Patients' Perceptions of Quality of Care for Colorectal Cancer by Race, Ethnicity, and Language
http://www.100md.com
▲還散笫雖悝◎
the Division of General Medicine and Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School
Department of Health Care Policy, Harvard Medical School
Division of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA
Center on Outcomes, Research and Education Evanston Northwestern Healthcare Research Institute, Evanston, IL
Cancer Surveillance Program, Sacramento
Public Health Institute, Berkeley
Northern California Cancer Center, Union City
Cancer Surveillance Section, California Department of Health Services, Sacramento, CA
ABSTRACT
PURPOSE: To identify opportunities for improving care, we evaluated patients' perceptions of the quality of their cancer care by race, ethnicity, and language.
PATIENTS AND METHODS: We surveyed a population-based cohort of 1,067 patients with colorectal cancer in northern California approximately 9 months after diagnosis. Adjusting for clinical and demographic factors, mean problem scores were analyzed on a 100-point scale for six domains of care.
RESULTS: Mean problem scores were highest for health information (47.8), followed by treatment information (32.3), psychosocial care (31.7), coordination of care (21.3), confidence in providers (13.1), and access to cancer care (12.7). In adjusted comparisons with white patients, African American patients reported more problems with coordination of care (difference, 9.8; P < .001), psychosocial care (difference, 7.2; P = .03), access to care (difference, 6.6; P = .03), and health information (difference, 12.5; P < .001). Asian/Pacific Islander patients reported more problems than did white patients with coordination of care (difference, 13.2; P < .001), access to care (difference, 15.5; P < .001), and health information (difference, 12.6; P = .004). Hispanic patients tended to report more problems with coordination of care (difference, 4.4; P = .06), access to care (difference, 5.8; P = .08), and treatment information (difference, 7.0; P = .06). Non每English-speaking white patients reported more problems than other white patients with coordination of care (difference, 21.9; P < .001), psychosocial care (difference, 16.1; P = .009), access to care (difference, 19.8; P = .003), and treatment information (difference, 17.8; P = .002). Non每English-speaking Hispanic patients reported more problems than other Hispanic patients with confidence in providers (difference, 16.9; P = .01).
CONCLUSION: Efforts to improve patients' experiences with cancer care should address disparities by race, ethnicity, and language, particularly in coordination of care, access to care, and the provision of relevant information.
INTRODUCTION
Patients' experiences with health care providers are a central component of their quality of care, complementing more technical aspects of care such as the appropriate use of medications and procedures.1 During the last 15 years, surveys have been developed to assess patients' experiences with care while hospitalized or enrolled in managed care plans.2-5 These experiences are particularly important for patients with cancer, because they commonly require access to multiple specialists, effective coordination of care, accurate information about the disease and treatment options, and timely attention to symptoms. Both the National Cancer Policy Board of the Institute of Medicine and the President's Cancer Panel have drawn attention to perceived deficiencies in patients' experiences with cancer care,6,7 but few studies have analyzed this topic in broadly representative cohorts of patients with cancer.8
Therefore, we surveyed a racially and ethnically diverse cohort of more than 1,000 patients diagnosed with colorectal cancer in northern California. Our primary objective was to evaluate whether patients' reported problems with care varied by race and ethnicity because prior studies demonstrated racial and ethnic differences in cancer treatment and outcomes.9,10 We also assessed variations in patients' problems with care by language, income, education, sex, age, and health status, and we analyzed how patients' experiences with care were related to their overall ratings of the quality of cancer care.
PATIENTS AND METHODS
Study Population
For patients diagnosed with cancer in California, hospitals, radiation treatment facilities, and physicians are required by state law to report their patients' cancer diagnosis, demographic characteristics, and first course of treatment to the California Cancer Registry, the largest population-based cancer registry in the world.11 Eligible patients for this study included adults age 40 to 84 years who were newly diagnosed with invasive colorectal cancer from April 1999 through April 2000 in nine counties of northern California (Alameda, Contra Costa, Placer, Sacramento, San Francisco, San Joaquin, San Mateo, Santa Clara, and Solano) served by the San Francisco/Oakland, San Jose/Monterey, and Sacramento regional registries.
Regional registry staff identified patients within 4 months of diagnosis through regular visits to hospitals and pathology laboratories. Patients were excluded if they were deceased, did not speak English or Spanish, had physical or mental impairments reported by their physicians or family members that prevented them from completing a telephone or mail survey, or their hospital lacked valid address or telephone information for them. The California Cancer Registry provided information on patients' age, sex, race and ethnicity, tumor stage,12 and receipt of surgery as obtained by the regional registries from medical records.
The study protocol was approved by institutional review boards of the California Department of Health Services, Public Health Institute, Northern California Cancer Center, and Harvard Medical School, in accordance with assurances filed with and approved by the US Department of Health and Human Services.
Survey Instrument
The survey instrument was designed to assess patients' experiences with cancer care, health-related quality of life, comorbid illnesses, and sociodemographic characteristics. Questions regarding patients' experiences with cancer care were obtained from a survey designed by the Picker Institute (Boston, MA).2-4 After adapting the self-administered Picker Institute cancer survey for telephone interviews and pilot testing it among 50 patients with colorectal cancer in northern California, we selected 34 questions to include in our instrument. Thirty-three questions addressed problems with specific aspects of care (see Appendix), and one question elicited patients' overall rating of care on a five-level scale ranging from poor to excellent. Patients were considered to have problems with care if they gave suboptimal responses, consistent with prior analyses of Picker Institute survey questions.2-4
To assess health-related quality of life of patients with colorectal cancer, we used the previously validated Functional Assessment of Cancer Therapy每Colorectal (FACT-C) survey.14 This 34-item instrument evaluates patients' emotional, social, physical, and functional well-being, and colorectal cancer每specific issues. Scores for the physical, functional, and colorectal subscales can be summed to create a Trial Outcomes Index. Higher scores on the FACT-C indicate better health-related quality of life.
Patients reported their demographic characteristics (age, sex, marital status, race, Hispanic ethnicity, language spoken at home, education, income) and whether they had received surgery, chemotherapy, or radiation therapy for their colorectal cancer. Patients also reported whether they had been diagnosed with 14 comorbid conditions; prior research has shown that patients are generally accurate in reporting conditions noted in their medical records.15 The full survey was translated into Spanish and back-translated into English to ensure accuracy.
Data Collection
The treating physician listed in the hospital record for each patient was notified by mail about the study. Patients then received a letter from the regional registry describing the study and offering a $20 stipend for completing the survey. Trained interviewers from the Public Health Institute attempted to contact patients up to 15 times during 1 month to conduct the survey in English or Spanish via computer-assisted telephone interviewing. Patients who did not respond and those with impaired hearing were mailed a self-administered version of the survey.
Statistical Analysis
To group questions into domains of care, we performed a principal factor analysis with promax (oblique) rotation. Examination of eigenvalues suggested solutions with four to eight factors. We reviewed these solutions in light of the substantive item content and groupings derived from patient focus groups conducted by the Picker Institute. We combined individual questions into six domains of care that had moderate to high internal consistency, including coordination of care (eight items; Cronbach's coefficient = .77), confidence in providers (four items; = .65), psychosocial care (five items; = .64), access to cancer care (three items; = .55), treatment information (four items; = .65), and health information (seven items; = .82). Average problem scores in specific domains were calculated as the percentage of answered questions for which patients reported problems with care. Two items on the management of pain and nausea were analyzed individually.
