Each year, Brown University awards Research Seed Funds—intended to support competitive research proposals, including preliminary work and collaboration—and Salomon Faculty Research Awards—designed to recognize excellence in scholarship.

Carolina Hass-Koffler, associate professor of psychiatry and human behavior and of behavioral and social sciences, along with co-PI Erica Eaton, assistant professor of psychiatry and human behavior and of behavioral and social sciences, will receive a Seed Award for their project Safety, Feasibility, and Acceptability of MDMA-Assisted Therapy for the Treatment of Co-Occurring Posttraumatic Stress Disorder and Alcohol Use Disorders in Combat Veterans. Richard B. Salomon Faculty Research Awards will provide funding to Kaley Hayes, assistant professor of health services, policy, and practice, for Leveraging New Databases to Understand Medication Use in the Post-Acute Care Setting among Older Adults with Hip Fracture, and to Arman Oganisian, assistant professor of biostatistics, for Bayesian Machine Learning for Sequential Decision-Making with Incomplete Information.

Learn more about these funded research projects below.

SEED AWARD

Safety, Feasibility, and Acceptability of MDMA-Assisted Therapy for the Treatment of Co-Occurring Posttraumatic Stress Disorder and Alcohol Use Disorders in Combat Veterans
PI: Carolina Hass-Koffler, associate professor of psychiatry and human behavior and of behavioral and social sciences
Co-PI: Erica Eaton, assistant professor of psychiatry and human behavior and of behavioral and social sciences

Co-occurring PTSD and alcohol use disorder (PTSD-AUD) is common following combat and associated with more severe symptomatology, increased suicidality, and poorer response to treatment than either disorder alone. Available PTSD-AUD treatments effectively treat only a fraction of people who engage in them for adequate dose and duration, leading to growing interest in alternative medications, including psychedelics. The combined neurobiological effects of MDMA increase compassion, reduce defenses and fear of emotional injury, and enhance communication and introspection, making MDMA-AT especially useful for treating PTSD-AUD. This pilot trial will be the first to assess feasibility and acceptability of MDMA-assisted psychotherapy (MDMA-AT) in veterans with combat-related PTSD and AUD (N=20) and will result in a new interdisciplinary collaboration at Brown. Participants will be recruited via social media and clinician referrals and will complete an initial screen and baseline appointment including informed consent. Eligible participants will receive MDMA-AT, including three Experimental Sessions with MDMA administration that will be conducted under established protocols. Follow-up data will be collected at post-treatment and at one-month. This project will allow us to: 1) assemble a research team including the training of two MDMA-AT clinicians, 2) determine the feasibility of recruitment, 3) determine the acceptability of and safety of MDMA-AT, 4) provide preliminary evidence of the effects of MDMA-AT, and 5) refine study procedures in preparation for a fully powered RCT to test the effectiveness of MDMA-AT for PTSD-AUD. This collaboration will help to position Brown at the forefront of psychedelic research for common and impairing mental health problems.

SALOMON FACULTY RESEARCH AWARDS

Leveraging New Databases to Understand Medication Use in the Post-Acute Care Setting among Older Adults with Hip Fracture
Kaley Hayes, assistant professor of health services, policy, and practice

Real-world data are critical to understand drug effects in populations that are excluded in randomized controlled trials (RCTs). US Medicare claims are among the most powerful real-world data, given their large size and inclusion of older adults who are often excluded from RCTs. An important area of real-world research for older adults is health outcomes that result from transitions of care (e.g., discharge from a hospital to a post-acute care skilled nursing facility [SNF]). Medication changes during this period are particularly important to understand, as drugs are often started or stopped due to competing health demands. However, Medicare data do not capture medication use in post-acute care SNFs because of bundled payments. Thus, to date, we have been unable to examine medication use and effects in the post-acute care SNF setting. Our team has recently acquired Omnicare long-term care pharmacy data (>60% of US nursing homes) that can be linked to Medicare claims to fill this gap. We propose a pilot project to use Omnicare data in a retrospective cohort study of older adults who experience a hip fracture and are discharged to a SNF for post-acute care. First, we will examine initiation of analgesic regimens post-fracture. Then, we will explore which of a patient’s medications used before the hip fracture are continued or discontinued in the post-acute care SNF setting. This project will provide proof of concept, preliminary data, and an established track record for the team to enhance an R21 proposal on discontinuing medications during transitions of care.

Bayesian Machine Learning for Sequential Decision-Making with Incomplete Information
Arman Oganisian, assistant professor of biostatistics

In our application of interest, we observe data on children diagnosed with acute myeloid leukemia (AML). Patients move through a sequence of treatment courses, with physicians deciding the next treatment given response to past treatments. Our goal is to answer important clinical questions: what would, say, 2-year survival rate have been had patients followed one treatment sequence versus another? Is there an “optimal” sequence that maximizes 2-year survival? In practice, confounding impedes direct attribution of survival improvements purely to the treatment sequence. For example, patients following less aggressive sequences may have better cardiac function throughout and, therefore, better survival prospects even had they followed another sequence. Moreover, adjusting for confounders (e.g. cardiac function) is difficult since they are irregularly measured over time – yielding incomplete information. We propose modeling the decision process via robust, state-of-the-art Bayesian machine learning (ML) methods which simultaneously adjust for confounding and sequentially impute missing values over time. Though motivated by AML, potential applications range from health policy to economics. For instance, public health policies are often rolled out sequentially (e.g. statewide vaccination done county-by-county) and we may be interested in estimating an optimal rollout schedule. Economists often model sequential pricing decisions as firms respond to competitors, with the goal of estimating an optimal pricing strategy. This proposal therefore has intrinsic merit and potential for broad impact in addressing complexities encountered across a range of interdisciplinary areas. Moreover, our findings will help launch a broader future research program developing Bayesian ML methods for sequential decision-making.