Can this be done 8 pm 4/22
Saltwata2012
factors.pdf
Are racial differences in patient-physician cancer communication and information explained by background, predisposing, and enabling factors?
Clara Manfredi, Ph.D**, Karen Kaiser, Ph.D, Alicia K. Matthews, Ph.D., and Timothy P. Johnson, Ph.D
Abstract Research shows that African Americans tend to have poorer and less informative patient-physician communication than Whites. We analyzed survey data from 248 African American and 244 White cancer patients to examine whether this disadvantage could be explained by race variability on several other variables commonly reported to affect communication. These variables were organized into background, enabling, and predisposing factors, based on the Precede-Proceed Model. Multivariate regressions were used to test whether race differences on communication and information variables persisted after successively controlling for background, enabling and predisposing factors. African American patients had higher interpersonal communication barriers than Whites, but this difference did not persist after controlling for background factors. African Americans also had higher unmet information needs and were less likely to receive the name of a cancer expert. These differences persisted controlling for all other factors. Future research should focus on the informational disadvantages of African American patients and how such disadvantages may affect cancer treatment decisions.
Introduction African Americans have poorer survival rates than Whites for almost all cancer sites even after controlling for stage at diagnosis (Ries et al., 2007). This disparity has been attributed to race differences in income, education, medical insurance and co-morbidity, as well as to differences in cancer treatment modalities (Bach et al., 2002). Some authors have argued that poor patient- physician communication may also be a contributing factor (Ashton et al., 2003; Cooper, Beach, Johnson & Inui, 2006; Gordon, Street, Sharf, Kelly & Souchel, 2006a). Poor communication could affect health outcomes through added barriers to coping with the disease and obtaining adequate information on which to base treatment decisions. Several literature reviews concluded that good patient communication and information are associated with reduced patient anxiety and greater sense of control, participation in medical decisions, and compliance with medical directives (Baile & Aaron, 2005; Bech, Daughtridge & Sloane, 2002; Mills & Sullivan, 1999). In other studies, patient information and participation were associated with better asthma management (Diette & Rand, 2007), better coping and adjustment to breast cancer (Rees & Bath, 2001) and a more active role in making decisions about prostate cancer treatment (Davison & Degner, 1997). Two studies have found that the quality of patient-physician communication affected breast cancer treatment (Liang et al. 2002) and end-of-life care (Zapka et al., 2006).
**Corresponding author: Clara Manfredi, Ph.D., [email protected], Professor Emerita, Program for Cancer Control and Population Science, Institute for Health Research and Policy, University of Illinois at Chicago, 1747 West Roosevelt Road, Room 558, (M/C 275), Chicago, IL, 60608-1264. Tel. (312) 996-2428, Fax (312) 996-0065.
NIH Public Access Author Manuscript J Health Commun. Author manuscript; available in PMC 2011 April 1.
Published in final edited form as: J Health Commun. 2010 April ; 15(3): 272–292. doi:10.1080/10810731003686598.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Research also indicates that African American patients tend to be less satisfied with their physician communication (Cooper-Patrick et al., 1999; Gordon et al., 2006a; Johnson, Saha, Arbelaez, Beach & Cooper, 2004; Saha, Arbelaez & Cooper, 2003) and report receiving less informative communication from physicians (Ayanian et al., 2005; Gordon et al., 2006a; Maly, Leake & Silliman, 2003; Zapka et al., 2006) than White patients. Similarly, observational studies of medical encounters found that patient-physician communication was poorer for African American than for White patients (Gordon et al., 2006b; Gordon, Street, Kelly, Soucheck & Wray, 2005; Johnson, Roter, Powe & Cooper, 2004; Levinson et al., 2008; Street, Gordon & Haidet, 2007; Street, Gordon, Ward, Krupat & Kravitz, 2005). However, other factors besides race influence patient-physician communication, such as patient preferences and demographic characteristics. Studies have found that cancer patients vary in the extent and nature of the information they desire about their disease (Czaja, Manfredi & Pierce, 2003; Degner et al., 1997; Mills and Sullivan, 1999; Finney-Rutten, Arora, Bakos, Aziz & Rowland, 2005) and in their ability to obtain information from physicians (Davison & Degner, 1997; Maly, Stein, Umezawa, Leake & Anglin, 2008; Rees & Bath, 2001; Siminoff, Graham & Gordon, 2006). In these studies, personal preferences and information acquisition were associated with demographic characteristics, such as education and income levels or marital and medical insurance status, that also vary by race (U.S. Census Bureau, 2003).
Studies examining whether the communication disadvantages of African American patients persist after controlling for demographic characteristics and other patient or physician factors have produced mixed results. Some studies show persistent race effects (e.g., Cooper et al., 1999; Levinson et al., 2008; Zapka et al., 2006), others find that race effects disappear when controlling for other factors (e.g., Gordon et al., 2005; 2006b), and some find a combination of persistent and vanishing effects (e.g., Johnson et al., 2004a, 2004b; Street et al., 2007). Moreover, drawing conclusions from these studies is difficult because of different methods, different measures of communication, and disparate arrays of control variables. More information is needed to better understand what accounts for the effect of race on patient- physician communication.
We address this weakness of the literature by organizing several factors commonly assumed to influence patient-physician communication within an established conceptual framework. We analyzed findings from a survey of African American and White cancer patients to determine: (a) whether race differences on patient-physician communication and information outcomes could be explained by variability on these other factors, and (b) whether race differences on information outcomes could be explained by additionally controlling for the communication variables.
Conceptual framework Factors influencing patient-physician communication
We used the Precede-Proceed Model (Greene & Kreuter, 1999) to organize the various correlates of patient-physician communication cited in the empirical literature. This broadly defined model has been a useful heuristic tool for studying a variety of behavioral outcomes (Carlsson, Gielen, & McDonald, 2002) including cancer patient information seeking (Czaja, Manfredi & Pierce, 2003). In the model, the effect of background factors on a given illness behavior (such as seeking information from physicians) is mediated by enabling and predisposing factors. The experiences incurred while engaging or trying to engage in the behavior can result in either positive or negative reinforcement of the behavior. The following is a brief summary of the factors examined in this study.
Background factors that could affect patient-physician communication besides patient demographic characteristics include hospital and disease characteristics. Type of hospital
Manfredi et al. Page 2
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
where a patient is diagnosed may influence the perceived need for additional information or consultations. Disease characteristics have included cancer type (Czaja et al., 2003; Mills & Sullivan, 1999), cancer stage, being currently in treatment (Rees & Bath, 2000; Squires, Rutten, Treiman, Bright & Hesse, 2005) and months elapsed since diagnosis (Vogel, Bengel & Helmes, 2008).
Enabling factors are those that affect a patient’s ability to carry out the behavior in question. Factors that may facilitate patient-physician communication include having medical insurance (Stepanikova & Cook, 2008) and having a companion present during medical visits (Czaja et al., 2003; Gordon et al., 2006a). The active involvement of one’s regular physician could also facilitate communication or obtaining and processing information from one’s cancer physicians.
Predisposing factors include beliefs and preferences that predispose patients to engage in a given behavior, including seeking and processing information. Two such factors are individual preferences regarding informed participation in treatment decisions and trust in physician knowledge about treatment options. In cancer patients, these factors have been associated with patients actively seeking information from other sources (Degner et al., 1997; Czaja et al., 2003) and discussing the information they found with their physicians (Manfredi, Czaja, Buis & Derk, 1993).
