Incident Cardiovascular Disease Risk among Older Asian, Native Hawaiian and Pacific Islander Breast Cancer Survivors

This article is featured in Selected Articles from This Issue, p. 5

Approximately 58,000 cancer cases were diagnosed in the Asian, Native Hawaiian and Pacific Islander (ANHPI) population in the United States in 2016 (1). Breast cancer is the most common cancer diagnosed among ANHPI women, with 11,090 cases diagnosed in 2016. The ANHPI population is incredibly heterogenous and the Asian population in the United States includes individuals with origins in East Asia, Southeast Asia and the Indian subcontinent. The prevalence of cancer risk factors such as tobacco smoking, alcohol drinking and overweight/obesity are generally lower in the ANHPI population than in Non-Hispanic White (NHW; ref. 1). However, issues of concern for ANHPIs are that they are less likely to be diagnosed with cancer at localized stage for certain cancers compared with the NHW population, and they have lower cancer screening rates for cervical and colorectal cancer. They are also the only race and ethnicity group in the United States that experiences cancer as the leading cause of death instead of heart disease (2).

Cancer survivors may experience incident adverse health outcomes or late effects caused by the cancer treatment such as cardiotoxicity and cardiovascular disease (CVD; ref. 3). Cardiotoxicity among breast cancer survivors is associated with anthracycline and HER2 inhibitor treatment, and radiation therapy. (4) There are very few large-scale cancer survivorship studies including ANHPIs, and previous studies have focused on important areas such as the impact of acculturation on quality of life (5–7), unmet needs of Asian cancer survivors (8, 9), and survival or cause of death as outcomes (10–13). Large-scale studies of breast cancer survivors, none of which focused on ANHPI cancer survivors, have been conflicting, with some reporting increased CVD risks (14), modestly increased CVD risks (15) or no CVD risk (16–18). In a SEER analysis, the risks of cardiovascular mortality were lower overall for ANHPI patients with breast cancer compared with NHW patients with breast cancer, but for Native Hawaiian patients with breast cancer, the cardiovascular mortality risk was higher than NHW patients with breast cancer (11). When comparing U.S.-born ANHPI patients with breast cancer to non–U.S.-born ANHPI patients with breast cancer, increased cardiovascular mortality risks were identified in the U.S.-born ANHPI patients. (11) There is a need to study the incidence instead of mortality due to CVD among ANHPI breast cancer survivors, and to study them by specific race and ethnicity groups among ANHPIs.

In terms of cancer treatment adherence among ANHPI breast cancer survivors, some studies reported no difference in adherence to treatment for breast cancer among ANHPI patients (19, 20), while one study reported higher rates of adherence to chemotherapy for ANHPI patients with breast cancer which was attributed to advanced cancer stage and high grade at diagnosis. (21) When comparing breast conserving surgery versus mastectomy, ANHPI patients with breast cancer were less likely to receive breast conserving surgery, regardless of education level of the patient or tumor size (22). However, they were more likely to initiate adjuvant endocrine therapy compared with NHW women (OR = 1.28; 95% CI, 1.03–1.58) in a SEER-Medicare database study (23). These treatment differences for the ANHPI population may also further contribute to disparities in late effects following breast cancer diagnosis and treatment among ANHPI cancer survivors. The aim of this study is to investigate incident CVD outcomes among older ANHPI breast cancer survivors in the SEER-Medicare data in comparison to older NHW breast cancer survivors, and to compare risks amongst specific ANHPI race and ethnicity groups.

We used the SEER-Medicare linked database from 2000 through 2020 to estimate the risk of CVD among breast cancer survivors with a population-based cohort study design. The SEER-Medicare data, consisting of two large population-based sources SEER and Medicare, contains information on demographics, cancer-related characteristics, and healthcare claims among Medicare beneficiaries with cancer. International Classification of Disease (ICD) codes are available in the Medicare claims files including claims from hospitalizations (Part A), physician/supplier bills (Part B), and institutional outpatient providers. The study was approved by the Institutional Review Board at the University of Utah and received approval for waiver of consent. Data from this study are available upon application to SEER-Medicare and IRB approval.

