- © 2010 by American Society of Clinical Oncology
Delays in Adjuvant Chemotherapy Treatment Among Patients With Breast Cancer Are More Likely in African American and Hispanic Populations: A National Cohort Study 2004-2006
- From Health Services Research, American Cancer Society, Atlanta, GA; American College of Surgeons, Chicago, IL; and Roswell Park Cancer Institute, University of Buffalo, Buffalo, NY.
- Corresponding author: Stacey A. Fedewa, MPH, Department of Surveillance and Health Policy, American Cancer Society, 250 Williams St, Atlanta, GA 30303; e-mail: Stacey.Fedewa{at}cancer.org.
Abstract
Purpose Previous studies have indicated poorer survival among women receiving adjuvant chemotherapy > 90 days after surgery compared with women receiving adjuvant chemotherapy within 90 days of surgery.
Patients and Methods Women diagnosed between 2004 and 2006 with invasive breast cancer (stages I to III) and treated with surgery and adjuvant chemotherapy were selected from the National Cancer Database (n = 107,587). We evaluated factors associated with prolonged time to start adjuvant chemotherapy (≥ 60 and ≥ 90 days after surgical resection) using multivariable log binomial models to estimate risk ratios (RRs) and 95% CIs.
Results The average time to adjuvant chemotherapy was 41.46 days (± 24.46 days). Overall, 85.2% and 95.8% of women received adjuvant chemotherapy within 60 and 90 days of surgery, respectively. African American and Hispanic patients had higher risk of 60-day delay (RR, 1.36; 95% CI, 1.30 to 1.41 and RR, 1.31; 95% CI, 1.23 to 1.39, respectively) and 90-day delay (RR, 1.56; 95% CI, 1.44 to 1.69 and RR, 1.41; 95% CI, 1.26 to 1.59, respectively) compared with white patients. Insurance type, stage, comorbidity, and facility type were also associated with adjuvant chemotherapy delay.
Conclusion The majority of women in our study received adjuvant chemotherapy within the time frame (90 days) for which there is no evidence of poorer outcome. However, the rate of delay varied by patient and by clinical and facility factors. Future studies on the role of structural, physician, clinical, and patient factors in adjuvant chemotherapy delay in populations of women with higher rates of delay and potential interventions are needed.
INTRODUCTION
The American Cancer Society (ACS) estimates that 192,370 women will be diagnosed with breast cancer and 40,610 deaths due to breast cancer will occur in 2009.1 Adjuvant chemotherapy has improved survival among patients with breast cancer.2,3 Clinical trials demonstrating this benefit generally required chemotherapy to be administered within 7 to 8 weeks (49 to 56 days) of definitive surgery. In practice, not all patients initiate treatment in this time frame. Several studies have examined the impact of early and late initiation of adjuvant chemotherapy on survival.2,4–6 The Danish Breast Cancer Consortium Group (DBCCG) did not observe improved survival among patients with early initiation of adjuvant chemotherapy among women who began chemotherapy at any point within 90 days of surgery.2 However, several studies demonstrate that substantial delays in chemotherapy initiation reduce survival. One previous study observed lower survival among the 5% of women whose therapy was initiated more than 90 days after surgery.7 A second study of elderly women showed worse overall and cause-specific survival among patients whose adjuvant chemotherapy was initiated more than 90 days after surgery.5 Based on these data, the National Quality Forum endorsed initiation of chemotherapy within 120 days of diagnosis as a quality measure for women with hormone receptor–negative breast cancer, similar to a time frame of 90 days from surgery based on an average 30 days from initial diagnosis to completing surgery.8
Studies examining the time from symptoms to diagnosis and diagnosis to initial treatment among patients with breast cancer indicate that African American patients are more likely to experience diagnostic and treatment delays, which are also associated with worse survival outcomes.9–11 The association between race/ethnicity and adjuvant chemotherapy delay has not been as widely examined and, to our knowledge, only one study has analyzed racial disparities in delays of adjuvant chemotherapy.5 Hershman et al5 did not observe a statistically significant difference in adjuvant chemotherapy delay between African American and white patients; however, this study included only Medicare patients who were age ≥ 65 years and did not consider other racial/ethnic groups. Our study extends previous findings by examining adjuvant chemotherapy among a national sample of women age 18 to 99 years, with diverse insurance status and racial/ethnic status.
