This study is among the first to comprehensively examine the role of physician communication and patient cognitive factors, along with socioeconomic status, on racial/ethnic disparities in breast cancer treatment and QOL within a single exploratory model derived from empirical findings from the extant literature. Compared to the other women, African Americans reported equally good QOL, whereas Latinas reported worse QOL, concurring with the results of previous studies (Spencer et al., 1999
). Further, both African-American women and Latinas received less BCS compared to Whites, which is also supported by previous literature (Mandelblatt et al., 2002
; C. R. Morris et al., 2000
). Thus, the first hypothesis that there would be racial/ethnic disparities in older women’s QOL and breast cancer treatment was confirmed, especially among Latinas.
Greater patient perception of interactive information-giving and shared treatment decision-making by physicians correlated with increased QOL and receipt of BCS in the bivariate analysis, confirming the study’s second hypothesis. However, the path model showed that the influence of patient-reported physician participatory style was indirect, operating through patient cognitive variables. The positive association of these patient-empowering communications with patient QOL was mediated by patients’ positive breast cancer coping strategies and their feelings of self-efficacy in patient-physician interactions. The latter finding is consistent with other cancer research supporting the role of self-efficacy in predicting better QOL (Maliski et al., 2004
), but expands on those results by highlighting the powerful role of physician communication in mediating these effects. Similarly, the model takes previous work demonstrating the relationship of formation of patient-physician partnerships through physician communication to BC patients’ coping (McWilliam et al., 2000
) one step further to show its positive influence on QOL.
In terms of BCS, the model links earlier findings that better physician communication in the form of information giving improves BC knowledge (R. C. Maly, Leake et al., 2004
; Whelan et al., 2004
) and that knowledge increases BCS rates (Whelan et al., 2004
). Overall, the model provides a comprehensive picture that illustrates how BC treatment decisions may be influenced by patient empowering communication. Self-efficacy posited to be engendered by physicians in the model, also was associated with that increased BCS rates, showing its independent effect on treatment. Moreover, since physicians’ emotional support had no effects on the outcomes, it appears that physician communication that may specifically empower patients, in terms of self-efficacy, knowledge, and coping, was a key statistical predictor of patient QOL and treatment received, controlling for sociodemographic characteristics, comorbidity, and BC stage. However, the role of participatory decision-making in breast cancer treatment merits further inquiry since some authors report a negative association between patient involvement and receipt of BCS (S. J. Katz et al., 2005
), while this study found that inclusion of patients in their treatment plans was associated indirectly with greater receipt of BCS.
The third study hypothesis that relationships between race/ethnicity and the outcomes of QOL and BCS would be mediated by provider- and system-level factors and patient cognitive factors was partially confirmed. Racial/ethnic disparities in breast cancer treatment were, in fact, mitigated to some degree by these factors, but Latina ethnicity was a direct statistical predictor of poorer QOL in the path model, bypassing all intermediate pathways. The direct link between Latina ethnicity and QOL indicates that, beyond mistrust and poor educational and financial resources included in the model, additional unmeasured factors likely played a part in older Latinas’ poorer reported QOL. Interestingly, although there was a positive and statistically significant association between the two study outcomes in bivariate analysis, QOL and receipt of BCS, as supported by recent research (Engel et al., 2004
), this relationship did not persist in the path model. Factors beyond those examined in the other studies, such as patient cognitive variables and medical mistrust, were incorporated in our comprehensive model and may have attenuated this association and thus suggest that further study is required.
The path model indicates that compared to Whites, African Americans expressed substantial mistrust of the health care system and Latinas reported somewhat less mistrust. Medical mistrust, in turn, was associated with less positive coping with breast cancer, leading to reduced QOL in the path model. Medical mistrust also predicted decreased feelings of self-efficacy in interacting with physicians, thus reducing both QOL and the likelihood of receiving BCS. These findings support the contention that medical mistrust accounts for at least some of the racial/ethnic disparities in medical outcomes generally (Bogart, Bird, Walt, Delahanty, & Figler, 2004
; LaVeist et al., 2000
; Matthews, Sellergren, Manfredi, & Williams, 2002
; O’Malley, Sheppard, Schwartz, & Mandelblatt, 2004
African-American race did not predict surgeon participatory style and information-giving, although it was a negative predictor of surgeon emotional support and breast cancer knowledge. Because perception of surgeons’ emotional support was highly correlated with report of their participatory style, it may be that even if adequate information had been provided, it could have been met with suspicion on the part of African-American patients because of their greater mistrust of the health care system.
SES had a profound relationship to patient report of surgeon participatory style and was significantly correlated with all of the provider-and system-level variables, including surgeon communication and patient self-efficacy, breast cancer knowledge and positive coping. While the life circumstances patients bring to a diagnosis of breast cancer cannot be altered, policy makers and providers can take steps to offset, or at least mitigate the effects of fewer educational and economic resources by increasing information-giving and providing low SES patients, in particular, with opportunities to participate in decisions about their treatment (Institute of Medicine, 2003
This study has several limitations. First, much of the data were collected by patient self-report and are subject to recall and social response biases. However, people who have undergone a major, life-threatening health crisis often manifest very clear recall of the details surrounding the event (Brown & Kulik, 1982
). For example, breast cancer patients can recall when they first noticed their symptoms (Burgess, Ramirez, Richards, & Love, 1998
), and can also accurately report details of their treatment (Maunsell, Drolet, Ouhoummane, & Robert, 2005
). A second limitation is that study participants were from the Los Angeles County area and may not reflect the greater population of older women in the United States, especially Latinas. The majority of Latinas in Los Angeles County are of Mexican origin (California Health Interview Survey, 2005
) and their cultural beliefs, preferences and attitudes may vary from those of Latinas in other parts of the country. A third limitation is that participants with low health literacy may not have been able to adequately complete the self-administered portion of the survey instrument. However all forms were intended for an eighth grade education or less and telephone consultation was available to all respondents. Finally, the predictive model is cross-sectional and causality cannot be assumed. Other models and directionality of influences might be equally reasonable. In particular, patient-level attributes may induce different physician-level communication styles, so influences may flow in a bidirectional manner. However, we have provided a rationale for the directionality of our particular model that influence normally flows from the physician to the individual patient. Future studies should obtain observational data on patient-physician communication style and/or longitudinal data over the course of diagnosis and treatment to better resolve such issues and provide for model improvement.
Clearly, medical mistrust, positive coping with breast cancer, knowledge about breast cancer and its treatment options, and feelings of self-efficacy derived from patients’ personal backgrounds and physician communication, played important roles in QOL and surgical treatment selection, independent of SES, stage of breast cancer, and comorbidity. Other factors not considered in this study may also have crucial influences. Further studies are needed to understand patient preferences and their origins, preconceptions about breast cancer treatment, and competing needs that may compel some women to select mastectomy as a shorter form of therapy, and the role of family and friends during decision-making. In addition, extraneous factors that were not measured can affect QOL, such as living circumstances, family caretaking responsibilities, and other sources of stress found among older women.
The study is encouraging in that neither minority ethnicity nor medical mistrust were directly associated with rates of BCS, but rather operated through the potentially malleable constructs of self-efficacy, coping, and knowledge, which, in turn, were associated with physician communication styles, also potentially malleable through education and training. However, the negative relationship of Latina ethnicity with QOL may be more difficult to address since it was independent of the patient-cognitive factors that were related to physician communication. Future studies should address medical mistrust among older minority breast cancer patients, overcoming gaps in breast cancer knowledge, enhancing self-efficacy in patient-provider discussions, and supporting positive coping, so that appropriateness of treatment decision-making and quality of life can be maximized in all older patients.