We included women who: (i) were age ≥66 years, (ii) were diagnosed with a first primary invasive breast cancer between 2000 and 2017, (iii) had continuous Medicare Parts A and B throughout the follow-up period, and (iv) did not have health maintenance organization enrollment (Supplementary Fig. S1). This study focused on patients with cancer who were ANHPI or NHW. We excluded women who had less than 1 year of follow-up or were diagnosed with cancer based on autopsy or death certificate. A breast cancer diagnosis was classified according to the SEER Site code 26000, which uses the International Classification of Diseases for Oncology, Version 3 (ICD-O-3 code: C50), excluding mesotheliomas, Kaposi's sarcoma, and lymphomas. To calculate baseline comorbidity scores, we started the eligibility at the age of 66 years old and required enrollment in Part A/B to be continuous a year before cancer diagnosis. Each ANHPI cancer patient was matched to up to three NHW patients with cancer by exact year of diagnosis and age at the time of cancer diagnosis.

Events of CVD were identified using codes from ICD version 9 and 10. ICD diagnosis codes were processed with the Chronic Conditions Data Warehouse (CCW) categorizations for acute myocardial infarction, heart failure, ischemic heart disease and stroke/transient ischemic attack (Chronic Conditions Warehouse. 27 CCW Chronic Conditions Algorithms: MBSF_CC_YYYY File. REVISED 02/2022; Chronic Conditions Warehouse. 30 CCW Chronic Conditions Algorithms: MBSF_CHRONIC_YYYY File. REVISED 02/2022). We also created a composite CVD outcome variable including all 4 of these diseases. Individuals diagnosed with the CVD diagnosis before the start of each analysis time period (1 year after cancer diagnosis) were considered as prevalent cases and were excluded for that specific outcome (Supplementary Table S1). Physicians may have recorded a “rule-out” diagnosis or unconfirmed diagnoses in physician and outpatient claims. To avoid overestimation of the outcomes, conditions ascertained by physicians and outpatient claims were required to occur more than once between 30 and 60 days.

Information on age at cancer diagnosis, year of cancer diagnosis, tumor characteristics, first-course cancer treatment (chemotherapy, hormone-based therapy, radiation, surgery), rural residence, census tract-level socioeconomic status (SES) index, and vital status was obtained from the SEER cancer file. We used the Rural–Urban Continuum codes (RUCC) to define rural counties for the county of residence at cancer diagnosis (https://www.ers.usda.gov/data-products/rural-urban-continuum-codes.aspx). The Yost index, a measure of SES, is calculated on the basis of a factor analysis of seven variables including average educational level, median income, poverty rate, median housing value, median rent, unemployment rate, and occupation (% working class; ref. 24). All the patients with breast cancer in our database were on Medicare but we excluded patients on supplemental insurance due to the lack of claims data on the portion of services covered by supplemental insurance. The baseline Charlson Comorbidity Index (CCI) was calculated for the 1-year period prior to the cancer diagnosis. (25) The ICD and CPT codes for tobacco use and obesity were also from the CCW (Chronic Conditions Warehouse. Other Chronic Health, Mental Health, and Potentially Disabling Chronic Conditions Algorithms: MBSF_OTCC_{YYYY} File. REVISED 02/2022).

Demographic and clinical characteristics between ANHPI and NHW breast cancer survivors were compared using Pearson χ² tests. We also compared the demographic characteristics of ANHPI and NHW patients with breast cancer, dropping the New York, Massachusetts and Idaho registries because patients from these registries did not have cancer treatment information to contribute to some of the analyses. Patients were scored as censored if they died or did not develop the CVD outcomes by the last follow-up date. We used two follow-up periods with a separate model for each, to evaluate the long-term risk of CVD, >1 to 5 years and >5 years after a cancer diagnosis. Cox proportional hazards models stratified on matched pairs were used to estimate HRs and 95% confidence intervals (CI) comparing ANHPI to NHW patients with cancer for each CVD outcome and for the risk of death with adjustment for potential confounders, including cancer registry, baseline CCI, residence (rural/urban), and SES. To explore differences among the ANHPI race/ethnicity groups, we chose Japanese patients with breast cancer as the reference group because they were the largest ANHPI group in the data. Risk factors for heart failure were also explored for ANHPI patients with breast cancer. Potential confounding factors were selected a priori based on the three properties of confounders determined by a causal model, a directed acyclic graph (DAG). Specifically, baseline CCI, rural residence, income, education and cancer registry region are risk factors for CVD (property 1), and are associated with race and ethnicity (property 2). While cancer registry region is not a mediator (property 3), the other covariates may be on the path for the association between race and ethnicity and CVD among patients with breast cancer. We were interested in the risk through alternate pathways; thus we adjusted for these factors. Baseline CCI included diagnosis of myocardial infarction, coronary heart disease, cerebrovascular accidents and transient ischemic attack before cancer diagnosis. Because we excluded prevalent events, when we estimate the risk of stroke as an example, patients who had prevalent stroke from the CCI are excluded but the model would adjust for the other previous CVD with the baseline CCI variable. The proportional hazards assumption was tested for each Cox proportional hazards model by including interactions between the predictors and time in the model. For models where the proportional hazards assumption was violated, we used flexible parametric survival models with restricted cubic splines and compared the resulting estimates to those from the Cox proportional hazards model (26, 27). Analyses were performed using SAS software (version 9.4, SAS Institute, Cary, NC).

