|Home | About | Journals | Submit | Contact Us | Français|
Disparities in care have been documented for foreign-born cancer patients in the US. However, limited data are available on lung and colorectal cancer. We assessed whether patient-reported quality and receipt of recommended care differed between US-born and foreign-born cancer patients.
We collected surveys and medical records for a population-based cohort including white, Hispanic, and Asian adults (2,205 US-born and 890 foreign-born) with lung or colorectal cancer diagnosed in California during 2003–2005. We used logistic regression to assess the association of nativity with patient-reported quality of care and receipt of recommended treatments (adjuvant chemotherapy for stage III colon cancer, adjuvant chemotherapy and radiation for stage II/III rectal cancer, and curative surgery for stage I/II non-small cell lung cancer). We also assessed whether language explained any differences in care by nativity.
Overall, 46% of patients reported excellent care, but foreign-born patients were less likely than US-born patients to report excellent quality of care (adjusted odds ratio (AOR)=0.80, 95% confidence interval [CI]=0.65–1.00), a difference partly explained by language of survey, an indicator of English proficiency. Rates of recommended therapies ranged from 64% to 85%; foreign-born patients were less likely to receive chemotherapy and radiation for stage II/III rectal cancer (AOR=0.35, 95% CI=0.12–0.99). Rates of other treatments did not differ significantly by nativity.
Foreign-born cancer patients reported lower quality of care and were less likely to receive some cancer therapies than US-born. Better coordination of care and communication about cancer treatments and expanded use of interpreters may lessen these disparities.
Disparities in cancer diagnosis, treatment, and survival have been documented for immigrants in the United States (US). Foreign-born individuals are diagnosed with cancer at more advanced stages or with larger tumors(1–4), receive less definitive cancer treatments(1) and have worse survival once diagnosed with cancer(4–6) than individuals of similar race/ethnicity who were born in the US. Such disparities may be related to access to screening services(7;8) or care, which may be mediated in part by language differences and communication difficulties(9–11).
The extent of disparities in cancer outcomes by nativity remains poorly understood. Prior studies have focused on breast(1–4;6;12), colorectal(5), and gastric(13) cancer and have variably included Hispanic(1;2) or Asian(3–6;12;13) patients. Not all studies reported worse care for foreign-born patients; one reported earlier-stage diagnosis, more complete surgical treatment and improved survival of Asians and Pacific Islanders with gastric cancer, which was most pronounced among the foreign-born subgroup(13). Moreover, prior studies have typically used cancer registry data, which are limited by incomplete data on birthplace(14–18) and no information on factors related to acculturation. Few data are available on disparities by nativity in the treatment of lung and colorectal cancer, despite these cancers representing the two leading causes of cancer mortality in the US(13).
In 2007 approximately 38 million foreign-born individuals resided in the US, of whom 18 million were Hispanic and 9 million were Asian(19). To document whether US- and foreign-born patients with lung or colorectal cancer differ in their perceived quality of care and receipt of guideline-based treatments, we studied a large, population-based cohort of white, Hispanic, and Asian patients. Specifically, we assessed differences by nativity in patients’ ratings of the quality of their care and in receipt of recommended treatments. In addition, we assessed whether such differences were influenced by English-language proficiency.
Patients were recruited as part of a large national study of cancer care and outcomes undertaken by the Cancer Care Outcomes Research and Surveillance (CanCORS) Consortium (20;21). The study was approved by the human subjects committees at all participating institutions. Patients aged 21 and older diagnosed with invasive lung or colorectal cancer during September 2003 through December 2005 were identified within approximately three months of diagnosis using rapid case ascertainment from population-based cancer registries or pathology and cytology records. Patients were living in Alabama, Iowa, Los Angeles County, Northern California, and central North Carolina, or received their care in one of 5 large health maintenance organizations or at one of 15 medical centers in the Veterans Affairs Health Care system.
