Abstract IA14: Multilevel sources of bias as determinants of treatment disparities

Abstract

Abstract Race disparities in cancer incidence and mortality continue to persist. Unequal treatment (IOM, 2003) for Blacks compared to Whites, is a key factor driving disparities in survivorship and mortality. Research documenting disparities in cancer treatment has shown, for example, that when compared to Whites, Black patients have less access and/or information about clinical trials, are more likely to be under dosed when receiving chemotherapy for breast cancer, are less likely to receive chemotherapy for leukemia, are given less information about hospice care, are less likely to receive adequate levels of pain medication, among others. Social psychological theories of intergroup relations have been helpful for understanding unequal treatment as an outcome of individual level factors associated with physician and patient race related thoughts and feelings, (e.g., White physicians' prejudice or explicit/implicit bias regarding Black patients and Black patients' mistrust or suspicion of White physicians and the healthcare system.) These individual level factors may be expressed in the treatment setting because most Black cancer patients are treated by non-Black oncologists and experience race-discordant clinical interactions. The interaction occurs at the relational level, which, thus, presents a second set of factors related to poor communication and errors in understanding that also potentially contribute to treatment disparities. This presentation will include research showing how social psychological theories of intergroup relations are helping to explain how individual level race-related thoughts and feelings and relational level communication factors that arise in race-discordant clinical interactions result in treatment disparities. The discussion will include initial results from a current study at two comprehensive cancer centers in an urban Midwestern city that included Black cancer patients (n=112) and non-Black oncologists (n=18). Baseline information collected before the initial clinic visit to discuss treatment options included measures of physicians' explicit and implicit racial bias and patients' perceived past discrimination, and group-based suspicion of healthcare. The clinic visit was then video recorded and digitally coded for verbal and nonverbal behavior, information exchange, physician/patient talk time, interaction length, patient centeredness, behavioral synchronicity, etc. Measures administered to patients after the visit included perceived patient centeredness, trust, distress, difficulty in remembering the content of the interaction, and expectations about recommended treatments. Physician measures included perceptions of patient involvement in treatment decisions, and the patients' personal attributes, and expectations about patient responses to treatments. Appropriateness of actual treatment received was later abstracted from the patient's medical record. Early findings show that higher oncologist implicit bias is associated with less reported patient involvement in treatment decisions, less time spent with the patient, lower blind observers' ratings of physician supportiveness, and less smiling and shorter gaze towards patient. Patients interacting with higher implicit oncologists reported less patient centeredness, less trust in the oncologist (one week later), and more difficulty remembering and understanding the conversation (one week later). Indirect effects show that higher implicit physician bias results in lower patient perceived patient centeredness (during the interaction), which subsequently causes patients to be more distressed following the interaction and to expect the recommended treatment to be more difficult to complete. Interventions to reduce the negative impacts of individual level race related thoughts and feelings and race-discordant interactions will be described during the talk. Unfortunately, race-discordant clinical interactions are unlikely to change in the near-term given the major shortage of Black oncologists (e.g., only about 172 Black medical oncologists practice in the U.S.) and low percentage of Black medical trainees. However, other multilevel approaches are being tested, including patient activation techniques, such as the use of question prompt lists to improve communication patterns in the visit, which may enhance the quality of the interaction and the treatment decision making process. In addition, physician-focused interventions are being created to increase relational communication skills and alliance building. One unique strategy is to increase mindfulness and behavior modeling. The study authors are testing this by using direct observations of video recorded oncology interactions. The project involves having professional actors re-enact the actual clinical interactions that we have video recorded of high-implicit bias physicians and low implicit bias physicians. Fidelity studies comparing our re-enacted (but de-identified) clinical interactions with the original video recordings show strong concordance. Using re-enactments to show physicians, especially less experienced physicians, how such verbal and nonverbal behavioral patterns differ may provide effective training. Citation Format: Terrance L. Albrecht. Multilevel sources of bias as determinants of treatment disparities. [abstract]. In: Proceedings of the Eighth AACR Conference on The Science of Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; Nov 13-16, 2015; Atlanta, GA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2016;25(3 Suppl):Abstract nr IA14.