We compared survey respondents and nonrespondents using the 2 test for categorical variables (race/ethnicity, sex, stage, receipt of surgery) and Student's t test for age. Respondents' race and ethnicity were based on their self-reported data. For nonrespondents, race and ethnicity were determined from reporting facilities and a Hispanic surname algorithm used by the California Cancer Registry to improve the identification of Hispanic patients.13
We compared unadjusted problem scores in each domain and overall ratings of cancer care by patients' demographic characteristics. Among patients with stage III colon cancer or stage II or III rectal cancer, for whom adjuvant chemotherapy is recommended by national guidelines,16 we evaluated the association of problem scores with use of chemotherapy. To assess how problems with specific aspects of care were related to overall ratings of care, we calculated Pearson correlation coefficients of the six domain scores with the single question, "Overall, how would you rate the quality of your health care since your diagnosis of cancer?" We also stratified these correlations by race and ethnicity.
We analyzed multivariable linear regression models to assess the independent predictors of patients experiencing problems with care in each domain, and analyzed logistic regression models to assess predictors of patients rating their overall cancer care as excellent or very good. These models controlled for sociodemographic characteristics, cancer stage, five comorbid illnesses reported by at least 15% of survey respondents (hypertension, coronary heart disease, diabetes mellitus, chronic lung disease, and depression), and FACT-C scores (emotional well-being, social/family well-being, and the Trial Outcomes Index). We also used generalized estimating equations to adjust for clustering of patients by hospitals where they initiated treatment. Variables for non每English-speaking white, Asian/Pacific Islander, and Hispanic patients were added to assess language as a potential mediator of observed differences in care. In a secondary analysis, we added variables for the proportions of patients at each hospital who were African American, Hispanic, and Asian/Pacific Islander to assess whether patient-level effects of race and ethnicity on problem scores were confounded by hospital-level compositional effects; we also tested interactions of patients' race/ethnicity with the racial and ethnic composition of the hospitals. All analyses were conducted with SAS statistical software, version 8.2 (SAS Institute, Cary, NC) and two-tailed P values.
RESULTS
Patient Characteristics
Among 2,139 patients with colorectal cancer diagnosed during the study period, we excluded 648 patients, including 159 who had died, 139 who did not speak English or Spanish, 29 with physical or mental impairments that precluded interviewing as reported by their physicians or family members, 273 with no valid address or telephone number in hospital records, 27 who had recurrent colorectal cancer or another type of cancer, and 21 who had only carcinoma-in-situ. Of the remaining 1,491 eligible patients, 259 refused to participate, 57 were too ill to complete the survey, 18 had impaired hearing, and 78 did not respond by phone or mail.
Surveys were completed by 1,079 patients (72.4%), including 1,002 telephone interviews and 77 mail surveys. Respondents were more likely than nonrespondents to be white (74.9% v 68.1%) or African American (8.0% v 7.1%), and less likely than nonrespondents to be Hispanic (9.7% v 10.8%) or Asian/Pacific Islander (7.5% v 14.0%; P = .001). Respondents were also somewhat younger (mean, 66.2 v 69.4 years; P < .001), more likely to be women (50.4% v 40.8%; P = .001), less likely to have metastatic cancer (10.1% v 15.8%; P = .002), and more likely to have undergone colorectal surgery (98.8% v 95.2%; P < .001). The median time from diagnosis to completing the survey was 8.9 months (interquartile range, 7.7 to 10.5 months). Because our study focused on racial and ethnic differences in care, we excluded five patients who reported their race as "other," five patients who provided minimal data on their experiences with care, and two patients whose age was beyond our designated range of 40 to 84 years.
The characteristics of the study cohort of 1,067 patients stratified by race and ethnicity are shown in Table 1. Seventy-five respondents (7.0%) reported speaking a primary language other than English at home, including 32 Hispanic patients, 28 Asian/Pacific Islander patients, and 15 white patients. About half of the study cohort (48.0%) had received chemotherapy, and 17.6% had received radiation therapy. The 1,067 study participants were initially treated in 75 hospitals (median, 12 patients; interquartile range, five to 21 patients).
Patients' Reported Problems With Care
Mean problem scores varied widely across the six domains of care (Table 2). The highest problem scores were reported for health information (47.8), followed by treatment information (32.3), psychosocial care (31.7), and coordination of care (21.3). Lower problem scores were reported for confidence in providers (13.1) and access to cancer care (12.7). On individual questions, for example, half of the patients reported receiving insufficient health information about changes in their work or usual activities (52.7%), relationship with their spouse or partner (50.7%), or sexual activity (49.6%). Other common problems included not being given enough information about cancer treatments (29.3%), not having providers work well together (25.1%), and not being involved in decisions as much as desired (24.7%). Smaller proportions of patients reported they did not receive sufficient pain control (12.8%), were discouraged from seeking needed cancer care (12.1%), believed their doctors did not know enough about cancer treatments (11.5%), or were not treated with respect and dignity (9.7%).
Unadjusted mean problem scores on the six domains of care differed significantly by race and ethnicity (Table 3), with more problems reported by Asian/Pacific Islander, Hispanic, and African American patients relative to white patients, and by patients who did not speak English at home relative to those who did. Asian/Pacific Islander patients also reported more problems with pain control (30.0%; P < .001) than white (10.5%), African American (14.1%), or Hispanic patients (15.5%). White patients who did not speak English at home had mean problem scores that were substantially higher than those of English-speaking white patients on each domain (all P < .07). Hispanic patients who did not speak English at home reported more problems than other Hispanic patients with coordination of care (P = .04), confidence in providers (P = .004), and treatment information (P = .07). In contrast, mean problem scores did not differ significantly by language among Asian/Pacific Islander patients (all P > .42).
Younger patients reported significantly more problems with confidence in providers and health information than older patients (Table 3). Patients who had not graduated from high school and those with low incomes reported significantly more problems than more educated patients and more affluent patients, respectively, in each domain except health information. Unadjusted problem scores did not differ significantly by sex or marital status, except married patients reported more problems with health information.
Among 352 patients who were potentially eligible for adjuvant chemotherapy, chemotherapy recipients tended to report lower mean problem scores than nonrecipients with access to cancer care (11.0 v 16.7; P = .07) and confidence in providers (12.9 v 18.4; P = .08). They did not differ significantly, however, on other domains of patients' experiences (P > .20).
In multivariable analyses, significant adjusted differences in total problem scores were evident by race and ethnicity in four of six domains (Fig 1). Relative to white patients, African American patients reported significantly more problems than white patients with coordination of care (difference, 9.8; P < .001), psychosocial care (difference, 7.2; P = .03), access to cancer care (difference, 6.6; P = .03), and health information (difference, 12.5; P < .001). Similarly, Asian/Pacific Islander patients reported more problems than white patients with coordination of care (difference, 13.2; P < .001), access to cancer care (difference, 15.5; P < .001), and health information (difference, 12.6; P = .004). Hispanic patients tended to report more problems with coordination of care (difference, 4.4; P = .06), access to cancer care (difference, 5.8; P = .08), and treatment information (difference, 7.0; P = .06).
