Racial differences in medical mistrust among men diagnosed with prostate cancer
Mistrust of healthcare providers and systems is a significant barrier to quality healthcare. However, limited empirical data are available on perceptions of medical mistrust among individuals who are diagnosed with cancer. The objective of this study was to identify sociodemographic, clinical, and cultural determinants of mistrust among men diagnosed with prostate cancer.
The authors conducted an observational study among 196 African-American men (n = 71) and white men (n = 125) who were newly diagnosed with prostate cancer during 2003 through 2007.
Race, education, healthcare experiences, and cultural factors had significant effects on mistrust. African-American men (P = .01) and men who had fewer years of formal education (P = .001) reported significantly greater levels of mistrust compared with white men and men who had more education. Mistrust also was greater among men who had been seeing their healthcare provider for a longer period (P = .01) and among men with lower perceptions of interdependence (P = .01).
The current findings suggested that efforts to enhance trust among men who are diagnosed with prostate cancer should target African-American men, men with fewer socioeconomic resources, and men with lower perceptions of interdependence. Reasons for deterioration in trust associated with greater experience with specialty providers should be explored along with the effects of interventions that are designed to address the concerns of individuals who have greater mistrust. Cancer 2009. © 2009 American Cancer Society.
Being diagnosed with cancer is an acute medical crisis1; effective communication between patients and physicians is critical to ensuring that men understand their diagnosis, options for treatment, and the pros and cons of each treatment option. However, lack of trust in healthcare providers and the medical system is a significant barrier to effective communication and use of clinical services2 and also may contribute to racial differences in outcomes after prostate cancer diagnosis.3 Trust is a multidimensional construct that reflects patient's expectations that healthcare providers will behave in ways that demonstrate that their interests are a priority.4-6 Key components of trust include perceptions of the healthcare provider's interpersonal skills and technical abilities, suspicion of healthcare systems and providers, expectations of unfair treatment, and perceived support from providers.4, 5, 7, 8 Trust is influenced by race and healthcare experiences; African Americans and individuals with less experience with providers are most likely to report low levels of trust.9-11 Fewer quality interactions with providers also are associated with low trust.9, 12 Recent research also has demonstrated that African-American men who are diagnosed with prostate cancer report lower levels of trust compared with white men3; however, empirical data are not available on predictors of mistrust in men who are diagnosed with this disease.
The objective of the current study was to identify factors that have significant independent associations with mistrust among African-American and white men who are diagnosed with prostate cancer. Because of the marked racial disparities in prostate cancer outcomes and consistent racial differences in trust,3, 9, 10, 13 mistrust that is rooted in an individual's racial experience and perception was of particular interest in this study. We predicted that African-American men would report significantly greater levels of mistrust compared with white men. To understand the context within which mistrust is manifested, we also evaluated the relationship between mistrust and healthcare experiences (eg, the length of the patient-provider relationship) and cultural factors. We hypothesized that mistrust would be associated with having a shorter relationship with healthcare providers. Culture is defined as a set of shared and socially transmitted ideas about the world that are passed down from generation to generation14; previous research has indicated that beliefs and values related to religiosity are associated with trust in minorities.15 However, it is unknown whether these factors are associated with mistrust among ethnically diverse samples of men facing an acute clinical diagnosis. Furthermore, although previous research has demonstrated that cultural values related to collectivism are associated with aspects of the patient-provider relationship that contribute to trust,16 endorsement of these values also may have a direct effect on mistrust. That is, because collectivism reflects the extent to which interdependence, relationships, and cooperation with others is valued17 and because mistrust is an indication of the quality of the relationship between providers and patients, mistrust should be greater among men with lower levels of collectivism. Conversely, individuals with greater religiosity may be more inclined to mistrust healthcare providers and systems because of a greater reliance on their spiritual faith and practices. A better understanding of mistrust among men newly diagnosed with prostate cancer is needed to identify patient and system factors that should be addressed to enhance trust among African-American men with the goal of improving their treatment trajectory in terms of decisions and outcomes.
