Respondents could select more than 1 reason for each question.
aData are reported as number (percentage) of women unless otherwise specified. Totals may not always equal 100% because of rounding.
bUsing χ2 tests for categorical variables and the Kruskal-Wallis test for the health literacy score.
cComorbidity was assessed by adding the number of self-reported medical conditions25,27 including other cancers, diabetes, heart attack, stroke, emphysema/chronic bronchitis/asthma/other chronic lung disease, kidney problems, and depression/other psychiatric illness.
dUsing the Fisher exact test because some cell counts (number for that category) were lower than 5.
eLower score indicates better literacy.
fUsing cancer registry data; 1 case reported as stage IV was categorized as stage III.
Abbreviation: HLGFT, Hosmer-Lemeshow goodness-of-fit test.
aFor each question, patients were instructed to select all responses that applied to them.
bBy χ2 testing for differences in responses by race/ethnicity.
cModels for surgeon selection had C statistics ranging from 0.62 to 0.72.
dModels for hospital selection had C statistics ranging from 0.63 to 0.71.
eParticipants may be included in more than 1 model. Models were not performed for “treated by surgeon in past” and “other” because of the rarity of these responses.
fUsing multivariable logistic regression, adjusting for age, comorbidity, marital status, mean health literacy score, insurance status, stage, and education. Bolded results are significant at P < .05.
gUsing multivariable logistic regression, adjusting for age, comorbidity, marital status, mean health literacy score, stage, and education. Bolded results are significant at P < .05.
hParticipants may be included in more than 1 model. Models were not performed for “treated by hospital in past” and “other” because of the rarity of these responses.
iUsing the Fisher exact test because some cell counts (number for that category) were lower than 5.
aBy χ2 testing, unadjusted.
bOne English-speaking Hispanic women did not answer this question.
aUsing multivariable logistic regression, adjusting for race/ethnicity, age, comorbidity, marital status, stage, mean health literacy score, and education. Insurance was not included because of the rarity of uninsured status. Bolded results are significant at P < .05.
bEach possible response with at least 10% of women answering “yes” for the reasons for surgeon selection was added to base model (patient variables listed above) and then a final model was performed which included all possible responses for surgeon selection.
cModel C statistics ranged from 0.64 to 0.66 for ratings of surgeon and 0.59 to 0.60 for ratings of hospital.
dEach possible response with at least 10% of women answering “yes” for the reasons for hospital selection was added to base model (patient variables listed above) and then a final model was performed which included all possible responses for hospital selection.
eOne English-speaking Hispanic woman did not answer this question.
eTable. Survey Questions on Surgeon and Hospital Selection
eFigure. Surgeon/Hospital Choice Survey
Customize your JAMA Network experience by selecting one or more topics from the list below.
Freedman RA, Kouri EM, West DW, Keating NL. Racial/Ethnic Differences in Patients’ Selection of Surgeons and Hospitals for Breast Cancer Surgery. JAMA Oncol. 2015;1(2):222–230. doi:10.1001/jamaoncol.2015.20
Racial differences in breast cancer treatment may result in part from differences in the surgeons and hospitals from whom patients receive their care. However, little is known about differences in patients’ selection of surgeons and hospitals.
To examine racial/ethnic differences in how women selected their surgeons and hospitals for breast cancer surgery.
Design, Setting, and Participants
We surveyed 500 women (222 non-Hispanic white, 142 non-Hispanic black, 89 English-speaking Hispanic, and 47 Spanish-speaking Hispanic) from northern California cancer registries with stage 0 to III breast cancer diagnosed during 2010 through 2011. We used multivariable logistic regression to assess the reasons for surgeon and hospital selection by race/ethnicity, adjusting for other patient characteristics. We also assessed the association between reasons for physician selection and patients’ ratings of their surgeon and hospital.
Main Outcomes and Measures
Reasons for surgeon and hospital selection and ratings of surgeon and hospital.