In a secondary analysis, the racial and ethnic composition of hospitals where patients were initially treated was not associated with problems in any domain, and differences in problem scores by patients' race and ethnicity remained significant and similar in magnitude. For several domains of problems, however, interactions of the patient-level and hospital-level variables were statistically significant. Smaller differences in problem scores for coordination of care, confidence in providers, access to care, and treatment information between white and African American patients were observed at hospitals with larger proportions of African Americans (all P < .05). The difference in problem scores for psychosocial care was similarly reduced for Asian/Pacific Islander patients treated in hospitals with a larger proportion of Asian/Pacific Islanders (P = .007). In each case, white patients' problem scores had little association with hospitals' racial composition, but problem scores for the nonwhite group were significantly lower at hospitals where that group constituted a larger proportion of the patients.
Several other demographic characteristics were associated with higher adjusted problem scores in specific domains. Patients with an annual income below $25,000 reported more problems than those with higher incomes in access to cancer care (difference, 4.4; P = .04). Married patients reported more problems than unmarried patients with health information (difference, 6.1; P = .006). Women reported higher problem scores than men for confidence in providers (difference, 2.2; P = .04), as did college graduates relative to high school graduates (difference, 3.6; P = .009). Age was not associated with adjusted problem scores in any domain.
Adding language spoken at home to these models, non每English-speaking white patients reported more problems than English-speaking white patients with coordination of care (difference, 21.9; P < .001), psychosocial care (difference, 16.1; P = .009), access to cancer care (difference, 19.8; P = .003), and treatment information (difference, 17.8; P = .002). Non每English-speaking Hispanic patients reported more problems than English-speaking Hispanic patients with confidence in providers (difference, 16.9; P = .01). Among Asian/Pacific Islander patients, adjusted problem scores did not differ significantly by language on any domain.
Worse emotional well-being and social/family well-being as measured by the FACT-C instrument were associated with significantly higher adjusted problem scores on each of the six domains of patients' experiences with care (Table 4). Worse physical, functional, and disease-specific well-being on the Trial Outcomes Index were associated with higher adjusted problem scores for coordination of care, confidence in providers, and health information. Patients with stage III tumors reported lower adjusted problem scores for treatment information than those with stage I tumors (difference, 5.8; P = .01), but stage was not associated with adjusted problem scores for other domains. Patients with chronic lung disease reported higher adjusted problem scores for health information than other patients (difference, 6.2; P = .005), but other comorbid conditions were not clearly associated with differences in problem scores.
Overall Ratings of Care
The proportions of patients rating their overall quality of cancer care as very good or excellent were significantly lower for nonwhite and non每English-speaking patients (Fig 2). Relative to white patients in an adjusted analysis, Asian/Pacific Islander patients (odds ratio [OR], 0.3; 95% CI, 0.2 to 0.6) and Hispanic patients (OR, 0.6; 95% CI, 0.4 to 1.0) remained less likely to rate their overall cancer care as very good or excellent. Non每English-speaking white patients were also less likely than English-speaking white patients to rate their care as very good or excellent (adjusted OR, 0.1; 95% CI, 0.04 to 0.5).
In the full cohort, patients' overall ratings of care had a strong negative correlation with their problem scores for each domain (all P < .001), including coordination of care (r = 每0.58), confidence in providers (r = 每0.46), treatment information (r = 每0.38), psychosocial care (r = 每0.33), access to cancer care (r = 每0.32), and health information (r = 每0.32). Problem scores for coordination of care were substantially higher among those rating their overall cancer care as poor (60.4), fair (60.6), good (41.2), or very good (21.5) than among those rating their care as excellent (9.4; P < .001). Problems with coordination of care also had the strongest correlation with overall ratings of care within each racial and ethnic group (white, 每0.57; African American, 每0.56; Hispanic, 每0.45; Asian/Pacific Islander, 每0.64; all P < .001).
DISCUSSION
Cancer care is often complex, involving numerous specialists and different types of curative, life-sustaining, and palliative therapies during the course of patients' illness. This complexity makes the interpersonal aspects of cancer care particularly important if patients are to achieve optimal clinical outcomes and quality of life. African American, Hispanic, Asian/Pacific Islander, and non每English-speaking white patients were significantly more likely than English-speaking white patients to report problems in multiple domains of care for which patients are the best or only source of information. Problems with coordination of care were most strongly correlated with worse ratings of the overall quality of cancer care.
How can clinicians and policy makers respond to the results of this study? Our findings suggest that standardized surveys of patients' experiences with cancer care could be used regularly to assess potential disparities by race, ethnicity, and language. In the United Kingdom, for example, a nationwide survey of experiences of 65,000 patients with cancer care was initiated by the National Health Service in 2000, including comparative assessments by white, African American, and South Asian patients.17 Such survey data can be used by health care organizations and providers to identify opportunities for improving their patients' experience with care.18
While striving to enhance the health care experiences of all patients with cancer, clinicians should devote special attention to providing effective communication, access to needed services, and coordination of care for nonwhite and non每English-speaking patients. More frequent problems with care reported by non每English-speaking patients indicate that language barriers may be an important contributor to disparities in care for cancer and other conditions.19,20 National standards proposed by the federal Office of Minority Health emphasize the importance of bilingual physicians, nurses, and professional interpreters to promote effective communication with patients who have limited or no English-language proficiency.21
The strong correlation of problems in coordination of care with lower overall ratings of the quality of care suggests that new initiatives to enhance coordination of care may have the greatest impact on improving patients' overall satisfaction with care and eliminating racial and ethnic disparities in this domain. Effective approaches for improving coordination may include multidisciplinary team meetings to address clinical and supportive needs of specific patients, navigators to help patients keep their medical appointments and obtain needed information, and computerized systems to aid clinicians in tracking consultations, test results, and treatments.22
Problems with interpersonal aspects of care for nonwhite and non每English-speaking patients may contribute to differences in cancer treatment and outcomes by race, ethnicity, and language.23 In our study, for example, patients who were eligible for adjuvant chemotherapy but did not receive it tended to report more problems with access to cancer care and confidence in providers, suggesting that technical quality of care is correlated with patients' experiences in these two domains. Future research should explore how patients' problems with care influence their outcomes and evaluate interventions to promote optimal technical and interpersonal care.
Another notable finding was the strong association between worse health-related quality of life, especially worse emotional and social well-being, and higher problem scores on all domains of care, consistent with prior research for other conditions.24 Emotionally and socially isolated patients may have greater unmet needs and worse experiences with care, or their worse health-related quality of life may result in part from unmeasured differences in disease severity and progression. Although these two plausible explanations cannot be distinguished in our cross-sectional study, regular assessments of health-related quality of life in patients with cancer may help to identify those at increased risk for problems with care.
Prior studies have assessed racial and ethnic differences in the interpersonal aspects of medical care, primarily through surveys of patients in primary care practices or health maintenance organizations.25 As in our study, Asian/Pacific Islander patients have reported more problems than white patients with access to care, communication with physicians, and coordination of care.26-30 Asian/Pacific Islander patients also have reported less satisfaction with care in most26-28,30 but not all of these studies.29 African American and Hispanic patients have reported worse access to needed specialty care than white patients in managed care plans.30 Relative to white patients, however, Hispanic and African American patients generally have reported similar satisfaction and ratings of communication with their physicians,26-30 although Hispanic patients who speak only Spanish have reported less satisfaction with care than those who speak English.31 One qualitative study found that problems with care were particularly concentrated among chronically ill African American patients with low incomes.32
A major strength of our study was the population-based sample, providing a highly representative and diverse cohort that was not limited to selected hospitals or health care systems. Questions related to quality of care were based on prior research that determined items most salient to patients,2-4 an approach that has been endorsed by a recent review of studies measuring interpersonal aspects of cancer care.8 Our survey was primarily administered by telephone, so biases related to racial and ethnic differences in literacy were minimized.