The datasets used to conduct this study are available upon approval of a research protocol from the NCI. Instructions for obtaining these data are available at https://healthcaredelivery.cancer.gov/seermedicare/obtain/

A total of 7,122 ANHPI breast cancer survivors and 21,365 NHW patients with breast cancer were identified in the SEER-Medicare database (Table 1). The majority of ANHPI patients with breast cancer were from the Hawaii, Greater California, Los Angeles, San Francisco and New York cancer registries. In terms of the specific race/ethnicity groups, the largest proportion of ANHPI breast cancer survivors were Japanese, Chinese, and Filipino. For census-tract income level measures, ANHPI breast cancer survivors had a higher proportion in the highest income category, but a lower proportion of highest education group than NHW patients with breast cancer. The highest SES category based on the Yost index had a higher proportion of ANHPI breast cancer survivors than that of NHW breast cancer survivors (31.1% vs. 26.4%), although this variable had a fairly high proportion of missing values. A baseline CCI score of 0 was less common among ANHPI breast cancer survivors (65.1% vs. 56.8%; P value for χ² < 0.001). Obesity and tobacco use disorders were less common among ANHPI patients with breast cancer compared with NHW patients with breast cancer. When we compared all ANHPI and NHW patients with breast cancer to the patient groups dropping the New York, Massachusetts and Idaho registries, we did not observe any differences in the demographic characteristics proportions nor with the χ² statistics.

Demographics by specific race/ethnicity groups are shown in Supplementary Fig. S2 and Supplementary Table S2. The highest proportion in the highest SES groups based on the Yost index, education and income were for Asian Indian and Pakistani patients with breast cancer. Lower proportions in the high SES groups were observed for Vietnamese, Native Hawaiian and other Pacific Islander patients with breast cancer. Filipino and Other Pacific Islander patients with breast cancer had the highest proportion of 2+ comorbidities. The proportion of patients with breast cancer with obesity and tobacco use were very low for most ANHPI groups compared with NHW patients with breast cancer, with the exception of other Pacific Islander and Native Hawaiian patients with breast cancer.

Compared with NHW patients with breast cancer, ANHPI patients with breast cancer were diagnosed with localized cancer at a slightly higher proportion, had a higher proportion of invasive ductal carcinoma, and a higher proportion of HER2-positive tumors (Table 2). Conversely, ANHPI patients with breast cancer had a lower proportion of ER/PR positive tumors than NHW patients with breast cancer. ANHPI patients with breast cancer underwent total mastectomies at a higher rate than partial mastectomies and had less radiation treatment compared with NHW patients with breast cancer. Partial mastectomies appeared to be low in particular for Filipino, Korean, and Vietnamese women (Supplementary Fig. S3; Supplementary Table S3). There were no differences in chemotherapy treatment comparing ANHPI and NHW patients with breast cancer.

Supplementary Figure S4 shows the incidence rates of the four CVD outcomes, ordered from higher to lower incidence. Japanese, Chinese, and Korean breast cancer survivors tended to have lower incidence of most of the CVD outcomes than the other ANHPI race/ethnicity groups. The only exception was Chinese patients with breast cancer who had a higher incidence of stroke than most ANHPI race/ethnicity groups at 10 per 1000 person-years. The other Pacific Islander, Asian Indian and Pakistani, and Native Hawaiian patients with breast cancer had the highest incidence of acute myocardial infarction and heart failure.