The 500 participants represented a response rate of 47.8% and a participation rate of 69%. The most frequently reported reason for surgeon selection was referral by another physician (78%); the most frequently reported reason for hospital selection was because it was a part of a patient’s health plan (58%). After adjustment, 79% to 87% of black and Spanish-speaking Hispanic women reported selecting their surgeon based on a physician’s referral vs 76% of white women (P = .007). Black and Hispanic patients were less likely than white patients to report selecting their surgeon based on reputation (adjusted rates, 18% and 22% of black and Hispanic women, respectively, vs 32% of white women; P = .02). Black and Hispanic women were also less likely than white women to select their hospital based on reputation (adjusted rates, 7% and 15% vs 23%, respectively; P = .003). Women who selected their surgeon based on reputation more often rated the care from their surgeon as excellent (adjusted odds ratio, 2.21; 95% CI, 1.24-3.93); those reporting their surgeon was one of the only surgeons available through the health plan less often reported excellent quality of surgical care (adjusted odds ratio, 0.56; 95% CI, 0.34-0.91).
Conclusions and Relevance
Compared with white patients with breast cancer, minority patients were less actively involved in physician and hospital selection, relying more on physician referral and health plans rather than on reputation. Interventions to promote involvement in surgeon and hospital selection may have potential for addressing disparities related to lower-quality care from surgeons and hospitals.
Racial/ethnic disparities in the use, quality, and delivery of medical care have been well described1-3 and may affect clinical outcomes for patients receiving cancer-directed treatments. Research suggests that differences in treatment and outcomes may occur in part because of differences in the clinicians and institutions where patients receive their care.4-14 Minority patients are more likely than white patients to receive medical care in lower-quality hospitals,4,5,7,8,15-17 from clinicians caring for higher proportions of minority patients,18,19 and from physicians who are less well trained than those treating white patients.19 Among patients with breast cancer, black women are more likely than white women to undergo breast surgery at hospitals with lower rates of radiation use following breast conservation.18 Hospital factors may also contribute to racial differences in delays in adjuvant breast cancer care.20
These racial/ethnic differences in where patients obtain oncology care may be due in part to differences in referral patterns among clinicians or patients’ involvement in selecting physicians and hospitals and preferences about those providers. A patient’s choice of hospital and physician may be influenced by past experiences, other medical conditions, type of surgery to be performed, hospital location, or recommendations by family and friends.21,22 Previous research has shown that patients with breast cancer who actively participate in selecting their surgeons are more likely to receive care in higher-volume hospitals and in hospitals with cancer programs.23 However, data are limited with regard to how women select their physician and hospital for cancer treatment and whether there are racial/ethnic differences in such decisions. Promoting thoughtful decision making when choosing a physician and hospital may be an important element in addressing treatment disparities.
In this study, we interviewed a diverse sample of women with breast cancer in northern California to understand racial/ethnic differences in how patients select their surgeons and hospitals for breast cancer surgery, accounting for relevant factors such as educational attainment, insurance, health literacy, and English-language proficiency.
The most frequently reported reason for breast surgeon selection among 500 women with early-stage breast cancer was referral by another physician. This response was provided more frequently by black and Spanish-speaking (vs white) women.
Black and Hispanic patients were less likely than white patients to report selecting their surgeon and hospital based on reputation.
Women who selected their surgeon based on reputation more often rated the care from their surgeon as excellent.
Our findings suggest less-active involvement in surgeon and hospital selection for minority patients, who rely more on physician referral and health plans rather than on reputation.
Interventions to promote involvement in surgeon and hospital selection may have potential for addressing disparities related to lower-quality care from surgeons and hospitals.
As previously described,24 we identified 1118 white, black, or Hispanic women from regions 1/8 (San Francisco/Santa Clara) and region 3 (Sacramento) of the California Cancer Registry (CCR) who were diagnosed as having stage 0 to III breast cancer in 2010 through 2011 and underwent primary surgery for their breast cancer. We obtained study approvals from the CCR, the California Health and Human Services Agency Committee for the Protection of Human Subjects, and Harvard Medical School’s Committee on Human Studies.