This study also had several potential limitations. Our response rate of 72.9% was consistent with other population-based surveys of chronically ill patients, but we may have underestimated problems with care because non每English-speaking patients who were either ineligible or nonrespondents to our survey were probably more likely to experience such problems. We studied patients in urban and suburban counties of northern California, so our findings might not generalize to other regions or rural areas. Patients who died within 9 months of their cancer diagnosis were ineligible for this study, so future research should assess the health care experiences of early decedents whose needs are primarily palliative in nature.
Remedying problems with interpersonal aspects of cancer care may be an important step toward eliminating racial and ethnic disparities in cancer treatment and outcomes. Although the marginal benefit of such efforts to improve cancer care may be greatest for nonwhite and non每English-speaking patients who reported the most problems in our study, these efforts have the potential to benefit patients of all racial and ethnic backgrounds.
Acknowledgment
We thank Gretchen Agha, Marti Felter, and Mary Anderson for coordinating patient enrollment; Bonnie Davis and the California Department of Health Services survey research unit for conducting telephone interviews; Mark Allen, Craig Grilley, Scott Riddle, and Robert Wolf for programming assistance; Paul D. Cleary and Neeraj Arora for helpful comments on the manuscript; National Research Corporation for authorizing use of the Picker cancer survey; and David Cella for authorizing use of the Functional Assessment of Cancer Therapy每Colorectal survey.
NOTES
Supported by Grants No. R01 HS09869 and U01 CA93324 from the Agency for Healthcare Research and Quality and the National Cancer Institute.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Eisenberg L: Good technical outcome, poor service experience: A verdict on contemporary medical care? JAMA 285:2639-2641, 2001
Cleary PD, McNeil BJ: Patient satisfaction as an indicator of quality care. Inquiry 25:25-36, 1988
Cleary PD, Edgman-Levitan S, Roberts M, et al: Patients evaluate their hospital care: A national survey. Health Aff (Millwood) 10:254-267, 1991
Cleary PD, Edgman-Levitan S: Health care quality: Incorporating consumer perspectives. JAMA 278:1608-1612, 1997
Fremont AM, Cleary PD, Hargraves JL, et al: Impact of patient-centered care processes on long-term myocardial infarction outcomes. J Gen Intern Med 16:800-808, 2001
Hewitt M, Simone JV: National Cancer Policy Board: Ensuring Quality Cancer Care. Washington, DC, National Academy Press, 1999
President's Cancer Panel: Voices of a Broken System: Real People, Real Problems. Bethesda, MD, National Cancer Institute, 2001
Darby C: Measuring the patient's perspective on the interpersonal aspects of cancer care, in Lipscomb J, Gotay CC, Snyder C (eds): Outcomes Assessment in Cancer. Cambridge, United Kingdom, Cambridge University Press, 2005, pp 290-304
Hodgson DC, Fuchs CS, Ayanian JZ: Impact of patient and provider characteristics on the treatment and outcomes of colorectal cancer. J Natl Cancer Inst 93:501-515, 2001
Shavers VL, Brown ML: Racial and ethnic disparities in the receipt of cancer treatment. J Natl Cancer Inst 94:334-357, 2002
Morris CR, Cohen R, Perkins C, et al: Cancer in California: 1988-1996. Sacramento, CA, Department of Health Services, Cancer Surveillance Section, 1999
American Joint Committee on Cancer: Manual for Staging of Cancer (5th ed): Philadelphia, PA, Lippincott-Raven, 1997
Stewart SL, Swallen KC, Glaser SL, et al: Comparison of methods for classifying Hispanic ethnicity in a population-based cancer registry. Am J Epidemiol 149:1063-1071, 1999
Ward WL, Hahn WA, Mo F, et al: Reliability and validity of the Functional Assessment of Cancer Therapy-Colorectal (FACT-C) quality of life instrument. Qual Life Res 8:181-195, 1999
Katz JN, Chang LC, Sangha O, et al: Can comorbidity be measured by questionnaire rather than medical record review? Med Care 34:73-84, 1996
National Institutes of Health: NIH Consensus Conference: Adjuvant therapy for patients with colon and rectal cancer. JAMA 264:1444-1450, 1990
Cancer: National Overview 1999/2000. http://www.dh.gov.uk/PublicationsAndStatistics/PublishedSurvey/NationalSurveyOfNHSPatients/NationalSurveyCancer/fs/en
Edgman-Levitan S, Shaller D, McInnes K, et al: The CAHPS Improvement Guide: Practical Strategies for Improving the Patient Care Experience. Boston, MA, Harvard Medical School, 2003, pp 23-34
Woloshin S, Bickell NA, Schwartz LM, et al: Language barriers in medicine in the United States. JAMA 273:724-728, 1995
Baker DW, Hayes R, Fortier JP: Interpreter use and satisfaction with interpersonal aspects of care for Spanish-speaking patients. Med Care 36:1461-1470, 1998
Office of Minority Health: National Standards for Culturally and Linguistically Appropriate Services in Health Care. Washington, DC, Department of Health and Human Services, 2001
Bickell NA, Young GJ: Coordination of care for early-stage breast cancer patients. J Gen Intern Med 16:737-742, 2001
Smedley BD, Stith AY, Nelson AR: Unequal Treatment: Confronting Racial and Ethnic Disparities in Healthcare. Washington, DC, National Academies Press, 2003
Hargraves L, Wilson IB, Zaslavsky AM, et al: Adjusting for patient characteristics when analyzing reports from patients about hospital care. Med Care 39:635-641, 2001
Zaslavsky AM, Zaborski LB, Ding L, et al: Adjusting performance measures to ensure equitable plan comparisons. Health Care Financ Rev 22:109-125, 2001
Meredith LS, Siu AL: Variation and quality of self-report health data: Asians and Pacific Islanders compared with other ethnic groups. Med Care 33:1120-1131, 1995
Taira DA, Safran DG, Seto TB, et al: Asian-American patient ratings of physician primary care performance. J Gen Intern Med 12:237-242, 1997
Murray-Garcia JL, Selby JV, Schmittdiel J, et al: Racial and ethnic differences in a patient survey: Patients' values, ratings, and reports regarding physician primary care performance in a large health maintenance organization. Med Care 38:300-310, 2000
Morales LS, Elliott MN, Weech-Maldonado R, et al: Differences in CAHPS adult survey reports and ratings by race and ethnicity: An analysis of the National CAHPS benchmarking data 1.0. Health Serv Res 36:595-617, 2001
Lurie N, Zhan C, Sangl J, et al: Variation in racial and ethnic differences in consumer assessments of health care. Am J Manag Care 9:502-509, 2003
Morales LS, Cunningham WE, Brown JA, et al: Are Latinos less satisfied with communication by health care providers? J Gen Intern Med 14:409-417, 1999
Becker G, Newsom E: Socioeconomic status and dissatisfaction with health care among chronically ill African Americans. Am J Public Health 93:742-748, 2003(John Z. Ayanian, Alan M. )
Department of Health Care Policy, Harvard Medical School
Division of Medical Oncology, Dana-Farber Cancer Institute, Boston, MA
Center on Outcomes, Research and Education Evanston Northwestern Healthcare Research Institute, Evanston, IL
Cancer Surveillance Program, Sacramento
Public Health Institute, Berkeley
Northern California Cancer Center, Union City
Cancer Surveillance Section, California Department of Health Services, Sacramento, CA
ABSTRACT
PURPOSE: To identify opportunities for improving care, we evaluated patients' perceptions of the quality of their cancer care by race, ethnicity, and language.
