The role of psychosocial factors in shared decision making (SDM) in cancer survivors: A risk prediction model.

541 Background: Cancer survivors make up more than 5% of the U.S. population, with nearly 19 million expected by 2025, and represent a group with unique healthcare needs. Shared decision-making (SDM) between patients and providers is essential to patient-centered care. We evaluated whether various factors—demographic, socioeconomic, behavioral, geographic, health, health information, and psychosocial—predicted cancer survivors’ risk of not engaging in SDM. Methods: We used data from the Health Information National Trends Survey (HINTS), a national survey on knowledge of, attitudes toward, and use of health information, covering years 2012-2024. Participants included those > 18 years old, diagnosed with cancer in the past 20 years, who had >1 non-emergency healthcare visit in the past year. SDM, the primary outcome, was assessed by the question, “In the past 12 months, how often did your health professional involve you in decisions about your healthcare as much as you wanted?” Responses were rated on a four-point Likert scale, with SDM coded “1” for always and 0 otherwise (aligned with the Healthy People 2030 SDM objective). Forty-five variables were evaluated. We developed a risk model to predict SDM using a random set of 60% of participants and k-fold cross-validation. The model was validated in the remaining 40%. Logistic regression was used. Results: A total of N = 3,622 participants [>65 years, 62.5%; female, 54.8%; Black race, 12.9%] were included. In the training dataset of n = 2174 respondents, a model with 4 variables was identified. Respondents expressing distrust of their provider (42.6% vs. 60.6%, p = .0002) or family (56.8% vs. 62.4%, p = .008), high anxiety or depression levels (54.0% vs. 65.8%, p < .0001), and perceived health no better than fair (51.4% vs. 62.7, p < .0001) were at lower risk of experiencing SDM. We summed the adverse risk factors for all respondents, splitting at the median score; those with 2 or more RFs (40.2%) were 46% less likely to experience SDM vs. those with 0 or 1 (59.8%; OR = 0.54, 95% CI, 0.46-0.65, p < .0001). Results were consistent in the test set (OR = 0.56, 95% CI, 0.45-0.70, p < .0001), confirming model validation. Among all individuals, the proportion of respondents experiencing SDM decreased from 70.4% to 61.9% to 52.3% to 46.2% as the number of adverse risk factors increased (in quartiles) from 0 to 1 to 2 to 3-4. Individuals with 3 or 4 risk factors (vs. those with 0 RFs) were 64% less likely to experience SDM (OR = 0.36, 95% CI, 0.28-0.46, p < .0001). Conclusions: We developed and validated a risk model comprised of psychosocial risk factors related to trust, perceived health, and anxiety/depression, which predicted low levels of SDM in cancer survivors. Conversely, demographic, socioeconomic, and other domains were not related to SDM. Identifying patients unlikely to engage in SDM could provide opportunities to intervene to enhance patient autonomy in care decisions.