Compared with NHW patients with breast cancer, overall ANHPI, Japanese, Chinese and other Asian patients with breast cancer had a lower risk of heart failure and ischemic heart disease (Table 3). The CVD event numbers by outcome and race and ethnicity are shown in Supplementary Table S4. The proportion of ANHPI patients with breast cancer diagnosed with the 4 CVD outcomes and composite CVD within 5 years of cancer diagnosis is also shown by age and race and ethnicity (Supplementary Table S5). However, CVD outcomes were not different between NHW patients with breast cancer and Filipino, Asian Indian and Pakistani, Korean, or Vietnamese patients with breast cancer. As a sensitivity analysis, we additionally adjusted for tobacco use disorders and obesity, but the inferences did not change for the CVD outcomes. We also conducted a sensitivity analysis excluding the Japanese patients with breast cancer to assure that the results were not dependent on one ANHPI group (Supplementary Table S6); the protective HRs for heart failure and ischemic heart disease for ANHPI compared with NHW patients with breast cancer were still observed.

When Japanese patients with breast cancer were taken as the reference group, we observed increased risks of CVD for NHW, Filipino, Asian Indian and Pakistani and Native Hawaiian patients with breast cancer >1 to 5 years. For specific CVDs, higher risks for heart failure and ischemic heart disease were observed for NHW and Filipino patients with breast cancer in both follow-up periods (Table 4). For heart failure risk >1 to 5 years after cancer diagnosis, increased risks were observed for Asian Indian and Pakistani, Native Hawaiian and Vietnamese patients with breast cancer compared with Japanese patients with breast cancer. For ischemic heart disease risk >1 to 5 years after cancer diagnosis, we observed increased risks for Asian Indian and Pakistani, and other Pacific Islander patients with breast cancer compared with Japanese patients with breast cancer. Some of these risks persisted into the >5-year follow-up period. Vietnamese patients with breast cancer had an almost 8-fold increase in the risk of acute myocardial infarction compared with Japanese patients with breast cancer.

Risk factors identified for heart failure among ANHPI breast cancer survivors included older age, higher CCI score, distant cancer stage, and chemotherapy treatment. Protective factors included higher income, higher SES with the Yost index, and breast cancer surgery (Table 5). Histology, HER2 status, ER status, PR status, and triple-negative breast cancer were not risk factors for heart failure among the ANHPI breast cancer survivors in this data. ANHPI breast cancer survivors who were diagnosed at 81 to 100+ years of age had a 3-fold increase in risk of heart failure compared with ANHPI breast cancer survivors diagnosed at 66 to 70 years of age. ANHPI breast cancer survivors diagnosed with distant cancer stage and having 2 or more comorbidities at baseline had a 2.5-fold increase in risk of heart failure.

ANHPI patients with breast cancer had an overall lower risk of death compared with NHW patients with breast cancer (Table 6). When separated out by ANHPI race/ethnicity groups, compared with NHW patients with breast cancer, Japanese, Chinese, Other Asian, and Korean patients with breast cancer had a lower risk of death. However, Native Hawaiian patients with breast cancer had a higher risk of death (HR = 1.33; 95% CI, 1.11–1.59) compared with NHW patients with breast cancer. When Japanese patients with breast cancer were taken as the reference group, Filipino, Asian Indian and Pakistani, Native Hawaiian, Vietnamese and Other Pacific Islander patients with breast cancer had a higher risk of death. Heart failure diagnoses were associated with the risk of death among ANHPI patients with breast cancer overall, and specifically among Japanese, Filipino, Other Asian, Asian Indian and Pakistani, and Korean patients with breast cancer (Supplementary Table S7).

While ANHPI race/ethnicity groups have been analyzed as a group in previous analyses of cancer survivorship, ANHPI are a heterogenous group in relation to CVD risk factors and SES. The risk of incident CVD diagnosis was lower among ANHPI breast cancer survivors overall, and specifically among Japanese, Chinese, and other Asian breast cancer survivors, when compared with NHW breast cancer survivors. However, comparing amongst the specific ANHPI race/ethnicity groups, we observed heterogeneity in the incidence rates of CVD. Compared with Japanese breast cancer survivors, Filipino, Asian Indian and Pakistani, and Native Hawaiian breast cancer survivors had higher risks of both heart failure and ischemic heart disease. Risk factors for heart failure included older age, higher CCI score, distant stage and chemotherapy treatment among ANHPI breast cancer survivors.