PATIENTS AND METHODS: We surveyed a population-based cohort of 1,067 patients with colorectal cancer in northern California approximately 9 months after diagnosis. Adjusting for clinical and demographic factors, mean problem scores were analyzed on a 100-point scale for six domains of care.
RESULTS: Mean problem scores were highest for health information (47.8), followed by treatment information (32.3), psychosocial care (31.7), coordination of care (21.3), confidence in providers (13.1), and access to cancer care (12.7). In adjusted comparisons with white patients, African American patients reported more problems with coordination of care (difference, 9.8; P < .001), psychosocial care (difference, 7.2; P = .03), access to care (difference, 6.6; P = .03), and health information (difference, 12.5; P < .001). Asian/Pacific Islander patients reported more problems than did white patients with coordination of care (difference, 13.2; P < .001), access to care (difference, 15.5; P < .001), and health information (difference, 12.6; P = .004). Hispanic patients tended to report more problems with coordination of care (difference, 4.4; P = .06), access to care (difference, 5.8; P = .08), and treatment information (difference, 7.0; P = .06). Non每English-speaking white patients reported more problems than other white patients with coordination of care (difference, 21.9; P < .001), psychosocial care (difference, 16.1; P = .009), access to care (difference, 19.8; P = .003), and treatment information (difference, 17.8; P = .002). Non每English-speaking Hispanic patients reported more problems than other Hispanic patients with confidence in providers (difference, 16.9; P = .01).
CONCLUSION: Efforts to improve patients' experiences with cancer care should address disparities by race, ethnicity, and language, particularly in coordination of care, access to care, and the provision of relevant information.
INTRODUCTION
Patients' experiences with health care providers are a central component of their quality of care, complementing more technical aspects of care such as the appropriate use of medications and procedures.1 During the last 15 years, surveys have been developed to assess patients' experiences with care while hospitalized or enrolled in managed care plans.2-5 These experiences are particularly important for patients with cancer, because they commonly require access to multiple specialists, effective coordination of care, accurate information about the disease and treatment options, and timely attention to symptoms. Both the National Cancer Policy Board of the Institute of Medicine and the President's Cancer Panel have drawn attention to perceived deficiencies in patients' experiences with cancer care,6,7 but few studies have analyzed this topic in broadly representative cohorts of patients with cancer.8
Therefore, we surveyed a racially and ethnically diverse cohort of more than 1,000 patients diagnosed with colorectal cancer in northern California. Our primary objective was to evaluate whether patients' reported problems with care varied by race and ethnicity because prior studies demonstrated racial and ethnic differences in cancer treatment and outcomes.9,10 We also assessed variations in patients' problems with care by language, income, education, sex, age, and health status, and we analyzed how patients' experiences with care were related to their overall ratings of the quality of cancer care.
PATIENTS AND METHODS
Study Population
For patients diagnosed with cancer in California, hospitals, radiation treatment facilities, and physicians are required by state law to report their patients' cancer diagnosis, demographic characteristics, and first course of treatment to the California Cancer Registry, the largest population-based cancer registry in the world.11 Eligible patients for this study included adults age 40 to 84 years who were newly diagnosed with invasive colorectal cancer from April 1999 through April 2000 in nine counties of northern California (Alameda, Contra Costa, Placer, Sacramento, San Francisco, San Joaquin, San Mateo, Santa Clara, and Solano) served by the San Francisco/Oakland, San Jose/Monterey, and Sacramento regional registries.
Regional registry staff identified patients within 4 months of diagnosis through regular visits to hospitals and pathology laboratories. Patients were excluded if they were deceased, did not speak English or Spanish, had physical or mental impairments reported by their physicians or family members that prevented them from completing a telephone or mail survey, or their hospital lacked valid address or telephone information for them. The California Cancer Registry provided information on patients' age, sex, race and ethnicity, tumor stage,12 and receipt of surgery as obtained by the regional registries from medical records.
The study protocol was approved by institutional review boards of the California Department of Health Services, Public Health Institute, Northern California Cancer Center, and Harvard Medical School, in accordance with assurances filed with and approved by the US Department of Health and Human Services.
Survey Instrument
The survey instrument was designed to assess patients' experiences with cancer care, health-related quality of life, comorbid illnesses, and sociodemographic characteristics. Questions regarding patients' experiences with cancer care were obtained from a survey designed by the Picker Institute (Boston, MA).2-4 After adapting the self-administered Picker Institute cancer survey for telephone interviews and pilot testing it among 50 patients with colorectal cancer in northern California, we selected 34 questions to include in our instrument. Thirty-three questions addressed problems with specific aspects of care (see Appendix), and one question elicited patients' overall rating of care on a five-level scale ranging from poor to excellent. Patients were considered to have problems with care if they gave suboptimal responses, consistent with prior analyses of Picker Institute survey questions.2-4
To assess health-related quality of life of patients with colorectal cancer, we used the previously validated Functional Assessment of Cancer Therapy每Colorectal (FACT-C) survey.14 This 34-item instrument evaluates patients' emotional, social, physical, and functional well-being, and colorectal cancer每specific issues. Scores for the physical, functional, and colorectal subscales can be summed to create a Trial Outcomes Index. Higher scores on the FACT-C indicate better health-related quality of life.
Patients reported their demographic characteristics (age, sex, marital status, race, Hispanic ethnicity, language spoken at home, education, income) and whether they had received surgery, chemotherapy, or radiation therapy for their colorectal cancer. Patients also reported whether they had been diagnosed with 14 comorbid conditions; prior research has shown that patients are generally accurate in reporting conditions noted in their medical records.15 The full survey was translated into Spanish and back-translated into English to ensure accuracy.
Data Collection
The treating physician listed in the hospital record for each patient was notified by mail about the study. Patients then received a letter from the regional registry describing the study and offering a $20 stipend for completing the survey. Trained interviewers from the Public Health Institute attempted to contact patients up to 15 times during 1 month to conduct the survey in English or Spanish via computer-assisted telephone interviewing. Patients who did not respond and those with impaired hearing were mailed a self-administered version of the survey.
Statistical Analysis
To group questions into domains of care, we performed a principal factor analysis with promax (oblique) rotation. Examination of eigenvalues suggested solutions with four to eight factors. We reviewed these solutions in light of the substantive item content and groupings derived from patient focus groups conducted by the Picker Institute. We combined individual questions into six domains of care that had moderate to high internal consistency, including coordination of care (eight items; Cronbach's coefficient = .77), confidence in providers (four items; = .65), psychosocial care (five items; = .64), access to cancer care (three items; = .55), treatment information (four items; = .65), and health information (seven items; = .82). Average problem scores in specific domains were calculated as the percentage of answered questions for which patients reported problems with care. Two items on the management of pain and nausea were analyzed individually.
