less likely than white patients to receive chemotherapy (54.1% vs 70.2%, difference = 16.1%, 95% CI = 6.0% to 26.2%, P =.006 ), radiation therapy (73.7% vs 83.4%, difference = 9.7%, 95% CI = 0.4% to 19.8%, P =.06), or both (60.6% vs 76.9%, difference = 16.3%, 95% CI = 4.3% to 28.3%, P =.008). Patient and provider characteristics had minimal influence on the racial disparity in the use of adjuvant therapy. Racial differences in oncologist consultation rates do not explain disparities in the use of adjuvant treatment for rectal cancer. A better understanding of patient preferences, patient provider interactions, and potential influences on provider decision making is necessary to develop strategies to increase the use of adjuvant treatment for rectal cancer among black patients. J Natl Cancer Inst 2008;100: Despite recent overall improvements in rectal cancer outcomes ( 1, 2 ), black patients continue to have substantially worse prognosis than white patients. Current data from the Surveillance, Epidemiology, and End Results (SEER) registry reveal that the long-term survival rates after rectal cancer surgery are 14% 20% lower for black patients than for white patients ( 3 ). Many non cancer-related reasons for this difference have been identified, such as increased burden of comorbid disease among black patients, later stage at presentation, and sociodemographic factors that are independently associated with decreased survival ( 4 6 ). However, our group and others ( 7, 8 ) have shown that disparities in survival are also partly attributable to lower use of adjuvant radiation and chemotherapy among black patients. It is unclear why blacks are less likely than whites to receive appropriate adjuvant therapy. One possibility is that black patients and their oncologists actively consider therapy but make an explicit decision not to proceed with treatment. Black patients may be more likely to decline adjuvant treatment than white patients because of a sense of fatalism toward disease, a general lack of trust Affiliations of authors: Department of Surgery, University of Michigan, Ann Arbor, MI (AMM, AJH, JDB); Michigan Surgical Collaborative for Outcomes Research and Evaluation, University of Michigan, Ann Arbor, MI (AMM, JDB); Department of Family Medicine, University of Washington, Seattle, WA (BM, LMB); Department of Surgery, Oregon Health and Science University, Portland, OR (KGB). Correspondence to: Arden M. Morris, MD, MPH, Department of Surgery, University of Michigan, 1500 East Medical Center Dr, TC-5343, Ann Arbor, MI ( ammsurg@umich.edu ). See Funding and Note following References. DOI: /jnci/djn The Author(s). This is an Open Access article distributed under the terms of the Creative Com - mons Attribution Non-Commercial License ( by-nc/2.0/uk/), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. 738 Articles JNCI Vol. 100, Issue 10 May 21, 2008

2 in health-care providers, or concerns about the risks and side effects of treatment ( 9 ). These concerns may stem from previous personal experiences or fundamental cultural beliefs. Alternatively, black patients may be more likely than white patients to decline therapy because of compelling personal circumstances, such as being the family caregiver ( 10 ) or being unable to travel for regular treatment. Oncologists may recommend against adjuvant therapy because of clinical considerations such as the presence of comorbid disease, which is more common among black patients ( 11, 12 ), or because, as suggested by limited retrospective data ( 13 ), they believe that therapy is less effective among black patients. Even in the absence of overt discrimination, race-related differences in patient provider interactions may also influence patients preferences or oncologists clinical recommendations regarding adjuvant treatment for rectal cancer. Another possible reason for the disparity in adjuvant treatment is that black patients are simply not being offered a choice. That is, it is possible that black patients with advanced rectal cancer have a lower rate of referral to oncologists in the first place and thus have no opportunity to choose adjuvant therapy. Previous studies ( ) indicate that black patients who are diagnosed with a host of diseases are less likely than white patients to be referred for treatment or simply to obtain access to specialty care. To better understand the role of access to consultation in subsequent use of adjuvant therapy, we used a large national database to compare the frequency of oncologist consultations between black and white patients with stage II or III rectal cancer. We then explored whether disparities in the use of adjuvant therapy persisted among patients who had an initial consultation. Methods Data We used the SEER-Medicare linked database ( 17 ) of the