PATIENTS AND METHODS
Data Set
Data from the National Cancer Data Base (NCDB), a hospital-based cancer registry jointly sponsored by ACS and the American College of Surgeons, were used in this study. The NCDB includes approximately 70% of all malignant cancers in the United States from more than 1,400 facilities accredited by the American College of Surgeons' Commission on Cancer (CoC) that collect and submit data to NCDB.12 The NCDB contains standardized data elements on patient demographics, insurance status, tumor characteristics, first course of treatment, census-level socioeconomic factors, and facility-level factors. Data are entered using the standardized data dictionary of the NCDB—the Facility Operations and Data Standards (FORDS) Manual.13 The NCDB uses an algorithm to identify multiple patient reports for the same cancers. A data quality and completeness score is computed for duplicate records, and only the record with the most complete data is retained in the analytic database. Because no patient, provider, or hospital identifiers were examined in this study and no protected health information was reviewed, institutional review board approval was not required for this study.
Study Population and Patient Selection
Women diagnosed with their first primary invasive stage I, II, or III breast cancer between January 1, 2004, and December 31, 2006, and received all or part of their first treatment at a CoC-accredited facility were selected from the NCDB (N = 337,021).
Our analysis was restricted to women age 18 to 99 years who had a surgical resection and adjuvant chemotherapy for initial treatment (Fig 1). Patients were excluded who did not receive surgery (n = 8,536) or chemotherapy (n = 181,891). Patients who had missing information on surgery or chemotherapy date (n = 12,989), who initiated treatment > 365 days following surgery, who had radiation before or on the same day as chemotherapy, and who received neoadjuvant chemotherapy were also excluded (n = 22,140). The final analytic cohort contained 107,587 patients.
Derivation of the study population, National Cancer Data Base 2003 to 2006.
We examined two levels of adjuvant chemotherapy delay (60- and 90-day delay). The time to adjuvant chemotherapy was defined by the days from the most definitive resection of the primary site to the first administration of chemotherapy. The definitive surgical procedure at the primary site captured the most invasive surgical procedure at the primary site and included excisional biopsy, lumpectomy, and mastectomy. The date of first chemotherapy was defined from one of two elements collected in the NCDB: first date of chemotherapy or first date of systemic therapy. We selected time from definitive surgery to initiating chemotherapy rather than time from diagnosis because this is the measure used in previous studies5 and because the latter is a composite measure reflecting both time from diagnosis to surgery and from surgery to chemotherapy.
Patient, area, and facility-level characteristics were included in our analyses. Patient race was categorized as non-Hispanic white, African American, Hispanic, Asian, and other/unknown. Primary payer and/or insurance type were grouped into the following categories: Medicaid, Medicare (including Medicare alone and with supplement), uninsured (which includes not insured, charity write-off, and self-pay), private insurance plans (health maintenance organizations, preferred provider organizations), and other/unknown. Because Medicare is available to essentially all US residents age 65 years and older, but only for permanently disabled individuals younger than age 65 years, the Medicare category was dichotomized for the analyses into Medicare for patients age 18 to 64 years and Medicare for patients age 65 to 99 years. Geographic region was based on the reported state of residence at diagnosis as classified in the United States Census Report,14 and American Joint Commission on Cancer (AJCC) Cancer stage (6th edition) was included in the analyses along with hormone receptor status.15 The NCDB calculates the Charlson-Deyo comorbidity index on the basis of the registry reporting of up to six preexisting comorbidities (excluding the patients' cancer diagnoses) as recorded in the hospital discharge summary, which was based on the International Classification of Disease, 9th Revision, Clinical Modification (ICD-9-CM).16,17 Year and age of diagnosis were also included in the analysis. An area-based indicator of socioeconomic status—the proportion of the population that did not have a high school diploma—was derived at the ZIP Code level from 2000 US Census data and was included as quartiles of the observed distribution in the general US population.