Reinforcing factors include the experiences that patients have when engaging in a given behavior. Regarding patient-physician interactions, the communication experiences that patients have during some medical encounters can influence how patients will approach communication in subsequent encounters. For example, research shows that patients who felt that physicians reacted favorably to their asking questions and discussing issues were more likely to seek additional information from other sources and later discuss that information with the physicians (Czaja et al., 2003; Reese & Bath, 2000). These reinforcing effects are particularly relevant for cancer patients, whose diagnosis and treatment often require contacts with multiple physicians and hospitals.
Communication variables and information outcomes Patient-physician communication has been categorized on two main dimensions: (a) interpersonal (relationship building) and (b) instrumental (mutual information exchange) (Ashton et al., 2003; Ong, deHaes, Hoos & Lammes, 1995). Interpersonal communication refers to qualitative aspects of patient-physician interaction, such as physician supportiveness or respectfulness (Gordon et al., 2006a), relationship building (Levinson et al., 2008) or patient- centeredness (Johnson et al., 2004b). Instrumental communication refers to the mutual exchange of information between the doctor and patient, such as patients describing their symptoms and concerns, and physicians explaining a diagnosis or addressing the patient’s concerns. These two communication dimensions are thought to shape the specific information patients obtain to enable them to make treatment and other disease-coping decisions (Ashton et al., 2003; Ong et al., 1995).
Study model and questions Our study contains multiple variables measuring both the interpersonal and instrumental dimensions of communication. These variables measured the experiences that patients had across all physicians seen since their cancer diagnosis, reflecting potential reinforcing effects across multiple visits. We also assessed whether patients received specific information, including referrals for a second opinion and information about clinical trials, cancer experts, and specialized cancer centers. As such, we comprehensively examine the communication experiences and receipt of information in a sample of cancer patients.
Manfredi et al. Page 3
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Figure 1 shows our study model. The model illustrates the conceptual relationships among the sets of variables that were empirically examined and does not imply causal pathways, which cannot be determined from cross-sectional survey data. Based on our model, the study addressed the following questions. 1) How do African American and White cancer patients differ on each of the study variables? 2) Are any observed race differences in the communication variables explained by race variability in background characteristics, enabling or predisposing factors? 3) Are any observed differences in information outcomes explained by race variability in background characteristics, enabling and predisposing factors, and communication variables?
Methods Sample design and data collection
Data for this paper come from a survey conducted by the authors to investigate disease experiences and quality of life among cancer patients. The study sample included 248 African American and 244 White patients diagnosed with breast, prostate, or colorectal cancer 1–2 years prior to being interviewed. To obtain a sample representative of African American cancer patients in Illinois, the sample included cases from all 79 Illinois hospitals with cancer registries and located in the 15 Illinois counties with at least 10 reported African American cancer cases (any site) per year. Included were three other hospitals located in bordering states but known to serve Illinois residents. The White sample was identified from the same hospital pool and equally stratified on hospital characteristics, cancer site, gender, age (below or above age 60), and months elapsed since diagnosis. To increase representation of non-metropolitan patients, the sample was stratified to include 30% of cases from hospitals located outside of Cook County (which includes Chicago and the majority of African Americans in Illinois).
A total of 753 patients meeting the above criteria were identified through 33 hospital registries and the Illinois State Cancer Registry. Upon telephone contact, 149 cases were found to be ineligible (more than 30 months since diagnosis, other cancer site, deceased, non-English speaking). Of the qualifying cases, 509 (80%) completed the interview. Seventeen additional cases were lost due to missing information. All interviews were conducted by telephone by the Survey Research Laboratory (SRL) at the University of Illinois at Chicago and lasted approximately one hour. There was no attempt to achieve race concordance between patients and interviewers; however, SRL interviewers are carefully trained and supervised and are experienced in surveying diverse populations. Respondents received $30.00 as compensation. The institutional review boards of the University of Chicago and the University of Illinois at Chicago approved all study protocols.
Study Measures Independent variables (See list in Table 1)—Background variables included hospital characteristics (urban, suburban, or rural location; bed size; presence of a cancer center or cancer program; teaching hospital), disease characteristics (cancer site; months elapsed since diagnosis; currently in treatment); co-morbidity (coded yes if a patient reported having any chronic condition such as diabetes, hypertension, etc.); and patient demographic characteristics (gender, age, education, income, marital status). Enabling variables included having any medical insurance, whether a companion was usually present at medical visits (defined as at least 75% of the visits), and whether a patient’s regular physician remained involved in the cancer treatment after the cancer diagnosis.
Predisposing variables included patient trust in physician knowledge and desire to be involved in treatment decisions. Trust in physician knowledge was a patient’s level of agreement on a 4-point scale with the following statement: “Most doctors find out about the most current cancer
Manfredi et al. Page 4
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
treatment before making decisions.” A scale measuring desire to be involved in treatment decisions was available from another study of cancer patients (Manfredi et al., 1993). The scale is constructed from 4-point level of agreement with five statements: (a) doctors should make completely clear to patients the risks of treatment or operations, (b) patients should always get a second opinion before starting any treatment plans, (c) it is all right for patients to ask their physicians to consult with known cancer experts, (d) it is the patient’s responsibility to learn as much as possible about his or her disease and treatment options, and (e) patients should trust their doctors and do what they say without asking a lot of questions. A higher final score means stronger desire to be involved. In our data the scale had a Cronbach Alpha reliability coefficient of .59.
Dependent variables—There were six communication variables. Three of these were assessed for each physician a patient reported seeing since the cancer diagnosis. (In this sample the number of physicians ranged from 2 to 8). For each patient, two of these variables were the percent of physicians he or she had seen who a) discussed the diagnosis and treatment with the patient (range from 0.00=no physician discussed to 1.00=all physicians discussed), and b) to whom the patient asked questions about the cancer diagnosis or treatment (from 0.00=did not ask of any physician to 1.00=asked of all physicians). The third variable was a 4-point scale assessing how comfortable the patient was when asking questions (from 1.00=very uncomfortable with all physicians to 4.00=very comfortable with all physicians).
The three other communication variables were assessed with reference to all physicians seen (i.e.: “thinking of all physicians you have seen since your cancer diagnosis….”). The variable ‘interpersonal communication barriers’ was constructed from the 4-item medical interactions subscale of the Cancer Rehabilitation Evaluation Form–Short Form (Coscarelli, Ganz & Heinrich, 1991). These items assessed how often (never, occasionally, often, very often) patients (a) felt that doctors did not explain what they were doing to them, (b) had difficulty expressing their feelings to doctors, (c) had difficulty telling their doctors about new symptoms, and (d) felt they needed more control over what the doctors were doing to them. Higher scores indicate more frequent experiences of communication barriers. The scale has a Cronbach Alpha coefficient of .62. The scale “unmet information needs” was obtained from another study of cancer patients (Manfredi et al., 1993b). It was constructed from four 4-point items indicating how often (never, occasionally, often, very often) patients: (a) had experienced difficulty understanding what the doctors told them about their cancer or treatment, (b) felt they needed more information about their illness or (c) about their treatment, and (d) felt that doctors had discussed all available treatment options. Higher scores indicate greater unmet information needs. The Cronbach’s Alpha coefficient was .76. A single question assessed overall information satisfaction (1=very dissatisfied to 4=very satisfied).
