Associations Between Psychosocial Attributes of Patients/Oncologists with Potentially Burdensome Interventions (PBIS) in The Final Month of Life: Secondary Analyses of Voice

Paul Duberstein

Paul Duberstein


Decreasing the use of potentially inappropriate interventions (PIIs) patients receive at the end of life (EoL) is a desirable but elusive goal. In the 1990s, the SUPPORT study underscored how hard it is to reach this goal; more recent interventions have not been more successful. This is hardly surprising, as few studies of risk factors for PIIs have been conducted. In this study, we tested the hypothesis that psychosocial attributes of patients and oncologists are independently associated with PIIs. 


Predictor variables in these secondary analyses of the Values and Options in Cancer Care (VOICE) study (April 2002-October 2015) were patient attitudes toward EoL care (palliative care, life-sustaining treatments, experimental treatments) and oncologist-reported comfort with medical paternalism (feels comfortable making decisions with little or no input from the patient). Outcome variables (PIIs) were chemotherapy use (≤14 days before death, 15-31 days before death, > 31 days), ED/inpatient admissions (0, 1,≥ 2) and not receiving hospice care in the last month of life. Covariates were patient age, gender and education; oncologist specialty; cancer aggressiveness; study arm and site (Western NY, Northern CA).


Data were abstracted from medical charts for all 151 patients who died during the course of the study. Median survival was 16 months. Patients of physicians who reported being more comfortable with medical paternalism were more likely to receive chemotherapy in the last month of life; these patients were also more likely to have multiple ED/hospital admissions. Patients who expressed a preference for experimental treatments were more likely to receive chemotherapy in the final month of life. Patient attitudes toward palliative care, patient preferences for life-sustaining treatments, and physician comfort with medical paternalism were not significantly associated with hospice use.


In this proof-of-concept study of psychosocial risk factors for PIIs, physician comfort with medical paternalism and patient preference for experimental treatments were independently associated with receipt of chemotherapy at the EoL. Patients of physicians who report being very comfortable with paternalism were more likely to use other costly services. Interventions that directly target or are tailored to certain decision-making norms (paternalism) and patient preferences for experimental treatments might decrease PIIs, improve quality of care at the EoL, and help address the crisis of affordability of care.