MATERIALS AND METHODS
This study was conducted at the University of Pennsylvania and was approved by the Institutional Review Board. Participants were African-American and white men who were newly diagnosed with prostate cancer. To be eligible for participation, men had to be diagnosed with biopsy-confirmed prostate cancer (tumor classification T1-T3) within the past 2 months to 5 months, and men had to have complete clinical data on stage, prostate-specific antigen (PSA), and Gleason score for inclusion in the analysis. Consistent with previous research in patients with newly diagnosed cancer,18 the study enrollment rate was 46% among all eligible men who were referred to the study (n = 470). There were no differences in study enrollment based on Gleason score (chi-square test, 0.05; P = .83) or disease stage (chi-square test, 0.43; P = .51)1; however, white men (chi-square test, 8.61; P = .003) and men with lower PSA values (t = 2.64; P = .01) were most likely to enroll in the study. Among men who enrolled in the study, 92% completed the baseline telephone interview.
Eligible men were recruited into the study at urology and radiation oncology practices located in the Philadelphia, Pennsylvania metropolitan area. Recruitment sites included the urology practices at the University of Pennsylvania Health System (UPHS) and community-based urology practices. Men were recruited into the study by clinic and research staff during a follow-up appointment after they had been informed about their biopsy result. It should be noted that some men who were recruited at the UPHS were identified at the presurgery class for radical prostatectomy, and others were being seen for a second opinion and/or treatment. At each recruitment site, eligible men received a verbal and written description of the study and the procedures involved in participation. Written informed consent was obtained from all men who enrolled in the study. Men were contacted for a baseline telephone interview about 1 week to 4 weeks after study enrollment and provision of written informed consent. The baseline interview was completed by trained research assistants at the University of Pennsylvania and took about 30 minutes to complete. The baseline was a structured survey that obtained sociodemographics, race, and treatment status and assessed cultural factors and mistrust. Men were contacted for follow-up telephone interviews at 3 months, 6 months, and 12 months after the baseline survey. The current study focused on data that were collected during the baseline telephone interview.
Race, age, marital status, education, income, and employment status were obtained by self-report during the baseline telephone interview. With the exception of age, these variables were recoded into dichotomous variables based on the distribution of responses.
PSA, Gleason score, and TNM classification at diagnosis were obtained at enrollment. With the exception of PSA, these items were recoded into dichotomous variables (eg, T1 tumors vs T2/T3 tumors) based on the distribution of responses. Treatment status was obtained by self-report during the baseline telephone interview by items that asked men whether they had received surgical, radiation, or expectant (eg, watchful waiting) treatment. Men who had completed surgical treatment or had initiated radiation or other types of therapy (eg, cryosurgery) were categorized as having initiated or completed treatment. Men who had not initiated any treatment were categorized as being pending for treatment.
Similar to what has been reported in previous research,9 healthcare experiences were evaluated in terms of the length of the relationship with the physician at each recruitment site. Specifically, men were asked to report how long they had been seeing the physician who was providing their prostate cancer care. This variable was recoded into a dichotomous variable (>3 months vs <3 months) based on the distribution of responses.
We used the religiosity and collectivism scales developed by Lukwago and colleagues19 to evaluate the extent to which men endorsed religious values (eg, when I am ill, I pray for healing; I have a personal relationship with God) and values related to group interdependence and support (eg, it is important for families to do everything they can to help others move ahead in life). The religiosity scale included 9 Likert-style items, and the collectivism scale had 6 Likert-style questions. These scales had acceptable internal consistency in our sample (Cronbach α: collectivism, α = .64; religiosity, α = .94). Higher scores indicated greater levels of collectivism and religiosity.
We used the Group-Based Medical Mistrust Scale (GBMMS)8 to evaluate mistrust in healthcare providers and systems. The GBMMS is a 12-item Likert-style scale that measures suspicion of healthcare providers and the healthcare system, expectations about racial discrimination by healthcare providers, and perceived support from providers. The GBMMS has been validated in samples that included African-American and Hispanic women8 and also has been administered to men and women in primary care and emergency department settings as part of research on treatment for migraines and adherence to colon cancer screening, respectively.20, 21 The GBMMS had good face validity with validated measures of trust6 and good internal consistency in this sample (Cronbach α = .86). Higher scores indicated greater mistrust.