The eligible study sample consisted of 11,629 patients with invasive lung cancer and 10,239 patients with colorectal cancer. After notifying the patient's physician, 8,340 lung and 7,841 colorectal patients were contacted, and 5,150 lung and 4,911 colorectal patients or their proxy were interviewed, for a response rate of 51.0% and cooperation rate among contacted participants of 59.9%(22). An additional 140 lung and 208 colorectal patients were excluded because their cancers were recurrent or they were later deemed to be ineligible, resulting in a total of 9,713 patients (5,010 lung and 4,703 colorectal).
For this study, we included only patients enrolled from the Northern California (Alameda, Contra Costa, Sacramento, San Francisco, San Joaquin, San Mateo, Santa Clara, and Solano counties) and Los Angeles county study sites, which accounted for the vast majority foreign-born individuals in CanCORS. We excluded 425 persons who did not identify country of birth. We restricted the cohort to whites, Hispanics and Asians due to small numbers of foreign-born individuals of other races. The final study population consisted of 2,205 US-born and 890 foreign-born individuals of white, Hispanic, or Asian race/ethnicity. Asian patients were most often of Chinese (45%) or Filipino (24%) descent.
Patients or surrogates of patients who were deceased or too ill to participate were interviewed by telephone approximately 3–6 months after diagnosis. Recruitment materials and patient and surrogate surveys were translated (with back-translation) into Spanish and Chinese (Mandarin and Cantonese) and administered by bilingual interviewers for respondents who preferred these languages. The surveys (available at www.cancors.org/public) included questions about cancer treatments, ratings of care, health status, and sociodemographic characteristics.
Patients’ medical records (available for 75% of US-born and 61% of foreign-born patients) and registry data were used to obtain information about cancer site, stage at diagnosis, and treatment.
Patients were asked, “Overall, how would you rate the quality of your health care since your diagnosis of [lung or colorectal] cancer?” Potential responses included ‘excellent,’ ‘very good,’ ‘good,’ fair,’ ‘poor’. We dichotomized the responses as “excellent” versus other.
We assessed receipt of guideline-recommended cancer treatments(23–25) for which sample sizes were sufficient for statistical comparisons, including adjuvant chemotherapy for stage III colon cancer, adjuvant chemotherapy and radiation therapy for stage II/III rectal cancer, and curative surgery for stage I/II non-small cell lung cancer. We documented receipt of chemotherapy and radiation therapy if a comprehensive medical record review identified receipt within 6 months of diagnosis or if the patient or surrogate reported that they had received or were planning to receive chemotherapy or radiation therapy at the time of the baseline survey (studies suggest that patients’ self-reports of these treatments are highly accurate(26–30)). Among patients with medical record data, we identified receipt of surgery for stage I/II non-small cell lung cancer if the medical records documented pneumonectomy, lobectomy, or wedge resection within 6 months of diagnosis (patients were not asked about these specific procedures in the interview).
The main independent variables were nativity and race/ethnicity, categorized as US-born or foreign-born white, US-born or foreign-born Hispanic, and US-born or foreign-born Asian. Control variables included age at diagnosis, sex, marital status, education, household income, number of self-reported comorbid illnesses, self-reported health status, survey type, cancer type, and cancer stage. Among foreign-born individuals, we also documented language of survey (English versus non-English) which reflects English language proficiency and may be more relevant for clinical care than language spoken at home. Only 73 patients (2.3%) lacked health insurance; therefore this variable was not included in analyses. Variables were categorized as depicted in Table 1.
We compared unadjusted proportions of each dependent variable by nativity, stratified by race/ethnicity using the Cochran-Mantel-Haenszel test. Next, we used logistic regression models to examine the association of nativity and race/ethnicity with each dependent variable, adjusting for the independent variables listed above. In a second set of models, we included language to assess if this variable explained any associations between nativity and the dependent variables. Finally, we investigated interactions of nativity and race/ethnicity.