We compared survey respondents and nonrespondents using the 2 test for categorical variables (race/ethnicity, sex, stage, receipt of surgery) and Student's t test for age. Respondents' race and ethnicity were based on their self-reported data. For nonrespondents, race and ethnicity were determined from reporting facilities and a Hispanic surname algorithm used by the California Cancer Registry to improve the identification of Hispanic patients.13
We compared unadjusted problem scores in each domain and overall ratings of cancer care by patients' demographic characteristics. Among patients with stage III colon cancer or stage II or III rectal cancer, for whom adjuvant chemotherapy is recommended by national guidelines,16 we evaluated the association of problem scores with use of chemotherapy. To assess how problems with specific aspects of care were related to overall ratings of care, we calculated Pearson correlation coefficients of the six domain scores with the single question, "Overall, how would you rate the quality of your health care since your diagnosis of cancer?" We also stratified these correlations by race and ethnicity.
We analyzed multivariable linear regression models to assess the independent predictors of patients experiencing problems with care in each domain, and analyzed logistic regression models to assess predictors of patients rating their overall cancer care as excellent or very good. These models controlled for sociodemographic characteristics, cancer stage, five comorbid illnesses reported by at least 15% of survey respondents (hypertension, coronary heart disease, diabetes mellitus, chronic lung disease, and depression), and FACT-C scores (emotional well-being, social/family well-being, and the Trial Outcomes Index). We also used generalized estimating equations to adjust for clustering of patients by hospitals where they initiated treatment. Variables for non每English-speaking white, Asian/Pacific Islander, and Hispanic patients were added to assess language as a potential mediator of observed differences in care. In a secondary analysis, we added variables for the proportions of patients at each hospital who were African American, Hispanic, and Asian/Pacific Islander to assess whether patient-level effects of race and ethnicity on problem scores were confounded by hospital-level compositional effects; we also tested interactions of patients' race/ethnicity with the racial and ethnic composition of the hospitals. All analyses were conducted with SAS statistical software, version 8.2 (SAS Institute, Cary, NC) and two-tailed P values.
RESULTS
Patient Characteristics
Among 2,139 patients with colorectal cancer diagnosed during the study period, we excluded 648 patients, including 159 who had died, 139 who did not speak English or Spanish, 29 with physical or mental impairments that precluded interviewing as reported by their physicians or family members, 273 with no valid address or telephone number in hospital records, 27 who had recurrent colorectal cancer or another type of cancer, and 21 who had only carcinoma-in-situ. Of the remaining 1,491 eligible patients, 259 refused to participate, 57 were too ill to complete the survey, 18 had impaired hearing, and 78 did not respond by phone or mail.
Surveys were completed by 1,079 patients (72.4%), including 1,002 telephone interviews and 77 mail surveys. Respondents were more likely than nonrespondents to be white (74.9% v 68.1%) or African American (8.0% v 7.1%), and less likely than nonrespondents to be Hispanic (9.7% v 10.8%) or Asian/Pacific Islander (7.5% v 14.0%; P = .001). Respondents were also somewhat younger (mean, 66.2 v 69.4 years; P < .001), more likely to be women (50.4% v 40.8%; P = .001), less likely to have metastatic cancer (10.1% v 15.8%; P = .002), and more likely to have undergone colorectal surgery (98.8% v 95.2%; P < .001). The median time from diagnosis to completing the survey was 8.9 months (interquartile range, 7.7 to 10.5 months). Because our study focused on racial and ethnic differences in care, we excluded five patients who reported their race as "other," five patients who provided minimal data on their experiences with care, and two patients whose age was beyond our designated range of 40 to 84 years.
The characteristics of the study cohort of 1,067 patients stratified by race and ethnicity are shown in Table 1. Seventy-five respondents (7.0%) reported speaking a primary language other than English at home, including 32 Hispanic patients, 28 Asian/Pacific Islander patients, and 15 white patients. About half of the study cohort (48.0%) had received chemotherapy, and 17.6% had received radiation therapy. The 1,067 study participants were initially treated in 75 hospitals (median, 12 patients; interquartile range, five to 21 patients).
Patients' Reported Problems With Care
Mean problem scores varied widely across the six domains of care (Table 2). The highest problem scores were reported for health information (47.8), followed by treatment information (32.3), psychosocial care (31.7), and coordination of care (21.3). Lower problem scores were reported for confidence in providers (13.1) and access to cancer care (12.7). On individual questions, for example, half of the patients reported receiving insufficient health information about changes in their work or usual activities (52.7%), relationship with their spouse or partner (50.7%), or sexual activity (49.6%). Other common problems included not being given enough information about cancer treatments (29.3%), not having providers work well together (25.1%), and not being involved in decisions as much as desired (24.7%). Smaller proportions of patients reported they did not receive sufficient pain control (12.8%), were discouraged from seeking needed cancer care (12.1%), believed their doctors did not know enough about cancer treatments (11.5%), or were not treated with respect and dignity (9.7%).
Unadjusted mean problem scores on the six domains of care differed significantly by race and ethnicity (Table 3), with more problems reported by Asian/Pacific Islander, Hispanic, and African American patients relative to white patients, and by patients who did not speak English at home relative to those who did. Asian/Pacific Islander patients also reported more problems with pain control (30.0%; P < .001) than white (10.5%), African American (14.1%), or Hispanic patients (15.5%). White patients who did not speak English at home had mean problem scores that were substantially higher than those of English-speaking white patients on each domain (all P < .07). Hispanic patients who did not speak English at home reported more problems than other Hispanic patients with coordination of care (P = .04), confidence in providers (P = .004), and treatment information (P = .07). In contrast, mean problem scores did not differ significantly by language among Asian/Pacific Islander patients (all P > .42).
Younger patients reported significantly more problems with confidence in providers and health information than older patients (Table 3). Patients who had not graduated from high school and those with low incomes reported significantly more problems than more educated patients and more affluent patients, respectively, in each domain except health information. Unadjusted problem scores did not differ significantly by sex or marital status, except married patients reported more problems with health information.
Among 352 patients who were potentially eligible for adjuvant chemotherapy, chemotherapy recipients tended to report lower mean problem scores than nonrecipients with access to cancer care (11.0 v 16.7; P = .07) and confidence in providers (12.9 v 18.4; P = .08). They did not differ significantly, however, on other domains of patients' experiences (P > .20).
In multivariable analyses, significant adjusted differences in total problem scores were evident by race and ethnicity in four of six domains (Fig 1). Relative to white patients, African American patients reported significantly more problems than white patients with coordination of care (difference, 9.8; P < .001), psychosocial care (difference, 7.2; P = .03), access to cancer care (difference, 6.6; P = .03), and health information (difference, 12.5; P < .001). Similarly, Asian/Pacific Islander patients reported more problems than white patients with coordination of care (difference, 13.2; P < .001), access to cancer care (difference, 15.5; P < .001), and health information (difference, 12.6; P = .004). Hispanic patients tended to report more problems with coordination of care (difference, 4.4; P = .06), access to cancer care (difference, 5.8; P = .08), and treatment information (difference, 7.0; P = .06).