Descriptive statistics were generated to characterize the men in terms of sociodemographics and clinical characteristics. Then, we used chi-square tests of association and t tests to identify racial differences in sociodemographics and clinical factors. Next, we conducted bivariate analyses to evaluate the relation between mistrust and sociodemographic, clinical, and cultural factors. These analyses were performed to identify variables for inclusion in the regression model of mistrust. We then used regression analysis to identify factors that had significant independent associations with mistrust. Because of the potential for confounding within recruitment sites, we evaluated the effects of these factors on mistrust using fixed-effects modeling. Recruitment site was treated as a fixed effect to control for clustering within site. Variables that had a bivariate association of P < .10 with mistrust were included in the model. Because race was associated significantly with sociodemographic and cultural factors in our previous research,22 we included interaction terms in the model to determine whether the effects of these factors were different for African-American men and white men. Only the interactions that were statistically significant were retained.
Table 1 shows the sample characteristics. There were no racial differences in age (t = 0.07; P = .95), employment status (chi-square test, 2.87; P = .09), Gleason score (chi-square test, 1.40; P = .24), stage (chi-square test, 3.04; P = .08), or length of care (chi-square test, 2.41; P = .12). However, white men reported significantly greater incomes (chi-square test, 12.57; P = .0004) and education (chi-square test, 16.86; P = .0001) and were more likely to be married (chi-square test, 11.17; P = .001). White men also were more likely than African-American men to have initiated or completed treatment (chi-square test, 7.63; P = .01). African-American men had higher PSA levels compared with white men (t = 2.97; P = .004).
Table 1. Sample Characteristics (n=196)
|Race|| || || || |
| African American||71 (36)|| || || |
| White||125 (64)|| || || |
|Marital status|| || || || |
| Married||155 (79)||47 (66)||108 (86)||11.17*|
| Not married||41 (21)||24 (34)||17 (14)|| |
|Education level|| || || || |
| ≥Some college||125 (64)||32 (45)||93 (74)||16.86†|
| ≤High school||71 (36)||39 (55)||32 (26)|| |
|Employment status|| || || || |
| Employed||93 (47)||28 (39)||65 (52)||2.87‡|
| Not employed||103 (53)||43 (61)||60 (48)|| |
|Income level§|| || || || |
| >$50,000||112 (57)||29 (41)||83 (67)||12.57†|
| <$50,000||83 (43)||42 (59)||41 (33)|| |
|Insurance status|| || || || |
| Yes||192 (98)||67 (94)||125 (100)||7.29†|
| No||4 (2)||4 (6)||—|| |
|Length of care, mo|| || || || |
| >3||96 (49)||40 (56)||56 (45)||2.41|
| <3||100 (51)||31 (44)||69 (55)|| |
|Tumor classification|| || || || |
| T2/T3||62 (32)||17 (24)||45 (36)||3.04†|
| T1||134 (68)||54 (76)||80 (64)|| |
|Gleason score|| || || || |
| >6||83 (42)||34 (48)||49 (39)||1.40|
| ≤6||113 (58)||37 (52)||76 (61)|| |
|Treatment status∥|| || || || |
| Initiated/completed||111 (57)||31 (44)||80 (64)||7.63†|
| Pending||85 (43)||40 (56)||45 (36)|| |
|Age: Mean±SD, y||63.6 ± 8||63.6 ± 8.8||63.6 ± 7.6||T = 0.07|
|PSA: Mean±SD, ng/mL||6.4 ± 4.9||7.99 ± 6.5||5.52 ± 3.4||T = 3.49†|
Table 2 shows the bivariate analysis of mistrust. Of the sociodemographic characteristics that were analyzed, race, education, and income were associated significantly with mistrust. Compared with white men, African-American men reported significantly greater mistrust. In addition, men who were high school graduates or had less education and those with incomes less than $50,000 were most likely to report greater mistrust. Being unmarried had a marginally significant association with greater mistrust. With respect to clinical factors, length of care and treatment status were associated significantly with mistrust. Men who had been seeing their healthcare provider for more than 3 months reported greater mistrust compared with men who had a shorter length of care. Compared with men who had completed or initiated treatment, those whose treatment was pending also reported significantly greater mistrust. In terms of cultural factors, both collectivism and religiosity were associated significantly with mistrust. Men with lower perceptions of interdependence and those with greater religiosity reported higher levels of mistrust.