National Cancer Institute, which contains clinical, demographic, and medical claims data on patients aged 65 years or older, to identify rectal cancer patients who were diagnosed between January 1, 1992, and December 31, During this time period, SEER registries that covered five states New Mexico (includes Arizona Indians), Connecticut, Hawaii, Iowa, and Utah and six county-based areas within an additional four states Atlanta rural Georgia, Detroit, Los Angeles, San Francisco Oakland, San Jose Monterey, and Seattle Puget Sound ascertained the incident cases of rectal cancer among a population that represents approximately 14% of persons living in the United States. The SEER data included information about tumor stage and patient demographics. The Medicare claims data included information about Medicare enrollment dates, health maintenance organization membership, and feefor-service claims for inpatient and outpatient services provided to patients, including surgical care and hospital stays, physician consultations, and treatment with chemotherapy and/or radiation. This study was approved by the University of Washington Institutional Review Board. Study Population We identified all patients aged 66 years or older who were diagnosed with rectal cancer of American Joint Committee on Cancer CONTEXT AND CAVEATS Prior knowledge Black patients with rectal cancer are considerably less likely than white patients to receive adjuvant therapy, but it is unclear whether this disparity is due to underreferral of blacks to medical and radiation oncologists. Study design Surveillance, Epidemiology, and End Results (SEER) Medicare data were used to identify elderly ( 66 years of age) patients who had been hospitalized for resection of stage II/III rectal cancer. Associations between race and consultation with an oncologist and receipt of adjuvant therapy were analyzed. Logistic regression was used to analyze the influence of sociodemographic and clinical characteristics on black white differences in the receipt of adjuvant therapy. Contribution There was no difference in the frequency of consultation with a medical or radiation oncologist between the black patients and the white patients, but blacks were less likely than whites to consult with both a medical oncologist and a radiation oncologist. Among patients who saw an oncologist, black patients were less likely than white patients to receive chemotherapy, radiation therapy, or both. Patient and provider characteristics had minimal influence on the racial disparity in the use of adjuvant therapy. Implications A better understanding of patient preferences, patient provider interactions, and potential influences on provider decision making is necessary to develop strategies to increase the use of adjuvant treatment for rectal cancer among black patients. Limitations Aggregate data were used to describe patient-level income. Patient and physician decision making was not explored. Medicare data are limited to patients older than 65 years; thus, the findings may not apply to all patients with rectal cancer. [AJCC ( 18 )] stage II or III between January 1, 1992, and December 31, 1999 (n = 4647). Patients were excluded if 1) their tumor histology (ie, a sarcoma or carcinoid tumor) required that they receive atypical treatment (n = 119); 2) they had a prior colorectal cancer diagnosis (n = 112); 3) their cancer was diagnosed at death or autopsy (n = 6); or 4) they had a simultaneous diagnosis of AJCC stage IV cancer (n = 8). We used three additional exclusion criteria to ensure adequate preoperative data to identify comorbid diseases and adequate postoperative time duration for implementation of adjuvant therapy. First, because we used Medicare data to ascertain comorbidity for appropriate risk adjustment, patients who were not enrolled in Parts A and B of the Medicare fee-for-service plan for the 12 months before the month of their rectal cancer diagnosis were excluded (n = 1005). Second, because AJCC stage II/III rectal cancer patients who have not undergone resection within 6 months of diagnosis are highly unlikely to undergo curative resection and therefore to receive adjuvant therapy (as opposed to salvage or palliative therapy), we excluded patients who had not undergone surgical resection within 6 months of the diagnosis month (n = 813). Third, to ensure that we could ascertain receipt of adjuvant jnci.oxfordjournals.org JNCI Articles 739