In a secondary analysis, the racial and ethnic composition of hospitals where patients were initially treated was not associated with problems in any domain, and differences in problem scores by patients' race and ethnicity remained significant and similar in magnitude. For several domains of problems, however, interactions of the patient-level and hospital-level variables were statistically significant. Smaller differences in problem scores for coordination of care, confidence in providers, access to care, and treatment information between white and African American patients were observed at hospitals with larger proportions of African Americans (all P < .05). The difference in problem scores for psychosocial care was similarly reduced for Asian/Pacific Islander patients treated in hospitals with a larger proportion of Asian/Pacific Islanders (P = .007). In each case, white patients' problem scores had little association with hospitals' racial composition, but problem scores for the nonwhite group were significantly lower at hospitals where that group constituted a larger proportion of the patients.
Several other demographic characteristics were associated with higher adjusted problem scores in specific domains. Patients with an annual income below $25,000 reported more problems than those with higher incomes in access to cancer care (difference, 4.4; P = .04). Married patients reported more problems than unmarried patients with health information (difference, 6.1; P = .006). Women reported higher problem scores than men for confidence in providers (difference, 2.2; P = .04), as did college graduates relative to high school graduates (difference, 3.6; P = .009). Age was not associated with adjusted problem scores in any domain.
Adding language spoken at home to these models, non每English-speaking white patients reported more problems than English-speaking white patients with coordination of care (difference, 21.9; P < .001), psychosocial care (difference, 16.1; P = .009), access to cancer care (difference, 19.8; P = .003), and treatment information (difference, 17.8; P = .002). Non每English-speaking Hispanic patients reported more problems than English-speaking Hispanic patients with confidence in providers (difference, 16.9; P = .01). Among Asian/Pacific Islander patients, adjusted problem scores did not differ significantly by language on any domain.
Worse emotional well-being and social/family well-being as measured by the FACT-C instrument were associated with significantly higher adjusted problem scores on each of the six domains of patients' experiences with care (Table 4). Worse physical, functional, and disease-specific well-being on the Trial Outcomes Index were associated with higher adjusted problem scores for coordination of care, confidence in providers, and health information. Patients with stage III tumors reported lower adjusted problem scores for treatment information than those with stage I tumors (difference, 5.8; P = .01), but stage was not associated with adjusted problem scores for other domains. Patients with chronic lung disease reported higher adjusted problem scores for health information than other patients (difference, 6.2; P = .005), but other comorbid conditions were not clearly associated with differences in problem scores.
Overall Ratings of Care
The proportions of patients rating their overall quality of cancer care as very good or excellent were significantly lower for nonwhite and non每English-speaking patients (Fig 2). Relative to white patients in an adjusted analysis, Asian/Pacific Islander patients (odds ratio [OR], 0.3; 95% CI, 0.2 to 0.6) and Hispanic patients (OR, 0.6; 95% CI, 0.4 to 1.0) remained less likely to rate their overall cancer care as very good or excellent. Non每English-speaking white patients were also less likely than English-speaking white patients to rate their care as very good or excellent (adjusted OR, 0.1; 95% CI, 0.04 to 0.5).
In the full cohort, patients' overall ratings of care had a strong negative correlation with their problem scores for each domain (all P < .001), including coordination of care (r = 每0.58), confidence in providers (r = 每0.46), treatment information (r = 每0.38), psychosocial care (r = 每0.33), access to cancer care (r = 每0.32), and health information (r = 每0.32). Problem scores for coordination of care were substantially higher among those rating their overall cancer care as poor (60.4), fair (60.6), good (41.2), or very good (21.5) than among those rating their care as excellent (9.4; P < .001). Problems with coordination of care also had the strongest correlation with overall ratings of care within each racial and ethnic group (white, 每0.57; African American, 每0.56; Hispanic, 每0.45; Asian/Pacific Islander, 每0.64; all P < .001).
DISCUSSION
Cancer care is often complex, involving numerous specialists and different types of curative, life-sustaining, and palliative therapies during the course of patients' illness. This complexity makes the interpersonal aspects of cancer care particularly important if patients are to achieve optimal clinical outcomes and quality of life. African American, Hispanic, Asian/Pacific Islander, and non每English-speaking white patients were significantly more likely than English-speaking white patients to report problems in multiple domains of care for which patients are the best or only source of information. Problems with coordination of care were most strongly correlated with worse ratings of the overall quality of cancer care.
How can clinicians and policy makers respond to the results of this study? Our findings suggest that standardized surveys of patients' experiences with cancer care could be used regularly to assess potential disparities by race, ethnicity, and language. In the United Kingdom, for example, a nationwide survey of experiences of 65,000 patients with cancer care was initiated by the National Health Service in 2000, including comparative assessments by white, African American, and South Asian patients.17 Such survey data can be used by health care organizations and providers to identify opportunities for improving their patients' experience with care.18
While striving to enhance the health care experiences of all patients with cancer, clinicians should devote special attention to providing effective communication, access to needed services, and coordination of care for nonwhite and non每English-speaking patients. More frequent problems with care reported by non每English-speaking patients indicate that language barriers may be an important contributor to disparities in care for cancer and other conditions.19,20 National standards proposed by the federal Office of Minority Health emphasize the importance of bilingual physicians, nurses, and professional interpreters to promote effective communication with patients who have limited or no English-language proficiency.21
The strong correlation of problems in coordination of care with lower overall ratings of the quality of care suggests that new initiatives to enhance coordination of care may have the greatest impact on improving patients' overall satisfaction with care and eliminating racial and ethnic disparities in this domain. Effective approaches for improving coordination may include multidisciplinary team meetings to address clinical and supportive needs of specific patients, navigators to help patients keep their medical appointments and obtain needed information, and computerized systems to aid clinicians in tracking consultations, test results, and treatments.22
Problems with interpersonal aspects of care for nonwhite and non每English-speaking patients may contribute to differences in cancer treatment and outcomes by race, ethnicity, and language.23 In our study, for example, patients who were eligible for adjuvant chemotherapy but did not receive it tended to report more problems with access to cancer care and confidence in providers, suggesting that technical quality of care is correlated with patients' experiences in these two domains. Future research should explore how patients' problems with care influence their outcomes and evaluate interventions to promote optimal technical and interpersonal care.
Another notable finding was the strong association between worse health-related quality of life, especially worse emotional and social well-being, and higher problem scores on all domains of care, consistent with prior research for other conditions.24 Emotionally and socially isolated patients may have greater unmet needs and worse experiences with care, or their worse health-related quality of life may result in part from unmeasured differences in disease severity and progression. Although these two plausible explanations cannot be distinguished in our cross-sectional study, regular assessments of health-related quality of life in patients with cancer may help to identify those at increased risk for problems with care.
Prior studies have assessed racial and ethnic differences in the interpersonal aspects of medical care, primarily through surveys of patients in primary care practices or health maintenance organizations.25 As in our study, Asian/Pacific Islander patients have reported more problems than white patients with access to care, communication with physicians, and coordination of care.26-30 Asian/Pacific Islander patients also have reported less satisfaction with care in most26-28,30 but not all of these studies.29 African American and Hispanic patients have reported worse access to needed specialty care than white patients in managed care plans.30 Relative to white patients, however, Hispanic and African American patients generally have reported similar satisfaction and ratings of communication with their physicians,26-30 although Hispanic patients who speak only Spanish have reported less satisfaction with care than those who speak English.31 One qualitative study found that problems with care were particularly concentrated among chronically ill African American patients with low incomes.32
A major strength of our study was the population-based sample, providing a highly representative and diverse cohort that was not limited to selected hospitals or health care systems. Questions related to quality of care were based on prior research that determined items most salient to patients,2-4 an approach that has been endorsed by a recent review of studies measuring interpersonal aspects of cancer care.8 Our survey was primarily administered by telephone, so biases related to racial and ethnic differences in literacy were minimized.