Table 2. Bivariate Analysis of Medical Mistrust
|Race|| || |
| African American||26.3 (7)||.0001|
| White||21.1 (5.9)|| |
|Marital status|| || |
| Married||22.5 (6.7)||.08|
| Not married||24.6 (6.8)|| |
|Education level|| || |
| ≥Some college||21.2 (6.2)||.0001|
| ≤High school||26.1 (6.7)|| |
|Employment status|| || |
| Employed||22.2 (6.7)||.14|
| Not employed||23.6 (6.8)|| |
|Income level|| || |
| >$50,000||21.7 (6.1)||.002|
| <$50,000||24.7 (7.3)|| |
|Length of care, mo|| || |
| >3||24.2 (6.8)||.01|
| <3||21.7 (6.6)|| |
|Tumor classification|| || |
| T2/T3||22.4 (7.2)||.48|
| T1||23.2 (6.6)|| |
|Gleason score|| || |
| >6||22.6 (6.8)||.56|
| ≤6||23.2 (6.8)|| |
|Treatment status|| || |
| Initiated/completed||22 (6.7)||.01|
| Pending||24.4 (6.8)|| |
|Age||R = 0.10||.16|
|PSA||R = 0.06||.37|
|Collectivism||R = −0.16||.03|
|Religiosity||R = 0.21||.003|
The multivariate model of mistrust is provided in Table 3. There were significant main effects for race, education, marital status, length of care, and collectivism. Specifically, African-American men, men who had a high school education or less, and men who had been seeing their healthcare provider for more than 3 months reported significantly greater levels of mistrust compared with white men, men with greater education, and men with a shorter length of care. In addition, lower levels of collectivism were associated significantly with greater mistrust. The only significant interaction was between race and marital status (estimate = −6.43; standard error [SE] = 2.30; P = .006). Marital status had a significant effect on mistrust among white men; mistrust was greater among men who were married (estimate = 3.68, SE = 1.72; P = .03). Other interactions were not significant (P > .15). The sample size was sufficient for detecting significant interactions, as reflected in the confidence intervals for these nonsignificant interactions, which were shorter than the interaction that was statistically significant.23
Table 3. Multivariate Model of Medical Mistrust*
|Race|| || |
| African American||3.86†||1.36|
| White (reference)|| || |
|Education|| || |
| ≥Some college||−3.82‡||1.08|
| ≤High school (reference)|| || |
|Marital status|| || |
| Not married (reference)|| || |
|Income|| || |
| <$50,000 (reference)|| || |
|Length of care, mo|| || |
| <3 (Reference)|| || |
|Treatment status|| || |
| Pending (reference)|| || |
To our knowledge, this is the first empirical study to evaluate perceptions of medical mistrust among men who were newly diagnosed with prostate cancer based on sociodemographics, healthcare experiences, and cultural factors. Consistent with previous research,9, 10 we observed that African-American men and men who had fewer years of formal education reported significantly greater mistrust compared with white men and men who had more education. Racial differences in mistrust may reflect differences in the quality of interactions with healthcare providers in oncology settings12, 24; however, other factors may influence mistrust among white men. We observed that marital status had a significant effect among white men; mistrust was greater among those who were married. Spouses and live-in partners are an important source of support for men who are diagnosed with prostate cancer; and the presence or absence of spouses at consultations also has implications for men's engagement with healthcare systems.25, 26 It is possible that spouses have less of an effect on mistrust among African-American men because their perceptions of healthcare providers and systems are based on their own negative experiences. Attempts to replicate the current findings would do well to examine whether the presence of the spouse at consultations with providers contributes to an explanation of the complex effects of marital status on trust.
In contrast to our hypothesis and our previous research,11 we observed that mistrust was significantly greater among men who had been seeing their healthcare provider for a longer time. One reason why our finding differs from what was reported in previous research may be because the current study was conducted with men who were seeing a specialist physician in an oncology setting, whereas most previous studies on trust were conducted in primary care settings.6, 11 It is possible that, with repeated interactions, men accumulate disappointments in terms of their provider listening to them, providing information that they can understand, and having their concerns addressed. This possibility could be examined readily in future research and is an important area for the exploration of how to reduce racial disparities in prostate cancer outcomes.
