The use of chemotherapy at the end-of-life (EOL) is increasing, despite growing concerns that it may be burdensome, expensive, and potentially harmful. Between 20-50% of terminally-ill cancer patients receive chemotherapy in the last month of life (henceforth referred to as EOL chemotherapy). Rates are highest in ovarian cancer where patients may respond to multiple lines of chemotherapy, increasing uncertainty about when to stop treatment. Research suggests that EOL chemotherapy is associated with more aggressive EOL care and lower rates of hospice utilization, both of which are associated with worse patient quality of life (QoL) near death, higher costs, and more distress in bereaved caregivers. Recently, the American Society of Clinical Oncology (ASCO) identified EOL chemotherapy first on its "Top Five List" of widely used practices that could improve patient care while reducing costs, if stopped. The determinants of EOL chemotherapy use are poorly understood. Researchers from the Dartmouth Health Care Atlas have argued that cancer patients' EOL care is determined by medical supply factors, not patient preferences. In contrast, we have found that patients' treatment preferences, providers' behaviors, and the therapeutic alliance between patients and oncologists predict the intensity of advanced cancer patients' EOL care, both adjusting for and stratifying by health systems. Most recently, we have found that late-line chemotherapy is one of the most powerful predictors of aggressive EOL care in patients with advanced lung and gastrointestinal cancers. In Coping with Cancer, patients who received late-line chemotherapy, a median of 4.5 months before death, had 8-fold higher odds of undergoing ventilation or resuscitation in the last week of life, compared with patients who didnot. These preliminary results highlight the need to determine outcomes of late-line chemotherapy in other cancers, to understand better what factors influence the receipt of EOL chemotherapy, and to develop an intervention to improve EOL care. The overall goals of this project are: a) to examine outcomes of late-line chemotherapy (defined as e2 non- platinum-based regimens) in patients with platinum-resistant ovarian cancer, including whether patients are receiving care that is congruent with their preferences (treatment goal attainment), b) to identify modifiable determinants of EOL chemotherapy use to target in an intervention, and c) to develop an intervention to decrease EOL chemotherapy use. Two complementary data sources will be used for Aims 1&2: 1) a prospective, longitudinal, multi-institutional study of 200 patient with recurrent, platinum-resistant ovarian cancer (Coping with Cancer 2~ CwC2)~ and 2) a population-based cohort of 10,310 older patients with advanced ovarian cancer from SEER-Medicare. In Aim 1, I will use CwC2 data to examine associations between late-line chemotherapy and patient QoL, intensity of EOL care, and goal attainment~ and SEER-Medicare to examine patients' EOL health care utilization and patients' cancer and/or treatment related morbidity. In Aim 2, I will use CwC2's comprehensive information on psychosocial factors to examine modifiable patient and caregiver factors associated with EOL chemotherapy use, paired with a new physician survey which I will develop, refine, test, implement, and link to CwC2 data to identify modifiable physician factors associated with EOL chemotherapy. In Aim 3, I will develop and pilot an intervention at the Dana-Farber Cancer Institute and Massachusetts General Hospital to decrease EOL chemotherapy use, based upon my findings in Aims 1 and 2, which will likely target physicians since they are the most influential determinants of decisions to forgo life-sustaining treatments in other contexts.39 Together, these data will provide an evidence base to improve chemotherapy decision-making, identify novel and modifiable determinants of EOL chemotherapy, and allow me to target the these determinants in an intervention designed to decrease EOL chemotherapy use. The portfolio of proposed analyses, combined with close mentoring and specific training in decision-making, longitudinal data analysis, survey methodology, and intervention research, will equip me with the necessary skills and experience to transition from a mentored research fellow to become a successful, independent researcher.