Although obesity and tobacco use disorders before cancer diagnosis were lower in prevalence among ANHPI breast cancer survivors compared with NHW breast cancer survivors, heterogeneity was observed amongst the ANHPI race and ethnicity groups. Native Hawaiian and other Pacific Islander patients with breast cancer had lower proportions of patients in the highest SES group, and higher proportion of obesity and tobacco use disorders than the other ANHPI patients with breast cancer. Pacific Islander patients with breast cancer also had a lower proportion of local stage and higher proportions receiving chemotherapy compared with Japanese breast cancer survivors. These risk factors may be correlated with the higher risks of heart failure and ischemic heart disease for Native Hawaiian and Pacific Islander breast cancer survivors compared with Japanese breast cancer survivors.

Filipino breast cancer survivors also had high risks of heart failure and ischemic heart disease than Japanese breast cancer survivors, along with higher proportions of 2+ comorbidities and obesity, but lower proportions of smoking. In terms of clinical factors, Filipino patients with breast cancer had a lower proportion of local stage, lower proportion with partial mastectomy, and higher proportion having received chemotherapy, which may contribute to their higher risks of heart failure and ischemic heart disease.

Asian Indian and Pakistani patients with breast cancer had the highest proportions in the highest education, income and SES categories. Obesity was fairly high and the proportion receiving chemotherapy was higher among Asian Indian and Pakistani patients with breast cancer. Obesity and chemotherapy are risk factors for both heart failure and ischemic heart disease and may play a role in the higher risks observed among Asian Indian and Pakistani breast cancer survivors compared with Japanese breast cancer survivors. Previous studies have shown that South Asians are a higher risk of CVD, with higher genetic susceptibility to CVDs. (28, 29) Unfortunately, we did not have data on genetic susceptibility to CVD risk.

Vietnamese breast cancer survivors had lower proportions in the highest education, income and SES categories as well as lower proportions diagnosed at local stage and higher proportion receiving chemotherapy compared with both NHW and Japanese breast cancer survivors. The 2+ comorbidities was higher but obesity and smoking were lower for Vietnamese breast cancer survivors than Japanese breast cancer survivors. Higher risks of heart failure and acute myocardial infarction were observed among Vietnamese breast cancer survivors relative to Japanese breast cancer survivors. The SES difference suggest that potential differences in healthcare access may contribute to higher risks of heart failure and acute myocardial infarction.

The risk of death among ANHPI patients with breast cancer were lower compared with NHW patients with breast cancer, but amongst the ANHPI race and ethnicity groups, we observed heterogeneity in risks of death. The higher risk of death for Native Hawaiian patients with breast cancer compared with NHW patients with breast cancer may be due to higher obesity, lower SES, higher baseline comorbidities, tobacco, although we did not observe higher risks of CVD for Native Hawaiians compared with NHW patients with breast cancer. The higher risk of deaths for Filipino, Asian Indian and Pakistani, Native Hawaiian, Vietnamese and Other Pacific Islander patients with breast cancer compared with Japanese patients with breast cancer may partly be due to incident heart failure risks.

Strengths of the study include minimal survival bias due to the longitudinal capture of the CVD outcomes in claims data. Because we do not rely on the patient to recall their disease diagnosis, the study is not subject to recall errors. This is the largest cohort study focusing on older ANHPI patients with breast cancer, to our knowledge. We were able to include 1,683 Japanese, 1,404 Chinese and 1,382 Filipino breast cancer survivors who were older and on Medicare insurance. The large sample size allowed us to analyze specific ANHPI race/ethnicity groups, which has been very limited in previous studies. We were also able to stratify on follow-up time and investigate longer-term adverse health outcomes among older Asian breast cancer survivors in the >5 years after cancer diagnosis time period.