Facility-level characteristics include the volume of patients with breast cancer and treatment facility type. Volume categories were created by dividing the number of patients with breast cancer treated by an institution into equal-size hospital-specific tertiles. Low-, medium-, and high-volume facilities treated 1 to 137,138 to 288, and 289 to 1,789 patients throughout the study period, respectively. Four types of treatment facilities were included in the classification scheme used by the accreditation program of the CoC: community cancer, comprehensive community cancer, teaching or research, and National Cancer Institute (NCI) –designated comprehensive cancer centers. Community centers treat at least 300 cancer patients a year and have a full range of services for cancer care, but patients need a referral for portions of their treatment. Comprehensive community cancer centers offer the same range of services as the community hospitals but treat at least 650 annual cancer patients and conduct weekly cancer conferences. Teaching/research facilities have residency programs and ongoing cancer research.
Chemotherapy data collected by both hospital and central cancer registries are known to be variably incomplete.18–20 Although the proportion of women coded as “missing/unknown” for chemotherapy treatment in the NCDB was low, it is possible that some women coded as having “no chemotherapy” actually received chemotherapy that was not ascertained by the registry. To estimate the completeness of chemotherapy data, we examined reported chemotherapy use among a subset of women among the NCDB cohort (pre-exclusion) that should clearly receive chemotherapy after surgery—those age 18 to 70 years with hormone receptor–negative cancers with positive lymph nodes or negative nodes and tumor > 1 cm in size. There were 33,412 (9.91%) of the 337,021 patients in the cohort that met these criteria.
To evaluate the representativeness of the NCDB in relative-to-population–based registries, we compared selected characteristics of patients with breast cancer reported to the NCDB (pre-exclusion) with those of patients reported to the Surveillance, Epidemiology, and End Results 17 (SEER17) population-based registries.21 Comparisons were restricted to variables that were available and interpretable for both registries.
Statistical Analyses
Analyses were performed with SAS software version 9.2 (SAS Statistical Institute, Cary, NC). χ2 tests were employed to analyze the relationship between race and all other covariates. Additionally, χ2 tests (α = .05) were used to examine associations between each of the aforementioned independent variables, with the two outcome variables—60-and 90-day delay. Because of the common outcome in our study (> 10%), multivariable log binomial models were used to generate risk ratio and 95% CI estimates for 60- and 90-day delays.22 Two-way interactions between race and all remaining covariates were tested by comparing −2 log likelihood χ2 values from chunk tests.23
RESULTS
The average time from primary surgery to adjuvant chemotherapy was 41.14 days (± 24.46 days). In total, 15,965 (14.84%) and 4,521 (4.20%) patients experienced 60-and 90-day delays, respectively (Fig 2). The proportion of patients experiencing 90-day delay ranged from 3.59% in white patients to 6.78% and 6.91% in African American and Hispanic patients, respectively.
Number of days from primary surgery to chemotherapy, National Cancer Data Base 2004 to 2006. Data for patients receiving chemotherapy > 120 days from primary surgery date not shown. Q, quartile.
Table 1 displays patient, area-level, and facility characteristics by race/ethnicity. African American and Hispanic patients were disproportionately uninsured and insured by Medicaid and reside in lower socioeconomic status (SES) ZIP Codes compared with white patients. On average, white patients tended to be older than nonwhite patients. Higher Charlson-Deyo comorbidity score and hormone-negative tumors were more common among African American patients compared with all other race groups A higher percentage of African American patients were treated at teaching/research facilities compared with white patients. There were no major differences in stage of diagnosis and facility volume across the five race/ethnicity categories.