We mailed letters to eligible patients in English and Spanish inviting them to participate in a survey study about their breast cancer care. Potential participants were contacted by telephone. Women who agreed to participate provided verbal informed consent before the interview and received a $20 incentive on interview completion. Interviews were conducted by bilingual trained study staff using computer-assisted telephone interview software.
Participants were asked which of several statements describe how they selected their breast cancer surgeons22,23 and hospitals where they underwent cancer-directed surgery,21,22 including statements such as “I was referred to the surgeon by another doctor” and “I chose this surgeon (or hospital) because of his/her (or its) reputation” (eTable in the Supplement). We also asked about patients’ priorities in making decisions about hospitals (eg, “Which was more important when choosing the hospital that you went to—the location…reputation…or both equally important?”),21 as well as patients’ ratings of the overall quality of care provided by their surgeon and hospital. In addition, we collected information on race/ethnicity, educational attainment,25 insurance coverage at diagnosis,25 health literacy,26 and comorbidity25,27; Hispanic women were asked what language(s) they read and speak.28 Prior to participant enrollment, cognitive testing with 10 patients (in English and Spanish) demonstrated survey feasibility and clarity. A copy of the survey instrument is shown in the eFigure in the Supplement.
As described previously,24 among 1118 patients, 231 refused participation, 317 could not be reached, and 68 were deceased or too ill. Among the 502 women surveyed, 2 self-identified as Asian and were excluded. The American Association for Public Opinion Research29 response rate was 47.8%; the participation rate among those for whom we had contact information was 69%. Fewer white women had no contact information (15%) than black (25%) and Hispanic (23%) women. However, among women with contact information (n = 891), respondents (n = 502) had similar baseline characteristics as nonrespondents (n = 389), except respondents were younger (mean age, 58 vs 64 years; P < .001). Of 136 Hispanic women, 70 were interviewed in Spanish.
The outcomes of interest were responses to questions about surgeon and hospital selection (eTable in the Supplement). Women identified all response(s) that applied to their selection of surgeon and hospital. Women replying “other” were asked to specify further. We reviewed each “other” response and recategorized them as appropriate. For example, “primary care provider chose surgeon” was recategorized as “referred to surgeon by another doctor”; responses for “Kaiser selected surgeon” and “surgeon assigned through health plan” were categorized as “surgeon was one of the only ones available through my health care plan.” Responses suggesting active investigation into a surgeon’s reputation were recategorized as “chose surgeon based on reputation” (eg, “researched him,” “second opinion”). Lastly, we created a new response category for the 14 women who reported selecting their surgeon because they had received prior treatment with that surgeon. Two “other” responses were not recategorized: “Spanish speaker” and “referral sent out to various surgeons—was the first to return call.” For hospital selection, the most common “other” responses reflected the need to go to a hospital within their health plan, and we added a category for this (n = 292); we also added a category for the 13 patients who reported prior treatment at that hospital.
Our independent variable of interest was self-reported race/ethnicity. Because of the potential impact of language spoken on surgeon/hospital selection, we further categorized the 136 Hispanic participants based on responses to the question “In general, what language(s) do you read and speak?”28 Women who responded “only Spanish” or “more Spanish than English” were categorized as Spanish-speaking Hispanic (n = 47), and women who responded “both equally,” “more English than Spanish,” or “only English” were categorized as English-speaking Hispanic (n = 89).
Control variables included age, marital status, insurance status at time of diagnosis, disease stage (using registry data), number of self-reported comorbidities25,27 (ie, past diagnosis of another cancer, diabetes, heart disease, stroke, chronic lung disease, kidney problem, depression/psychiatric problems), educational attainment, and mean health literacy score.26 We assessed health literacy using a 3-item screening tool26: (1) “How confident are you filling out medical forms”; (2) “How often do you have problems learning about your medical condition?”; and (3) “How often do you have someone help you read hospital materials?” Responses used a 5-item Likert scale. After reversing responses for the first item, we assigned each answer a score of 1 to 5 (lower numbers reflect most confidence/fewest problems) and averaged the 3 scores. One participant did not answer question 2; we averaged her 2 other responses. Variables were categorized as in Table 1. We also examined the time from diagnosis to survey administration (median. 2.8 years; interquartile range, 2.1-2.7 years); this variable was not significantly associated with any responses for surgeon or hospital selection and was not included in multivariable models.