3 treatment with radiation or chemotherapy, we excluded all patients who were not alive and fully enrolled in parts A and B of the Medicare fee-for-service plan for the 9-month period that began with the month of diagnosis (n = 1457). (Some patients met more than one exclusion criteria; thus, the total number excluded was less than the sum of each category.) Nearly identical proportions of black and white patients were excluded for histology, previous cancer, or diagnosis at death. Different proportions of black and white patients were excluded for incomplete prior enrollment (22% white vs 30% black), no resection (18% white vs 27% black), and incomplete subsequent enrollment (32% white vs 44% black). Overall, 41% of white patients and 54% of black patients were excluded. The final study sample included 2716 patients, of whom 134 were black and 2582 were white. Study Variables The primary outcome of interest was administration of radiation or chemotherapy within 9 months of diagnosis of stage II or III rectal cancer among patients who had consulted with a radiation on - cologist and/or a medical oncologist. The secondary outcome was having an initial consultation with a radiation or medical oncologist. The independent variable was black or white patient race, as designated by the SEER database. We defined the administration of chemotherapy as the presence of inpatient and outpatient hospital claims or physician office claims for chemotherapeutic agents that are specifically used to treat colon and rectal cancers, that is, 5-fluorouracil, leucovorin, or floxuridine (Healthcare Common Procedure Coding System codes J9190 and J0640) and by general codes for chemotherapy during the appropriate time period (Current Procedural Terminology [CPT] codes , 96520, 96530, 96545, and and International Classification of Diseases, Ninth Revision, Clinical Modification [ ICD-9-CM ], diagnosis codes E0781, Q0083 Q0085, and V58.1 and procedure code 99.25). We defined administration of radiation therapy as any of the following codes for management or administration of radiation therapy: CPT codes , , 77470, 77499, 77750, , , or 77799; ICD-9-CM codes , 92.28, or 92.29, ICD-9-CM diagnosis code V58.0; or revenue center code We defined the initial medical oncologist consultation as the first Medicare claim from a medical oncologist (identified by carrier file Health Care Financing Administration [HCFA] provider specialty code 90) who was seen during the 1 month before or the 9 months after (and including) the month of diagnosis. We defined radiation oncologist (HCFA provider specialty code 92) consultations in an identical fashion. Patient Characteristics We obtained information about each patient s age at diagnosis, sex, and marital status from the SEER database. We collected information about each patient s income and education as detailed below using data from the year 1990 US Census for patients who were diagnosed through December 31, 1995, and data from the year 2000 US Census for patients who were diagnosed on January 1, 1996, or later. US Census data include median household income and median education level (according to race and ethnicity) for each census tract and zip code. Each patient was assigned the median income in their zip code of residence. Education level from 1990 US Census data was used to establish the race-specific percentage of persons older than 25 years with more than 12 years of education within each patient s census tract. Patients were classified according to their comorbidity index scores (0, 1, or 2). To assign the comorbidity score, we examined inpatient and outpatient Medicare claims during the 11 months before the month before the diagnosis month and used Romano s adaptation of the Charlson comorbidity index, as previously described ( 19 ). We used inpatient Medicare claims to determine the length of stay for the cancer resection hospitalization and, per common practice, defined prolonged hospitalization as a hospital stay that lasted more than 14 days. For some analyses, length of hospital stay was divided into three categories: hos pitalization up to 7 days (a normal length of stay); 8 14 days, a prolonged stay after colectomy but one that does not definitively imply a complication; and more than 14 days, a prolonged stay that implies some sort of complication. Provider Characteristics We linked the Unique Physician Identifier Numbers in the Medicare files to the 1993 and 1997 American Medical Association Masterfiles to determine characteristics of each patient s surgeon, medical oncologist, and/or radiation oncologist. Physician characteristics included race, age, number of years in practice, board certification (American Board of Surgery, American Board of Colon and Rectal Surgery, American Board of Internal Medicine, and/or American Board of Radiology), and practice type (solo or group). Case volume was calculated for all surgeons who treated the patients included in this study based on tertiles of total number of cases during the study period. Practice volume was calculated for each oncologist from the number of initial consultation claims for rectal cancer patients in the