Patient Characteristics by Race, NCDB, 2004-2006 (n = 107,587)
Table 2 displays adjusted risk ratios for 60- and 90-day delays by patient demographics, clinical factors, area-level SES, and facility-level characteristics. Hispanic and African American patients had a 31% and 36% increased risk of 60-day delay compared with white patients, respectively. In the 90-day delay analysis, Hispanic and African American patients had a 41% and 56% increased risk of delay compared with white patients, respectively. Uninsured, Medicaid and younger Medicare patients had a higher risk of 60- and 90-day delay compared with privately insured patients. Age was also a significant predictor of both measures of delay: older women were more likely to experience delay compared with younger women. Stage I and II patients had a 27% and 21% increased risk of 60-day delay compared with stage III patients, respectively. However, there were no major differences across stage groups in the 90-day delay analysis. The risk of delay also varied by facility characteristics: women treated at low- and medium-volume facilities and women treated at teaching/research and NCI cancer centers were at an increased risk of 60- and 90-day delays.
Multivariate Regression Analysis of Factors Related to 60- and 90-Day Chemotherapy Delay, Patients With Breast Cancer, NCDB, 2004-2006 (n = 107,587)
Of the 33,412 patients with hormone receptor–negative cancers examined to estimate the completeness of NCDB chemotherapy data, 25,921 (77.6%) patients received adjuvant chemotherapy. Among stage III patients, 4,475 (84.5%) of 5,296 received chemotherapy. There were documented reasons for nonadministration of chemotherapy for 5,504 (73.5%) of the 7,491 patients who did receive it (“chemotherapy not recommended” for 4,527 patients and “refused by patient” for 977 patients). The reason was not recorded/unknown for 1,871 patients. Therefore, overall chemotherapy was administered or a specific reason for nonadministration was provided for 94.4% of patients. In a multivariate analysis within this subcohort, treatment corresponded to known patterns of care with patients who were diagnosed at lower stages, were treated at community cancer centers, and were older, thus being less likely to receive adjuvant chemotherapy. Race and insurance were associated with receipt of adjuvant chemotherapy but not to the extent of stage, age, and facility type.
Table 3 displays patient-level characteristics of our pre-exclusion cohort (NCDB) and SEER17. Approximately 72% and 74% were coded as non-Hispanic white in the NCDB and SEER17, respectively. The age distribution of SEER patients and NCDB patients were similar: approximately half the women were younger than age 60 years. The stage and hormone receptor status distribution between the two data sources were also similar: 50% of patients were diagnosed with stage I disease and 74% of women had hormone-positive tumors.
Comparison of Clinical and Demographic Characteristics of Patients With Breast Cancer Reported to SEER17 and NCDB, 2004-2006
DISCUSSION
This study reports the factors associated with adjuvant chemotherapy delay in a large national cohort from the United States comprising nonelderly and elderly patients. The majority (95.8%) of women in our study received adjuvant chemotherapy within the 90-day time frame defined as appropriate and for which there is no evidence of degradation of outcome. A previous study of SEER-Medicare patients5 reported 60- and 90-day delay rates 1% to 2% lower than those in our study, and they did not observe an association between race/ethnicity and delay. Differences in the findings of the two studies are likely due to our inclusion of nonelderly patients with greater variation in insurance type.
The increased risk of delay in adjuvant chemotherapy among Hispanics and African Americans observed in this study may be attributable to lack of access to care, geographic distance to the treatment facility, and availability of transportation.24–28 Difficulty navigating the health system and language barriers may also contribute to the disparities in treatment delay.29,30 Insurance was also an important independent factor in predicting 60- and 90-day delay with uninsured, Medicaid insured, and younger Medicare patients experiencing higher rates of 60- and 90-day delay compared with privately insured patients. Cost of care and difficulty obtaining timely care are contributing factors among these patient groups. Previous studies report an increased likelihood of failure to appear at appointments and delayed appointments among Medicaid and uninsured patients.26
Patients age 69 years and older were much more likely to experience 60- and 90-day delays independent of other demographic, clinical, area, and facility factors, possibly related to higher rates of postoperative complications and comorbidities. Although we adjusted for comorbidities in multivariate modeling, the Charlson-Deyo index is limited to 17 specific comorbid conditions and may not have fully accounted for conditions that delay adjuvant chemotherapy. Additionally, comorbidity could be underestimated in cancer registries; however, a recent study of colorectal cancer from the NCDB showed that the NCDB-estimated comorbidity burden was similar to comorbidity derived from administrative data sets.31
Our finding that earlier stage of diagnosis was associated with an increased risk of 60-day delay is consistent with previous research5,10,32 and may be due to greater emphasis on timely therapy for patients with higher probability of recurrence. Delays in adjuvant chemotherapy were observed among patients treated at teaching/research facilities, NCI comprehensive cancer centers, and comprehensive community cancer centers when compared with community cancer centers. This may reflect a higher proportion of patients receiving referral to these comprehensive centers after surgery elsewhere, with the referral adding to delay. The lower rates of delay among high-volume facilities could be attributed to improved capacity to meet the demands of patients needing adjuvant chemotherapy.