In summary, the six communication variables included two variables measuring interpersonal communication (comfort asking physician questions and interpersonal communication barriers) and four variables measuring instrumental communication (physicians discussed diagnosis and treatment, patient asked questions, unmet information needs, and overall satisfaction with information). Finally, four information outcomes were coded ‘yes’ if a patient reported receiving from physicians 1) a referral for a second opinion, 2) the name of a cancer expert, 3) the name of a specialized cancer center, and 4) clinical trials information.
Data analysis Comparisons of the survey results for African American and White patients used Chi-square tests for categorical variables and t-tests for continuous variables. Multivariate (for continuous dependent variables) or logistic (for dichotomous dependent variables) regression models were used to examine race differences on each of the six communication variables and information
Manfredi et al. Page 5
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
outcomes, controlling for other factors. In developing our regression models, we first considered the communication variables as dependent variables. Each communication variable was regressed against race alone (Model 1), race and the background variables (Model 2), and race, background, predisposing, and enabling variables (Model 3). Next, we considered the information outcomes as dependent variables, using Models 1–3 as above and then adding the communication variables in a final regression model (Model 4). Data were analyzed using Stata/SE version 9.2 (StataCorp, 2005).
Results Race differences in survey variables
Table 1 shows the survey findings by race. Despite the sampling procedures, African American patients were more likely to be diagnosed in urban hospitals and less likely to be diagnosed in rural hospitals and to have colon cancer than White patients. African Americans were also overall younger, had lower education and lower income, were less likely to be married, have medical insurance, or to have a companion during medical visits, and were more likely to have co-morbidity and a regular physician involved in their cancer treatment. There were no differences by race on desire to be involved and trust in physician knowledge. African Americans reported significantly more interpersonal communication barriers and unmet information needs and had lower satisfaction with the information received from physicians. However, African Americans were more likely to ask questions of physicians and as likely to be comfortable asking questions. The only difference by race on the information outcomes was that African Americans were less likely than Whites to have received the name of a cancer expert.
Multivariate regressions Preliminary analysis (not shown) indicated that five independent variables were not significantly associated with any communication or information variables and/or had high collinearity with other variables. These variables were: hospital bed size, teaching hospital, months since cancer diagnosis, currently in treatment, and patient marital status. These variables were not included in the final multivariate regressions.
Table 2 shows the regression results for each communication variable. African Americans remained as likely as Whites to report that physicians had discussed diagnosis and treatment and to be comfortable asking questions after controlling for all background, enabling and predisposing factors (Models 2 and 3). The association of African American race with having more interpersonal communication barriers did not persist after controlling for background factors (Model 2), apparently explained by the lower income of African Americans.
African Americans remained more likely than Whites to have asked questions of physicians and to have more unmet information needs even after controlling for all background, enabling and predisposing factors. In Table 2, the regression coefficients for race under these two communication variables were essentially unchanged across Models 1, 2, and 3. Patients who asked questions were also younger, had higher income and a greater desire to be involved in treatment decisions and were more likely to have a companion at medical visits and a regular physician involved in the cancer treatment. Uninsured patients were also more likely to ask questions. This unexpected finding could be explained by uninsured patients having additional questions related to cost and access to cancer treatment. Patients with higher unmet information needs had lower trust in physician knowledge.
The association of race with overall information satisfaction persisted controlling for background factors (Model 2), but not when additionally controlling for predisposing and
Manfredi et al. Page 6
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
enabling factors (Model 3). In Model 3, overall satisfaction was associated with trust in physician knowledge and desire to be involved. Since these two variables did not vary by race, we repeated Model 3 with interaction terms for these variables and race. The interaction terms were not significant (African American X Trust: B. −0.03, S.E. 0.09, p. = .71; African American X Desire: B. −.199, S.E. 0.15, p. = .17, data not shown in table).
Table 3 shows the regression results for the information outcomes. After controlling for all background, enabling and predisposing factors, African Americans emerged as more likely than Whites to receive a referral for a second opinion (Model 3). Patients who received this information had higher income, a higher desire to be involved and were more likely to have a companion during visits and an involved regular physician. The association of race and receiving a second opinion referral did not persist controlling for communication variables (Model 4). The change could be explained by the African American patients’ higher propensity to ask questions, since having asked question was the only significant communication variable in this regression.
African Americans remained less likely than Whites to receive the name of a cancer expert after controlling for other variables. The regression coefficients for this association remained essentially unchanged from Model 1 through Model 4. Patients who received the name of a cancer expert were also more likely to be diagnosed in a hospital with a cancer center and to have breast cancer, and to have higher income, a higher desire to be involved, and a regular physician involved in their cancer treatment.
African Americans emerged as less likely than Whites to receive the name of a cancer center after controlling for background variables (Model 2). This association persisted controlling for predisposing and enabling factors (Model 3), but was reduced to marginal (p. = .06) non- significance after controlling for communication variables (Model 4). The reason for this change is unclear since none of the communication variables was significantly associated with receiving the name of a cancer center. Finally, race was not associated with receiving clinical trial information in any of the regression models.
Notably, regular physician involvement increased the likelihood of receiving the names of cancer experts or cancer centers. African Americans were less likely to receive this information but more likely to report regular physician involvement than Whites. To explore possible interaction effects, we repeated Model 4 for these information outcomes with interaction terms for race and regular physician involvement added. The interaction terms were not significant. The Beta coefficients for the interaction terms were B. 0.152, S.E. 0.454, p. = .74 for race X cancer expert name and B. −0.222, S.E. 0.470, p. = .64 for race X cancer center name (data not shown in table).
Discussion Good patient-physician communication is essential for cancer patients, given that cancer requires extensive contact with the health care system and the importance of making informed treatment decisions. Prior studies, reviewed in the introduction, found that such communication is poorer for African Americans than for Whites. We examined whether this disadvantage could be accounted for by race variability on other factors commonly associated with patient- physician communication.
Our first study question was how African American and White patients differed on the study variables. The survey found the expected race differences on background factors and medical insurance, but also had new findings. Compared to White patients, African American patients were less likely to have a companion present at medical visits and more likely to have their regular physicians involved in the cancer treatment. African Americans and Whites had similar
Manfredi et al. Page 7
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
levels of desire to be involved in treatment decisions and trust in physician knowledge. Little information is currently available in the literature about these four variables among African Americans. Thus, these findings contribute to our understanding of key racial differences and similarities. Finally, a surprising survey finding was how few patients in either race group received a referral for a second opinion or information about cancer experts and specialized cancer centers. Seeking a second opinion and consultation with cancer experts are steps recommended by current cancer care standards (ASCO-ESMO, 2006). However, the ability to do so may depend on physician information and referral.
As expected, African American patients in this study reported poorer experiences than White patients on several communication and information variables. Our second study question was whether these differences would be explained controlling for background, enabling, and predisposing factors. The findings indicated that this was clearly the case only for one of the interpersonal communication variables; background factors, particularly income, appeared to explain why African American patients had more interpersonal communication barriers than White patients. This finding is consistent with another study in which a race difference on perceived physician cultural bias did not persist controlling for multiple variables that included income (Johnson et al., 2004). Conversely, a race difference on participatory physician style persisted in another study whose multiple control variables did not include income (Cooper- Patrick et al., 1999). Income is a strong indicator of socioeconomic status, which has consistently been found to be associated with poor patient-physician communication (Willems et al., 2005). It may be that low patient socioeconomic status is a stronger barrier to establishing good interpersonal communication with physicians than race.