SEER database that were billed to Medicare in each study year. Statistical Analysis Initially, we used 2 tests to compare overall differences in demographic and clinical characteristics between black and white patients. We assessed the relative unadjusted rates of undergoing an adjuvant therapy consultation with either a radiation or a medical oncologist. For those patients who underwent an initial con sultation, we used the test of binomial proportions to compare unadjusted rates of adjuvant therapy administration by patient race. We then used a multiple logistic regression model to examine the degree to which patient and provider variables were associated with racial differences in the receipt of adjuvant treatment. Sociodemographic and clinical variables (patient age at diagnosis, sex, marital status, median household income in zip code of residence, Charlson Romano comorbidity score, and AJCC stage) were included in the model a priori. Cut points for categorical variables were chosen based on common use (eg, 5-year increments for age) or for generally equitably sized comparison groups (eg, median income); other variables were included in the model if they were statistically significantly associated ( P.05) with the outcome or improved the goodness of fit in either model (ie, of the association between race and chemotherapy administration or radiation therapy administration). Because it is common for adjuvant therapy to be used to treat stage II or III rectal 740 Articles JNCI Vol. 100, Issue 10 May 21, 2008

4 cancer, we calculated estimated relative risks (RRs) of adjuvant therapy by correction of the adjusted odds ratios, as previously described ( 20 ). All analyses were repeated with correction for clustering within hospitals using the general estimating equation; the results of these analyses are not reported because neither the point estimates nor the 95% confidence intervals (CIs) changed considerably. All tests of statistical significance were two-sided, and statistical significance was defined as P.05. All analyses were performed with SAS 9.1 software (SAS Institute, Inc, Cary, NC). Results Among patients who met the inclusion criteria, the 134 black patients did not differ statistically significantly from the 2582 white patients with respect to the most relevant clinical variables, that is, age at diagnosis and stage of disease ( Table 1 ). However, the two patient groups differed statistically significantly in every other demographic category. Black patients were more likely than white patients to be divorced, separated, or single (30.3% vs 11.1%, difference = 19.2%, 95% CI = 11.3% to 27.1%, P <.001); to reside in a zip code with a median household income less than $ per year (50.4% vs 11.1%, difference = 39.3%, 95% CI = 30.7% to 47.9%, P <.001); and to reside in a census tract in which less than 80% of persons aged 25 years or older of the same race had completed 12 years of education (67.2% vs 19.5%, difference = 47.7%, 95% CI = 39.4% to 56.0%, P <.001). During the initial hospitalization for rectal cancer resection, black patients were also statistically significantly more likely than white patients to have a prolonged hospital stay (ie, more than 14 days; 33.1% vs 16.7%, difference = 16.4%, 95% CI = 8.2% to 24.6%, P <.001) and to be operated on by a surgeon in the lowest tertile of annual case volume (65.0% vs 45.3%, difference = 19.7%, 95% CI = 11.0% to 28.4%, P <.001). A greater proportion of black patients than white patients saw a general surgeon rather than a colorectal surgeon, but the difference was not statistically significant (74.0% vs 66.8%, difference = 7.2%, 95% CI = 0.8% to 15.1%, P =.096). Likelihood of Initial Consultation With an Oncologist Although we found no statistically significant difference between black and white patients in the frequency of an initial consultation with either a medical oncologist (73.1% vs 74.9%, difference = 1.8%, 95% CI = 5.9% to 9.5%, P =.64; Table 2 ) or a radiation oncologist (56.7% vs 64.8%, difference = 8.1%, 95% CI = 0.5% to 16.7%, P =.06), black patients were less likely than white patients to consult with both a medical oncologist and a radiation oncologist (49.2% vs 58.8%, difference = 9.6%, 95% CI = 0.9% to 18.2%, P =.03). The medical and radiation oncologists seen by black patients were similar to those seen by white patients with respect to age, sex, number of years in practice, and board certification (data not shown). A greater proportion of black patients than white patients saw a black medical oncologist (14.8% vs 0.8%, difference = 14.0%, 95% CI = 4.5% to 23.5%, P =.004) or a black radiation oncologist (10.9% vs 2.8%, difference = 8.1%, 95% CI = 0.2% to 16.4%, P =.056), but the latter difference was not statistically significant. Likelihood of Subsequent