This study did not examine the type, duration, or dose of chemotherapy because these factors are not captured in cancer registries. Other studies33 have demonstrated disparities related to SES in the appropriate dose and dose reductions of chemotherapy among patients with breast cancer, which may negatively impact outcomes. Although the NCDB is not population-based and does not contain data for patients treated at non-CoC–accredited facilities, comparison of patients show that the clinical and demographic characteristics of patients in NCDB and the SEER17 registries are similar with the exception that SEER contains a higher percentage of Hispanic and Asian patients while the NCDB had a higher fraction classified as “unknown.”
Another limitation of this study is the restriction of the study cohort to patients who were documented in registry records as having had adjuvant chemotherapy. Concerns have been raised over the completeness and quality of treatment data in cancer registries, particularly for treatments administered in outpatient settings.34–36 Although only a small proportion (0.22%) of patients had missing/unknown chemotherapy data, it is likely that some individuals coded as having no chemotherapy actually had chemotherapy that was not documented in registry records. If there is significant misclassification that varies by race/ethnicity or insurance status and/or if patients with longer delays were less likely to have their treatment documented, the results of our study may be biased. Although it was not possible to evaluate these issues directly, several observations suggest that it is not likely to be a serious problem. Our analytic cohort (postexclusion) had distributions of race, geographic region, comorbidity, and facility characteristics similar to those of the pre-exclusion cohort, with the exception of fewer older patients for whom chemotherapy is not recommended (Tables 1 and 3). An analysis of chemotherapy rates in a subgroup of women for whom there is a high level of consensus that chemotherapy should be administered found that rates of chemotherapy were high (> 75%), and there was a documented reason for omission of chemotherapy in all but 5.6% of those women. The rates of chemotherapy among NCDB patients age 65 and older were similar to those in a published SEER-Medicare study.37 Among SEER-Medicare patients, 37.5% of patients with stage II and IIA node-positive breast cancer age 65 years and older received adjuvant chemotherapy compared with 40.70% of NCDB patients.37 Finally, a multivariate analysis of factors associated with receipt of chemotherapy in the pre-exclusion cohort found the expected strong associations with age and stage, but no or slight associations with race/ethnicity and being uninsured or Medicaid-insured (data available on request).
In conclusion, the majority of women who received adjuvant chemotherapy initiated treatment within the time frame (90 days) defined as appropriate and for which there is no evidence of degradation of outcome. However, African American and Hispanic patients were more likely to experience both 60- and 90-day delays. Tracking and patient navigation programs should be implemented to minimize such treatment delays.
AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
The author(s) indicated no potential conflicts of interest.
AUTHOR CONTRIBUTIONS
Conception and design: Stacey A. Fedewa, Elizabeth M. Ward, Andrew K. Stewart, Stephen B. Edge
Administrative support: Stacey A. Fedewa
Collection and assembly of data: Andrew K. Stewart
Data analysis and interpretation: Stacey A. Fedewa, Elizabeth M. Ward, Andrew K. Stewart, Stephen B. Edge
Manuscript writing: Stacey A. Fedewa, Elizabeth M. Ward, Andrew K. Stewart, Stephen B. Edge
Final approval of manuscript: Stacey A. Fedewa, Elizabeth M. Ward, Andrew K. Stewart, Stephen B. Edge
Footnotes
-
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
- Received December 17, 2009.
- Accepted June 25, 2010.