In contrast, the disadvantage of African American patients on instrumental communication and information outcomes remained unexplained. The differences by race on unmet information needs and receiving the name of a cancer expert or a specialized cancer center persisted or emerged controlling for all background, enabling and predisposing factors. This is consistent with two other studies based on patient interviews. These studies found that race differences in receiving information about cancer treatment (Ayanian et al., 2005) and end-of-life pain and symptom management (Zapka et al., 2006) persisted controlling for patient demographic characteristics and several disease and physician factors. However, the negative association of African American race and overall information satisfaction did not persist controlling for predisposing and enabling factors.
Our third question was whether race variability on communication variables could account for race differences in receiving specific information. The findings indicated that this was not generally the case. Controlling for communication variables did not diminish the negative association between being African American and receiving the names of cancer experts and only marginally reduced the association of race and receiving the name of a specialized cancer center. Other authors have suggested that African Americans may receive less information from their physicians because they less often engage in communication behavior that would typically elicit additional information, such as asking questions or expressing interest (Gordon et al., 2006b). However, we found that African American patients had the same desire to participate in treatment decisions and were more likely to ask questions of their physicians than White patients. Yet, they had overall poorer information outcomes. These discrepant findings could be due to different study methods (i.e., observation versus patient reports). African American patients may have felt they had asked questions and shown desire to be involved, but in actuality could have formulated the questions and expressed the interest in ways that were not well-conveyed to their physicians.
In considering the above findings it is important to note that all the study variables combined, including race, accounted for little variance in the communication and information variables.
Manfredi et al. Page 8
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
The strongest model R-square was .16. It is unlikely that this limited explanatory power was due to inappropriate choice of independent variables. These variables are commonly cited in the literature as correlates of patient-physician communication and they generally behaved as expected in the regression models. We tested and ruled out interaction effects of race and independent variables where the data suggested that such effects could be present. A more likely explanation is that the study model and variables explain only the portion of variance that depends on the patient’s end of the interaction. A greater portion of patient-physician communication and information may depend on other factors. For example, several observational studies found that physician communication with African American cancer patients was affected by physician’s behavior and biased perceptions (Gordon et al., 2006b; Gordon et al., 2005; Levinson et al., 2008; Street et al., 2005). Another study indicated that systemic-level factors may account for race disparities in patient information. Compared to physicians who treated mostly White patients, physicians who treated mostly African American patients had greater difficulties in obtaining access for their patients to hospitals, high-quality sub-specialists, and high-quality diagnosing procedures (Bach, Pham, Schrag, Tate & Hargraves, 2004).
In summary, our findings suggest that race disadvantages in interpersonal dimensions of patient-physician communication are largely explained by race variability on income. However, being African American appears to be uniquely associated with disadvantages in instrumental communication dimensions and in receiving information about cancer experts and specialized cancer centers. Receiving this information could be an important link in connecting patient-physician communication to race disparities in cancer treatment and survival. However, improving information outcomes for African American cancer patients may require addressing systemic barriers besides improving the quality of interpersonal communication.
Study strengths and limitations The study had several strengths. We utilized a representative sample of Illinois African American cancer patients, including cases from all Illinois hospitals reporting such cases. A comparable sample of White cancer patients came from the same hospital pool and was equally stratified on important hospital characteristics. This is important because hospital at diagnosis may affect the perceived need for additional medical opinions and referral to cancer expertise. Other strengths were the study conceptual framework and design, which allowed examining whether race disparity in patient-physician communication involved different explanations depending on different communication dimensions.
The study also had several limitations. Identification of population-based cancer patient samples requires working with cancer registries, which are subject to stringent subject protection regulations (Beskow, Sandler & Weinbergerer, 2006). Constrains inherent to these regulations could have biased in unknown ways the sample that the registries released to the study. Our White sample came from the 15 counties with African American cases; comparison by race might have produced different results if we had selected a sample more representative of all Illinois White cancer patients. Given the importance of where patients are treated, we chose comparability on hospitals at diagnosis over greater generalizability of the White sample. Another limitation was the relatively low reliability of the scales measuring patient desire to be involved and interpersonal communication barriers. However, both scales were previously published measures. Although low, their Alpha reliability with our data was still within the range deemed sufficient for exploratory research (Pedhazur & Schmelkin, 1991).
Manfredi et al. Page 9
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Conclusion The African American cancer patients in our survey reported poorer patient-physician communication and information outcomes than White cancer patients. The study results suggest that explanations for these race disparities differ depending on communication dimensions and information outcomes. The higher interpersonal communication barriers of African American patients were explained by background factors, mainly race variability on income. Their higher unmet information needs and poorer information outcomes, on the other hand, remained unexplained. These disadvantages persisted controlling for a variety of background, enabling and predisposing factors. However, regardless of race, all these factors combined also accounted for only a small portion of variance in the communication variables. A better understanding of race disparity on information outcomes may need to focus on physician and medical system factors besides patient factors.
Acknowledgments This research was supported by a grant from the National Cancer Institute, Bethesda, MD (Grant No. NCI R01CA77525-U). We thank the Illinois State Cancer Registry and the Tumor Boards at 33 participating hospitals for their cooperation with this study. We are grateful to two anonymous reviewers for their generous comments and suggestions.
References ASCO-ESMO: American Society of Clinical Oncology and the European Society for Medical Oncology.
ASCO-ESMO consensus statement on quality cancer care. Annals of Oncology 2006;17:1063–1064. [PubMed: 16751232]
Ashton CM, Haidet PM, Paternity DA, Collins TC, Gordon HS, et al. Racial and Ethnic disparities in the use of health services: bias, preferences, or poor communication? Journal of General Internal Medicine 2003;18:146–152. [PubMed: 12542590]
Ayanian JZ, Zaslavsky AM, Guadognoli E, Fuchs CS, Yost KJ, et al. Patients’ perceptions of quality of care for colorectal cancer by race, ethnicity, and language. Journal of Clinical Oncology 2005;23(27): 6576–6586. [PubMed: 16116149]
Bach PB, Pham HH, Schrag D, Tate RC, Hargraves JL. Primary care physicians who treat Blacks and Whites. New England Journal of Medicine 2004;35(6):575–584. [PubMed: 15295050]
Baile FW, Aaron J. Patient-physician communication in oncology: past, present, and future. Current Opinion in Oncology 2005;17(4):331–335. [PubMed: 15933462]
Beck SR, Daughtridge R, Sloane DP. Physician-Patient Communication in the Primary Care Office: A Systematic Review. Journal of the American Board of Family Practice 2002;15(1):25–38. [PubMed: 11841136]
Beskow LM, Sandler RS, Weinberger M. Research recruitment through U.S. central cancer registries: balancing privacy and scientific issues. American Journal of Public Health 2006;96(11):1920–1926. [PubMed: 16571700]
Coscarelli Schag CA, Ganz AP, Heinrich LR. Cancer Rehabilitation Evaluation System Short Form (CARES-SF). Cancer 1991;68:1406–1413. [PubMed: 1873793]
Carlson, GA.; McDonald, EM. Using the PRECEDE-PROCEED Planning Model to Apply Health Behavior Theories. In: Glanz, K.; Rimer, BK.; Marcus Lewis, F., editors. Health Behavior and health Education: Theory, Research, and Practice. San Francisco, CA: John Wiley & Sons, Inc; 2002. p. 409-436.