Adjuvant Therapy Although black patients and white patients were equally likely to have an initial oncologist consultation, the use of adjuvant therapy Table 1. Demographic and clinical characteristics by patient race * Characteristics White (n = 2582) differed statistically significantly between blacks and whites who had an initial consultation ( Table 2 ). Black patients were less likely than white patients to receive chemotherapy (54.1% vs 70.2%, difference = 16.1%, 95% CI = 6.0% to 26.2%, P =.006) or radiation therapy (73.7% vs 83.4%, difference = 9.7%, 95% CI = 0.4% to 19.8%, P =.06 ). Among the patients who saw both a medical % Black (n = 134) Patient-related variables Age, y Sex.008 Male Female Marital status <.001 Divorced, separated, or single Married Widowed Median household income in <.001 zip code of residence, $ Race-specific percentage of <.001 persons 25 years old who completed 12 years of education in census tract of residence Modified Charlson comorbidity.021 score AJCC stage Care-related variables Length of hospital stay for <.001 surgical resection, days > Surgeon case volume (annual) < Surgeon board certification.096 None ABS certified ABCRS certified * P values from two-sided 2 tests. Some categories do not total 100% due to rounding. AJCC = American Joint Committee on Cancer; ABS = American Board of Surgery; ABCRS = American Board of Colon and Rectal Surgery. P jnci.oxfordjournals.org JNCI Articles 741

5 Table 2. Likelihood (%) of adjuvant therapy consultation and treatment by patient race * Oncology service received White Black P Medical oncology consultation Use of chemotherapy after consultation Radiation oncology consultation Use of radiation after consultation Both radiation and medical oncology consultations Use of both therapies after consultations * 2 tests and the test of binomial proportions were used for comparisons. All statistical tests were two-sided. oncologist and a radiation oncologist, fewer black patients than white patients underwent both chemotherapy and radiation (60.6% vs 76.9%, difference = 16.3%, 95% CI = 4.3% to 28.3%, P =.008). Accounting for Differences in Adjuvant Therapy Use We used a logistic regression model to examine patient and provider characteristics that were associated with the differential use of adjuvant therapy by black and white patients and derived estimated relative risks from the calculated odds ratios. Among patients who had an initial oncologist consultation, the unadjusted relative risk of chemotherapy use for black vs white patients was 0.76 (95% CI = 0.62 to 0.92) and the unadjusted relative risk of radiation use was 0.87 (95% CI = 0.73 to 0.98) ( Table 3 ). Adjusting for specific patient and provider characteristics in the multivariable analysis had little effect on racial differences in the use of radiation or chemotherapy. After adjusting sequentially for demographic, clinical, and oncologist characteristics, black patients continued to have a statistically significantly lower likelihood of chemotherapy use than white patients (RR = 0.83, 95% CI = 0.67 to 0.98) ( Table 3 ). We found that the largest disparities in the use of chemotherapy between black and white patients were among patients who had characteristics associated with better health ( Table 4 ). For example, among patients with no comorbidities (ie, those with a Charlson score of 0), 72.6% of whites but only 48.9% of blacks received adjuvant chemotherapy (difference = 23.7%, 95% CI = 9.2% to 38.2%, P <.001). Similarly, black white differences in chemotherapy use were greater among patients who were younger at diagnosis or had shorter hospitalizations than among patients who were older or had prolonged hospitalizations. Discussion In recent years, rectal cancer survival rates overall have improved by 15% 20% due to the increasing use of adjuvant chemotherapy and radiation ( 2, ). However, black patients are less likely than white patients to survive rectal cancer, partly because they are less likely to receive adjuvant therapy ( 8 ). In this study, we found that racial differences in receipt of chemotherapy and radiation were not accounted for by racial differences in the initial access to a medical or radiation oncologist. Among patients who had an initial consultation with an oncologist, black patients were 22.9% less likely to have chemotherapy and 11.6% less likely to have radiation than white patients. Several studies ( ) have examined the reasons for nonuse of chemotherapy by rectal cancer patients in the general population without examining reasons for racial disparities in treatment rates. Previously identified factors associated with nonuse of chemotherapy in the general population include older age at diagnosis, excessive comorbidity, lack of clinical indications, and the most commonly