We conducted several sensitivity analyses. First, we repeated analyses of ratings of care combining “excellent” with “very good.” Second, because surrogates may rate care differently than patients and because they were surveyed when patients were deceased or very ill and these poorer outcomes could influence ratings, we repeated the analyses of ratings of care excluding surrogate interviews. Third, we repeated all analyses using a variable reflecting language spoken at home instead of language of the survey which may not accurately reflect language skills for patients whose surrogates completed the survey. Fourth, we assessed adjuvant chemotherapy and/or radiation therapy if the patient reported treatment was given but not if they reported it was planned. Finally, for assessing rectal cancer treatments, we ran models assessing receipt of chemotherapy or radiation versus neither treatment.
Item non-response was <2–3% for most variables. We used multiple imputation to impute missing survey data for items other than the dependent and independent variables of interest(31). Analyses were conducted using SAS statistical software, version 9.2.
The distribution of the patient characteristics stratified by race/ethnicity and nativity is presented in Table 1. Foreign-born Hispanic and Asian patients were younger than foreign-born whites and US-born groups. Asian patients were more often college graduates and Hispanic patients were more often in the low-income group. More US-born individuals completed the interview themselves. Many foreign-born Hispanic and Asian patients completed the survey in a language other than English.
Overall, 46% of patients rated their cancer care as excellent. In unadjusted analyses, the proportion rating quality of care as excellent varied between US-born and foreign-born patients within each race/ethnicity group (P=0.01). US-born patients more often reported excellent quality of care in all race/ethnicity groups compared with foreign-born patients (Table 1). In adjusted analyses, foreign-born patients were less likely than US-born patients to report excellent quality of care (adjusted odds ratio [OR]=0.80, 95% confidence interval [CI]=0.65–0.99, Table 2). Asian and Hispanic patients were also less likely than white patients to report excellent quality of care. When adjusting for language of survey (model II), the size of the effect of foreign birth on ratings decreased by about half (OR=0.89), and the association was no longer statistically significant. Language also explained some of the association of lower quality for Hispanic patients, but not for Asian patients. The interaction of nativity by race/ethnicity was not statistically significant (P=0.46 and 0.59 for Hispanic and Asian, respectively).
Overall, 77% of patients with stage III colon cancer received adjuvant chemotherapy. In unadjusted analyses, we found no differences in rates of adjuvant chemotherapy by nativity or race/ethnicity (P=0.79). Rates were highest among US-born Asian patients (92%) and lowest among foreign-born Asian patients (76%)(Table 1), but the numbers in these groups were small and this difference was not statistically significant (P=0.19). After adjustment, there was no statistically significant association between nativity or race/ethnicity and adjuvant chemotherapy (Table 3). Adjusting for language of survey did not change the results. The interaction of nativity by race/ethnicity was not statistically significant (P=0.71 and P=0.11 for Hispanic and Asian patients, respectively).
Overall, 64% of patients with stage II/III rectal cancer received adjuvant chemotherapy and radiation therapy. In unadjusted analyses, we observed no differences in rates of both adjuvant chemotherapy and radiation by nativity, stratified by race/ethnicity (P=0.61) (Table 1). In adjusted analyses, foreign-born patients had lower rates of adjuvant chemotherapy and radiation therapy (OR=0.35, 95% CI=0.12–0.99, Table 3). Rates of adjuvant therapy did not differ by race/ethnicity. After adjusting for language, the association of nativity and treatment decreased somewhat (OR=0.43) and was no longer statistically significant. Language was not statistically associated with receipt of adjuvant therapy, although sample sizes were small. The interaction of nativity by race/ethnicity was not statistically significant (P=0.24 for Hispanic and P=0.36 for Asian patients).
Overall, 85% of patients with stage I/II lung cancer underwent curative surgery. In unadjusted analyses, we observed no differences in rates of primary lung surgery by nativity or race/ethnicity (P=0.64, Table 1). After adjustment, we observed no association between nativity or race/ethnicity and receipt of surgery (Table 3). Similarly, adjusting for language of survey did not alter the results. The interaction of nativity by race/ethnicity was not statistically significant (P=0.66 for Hispanic and P=0.97 for Asian patients).