Cooper Patrick L, Gallo J, Gonzales JJ, Vu Hong T, Powe N, Nelson C, Ford DE. Race, gender and partnership in the patient-physician relationship. Journal of the American Medical Association 1999;282(6):583–589. [PubMed: 10450723]
Cooper LA, Beach MC, Johnson RL, Inui TS. Delving below the surface: Understanding how race and ethnicity influence relationships in health care. Journal of General Internal Medicine 2006;21:S21– 27. [PubMed: 16405705]
Manfredi et al. Page 10
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Czaja R, Manfredi C, Pierce J. The determinants and consequences of information seeking among cancer patients. Journal of Health Communication 2003;8:529–562. [PubMed: 14690888]
Davison BJ, Degner LF. Empowerment of men newly diagnosed with prostate cancer. Cancer Nursing 1997;20(3):187–196. [PubMed: 9190093]
Degner LF, Kristijanson LJ, Bowman D, Sloan JA, Carriere KC, O’Neil J, Bilodeau D, Watson P, Mueller B. Information needs and decisional preferences in women with breast cancer. Journal of the American Medical Association 1997;277(19):1485–1492. [PubMed: 9145723]
Diette GB, Rand C. The contributing role of health-care communication to health disparities for minority patients with asthma. Chest 2007;132(5-S):8025–8095.
Finney Rutten LJ, Arora NK, Bakos AD, Aziz N, Rowland J. Information needs and sources of information among cancer patients: a systematic review of research (1980–2003). Patient Education and Counseling 2005;57:250–261. [PubMed: 15893206]
Gordon HS, Street RL, Sharf BF, Kelly A, Souchel J. Racial differences in trust and lung cancer patients’ perceptions of physician communication. Journal of Clinical Oncology 2006a;24(6):904–909. [PubMed: 16484700]
Gordon HS, Street RL, Sharf BF, Kelly A, Souchel J. Racial differences in doctors’ information-giving and patients’ participation. Cancer 2006b;107:1313–20. [PubMed: 16909424]
Gordon HS, Street RL, Kelly AP, Souchek J, Wray NP. Physician-patient communication following invasive procedures: an analysis of post-angiogram consultations. Social Science & Medicine 2005;61:1015–1025. [PubMed: 15955403]
Green, LW.; Kreuter, MW. Health Promotion Planning: An Educational and Environmental Approach. 2. Mountain View, CA: Mayfield; 1999.
Johnson RL, Saha S, Arbolaez JJ, Beach MC, Cooper LA. Racial and ethnic differences in patient perceptions of bias and cultural competence in health care. Journal of General Internal Medicine 2004a;19(2):101–110. [PubMed: 15009789]
Johnson RL, Roter D, Powe N, Cooper L. Patient Race/Ethnicity and Quality of Patient-Physician Communication During Medical Visits. American Journal of Public Health 2004b;94(12):2084– 2090. [PubMed: 15569958]
Levinson W, Hudak P, Feldman J, Frankel R, Kuby A, et al. “It’s not what you say….”: Racial disparities in communication between orthopedic surgeons and patients. Medical Care 2008;46(4):410–416. [PubMed: 18362821]
Liang W, Burnett CB, Rowland JH, Metropol NJ, Eggert L, Hwang YT, Silliman RA. Communication between physicians and older women with localized breast cancer: implications for treatment and patient satisfaction. Journal of Clinical Oncology 2002;20(4):1008–1016. [PubMed: 11844824]
Maly RC, Stein JA, Umezawa Y, Leake BL, Anglin MD. Racial/ethnic differences in breast cancer outcomes among older patients: effects of physician communication and patient empowerment. Health Psychology 2008;27(6):728–736. [PubMed: 19025268]
Maly RC, Leake B, Silliman RA. Health care disparities in older patients with breast carcinoma. Cancer 2003;97:517–27. [PubMed: 12518377]
Manfredi C, Czaja R, Buis M, Derk D. Patient use of treatment-related information received from the Cancer Information Service. Cancer 1993;71:1326–1337. [PubMed: 8435809]
Mills M, Sullivan K. The importance of information giving for patients newly diagnosed with cancer: a review of the literature. Journal of Clinical Nursing 1999;8(6):631–642. [PubMed: 10827609]
Ong LM, deHaes JC, Hoos AM, Lammes FB. Docto-patient communication: A review of the literature. Social Science & Medicine 1995;40(7):903–918. [PubMed: 7792630]
Pedhazur, EJ.; Schmelkin, LP. Measurement, Design, and Analysis: An Integrated Approach. Hillsdale, N.J: Lawrence Erlbaum; 1991.
Rees C, Bath P. The information needs and source preferences of women with breast cancer and their family members: A review of the literature published between 1988 and 1998. Journal of Advanced Nursing 2000;31:833–41. [PubMed: 10759979]
Rees C, Bath P. Information-seeking behaviors of women with breast cancer. Oncology Nursing Forum 2001;28:691–98.
Manfredi et al. Page 11
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Ries, LAG.; Young, JL.; Keel, GE.; Eisner, MP.; Lin, YD.; Horner, MJ. SEER Survival Monograph: Cancer Survival Among Adults: U.S. SEER Program, 1988–2001, Patient and Tumor Characteristics. National Cancer Institute; Bethesda, MD: 2007. SEER Program, NIH Publ. No. 07-6215
Saha S, Arbelaez JJ, Cooper L. Patient-physician relationships and racial disparities in the quality of health care. American Journal of Public Health 2003;93:1713–1719. [PubMed: 14534227]
Siminoff LA, Graham GC, Gordon NH. Cancer communication patterns and the influence of patient characteristics: Disparities in information-giving and affective behaviors. Patient Education and Counseling 2006;6:355–360. [PubMed: 16860520]
Squires L, Rutten F, Treiman K, Bright MA, Hesse B. Cancer patients’ information needs across the cancer care continuum: evidence from the Cancer Information Service. Journal of Health Communication 2005;10(S-1):15–34. [PubMed: 16377598]
StataCorp. Stata Statistical Software: Release 9. College Station, Tx: StataCorp LP; 2005. Stepanikova I, Cook KS. Effects of poverty and lack of insurance on perceptions of racial and ethnic bias
in health care. Health Services Research 2008;43(3):915–930. [PubMed: 18546546] Street RL, Gordon H, Haidet P. Physicians’ communication and perceptions of patients: it is how they
look, how they talk, or is it just the doctor? Social Science & Medicine 2007;65(3):586–598. [PubMed: 17462801]
Street RL, Gordon HS, Ward MM, Krapat E, Kravitz L. Patient participation in medical consultation: why some patients are more involved than others. Medical Care 2005;43(10):960–969. [PubMed: 16166865]
U.S. Census Bureau. Statistical Abstract of the United States: 2003. 123. Washington, DC: U.S. Government Printing Office; 2003.
Vogel BA, Bengel J, Helmes AW. Information and decision making: patients’ needs and experiences in the course of breast cancer treatment. Patient Education and Counseling 2008;71:79–85. [PubMed: 18191933]
Willems S, De Maesschalck S, Deveugele M, Derese A, De Maeseneer J. Socioeconomic status of the patient and doctor-patient communication: does it make a difference? Patient Education and Counseling 2005;56(2):139–146. [PubMed: 15653242]
Zapka JG, Carter R, Carter CL, Hemessy W, Kurent JE, Deshrnais S. Care at the end of life: Focus on communication and race. Journal of Aging & Health 2006;18(6):791–813. [PubMed: 17099134]
Manfredi et al. Page 12
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Figure 1. Conceptual model
Manfredi et al. Page 13
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Manfredi et al. Page 14
Table 1
Study variables and variable comparisons by race
African Americans (=1) N=248 Whites (=0) N=244 Comparison Test
BACKGROUND VARIABLES % % Chi2
Characteristics of hospital at diagnosis
Urban hospital (% yes)a b 50.00 39.75 5.22*
Rural hospital (% yes)a 20.56 31.15 7.19**
Bed size (%)a c
1=200 or less 15.73 13.52 n.s.