Limitations of the study include the restriction to patients with cancer 66 years of age and older due to the use of the SEER-Medicare data. The results are not generalizable to younger ANHPI patients with breast cancer, because they may have different characteristics and generational differences in terms of smoking and obesity rates, SES, cancer treatment distributions, as well as acculturation. The results are also not generalizable to ANHPI patients with breast cancer who are on supplemental insurance plans, because we restricted to our eligibility to patients who were on Medicare only, to assure that events can be captured in the Medicare claims. Even within Medicare claims, it is possible to have missed claims or inaccurate ICD coding. However, in our previous studies, we validated identification of CVD outcomes in claims data by comparing them to self reported CVD (30–34). Finally, we are unable to take into account factors such as physical activity, which cannot be captured even with a proxy as self-report would be needed. We added a smoking proxy variable with the tobacco use disorder and obesity as a variable with use of CPT codes. We expect that these proxy variables undercount smoking and obesity and would not capture any overweight cancer survivors. We also cannot evaluate whether there were differences in missing for smoking and obesity. However, we were able to show that obesity was a risk factor for heart failure among ANHPI breast cancer survivors. Some of the registries did not have cancer treatment information and were not included in the analysis of CVD risk factors. We compared the demographic characteristics of all patients with cancer to the patient group dropping these registries and did not observe a difference. There could still be differences in cancer treatment patterns by registry that may contribute further to the CVD risk factor analysis. Finally, we do not have clinical values for risk factors such as blood pressure or cholesterol, which would have added important information to our study.

In conclusion, this is the first in-depth study of four major CVD outcomes among ANHPI breast cancer survivors with risk estimation for specific ANHPI race and ethnicity groups in the SEER Medicare data, to our knowledge. While ANHPI breast cancer survivors as a whole appear to have lower risks of CVD outcomes than NHW breast cancer survivors, the lower risks were observed only in specific ANHPI groups. In addition, heterogeneity was observed in CVD risks, as expected, due to the heterogenous nature of ANHPI race and ethnicity groups. Our results support the expected heterogeneity in baseline and clinical characteristics, and adverse health outcomes among cancer survivors in specific ANHPI race and ethnicity groups. Our findings support that subgroups of ANHPI breast cancer survivors may face greater risks for development of CVD outcomes, which can guide targeted prevention, education, and clinical management of ANHPI breast cancer survivors. Further research is needed to elucidate the disparities experienced in CVD risks among cancer survivors in specific ANHPI race and ethnicity groups.

M. Hashibe reports grants from NCI during the conduct of the study. M. Wei reports personal fees from Gilead Science, Cardinal Health, Aptitude Health, Curio Science, Binaytara Foundation; and personal fees from OncLive outside the submitted work. R. Tao reports personal fees from Binaytara Foundation; and personal fees from Helsinn outside the submitted work. A. Daud reports personal fees from AbbVie outside the submitted work. D. Tay reports an award from Flatiron Health for research that is unrelated to this study. The award is to the University of Utah and not to the PI. C.E. Chang reports grants from NCI during the conduct of the study; grants from NCI outside the submitted work. No disclosures were reported by the other authors.

The interpretation and reporting of these data are the sole responsibility of the authors. The ideas and opinions expressed herein are those of the author(s) and endorsement by the State of California Department of Public Health, the NCI, and the Centers for Disease Control and Prevention or their Contractors and Subcontractors is not intended nor should be inferred.

M. Hashibe: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. M. Wei: Supervision, investigation, methodology, writing–review and editing. C.J. Lee: Supervision, investigation, methodology, writing–review and editing. R. Tao: Supervision, investigation, methodology, writing–review and editing. A. Koric: Investigation, writing–review and editing. J. Wang: Supervision, writing–review and editing. A. Daud: Supervision, methodology, writing–review and editing. D. Tay: Investigation, writing–review and editing. J. Shen: Investigation. Y.C.A. Lee: investigation, writing–review and editing. C.-P.E. Chang: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing.

Research reported in this publication was supported by the NCI of the NIH under award number R01 CA244326 (M. Hashibe). This study used the linked SEER-Medicare database. The authors acknowledge the efforts of the NCI; the Office of Research, Development and Information, CMS; Information Management Services (IMS), Inc.; and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database. The collection of cancer incidence data used in this study was supported by the California Department of Public Health as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885; the NCI's Surveillance, Epidemiology and End Results Program under contract HHSN261201000140C awarded to the Cancer Prevention Institute of California, contract HHSN261201000035C awarded to the University of Southern California, and contract HHSN261201000034C awarded to the Public Health Institute; and the Centers for Disease Control and Prevention's National Program of Cancer Registries, under agreement # U58DP003862–01 awarded to the California Department of Public Health. The authors acknowledge the efforts of the NCI; the Office of Research, Development and Information, CMS; Information Management Services (IMS), Inc.; and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries in the creation of the SEER-Medicare database.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.