In sensitivity analyses, when combining excellent with very good in patients’ ratings of care, results were similar. In analyses of ratings of care excluding surrogate respondents; the beta coefficients were unchanged, but the effect of foreign-born status was no longer statistically significant due to the smaller sample size.
For all dependent variables, we repeated the second model including language spoken at home instead of the language of survey, and results were similar. We also conducted sensitivity analyses assessing adjuvant chemotherapy and/or radiation therapy where we did not consider the treatment to have been received if the patient reported the treatment was “planned” but not already given; results were similar. Finally, results of adjuvant therapy for stage II/III rectal cancer were similar when the dependent variable was dichotomized as either chemotherapy or radiation therapy versus none
In this population-based cohort of patients in California with lung or colorectal cancer, we observed lower patient ratings of cancer care among foreign-born patients relative to U.S.-born patients as well as among Hispanic and Asian individuals. These differences were partially explained by English language proficiency. We also observed lower rates of adjuvant chemotherapy and radiation therapy for stage II/III rectal cancer among foreign-born patients compared with US-born patients. We did not observe differences in rates of adjuvant chemotherapy for stage III colon cancer or curative surgery for stage I/II lung cancer. However, moderate-sized differences may have been missed due to relatively small sample sizes in these groups.
The Institute of Medicine has emphasized the importance of patient-centered care, declaring patient-centeredness to be one of six key aims of high-performing health care organizations(32). Previous research has found that racial or ethnic minorities with cancer and other conditions rate their care less favorably than white patients(33;34); this difference is particularly pronounced among non-English speaking patients(33–35). Asian and Hispanic patients have more problems than whites with coordination of care, access to care, and access to health and treatment information(35). The complexity of cancer care and the need for timely information about the disease and treatment options, which often require substantial coordination of care among multiple specialists, may be particularly challenging for foreign-born patients if they have limited social support to help with navigating the health care system. These challenges can be compounded by communication difficulties between non-English speaking patients and their clinicians, particularly if trained interpreters are unavailable and the clinicians do not speak the patients’ language.
Our finding of lower rates of adjuvant chemotherapy and radiation therapy for stage II/III rectal cancer among foreign-born patients is particularly notable because receipt of adjuvant therapies for rectal cancer typically requires coordination of multi-modality therapy by different providers, and sometimes requires neoadjuvant treatment. The lower rates of effective adjuvant therapy for rectal cancer by nativity is consistent with the findings of lower rates of guideline-recommended treatment for breast cancer among foreign-born Hispanic(1) and Asian(36;37) patients. Another small study also demonstrated less combined modality therapy for foreign-born Asian patients with lung cancer compared with non-Asian patients(38). The lack of differences by nativity in receipt of other treatments we studied may be because single modality therapy is less complex. One study found higher rates of complete surgical treatment of gastric cancer among foreign-born Asian and Pacific Islander patients than non-Hispanic white patients(13).
Issues of language and acculturation are important considerations in understanding differences in patients’ experiences with care. For newly diagnosed cancer patients who face a series of complex decisions regarding treatment, including trade-off between chances of cure, side effects and quality of life, good communication between the patient and the health care providers is crucial. Limited English proficiency may hamper communication as well as coordination and access to recommended care(33;39). Trained medical interpreters as well as patient navigator programs can improve care for patients with limited English proficiency(40), yet trained interpreters are available in person or by phone less than half of the times they are needed for patients with cancer, particularly in solo practices and single-specialty medical groups(41), and navigator programs are even less commonly available.
Language explained some of the lower ratings of care observed for Hispanic versus white patients, but not that for Asian versus white patients. Additional research is needed to understand what care beyond interpreter services is needed to improve experiences with care, particularly for Asian patients. While the observed findings point to language barriers as a prominent reason for disparities, the degree of adherence to native cultural beliefs and norms may also play a strong role, particularly in decisions about treatment options. In a prior study of providers who treat Asian patients with breast cancer, fear and cultural beliefs were among the most common reasons these patients chose mastectomy over breast-conserving surgery(42). Further research is needed to understand the role of acculturation, language, and cultural factors in treatment decision-making among foreign-born patients with cancer. Few patients in our study were recent immigrants, limiting our ability to assess the influence of years in the U.S. on treatment decisions.