We used χ2 tests to assess racial/ethnic differences in baseline characteristics and reasons for surgeon and hospital selection, the Fisher exact test to examine differences in insurance at diagnosis, and Kruskal-Wallis tests to assess differences in mean health literacy scores. We used multivariable logistic regression to assess the probability of providing each possible response to the choice of surgeon or hospital (eTable in the Supplement) by race/ethnicity after recategorizing “other” responses, adjusting for age, marital status, insurance, stage, education, mean health literacy score, and comorbidity. We calculated the adjusted proportion of patients’ reporting each reason for selection of surgeon or hospital by race/ethnicity, adjusting for covariates, based on the regression model.30 For the models examining “referred by a friend or family member” to the surgeon and hospital, we omitted the insurance variable because no uninsured participants selected these reasons. Because of small numbers in some education categories for those reporting referral to the hospital by a relative or friend, we created a binary variable for education (college graduate vs non–college graduate) for this model.
Finally, we described patients’ priorities for hospital location vs reputation, ratings of surgical and hospital care, and likelihood of recommending their surgeon and/or hospital to family and friends by race/ethnicity.21 Using a second set of multivariable logistic regression models, we assessed whether (1) reported reasons for selecting physicians were associated with excellent ratings of surgeons and (2) reported reasons for selecting hospitals were associated with excellent ratings of hospital care. We performed separate models for each reason endorsed by at least 10% of women, adjusting for race/ethnicity, age, comorbidity, mean health literacy score, stage, and education. Other than the literacy item and 1 woman who did not rate hospital care quality, there was no item nonresponse.
Baseline characteristics for the 500 participants (222 white, 142 black, 89 English-speaking Hispanic, and 47 Spanish-speaking Hispanic) are given in Table 1. In general, Hispanic women were younger and had poorer mean health literacy scores compared with others and black women were less likely to be married than other groups. Spanish-speaking Hispanic women had the highest rates of no insurance at breast cancer diagnosis and the lowest educational attainment.
Of 7 response options for selection of their surgeon, most women selected 1 (56%) or 2 (31%) responses. Approximately 81% of white women, 93% of black women, 88% of English-speaking Hispanic women, and 92% of Spanish-speaking Hispanic women selected 1 or 2 responses, with the remaining participants selecting more than 2 reasons for selecting surgeons. Rates for each response are displayed in the Figure, A, and stratified by race/ethnicity in Table 2. Most patients (78%) reported being referred by another physician. Approximately 25% of women selected their surgeon based on reputation. Less than one-quarter of women provided other response options for surgeon selection.
In adjusted analyses (Table 2), black (87%) and Spanish-speaking Hispanic (79%) women had higher adjusted rates of selecting their surgeon based on physician referral than white (76%) and English-speaking Hispanic (67%) women (P = .007). Black (18%), Spanish-speaking (22%), and English-speaking Hispanic (22%) women were less likely to report selecting their surgeons based on reputation than white women (32%) (P = .02) and to select surgeons based on recommendations by relatives or friends (3% to 14% for minority women vs 17% for white women) (P = .007).
Of 7 response options for hospital selection for their breast cancer surgery, most women selected 1 (53%) or 2 (32%) reasons. Approximately 82% of white, 89% black, 84% of English-speaking Hispanic, and 94% of Spanish-speaking Hispanic women selected 1 or 2 responses, with the remaining women selecting more than 2 responses. Rates for each response are displayed in the Figure, B, and stratified by race/ethnicity in Table 2, with the most frequent reason for hospital selection that the hospital was available through the patient’s health plan (58%).