cited reason a simple lack of referral to an oncologist. However, none of these factors was associated with the lower receipt of adjuvant therapy among black patients in our cohort of rectal cancer patients. Our examination of baseline patient characteristics that might account for racial disparities in treatment revealed that individual demographic and clinical variables had minimal influence. Predictably, we found that the use of chemotherapy by white patients declined with increasing age at diagnosis and increasing comorbidity. By contrast, among the black patients in our cohort, those with the lowest comorbidity scores, that is, those who were relatively healthier, were the least likely to receive treatment. Although we were unable to discern why our findings were magnified among the healthier black patients, previous studies Table 3. Influence of patient and provider characteristics on RR of adjuvant therapy receipt after initial oncologist consultation for black vs white patients * Model RR, radiation (95% CI) RR, chemotherapy (95% CI) Unadjusted 0.87 (0.73 to 0.98) 0.76 (0.62 to 0.91) Adjusted for age at diagnosis 0.92 (0.82 to 1.00) 0.84 (0.72 to 0.95) Previous model + adjustment for sex 0.87 (0.72 to 0.99) 0.77 (0.62 to 0.92) Previous model + adjustment for marital status 0.90 (0.76 to 1.00) 0.78 (0.63 to 0.93) Previous model + adjustment for median income 0.88 (0.71 to 1.01) 0.76 (0.59 to 0.93) Previous model + adjustment for comorbidity 0.93 (0.80 to 1.02) 0.80 (0.64 to 0.95) Previous model + adjustment for cancer stage 0.90 (0.74 to 1.03) 0.75 (0.57 to 0.93) Previous model + adjustment for length of stay 0.94 (0.80 to 1.03) 0.81 (0.65 to 0.96) Previous model + adjustment for oncologist annual case volume 0.93 (0.79 to 1.03) 0.80 (0.62 to 0.97) Previous model + adjustment for oncologist board certification 0.94 (0.81 to 1.03) 0.80 (0.62 to 0.98) Previous model + adjustment for oncologist number of years in practice 0.93 (0.78 to 1.03) 0.83 (0.67 to 0.98) * RR = relative risk; CI = confidence interval. 742 Articles JNCI Vol. 100, Issue 10 May 21, 2008

6 have found that poorer adherence to screening and other health recommendations among healthier patients is associated with lack of patient experience in medical encounters or increased family caretaker responsibilities ( 10, 30, 31 ). In addition, despite having access to an initial consultation, black patients overall may have been less likely to undergo adjuvant treatment because of difficulty or unfamiliarity with navigating the medical system ( 32 ). Finally, it is possible that black patients overall are more likely than white patients to refuse adjuvant therapy because of social or cultural reasons, such as reduced risk tolerance, lack of faith in medical treatment, or simple resignation to a disease state ( 9 ). Health-care providers play a crucial role in recommending and helping patients navigate treatment options. However, none of the provider characteristics that we examined was associated with racial differences in receipt of adjuvant therapy. It is possible that oncologists recommendations about adjuvant treatment are influenced in unmeasured ways by the race of the patient. For example, on the basis of retrospective cohort data ( 13 ), oncologists may believe that the efficacy of adjuvant therapy for colon and rectal cancers differs by patient race. However, large randomized prospective studies have not shown a difference in efficacy according to race ( ). Alternatively, patient provider interactions or race concordance between patients and their physicians may influence provider behavior, as has been previously reported ( 36, 37 ). In this study, black patients were more likely than white patients to see a nonwhite oncologist. However, the number of nonwhite oncologists was too small to draw conclusions about whether provider race was statistically significantly associated with the use of adjuvant chemotherapy. It is also possible that black patients may receive care from providers or in settings with fewer resources to ensure follow-up or initiation of adjuvant care ( 14 ). Our findings support those of Baldwin et al. ( 38 ), who evaluated the use of chemotherapy among stage III colon cancer patients. In that study, although black and white patients were equally likely to be referred to an oncologist, the black patients in the youngest age group were statistically significantly less likely than their white counterparts to undergo chemotherapy for stage III colon cancer. Taken together, data from Baldwin et al. ( 38 ) and from this study indicate that black patients who are younger and healthier that is, those who have the most potential life years to gain from chemotherapy are the least likely to receive chemotherapy. However, other data ( 39 ) indicate that