Strengths of our study included a large population of foreign-born patients of different racial and ethnic groups surveyed in one of 3 languages and with survey data supplemented by registry and medical record data. In addition, we assessed patients’ ratings of quality of care as well as receipt of cancer treatment among patients with lung and colorectal cancer who were US-born and foreign-born.
Several limitations of our study should also be noted. First, despite the relatively large numbers of patients, we had limited power to detect differences of moderate size for some measures that still may be clinically important, as evidenced by the relatively wide confidence intervals for some treatments. For example, we had >80% power to detect an absolute difference of 12% in adjuvant chemotherapy for stage III colon cancer between US-born and foreign-born patients, and an absolute difference of 15% for curative surgery for stage I/II lung cancer between these groups. We had less statistical power to detect potential interactions of nativity by race/ethnicity. Thus, studies with larger sample sizes may be needed to detect smaller differences in receipt of recommended care by nativity. Second, patient-reported quality of care measured by a single item might be conceptually or psychometrically problematic if individuals in different race/ethnic groups or with different levels of acculturation do not ascribe the same meaning to the response categories. Some evidence from the Consumer Assessments of Healthcare Providers and Systems (CAHPS) survey suggest that Hispanic and Asian individuals report care experiences that are similar to or less positive than care for non-Hispanic whites, yet they have more positive ratings of care(43), and Hispanic patients are more likely to use the extremes of response scales(44). Our results were robust to alternative definitions of high-quality care, suggesting that differences in reporting the highest levels of quality are unlikely to explain our results; nevertheless, additional validation of single-item quality assessments in varied racial/ethnic groups is needed.
Third, we have no information about how non-respondent foreign-born patients may differ from respondents. Fourth, we studied patients in California, limiting the generalizability of the findings to immigrants in other parts of the US, where the characteristics of immigrant populations may differ (45). Moreover, we were unable to assess specific racial/ethnic subgroups that may differ in their culture, religion, beliefs, and practices concerning health; this may especially be true for Asians who constitute a heterogeneous group; however, nearly half of the Asians in our sample were of Chinese origin. Finally, foreign-born patients were less likely than others to consent to medical record abstraction. Nevertheless, we supplemented information from the medical records with self-reports of chemotherapy and radiation, which prior studies demonstrate are quite accurate (26–30). Since the analyses for lung cancer surgery required medical record data, foreign-born patients who did not consent to medical record abstraction were not included in this analysis.
In conclusion, foreign-born individuals with lung or colorectal cancer reported lower quality of care than US-born patients and were less likely to receive some complex cancer treatments that are recommended by clinical guidelines. These differences were at least partly explained by English language proficiency. Cancer experiences and outcomes may be improved with a greater emphasis on coordination of care and improving communication with foreign-born patients, including providing additional services for these patients, such as interpreters and patient navigators.
We would like to acknowledge the patients and surrogates who responded to the survey as well as Garrett Kirk for administrative support and Hocine Azeni for assistance with statistical analyses, and to the CanCORS Statistical Coordinating Center and David Nerenz for assistance with coding of primary cancer treatments based on the medical record data.
This work of the Cancer Care Outcomes Research and Surveillance (CanCORS) Consortium was supported by grants from the National Cancer Institute (NCI) to the Statistical Coordinating Center (U01 CA093344) and the NCI-supported Primary Data Collection and Research Centers (Dana Farber Cancer Institute/Cancer Research Network U01 CA093332, Harvard Medical School/Northern California Cancer Center U01 CA093324, RAND/UCLA U01 CA093348, University of Alabama at Birmingham U01 CA093329, University of Iowa U01 CA01013, University of North Carolina U01 CA093326) and by a Department of Veteran’s Affairs grant to the Durham VA Medical Center CRS 02-164.
There are no financial disclosures from any authors.