In adjusted analyses, white women (23%) were more likely than black (8%), English-speaking Hispanic (15%), and Spanish-speaking Hispanic (7%) women to report choosing their hospital because of its reputation (P = .003) (Table 2). White women (48.9%) were also more likely than minority women (17% to 38%) to report selecting the hospital because they wanted to be treated at the hospital where their physician worked (P = .001). Spanish-speaking Hispanic (78%), English-speaking Hispanic (64%), and black (61%) women were more likely than white women (51%) to report that they chose the hospital because it was a part of their health plan (P = .02). There were no racial/ethnic differences in reporting being referred to hospitals by their physicians. All models suggested good fit except the model assessing the outcome of choosing a hospital because “it was near my home” (Hosmer-Lemeshow goodness-of-fit test, P = .02).
Most participants (64%) stated that they had other hospitals in their area to choose from for surgery (Table 3), and 44% reported that there were hospitals closer to their home than the one they selected. Overall, 51% of women reported that the hospital’s reputation was more important than location and reputation was more important for white women (60%) than black (45%), English-speaking Hispanic (55%), and Spanish-speaking Hispanic (23%) women (P < .001).
Overall, most women, regardless of race/ethnicity, reported excellent ratings of the quality of care delivered by their surgeon (77%) and hospital (63%), and most women stated that they would be extremely likely to recommend their surgeon or hospital to family members or friends (68% and 56%, respectively) (Table 3). Approximately 21% of women reported that their hospital was the “best” compared with others in the area.
In adjusted analyses examining associations of patients’ reasons for selecting their surgeon with excellent ratings of care for the surgeon (Table 4), choosing a surgeon based on reputation was significantly associated with higher odds of reporting excellent care from their surgeon (adjusted odds ratio, 2.21; 95% CI, 1.24-3.93), while choosing a surgeon because of one’s health plan was associated with lower odds of reporting excellent care from the surgeon (adjusted odds ratio, 0.56; 95% CI, 0.34-0.91) (Table 4). No other responses for the selection of surgeons or hospitals were significantly associated with excellent ratings, and race/ethnicity was not significantly associated with reporting excellent ratings for surgeon or hospital in any of the models.
In a large, diverse cohort of women with breast cancer, we observed variability in how women selected their surgeons and hospitals for their breast cancer–directed surgery. Most women relied on referrals from their physicians for selecting surgeons, particularly black women and Spanish-speaking Hispanic women. In addition, minority patients were less likely to report reputation as an important component of their decisions about surgeons and hospitals and were more likely to select a hospital because it was part of their health plan. These findings suggest less-active involvement of minority patients with regard to selecting physicians and hospitals for their care.
Prior studies have documented a more passive approach to medical care and decision making for black and Hispanic vs white women, particularly for those who do not speak English.31-33 These studies have observed less-active participation by minority and non-English speakers, with more reliance on physicians to make decisions and to provide information about health conditions.31-33 This decision-making approach may be even more evident when a patient is anxious, vulnerable, and overwhelmed in the setting of a new breast cancer diagnosis or if a patient is not proficient in English. In our study, we could not assess whether a less-active approach to physician and hospital selection led to worse quality of cancer care; however, women who selected the surgeon based on reputation rated the quality of care delivered by their surgeons more highly than others, and women who were directed to their surgeon via their health plan rated the quality of care from their surgeon lower. Although more research is needed to fully understand the impact of patient engagement in physician and hospital selection on quality of care, outcomes, and patient experiences, interventions that promote more active patient involvement in this process may have potential for addressing disparities related to lower-quality care from surgeons and hospitals.
Alternatively, since most patients in our cohort relied on referrals from physicians when selecting breast surgeons and hospitals, interventions could instead be targeted to referring physicians to ensure referrals to provider systems that deliver coordinated and high-quality cancer care. Surgeons are often the first cancer specialist a woman with a new diagnosis of breast cancer will encounter and thus may play a major role in securing high-quality medical oncology and radiation oncology referrals. Thus, an initial referral to a high-quality surgeon and hospital may lead to high-quality care throughout the cancer care continuum. However, we need more research on the potential associations of surgeons and other cancer care clinicians and how preexisting patterns34-36 of referrals could be optimally modified.