once therapy for colon cancer is initiated, the completion rates for blacks and whites are the same. Given the persistent racial disparity in colorectal cancer survival, data from this and the previous ( 38, 39 ) studies highlight the importance of initiating adjuvant treatment for black patients. Our study has several limitations that are worth noting. First, we used aggregate data to describe patient-level income, which could have contributed to the finding that income was not associated with the black white disparity in the use of chemotherapy. However, results of one study ( 40 ) indicate that geographic setting or neighborhood aggregate data may be as important as individual patient attributes in terms of access to care. Second, although our administrative data included a record of a first oncologist visit, it inherently lacked the depth necessary to explore patient or physician decision making during that visit. Similarly, administrative data permit a limited assessment of comorbidity and the risk of Table 4. Age, comorbidity, length of hospital stay, contextual income, and the use of chemotherapy among black and white stage II or III rectal cancer patients who had an initial medical oncology consultation * Variables potential complications of chemotherapy. If better risk adjustment were possible using administrative data, more inference about physician decisions might be feasible. Third, Medicare data are limited to patients older than 65 years; thus, our findings may not apply to all patients with rectal cancer. However, through Medicare, all of the patients included in this study had comparable health coverage. It is probable that younger black patients, who are more likely to lack medical insurance than white patients ( 41 ), undergo appropriate adjuvant therapy at even lower rates. In summary, our data indicate that racial disparities in the use of adjuvant therapy for rectal cancer exist despite equal access to an initial oncologist consultation. Moreover, disparities in the use of recommended adjuvant treatment cannot be attributed to easily identifiable attributes of the patient or the provider. Until targets for intervention can be clarified, such as patient physician communication and physician decision making, or the distribution of resources to settings where black patients are most likely to obtain care is improved, it is unlikely that racial disparities in the care of rectal and, possibly, other cancers will be improved in a meaningful way. References Use of chemotherapy, % White Black Modified Charlson score < Age, y Length of stay, d > Median household income in zip code of residence, $ * P values from two-sided 2 tests. 1. Dahlberg M, Pahlman L, Bergstrom R, Glimelius B. Improved survival in patients with rectal cancer: a population-based register study. Br J Surg ; 85 ( 4 ): Vauthey JN, Marsh RW, Zlotecki RA, et al. Recent advances in the treatment and outcome of locally advanced rectal cancer. Ann Surg ; 229 ( 5 ): ; discussion Ries LAG, Melbert D, Krapcho M, et al. SEER Cancer Statistics Review, Bethesda, MD : National Cancer Institute ; Based on November 2006 SEER data submission, posted to the SEER Web site. Available at: [Last accessed: August 20, 2007]. P jnci.oxfordjournals.org JNCI Articles 743

7 4. Chien C, Morimoto LM, Tom J, Li CI. Differences in colorectal carcinoma stage and survival by race and ethnicity. Cancer ; 104 ( 3 ): Wudel LJ Jr, Chapman WC, Shyr Y, et al. Disparate outcomes in patients with colorectal cancer: effect of race on long-term survival. Arch Surg ; 137 ( 5 ): ; discussion Marcella S, Miller JE. Racial differences in colorectal cancer mortality. The importance of stage and socioeconomic status. J Clin Epidemiol ; 54 ( 4 ): Bach PB, Schrag D, Brawley OW, Galaznik A, Yakren S, Begg CB. Survival of blacks and whites after a cancer diagnosis. JAMA ; 287 ( 16 ): Morris AM, Wei Y, Birkmeyer NJ, Birkmeyer JD. Racial disparities in late survival after rectal cancer surgery. J Am Coll Surg ; 203 ( 6 ): Katz JN. Patient preferences and health disparities. JAMA ; 286 ( 12 ): Kim C, Kabeto MU, Wallace RB, Langa KM. Quality of preventive clinical services among caregivers in the health and retirement study. J Gen Intern Med ; 19 ( 8 ): Hertz RP, Unger AN, Cornell JA, Saunders E. Racial disparities in hypertension prevalence, awareness, and management. Arch Intern Med ; 165 ( 18 ): Harris MI. Diabetes in America: epidemiology and scope of the problem. Diabetes Care ; 21 ( suppl 3 ): C11 C Jessup JM, Stewart A, Greene FL, Minsky BD. Adjuvant chemotherapy for stage III colon cancer: implications of race/ethnicity, age, and differentiation. JAMA ; 294 ( 21 ): Bach PB, Pham HH, Schrag D, Tate RC, Hargraves JL. Primary care physicians who treat blacks and whites. N Engl J Med ; 351 ( 6 ): Schneider EC, Zaslavsky AM, Epstein AM. Racial disparities in the quality of care for enrollees in Medicare managed care. JAMA ; 287 ( 10 ): Polednak AP. Segregation, discrimination and mortality in U.S. blacks. Ethn Dis ; 6 ( 1 2 ): Warren JL, Klabunde CN, Schrag D, Bach PB, Riley GF. Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population. Med Care ; 40 ( suppl 8 ): IV Greene F, Page D, Fleming I eds. AJCC Cancer Staging Manual. 6th ed. New York, NY : Springer ; Baldwin LM, Klabunde CN, Green P, Barlow W, Wright G. In search of the perfect comorbidity measure for use with administrative claims data: does it exist? Med Care ; 44 ( 8 ): Zhang J, Yu KF. What s the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA ; 280 ( 19 ): NIH Consensus Conference. Adjuvant therapy for patients with colon and rectal cancer. JAMA ; 264 ( 11 ): Baxter NN, Rothenberger DA, Morris AM, Bullard KM. Adjuvant radiation for rectal cancer: do we measure up to the standard of care? An epidemiologic analysis of trends over 25 years in the United States. Dis Colon Rectum ; 48 ( 1 ): Krook JE, Moertel CG, Gunderson LL, et al. Effective surgical adjuvant therapy for high-risk rectal carcinoma. N Engl J Med ; 324 ( 11 ): Martling A, Holm T, Johansson H, Rutqvist LE, Cedermark B. The Stockholm II trial on preoperative radiotherapy in rectal carcinoma: longterm follow-up of a population-based study. Cancer ; 92 ( 4 ): Jessup JM, Loda M, Bleday R. Clinical and molecular prognostic factors in sphincter-preserving surgery for rectal cancer. Semin Radiat Oncol ; 8 ( 1 ): Ayanian JZ, Zaslavsky AM, Fuchs CS, et al. Use of adjuvant chemotherapy and radiation therapy for colorectal cancer in a population-based cohort. J Clin Oncol ; 21 ( 7 ): Schrag D, Cramer LD, Bach PB, Begg CB. Age and adjuvant chemotherapy use after surgery for stage III colon cancer. J Natl Cancer Inst ; 93 ( 11 ): Mahoney T, Kuo YH, Topilow A, Davis JM. Stage III colon cancers: why adjuvant chemotherapy is not offered to elderly patients. Arch Surg ; 135 ( 2 ): Oliveria SA, Yood MU, Campbell UB, Yood SM, Stang P. Treatment and referral patterns for colorectal cancer. Med Care ; 42 ( 9 ): Carlos RC, Fendrick AM, Patterson SK, Bernstein SJ. Associations in breast and colon cancer screening behavior in women. Acad Radiol ; 12 ( 4 ): Carlos RC, Underwood W 3rd, Fendrick AM, Bernstein SJ. Behavioral associations between prostate and colon cancer screening. J Am Coll Surg ; 200 ( 2 ): Freeman HP. Patient navigation: a community based strategy to reduce cancer disparities. J Urban Health ; 83 ( 2 ): Dignam JJ, Colangelo L, Tian W, et al. Outcomes among African- Americans and Caucasians in colon cancer adjuvant therapy trials: findings from the National Surgical Adjuvant Breast and Bowel Project. J Natl Cancer Inst ; 91 ( 22 ): Dignam JJ, Ye Y, Colangelo L, et al. Prognosis after rectal cancer in blacks and whites participating in adjuvant therapy randomized trials. J Clin Oncol ; 21 ( 3 ): McCollum AD, Catalano PJ, Haller DG, et al. Outcomes and toxicity in African-American and Caucasian patients in a randomized adjuvant chemotherapy trial for colon cancer. J Natl Cancer Inst ; 94 ( 15 ): Cooper-Patrick L, Gallo JJ, Gonzales JJ, et al. Race, gender, and partnership in the patient-physician relationship. JAMA ; 282 ( 6 ): Cooper LA, Roter DL, Johnson RL, Ford DE, Steinwachs DM, Powe NR. Patient-centered communication, ratings of care, and concordance of patient and physician race. Ann Intern Med ; 139 ( 11 ): Baldwin LM, Dobie SA, Billingsley K, et al. Explaining black-white differences in receipt of recommended colon cancer treatment. J Natl Cancer Inst ; 97 ( 16 ): Dobie SA, Baldwin LM, Dominitz JA, Matthews B, Billingsley K, Barlow W. Completion of therapy by Medicare patients with stage III colon cancer. J Natl Cancer Inst ; 98 ( 9 ): Braveman PA, Cubbin C, Egerter S, et al. Socioeconomic status in health research: one size does not fit all. JAMA ; 294 ( 22 ): Cohen RA, Martinez ME. Health insurance coverage: early release of estimates from the National Health Interview Survey. Hyattsville, MD : National Center for Health Statistics ; Funding American Cancer Society, Atlanta, GA (Mentored Research Scholar Grant #MRSGT CHPHS); National Cancer Institute, National Institutes of Health, Bethesda, MD (R01CA089544). Note The views expressed in this article are the authors and do not necessarily represent the views of the American Cancer Society or the National Cancer Institute. The study sponsors had no role in the design of the study; the collection, analysis, or interpretation of the data; the writing of the manuscript; or the decision to submit the manuscript for publication. Manuscript received November 28, 2007 ; revised March 20, 2008 ; accepted April 7, Articles JNCI Vol. 100, Issue 10 May 21, 2008