This study also had several potential limitations. Our response rate of 72.9% was consistent with other population-based surveys of chronically ill patients, but we may have underestimated problems with care because non每English-speaking patients who were either ineligible or nonrespondents to our survey were probably more likely to experience such problems. We studied patients in urban and suburban counties of northern California, so our findings might not generalize to other regions or rural areas. Patients who died within 9 months of their cancer diagnosis were ineligible for this study, so future research should assess the health care experiences of early decedents whose needs are primarily palliative in nature.
Remedying problems with interpersonal aspects of cancer care may be an important step toward eliminating racial and ethnic disparities in cancer treatment and outcomes. Although the marginal benefit of such efforts to improve cancer care may be greatest for nonwhite and non每English-speaking patients who reported the most problems in our study, these efforts have the potential to benefit patients of all racial and ethnic backgrounds.
Acknowledgment
We thank Gretchen Agha, Marti Felter, and Mary Anderson for coordinating patient enrollment; Bonnie Davis and the California Department of Health Services survey research unit for conducting telephone interviews; Mark Allen, Craig Grilley, Scott Riddle, and Robert Wolf for programming assistance; Paul D. Cleary and Neeraj Arora for helpful comments on the manuscript; National Research Corporation for authorizing use of the Picker cancer survey; and David Cella for authorizing use of the Functional Assessment of Cancer Therapy每Colorectal survey.
NOTES
Supported by Grants No. R01 HS09869 and U01 CA93324 from the Agency for Healthcare Research and Quality and the National Cancer Institute.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Eisenberg L: Good technical outcome, poor service experience: A verdict on contemporary medical care? JAMA 285:2639-2641, 2001
Cleary PD, McNeil BJ: Patient satisfaction as an indicator of quality care. Inquiry 25:25-36, 1988
Cleary PD, Edgman-Levitan S, Roberts M, et al: Patients evaluate their hospital care: A national survey. Health Aff (Millwood) 10:254-267, 1991
Cleary PD, Edgman-Levitan S: Health care quality: Incorporating consumer perspectives. JAMA 278:1608-1612, 1997
Fremont AM, Cleary PD, Hargraves JL, et al: Impact of patient-centered care processes on long-term myocardial infarction outcomes. J Gen Intern Med 16:800-808, 2001
Hewitt M, Simone JV: National Cancer Policy Board: Ensuring Quality Cancer Care. Washington, DC, National Academy Press, 1999
President's Cancer Panel: Voices of a Broken System: Real People, Real Problems. Bethesda, MD, National Cancer Institute, 2001
Darby C: Measuring the patient's perspective on the interpersonal aspects of cancer care, in Lipscomb J, Gotay CC, Snyder C (eds): Outcomes Assessment in Cancer. Cambridge, United Kingdom, Cambridge University Press, 2005, pp 290-304
Hodgson DC, Fuchs CS, Ayanian JZ: Impact of patient and provider characteristics on the treatment and outcomes of colorectal cancer. J Natl Cancer Inst 93:501-515, 2001
Shavers VL, Brown ML: Racial and ethnic disparities in the receipt of cancer treatment. J Natl Cancer Inst 94:334-357, 2002
Morris CR, Cohen R, Perkins C, et al: Cancer in California: 1988-1996. Sacramento, CA, Department of Health Services, Cancer Surveillance Section, 1999
American Joint Committee on Cancer: Manual for Staging of Cancer (5th ed): Philadelphia, PA, Lippincott-Raven, 1997
Stewart SL, Swallen KC, Glaser SL, et al: Comparison of methods for classifying Hispanic ethnicity in a population-based cancer registry. Am J Epidemiol 149:1063-1071, 1999
Ward WL, Hahn WA, Mo F, et al: Reliability and validity of the Functional Assessment of Cancer Therapy-Colorectal (FACT-C) quality of life instrument. Qual Life Res 8:181-195, 1999
Katz JN, Chang LC, Sangha O, et al: Can comorbidity be measured by questionnaire rather than medical record review? Med Care 34:73-84, 1996
National Institutes of Health: NIH Consensus Conference: Adjuvant therapy for patients with colon and rectal cancer. JAMA 264:1444-1450, 1990
Cancer: National Overview 1999/2000. http://www.dh.gov.uk/PublicationsAndStatistics/PublishedSurvey/NationalSurveyOfNHSPatients/NationalSurveyCancer/fs/en
Edgman-Levitan S, Shaller D, McInnes K, et al: The CAHPS Improvement Guide: Practical Strategies for Improving the Patient Care Experience. Boston, MA, Harvard Medical School, 2003, pp 23-34
Woloshin S, Bickell NA, Schwartz LM, et al: Language barriers in medicine in the United States. JAMA 273:724-728, 1995
Baker DW, Hayes R, Fortier JP: Interpreter use and satisfaction with interpersonal aspects of care for Spanish-speaking patients. Med Care 36:1461-1470, 1998
Office of Minority Health: National Standards for Culturally and Linguistically Appropriate Services in Health Care. Washington, DC, Department of Health and Human Services, 2001
Bickell NA, Young GJ: Coordination of care for early-stage breast cancer patients. J Gen Intern Med 16:737-742, 2001
Smedley BD, Stith AY, Nelson AR: Unequal Treatment: Confronting Racial and Ethnic Disparities in Healthcare. Washington, DC, National Academies Press, 2003
Hargraves L, Wilson IB, Zaslavsky AM, et al: Adjusting for patient characteristics when analyzing reports from patients about hospital care. Med Care 39:635-641, 2001
Zaslavsky AM, Zaborski LB, Ding L, et al: Adjusting performance measures to ensure equitable plan comparisons. Health Care Financ Rev 22:109-125, 2001
Meredith LS, Siu AL: Variation and quality of self-report health data: Asians and Pacific Islanders compared with other ethnic groups. Med Care 33:1120-1131, 1995
Taira DA, Safran DG, Seto TB, et al: Asian-American patient ratings of physician primary care performance. J Gen Intern Med 12:237-242, 1997
Murray-Garcia JL, Selby JV, Schmittdiel J, et al: Racial and ethnic differences in a patient survey: Patients' values, ratings, and reports regarding physician primary care performance in a large health maintenance organization. Med Care 38:300-310, 2000
Morales LS, Elliott MN, Weech-Maldonado R, et al: Differences in CAHPS adult survey reports and ratings by race and ethnicity: An analysis of the National CAHPS benchmarking data 1.0. Health Serv Res 36:595-617, 2001
Lurie N, Zhan C, Sangl J, et al: Variation in racial and ethnic differences in consumer assessments of health care. Am J Manag Care 9:502-509, 2003
Morales LS, Cunningham WE, Brown JA, et al: Are Latinos less satisfied with communication by health care providers? J Gen Intern Med 14:409-417, 1999
Becker G, Newsom E: Socioeconomic status and dissatisfaction with health care among chronically ill African Americans. Am J Public Health 93:742-748, 2003(John Z. Ayanian, Alan M. )