The results of the current study demonstrate that race and healthcare experiences are not the only determinants of mistrust among men who are diagnosed with prostate cancer. We observed that collectivism had a significant effect on mistrust; as collectivism decreased, mistrust increased. Collectivism is a cultural factor that reflects the extent to which interdependence, cooperation, and relationships with others are valued; individuals with lower levels of collectivism may place greater importance on autonomy, independence, and self-direction.16 Previous research has demonstrated that greater self-direction is associated negatively with trust in institutions, such as healthcare systems27; it is possible that men with lower levels of collectivism are more skeptical about the healthcare system and whether or not providers act in their best interest. Alternatively, men with lower collectivism may have communication styles and preferences that are unsupported during clinical encounters. Previous research has indicated that individuals who value autonomy and independence are likely to have a direct and open communication style,28 are willing to be assertive during interactions with healthcare providers,16 and expect positive responses from these efforts.16 However, Street and colleagues29 observed that only 16% of physicians prompted patient participation during the clinical encounter through supportive talk or partnership-building communication.
Although this study provides novel information on predictors of medical mistrust among men with prostate cancer, some limitations should be noted. First, only about 50% of men enrolled in the study, and it is possible that men who had greater distrust were likely to decline study enrollment. However, our enrollment rates are similar to those reported in other research within newly diagnosed cancer patients,18 and >90% of men who enrolled in the study completed the baseline telephone interview. Thus, it is not likely that distrust is a barrier to completing study procedures. It also should be noted that, in contrast to previous prostate cancer research,30, 31 African-American men made up a substantial proportion of the participants in the current study. The observational nature of our study is an additional limitation that prevents us from establishing causality with respect to mistrust, healthcare experiences, and cultural factors. It is also possible that evaluating mistrust shortly after being diagnosed with prostate cancer may influence men's responses, because they are reluctant to report perceptions of unfair treatment, suspicion, and lack of support when they are actively seeking medical care for an acute illness. However, similar methods have been used to evaluate trust in healthcare providers in other research with cancer patients.12 It is also important to acknowledge that we did not evaluate whether or not mistrust differs depending on racial concordance with providers. This is an important area for future research on mistrust in oncology settings. Studies also should examine the relation between racial differences in mistrust and adherence to treatment recommendations among patients with prostate cancer.
Despite these potential limitations, our results have important implications for strategies to enhance trust among individuals who are facing an acute medical crisis. Trust is a reflection of the quality of the patient-provider relationship; previous research has suggested that efforts to improve trust should focus on enhancing the provider's communication and partnership-building skills.9, 12 Our finding that mistrust was greater among men who had a longer relationship with providers further supports these recommendations. However, there is limited evidence that such efforts increase trust among patients.32 Because communication between patients and providers is a reciprocal process,33 it may be important to direct strategies for enhancing trust to individuals who are at increased risk for mistrust in addition to developing strategies that improve physician behaviors. Recent research has demonstrated that healthcare providers and African-American patients exhibit less positive affect (eg, responsiveness, engagement, assertiveness, and attentiveness) during clinical encounters,34 and approximately 25% of low-income men who were diagnosed with prostate cancer reported low self-efficacy in communicating with healthcare providers.1 Poor communication with providers had a significant effect on trust among patients with lung cancer, especially those who were African American.12 Thus, it may be important to enhance the communication skills of African-American men and those with fewer socioeconomic resources (eg, low education) to mitigate patients' negative perceptions of providers and healthcare systems. Previous research has indicated that communication skills training that consisted of consultation planning for breast cancer treatment (eg, identification and organization of questions, role playing questions) reduced communication barriers with oncologists and also increased satisfaction with communication among healthcare providers.35 In other research, community-based communication skills training that focused on enhancing assertiveness increased participants' confidence in their ability to communicate with healthcare providers.36 Communication skills training that helps African-American men and those with fewer socioeconomic resources to identify and organize questions about their diagnosis and treatment also may be effective at reducing suspicion of healthcare providers. Our findings also suggest that, within these approaches, it may be important to consider men's communication preferences that are shaped by cultural factors. Future studies also should evaluate the effects of these interventions among men with different sociodemographic profiles.
Conflict of Interest Disclosures
Supported by National Cancer Institute grants P50CA105641-010004 and R01CA100254.
David Lee provides proctoring for Intuitive Surgical and lectures for SureRx.