2=201–400 38.31 47.54
3=401–600 40.32 34.43
4=More than 600 5.65 4.51
Cancer center (% yes)a 64.11 58.20 n.s.
Teaching hospital (% yes)a c 71.77 64.34 n.s.
Disease characteristics
Cancer sitea
Colorectal (% yes) 18.55 26.23 4.18*
Breast (% yes) 53.23 46.31 n.s
Prostate (% yes) 28.23 27.46 n.s.
Time since diagnosisa c
1=0–6 months 2.82 2.46 n.s
2=7–12 months 13.71 17.21
3=13–18 months 30.65 27.87
4=19 to 24 months 34.68 34.84
5=24+ months 18.15 17.62
Currently receiving treatment (% yes)a 22.58 17.21 n.s
Co-morbidity (% yes) 65.73 51.64 10.07**
Patient demographic characteristics
Gendera,c
1=Male 36.69 40.57 n.s.
0=Female 63.31 59.43
Age at diagnosisa
1=26–49 22.58 15.98 9.10*
2=50–64 44.76 39.34
3=65–74 23.39 29.10
4=75+ 9.27 15.57
Education
1=Less than high school 21.37 9.43 18.15***
2=High school 22.98 31.97
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Manfredi et al. Page 15
African Americans (=1) N=248 Whites (=0) N=244 Comparison Test
3=Some college or associates degree 29.44 25.00
4=Bachelors degree 12.10 15.98
5=Graduate degree 14.11 17.62
Income
1=Less than $30,000 49.19 26.64 26.59***
2=$30,000–$50,000 25.00 35.25
3=$50,000 + 25.81 38.11
Married or with partner (% yes)c 47.98 70.90 26.78***
ENABLING VARIABLES
Uninsured
0=public, private or other insurance 84.27 93.03 9.35**
1=uninsured 15.73 6.97
Companion at medical visits (% yes) 47.98 60.25 7.45**
Regular physician involvement (% yes) 43.55 31.15 8.08**
PREDISPOSING VARIABLES Mean (S.D.) Mean (S.D.) T-value
Trust in doctors’ knowledge (Most doctors have current treatment knowledge) (range 1 to 4; 4=higher trust)
2.95 (0.57) 2.99 (0.64) n.s.
Desire to be involved in decisions (range 1 to 4; 4=higher desire) 3.38 (0.38) 3.39 (0.39) n.s.
COMMUNICATION VARIABLES
Interpersonal Communication
Interpersonal communication barriers (range 1 to 4; 4=more barriers) 1.61 (0.70) 1.47 (0.56) 2.33*
Comfort level asking questions (mean across all physicians seen, 1=all very uncomfortable, 4=all very comfortable)
3.34 (0.58) 3.43 (0.52) n.s.
Instrumental communication
Unmet information needs (range 1 to 4; 4=more unmet needs) 2.09 (0.62) 1.86 (0.54) 4.24***
Overall satisfaction with information (range 1 to 4; 4=more satisfied) 3.37 (0.62) 3.52 (0.59) -2.66**
% physicians seen who discussed disease and treatment (0=none discussed, 1=all discussed)
.90 (0.20) .88 (0.22) n.s.
% physicians seen to whom patient asked questions (0=asked none, 1=asked questions of all)
.86 (0.25) .80 (0.28) 2.74**
INFORMATION RECEIVED % % Chi2
Referral for 2nd opinion (% yes) 36.29 31.15 n.s.
Name of cancer expert (% yes) 24.19 35.25 7.20**
Name of cancer center (% yes) 19.76 27.05 n.s.
Clinical trials information (% yes) 18.95 21.31 n.s.
a These are the variables on which the samples were stratified.
b All variables shown as “% yes” were coded 0=no, 1=yes.
c These variables were excluded from the regression analyses in Tables 2 and 3.
* p. ≤ .05;
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Manfredi et al. Page 16
** p. ≤ .01;
*** p. ≤. .001
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Manfredi et al. Page 17
Ta bl
e 2
M ul
tip le
re gr
es si
on m
od el
s of
c om
m un
ic at
io n
va ri
ab le
s on
b ac
kg ro
un d,
p re
di sp
os in
g an
d en
ab lin
g va
ri ab
le s
(N =4
92 )
P at
ie nt
c om
fo rt
a sk
in g
qu es
ti on
s In
te rp
er so
na l C
om m
un ic
at io
n B
ar ri
er s
D oc
to rs
d is
cu ss
ed d
ia gn
os is
/t re
at m
en t
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
A fr
ic an
A m
er ic
an −.
09 (.0
5) −.
07 (.0
5) −.
06 (.0
5) .1
3( .0
6) *
.1 1(
.0 6)
.1 0(
.0 6)
.0 2(
.0 2
.0 2(
.0 2)
.0 2(
.0 2)
B ac
kg ro
un d
B re
as t
.1 2(
.0 6)
.1 0(
.0 6)
−. 10
(.0 7)
−. 09
(.0 7)
.0 1(
.0 2)
.0 2(
.0 2)
Pr os
ta te
−. 02
(.0 7)
−. 01
(.0 7)
.0 3(
.0 8)
.0 3(
.0 8)
.0 0(
.0 3)
.0 1(
.0 3)
C om
or bi
di ty
.0 2(
.0 5)
−. 01
(.0 5)
−. 00
(.0 6)
.0 1(
.0 6)
−. 00
(.0 2)
−. 01
(.0 2)
C an
ce r C
en te
r .0
1( .0
5) .0
0( .0
5) −.
02 (.0
6) −.
03 (.0
6) −.
02 (.0
2) −.
02 (.0
2)
U rb
an H
os pi
ta l
.1 5(
.0 6)
* .1
7( .0
6) **
−. 10
(.0 7)
−. 10
(.0 7)
−. 00
(.0 2)
−. 00
(.0 2)
R ur
al H
os pi
ta l
.0 2(
.0 7)
.0 4(
.0 7)
.0 1(
.0 8)
.0 0(
.0 8)
−. 03
(.0 3)
−. 03
(.0 3)
A ge
−. 03
(.0 3)
.0 1(
.0 3)
−. 01
(.0 3)
−. 04
(.0 4)
−. 00
(.0 1)
−. 00
(.0 1)
E du
ca tio
n .0
3( .0
2) .0
3( .0
2) −.
03 (.0
2) −.
02 (.0
2) −.
02 (.0
1) **
−0 2(
.0 1)
**
In co
m e
.1 1(
.0 3)
** .0
9( .0
3) **
−. 09
(.0 4)
* −.
09 (.0
4) *
.0 3(
.0 1)
* .0
3( .0
1) *
P re
di sp
os in
g
T ru
st in
d oc
to rs
’ k no
w le
dg e
.0 3(
.0 4)
−. 16
(.0 5)
** .0
2( .0
2)
D es
ir e
to b
e in
vo lv
ed .3
2( .0
7) **
* −.
23 (.0
8) **
.0 0(
.0 3)
E na
bl in
g
N o
in su
ra nc
e .1
1( .0
8) −.
04 (.0
9) .0
4( .0
3)
C om
pa ni
on a
t m ed
ic al
v is
it .1
2( .0
5) *
.0 1(
.0 6)
.0 0(
.0 2)
R eg
ul ar
p hy
si ci
an in
vo lv
ed .1
0( .0
5) .0
2( .0
6) .0
5( .0
2) **
M od
el f-
va lu
e 3.
16 3.
84 **
* 5.
13 **
* 5.
44 *
2. 10
* 2.
84 **
* 1.
01 1.
57 1.
82 **
*
R -s
qu ar
ed .0
1 .0
7 .1
4 .0
1 .0
4 .0
8 .0
0 .0
3 .0
5
P at
ie nt
a sk
ed q
ue st
io ns
U nm
et I
nf or
m at
io n
N ee
ds O
ve ra
ll sa
ti sf
ac ti
on w
it h
in fo
rm at
io n
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
A fr
ic an
A m
er ic
an .0
7( .0
2) *
.0 6(
.0 3)
* .0
6( .0
3) *
.2 2(
.0 5)
** *
.2 0(
.0 6)
** *
.2 0(
.0 6)
** −.
15 (.0
5) **
−. 12
(.0 6)
* −.
11 (.0
6)
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Manfredi et al. Page 18 P
at ie
nt a
sk ed
q ue
st io
ns U
nm et
I nf
or m
at io
n N
ee ds
O ve
ra ll
sa ti
sf ac
ti on
w it
h in
fo rm
at io
n
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
B ac
kg ro
un d
B re
as t
.0 2(
.0 3)
.0 2(
.0 3)
−. 05
(.0 7)
−. 06
(.0 7)
.0 3(
.0 7)
.0 2(
.0 7)
Pr os
ta te
.0 3(
.0 3)
.0 5(
.0 3)
.1 1(
.0 8)
.1 0(
.0 8)
−. 01
(.0 8)
−. 01
(.0 8)
C om
or bi
di ty
.0 2(
.0 3)
.0 1(
.0 3)
−. 01
(.0 6)
−. 01
(.0 6)
−. 01
(.0 6)
−. 01
(.0 6)
C an
ce r C
en te
r .0
2( .0
3) .0
2( .0
2) −.
02 (.0
6) −.
03 (.0
6) −.
00 (.0
6) .0
1( .0
6)
U rb
an H
os pi
ta l
−. 02
(.0 3)
−. 01
(.0 3)
.0 7(
.0 6)
.0 8(
.0 6)
−. 03
(.0 7)
−. 04
(.0 7)
R ur
al H
os pi
ta l
−. 05
(.0 3)
−. 03
(.0 3)
−. 00
(.0 7)
.0 1(
.0 7)
.0 4(
.0 8)
.0 3(
.0 8)
A ge
−. 05
(.0 1)
** *
−. 03
(.0 1)
* −.
02 (.0
3) −.
03 (.0
3) .0
4( .0
3) .0
6( .0
3)
E du
ca tio
n −.
00 (.0
1) −.
00 (.0
1) −.
00 (.0
2) −.
00 (.0
2) −.
04 (.0
2) −.
04 (.0
2)
In co
m e
.0 4(
.0 2)
* .0
4( .0
2) *
−. 03
(.0 4)
−. 05
(.0 4)
.0 6(
.0 4)
.0 6(
.0 4)
P re
di sp
os in
g
T ru
st in
d oc
to rs
’ k no
w le
dg e
.0 2(
.0 2)
−. 16
(.0 4)
** *
.1 4(
.0 5)
** *
D es
ir e
to b
e in
vo lv
ed .1
0( .0
3) **
−. 02
(.0 8)
.1 6(
.0 8)
*
E na
bl in
g
N o
in su
ra nc
e .0
7( .0
4) *
−. 08
(.0 9)
.0 7(
.0 9)
C om
pa ni
on a
t m ed
ic al
v is
it .0
8( .0
2) **
* .0
3( .0
5) −.
00 (.0
6)
R eg
ul ar
p hy
si ci
an in
vo lv
ed .0
6( .0
2) * .
.0 2(
.0 6)
−. 03
(.0 6)
M od
el f-
va lu
e 7.
49 **
4. 05
** *
5. 42
** *
17 .9
5* **
2. 61
** 2.
74 **
* 7.
05 **
1. 30
1. 82
*
R -s
qu ar
ed .0
2 .0
8 .1
5 .0
4 .0
5 .0
8 .0
1 .0
3 .0
5
* p . <
.0 5;
** p.
< .0
1;
** * p
. < . .
00 1
* p . <
.0 5;
** p.
< .0
1;
** * p
. < . .
00 1
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Manfredi et al. Page 19
Ta bl
e 3
L og
is tic
re gr
es si
on m
od el
s of
in fo
rm at
io n
re ce
iv ed
fr om
p hy
si ci
an s
(N =4
92 )
R ec
ei ve
d re
fe rr
al fo
r se
co nd
o pi
ni on
R ec
ei ve
d N
am e
of C
an ce
r E
xp er
t
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 4
B (S
.D .)
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 4
B (S
.D .)
A fr
ic an
A m
er ic
an .2
3( .1
9) .4
1( .2
1) .4
5( .2
3) *
.4 1(
.2 3)
−. 53
(.2 0)
** −.
51 (.2
2) *
−. 52
(.2 3)
* −.
52 (.2
4) *
B ac
kg ro
un d
B re
as t
.0 4(
.2 7)
−. 01
(.2 7)
−. 02
(.2 8)
.6 7(
.2 9)
* .6
6( .3
0) *
.6 5(
.3 1)
*
Pr os
ta te
.3 5(
.2 9)
.4 3(
.3 0)
.4 0(
.3 1)
.5 6(
.3 2)
.5 9(
.3 3)
.6 0(
.3 3)
C om
or bi
di ty
−. 27
(.2 1)
−. 40
(.2 2)
−. 41
(.2 2)
.0 9(
.2 3)
.0 0(
.2 3)
.0 2(
.2 4)
C an
ce r C
en te
r .0
6( .2
1) .0
1( .2
2) −.
04 (.2
2) .7
5( .2
3) **
* .7
3( .2
4) **
.7 3(
.2 4)
**
U rb
an H
os pi
ta l
−. 08
(.2 4)
.0 2(
.2 5)
.0 2(
.2 6)
−. 08
(.2 6)
.0 4(
.2 7)
.0 8(
.2 8)
R ur
al H
os pi
ta l
.0 6(
.2 7)
.1 8(
.2 8)
.1 8(
.2 8)
.2 5(
.2 8)
.4 1(
.2 9)
.4 4(
.3 0)
A ge
−. 29
(.1 3)
* −.
18 (.1
3) −.
17 (.1
4) −.
21 (.1
3) −.
13 (.1
4) −.
15 (.1
4)
E du
ca tio
n .0
1( .0
8) .0
0( .0
9) −.
01 (.0
9) .0
6( .0
9) .0
4( .0
9) .0
6( .0
9)
In co
m e
.4 9(
.1 4)
** *
.4 5(
.1 5)
** .4
1( .1
5) **
.5 1(
.1 5)
** *
.4 2(
.1 6)
** .3
6( .1
6) *
P re
di sp
os in
g
T ru
st in
d oc
to rs
’ k no
w le
dg e
−. 04
(.1 7)
−. 10
(.1 7)
−. 11
(.1 7)
−. 21
(.1 8)
D es
ir e
to b
e in
vo lv
ed .9
6( .3
0) **
.8 0(
.3 2)
* .7
6( .3
2) *
.6 7(
.3 3)
*
E na
bl in
g
N o
in su
ra nc
e .1
9( .3
4) .1
3( .3
4) −.
44 (.4
1) −.
51 (.4
1)
C om
pa ni
on a
t m ed
ic al
v is
it .4
6( .2
1) *
.3 8(
.2 2)
.3 2(
.2 2)
.2 8(
.2 3)
R eg
ul ar
p hy
si ci
an in
vo lv
ed .5
1( .2
1) *
.4 9(
.2 2)
* .6
0( .2
3) **
.5 7(
.2 3)
*
C om
m un
ic at
io n
D oc
to r d
is cu
ss ed
−. 66
(.5 8)
.8 5(
.6 6)
Pa tie
nt a
sk ed
q ue
st io
ns 1.
15 (.5
6) *
.6 0(
.5 6)
Pa tie
nt c
om fo
rt a
sk in
g −.
08 (.2
5) −.
12 (.2
7)
C om
m un
ic at
io n
ba rr
ie rs
−. 23
(.2 3)
−. 08
(.2 4)
U nm
et in
fo rm
at io
n ne
ed s
.1 4(
.2 3)
−. 19
(.2 5)
Sa tis
fa ct
io n
w ith
in fo
rm at
io n
.2 8(
.2 1)
.2 4(
.2 2)
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Manfredi et al. Page 20
R ec
ei ve
d re
fe rr
al fo
r se
co nd
o pi
ni on
R ec
ei ve
d N
am e
of C
an ce
r E
xp er
t
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 4
B (S
.D .)
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 4
B (S
.D .)
M od
el C
hi 2
(d f)
1. 46
(1 )
39 .5
0( 10
)* **
61 .5
7( 15
)* **
70 .1
5( 21
)* **
7. 23
(1 )*
* 53
.0 1(
10 )*
** 70
.5 1(
15 )*
** 80
.3 9(
21 )*
**
Ps eu
do R
-s qu
ar ed
.0 0
.0 6
.1 0
.1 1
.0 1
.0 9
.1 2
.1 3
R ec
ei ve
d N
am e
of C
an ce
r C
en te
r R
ec ei
ve d
C lin
ic al
T ra
il In
fo rm
at io
n
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 4
B (S
.D .)
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 4
B (S
.D .)
A fr
ic an
A m
er ic
an −.
41 (.2
1) −.
48 (.2
4) *
−. 52
(.2 5)
* −.
47 (.2
5) −.
15 (.2
3) .0
0( .2
5) .1
1( .2
6) .0
3( .2
7)
B ac
kg ro
un d
B re
as t
.3 5(
.2 9)
.3 1(
.3 0)
.3 0(
.3 0)
.4 1(
.3 1)
.3 4(
.3 1)
.3 3(
.3 2)
Pr os
ta te
.0 1(
.3 4)
.0 0(
.3 5)
.0 4(
.3 5)
−. 55
(.3 8)
−. 59
(.3 8)
−. 67
(.3 9)
C om
or bi
di ty
−. 19
(.2 4)
−. 26
(.2 4)
−. 28
(.2 4)
−. 56
(.2 5)
* −.
51 (.2
6) *
−. 51
(.2 6)
C an
ce r C
en te
r .1
8( .2
4) .1
6( .2
4) .1
3( .2
5) .3
2( .2
6) .2
9( .2
6) .2
6( .2
7)
U rb
an H
os pi
ta l
.0 6(
.2 7)
.1 6(
.2 8)
.2 0(
.2 9)
−. 03
(.2 9)
−. 03
(.2 9)
−. 09
(.3 1)
R ur
al H
os pi
ta l
.2 1(
.3 0)
.3 1(
.3 1)
.3 0(
.3 1)
.2 3(
.3 1)
.2 8(
.3 2)
.2 7(
.3 2)
A ge
−. 42
(.1 4)
** −.
38 (.1
5) *
−. 42
(.1 5)
** −.
18 (.1
5) −.
16 (.1
5) −.
19 (.1
6)
E du
ca tio
n −.
22 (.1
0) *
−. 23
(.1 0)
* −.
23 (.1
0) *
.2 3(
.1 0)
* .2
2( .1
0) *
.2 1(
.1 1)
*
In co
m e
.2 4(
.1 6)
.1 8(
.1 6)
.1 6(
.1 7)
.2 7(
.1 7)
.1 5(
.1 7)
.1 4(
.1 8)
P re
di sp
os in
g
T ru
st in
d oc
to rs
’ k no
w le
dg e
−. 35
(.1 8)
−. 47
(.1 9)
* .0
4( .2
0) .0
1( .2
0)
D es
ir e
to b
e in
vo lv
ed .4
0( .3
3) .3
4( .3
4) .5
8( .3
5) .4
7( .3
7)
E na
bl in
g
N o
in su
ra nc
e .0
5( .3
7) .0
3( .3
8) −1
.2 1(
.5 6)
* −1
.1 3(
.5 6)
*
C om
pa ni
on a
t m ed
ic al
v is
it .1
5( .2
3) .1
4( .2
4) .1
8( .2
5) .1
8( .2
5)
R eg
ul ar
p hy
si ci
an in
vo lv
ed .4
4( .2
4) .4
8( .2
4) *
−. 18
(.2 6)
−. 18
(.2 7)
C om
m un
ic at
io n
D oc
to r d
is cu
ss ed
−. 36
(.6 5)
.0 7(
.6 9)
Pa tie
nt a
sk ed
q ue
st io
ns .2
2( .5
5) −.
16 (.6
1)
Pa tie
nt c
om fo
rt a
sk in
g −.
22 (.2
8) −.
18 (.3
2)
C om
m un
ic at
io n
ba rr
ie rs
−. 27
(.2 6)
−. 73
(.3 0)
*
J Health Commun. Author manuscript; available in PMC 2011 April 1.
N IH
-P A
A uthor M
anuscript N
IH -P
A A
uthor M anuscript
N IH
-P A
A uthor M
anuscript
Manfredi et al. Page 21 R
ec ei
ve d
N am
e of
C an
ce r
C en
te r
R ec
ei ve
d C
lin ic
al T
ra il
In fo
rm at
io n
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 4
B (S
.D .)
M od
el 1
B (S
.D .)
M od
el 2
B (S
.D .)
M od
el 3
B (S
.D .)
M od
el 4
B (S
.D .)
U nm
et in
fo rm
at io
n ne
ed s
−. 25
(.2 6)
.7 1(
.2 7)
**
Sa tis
fa ct
io n
w ith
in fo
rm at
io n
.3 0(
.2 4)
.2 4(
.2 5)
M od
el C
hi 2
(d f)
3. 66
(1 )
31 .1
4( 10
)* **
39 .8
5( 15
)* **
48 .2
6( 21
)* **
0. 43
(1 )
37 .9
8( 10
)* **
47 .4
6( 15
)* **
57 .9
6( 21
)* **
Ps eu
do R
-s qu
ar ed
.0 1
.0 6
.0 7
.0 9
.0 0
.0 8
.1 0
.1 2
* p . <
.0 5;
** p.
< .0
1;
** * p
. < . .
00 1
* p . <
.0 5;
** p.
< .0
1;
** * p
. < . .
00 1
J Health Commun. Author manuscript; available in PMC 2011 April 1.