Although location was important to some women in selecting their surgeon (10%) and hospital (20%) in our study, most women reported other hospitals in their area where they could have had their surgery, and nearly half of the women reported that there were other hospitals closer to their home. Other evidence suggests that, particularly in areas with high levels of racial segregation, black patients are more likely than white patients to undergo major surgical procedures at lower-quality hospitals despite living nearer to higher-quality hospitals.37 It is likely that factors such as comfort with receiving care in certain hospitals or physician referral patterns play a role in these decisions.
We recognize several study limitations. First, although population based, we studied women in northern California, where a relatively high proportion of women are insured by Kaiser Permanente.38 This may explain the large number of patients reporting that their health plan influenced selection. Second, we cannot rule out nonresponse bias, although responders were similar to nonresponders. Third, recall bias may have affected findings; however, the time to survey administration was not associated with responses for surgeon and hospital selection. Fourth, we lacked information about the sources (eg, other clinicians, family members) patients used to classify surgeons and hospitals as “reputable” when they provided this response; we also did not ask about women’s employment status. Finally, we did not ask women to provide a single most important reason for their choice of surgeons and hospitals, which may have enabled us to better identify the most important factors. However, the most women only selected 1 or 2 responses, and women also provided information on the relative importance of hospital reputation vs location.
Among a diverse group of patients with breast cancer, we observed differences by race/ethnicity in the level of involvement in selecting surgeons and hospitals, with white women more likely than minority women to select surgeons and hospitals based on reputation and less likely to be directed to surgeons and hospitals by their health plan or other physicians. These findings may explain some of the segregation in care that has been observed in other studies.4,5,7,8,15-17 More research is needed on how these selection strategies (or lack thereof) affect the quality of care patients receive and how we can best intervene to guarantee excellent and equal care for all patients with breast cancer. Assuring thoughtful decision making for provider and hospital selection may be an important element in addressing treatment disparities in cancer and has the potential to be an important focus for interventions.
Accepted for Publication: January 2, 2015.
Corresponding Author: Rachel A. Freedman, MD, MPH, Dana-Farber Cancer Institute, 450 Brookline Ave, Boston, MA 02215 (email@example.com).
Published Online: March 19, 2015. doi:10.1001/jamaoncol.2015.20.
Author Contributions: Dr Freedman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Freedman, Kouri, Keating.
Critical revision of the manuscript for important intellectual content: Freedman, West, Keating.
Statistical analysis: Freedman, Kouri, Keating.
Obtained funding: West, Keating.
Administrative, technical, or material support: Freedman, Kouri, West.
Study supervision: Freedman, West, Keating.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was sponsored by the Komen for the Cure Foundation. Dr Keating is also supported by K24CA181510 from the National Cancer Institute. The collection of cancer incidence data used in this study was supported by the California Department of Public Health as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885; the National Cancer Institute’s Surveillance, Epidemiology and End Results Program under contract HHSN261201000140C awarded to the Cancer Prevention Institute of California, contract HHSN261201000035C awarded to the University of Southern California, and contract HHSN261201000034C awarded to the Public Health Institute; and the Centers for Disease Control and Prevention’s National Program of Cancer Registries, under agreement U58DP003862-01 awarded to the California Department of Public Health.
Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The ideas and opinions expressed herein are those of the authors and endorsement by the State of California, Department of Public Health, the National Cancer Institute, and the Centers for Disease Control and Prevention or their contractors and subcontractors is not intended nor should be inferred.
Additional Contributions: We thank all women who participated in interviews, the Cancer Registry of Greater California, Ana Guerrero, BS, for assistance with interviews, and Joyce Lii, MA, MS, Harvard Medical School, for her programming expertise. Mss Guerrero and Lii received financial compensation for their contributions.
Create a personal account or sign in to: