Methods for studying treatment heterogeneity using large observational databases

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): We propose a multi-phase program of research to develop, validate, apply and disseminate methods for estimating and communicating individual level heterogeneity in effectiveness and safety (for a range of patient/clinician identified outcomes of interest) for non-randomized comparative effectiveness research. The research program will translate theory to practice, and produce a template for clinical decision support tools that could be used to facilitate personalized medicine and informed decision making while addressing the personal characteristics, conditions and preferences of individual patients. Recent innovations in statistical methods for analyzing randomized clinical trials involve prediction of individual patient probabilities of experiencing benefit and/or harm. During the K99 phase of the award, I will work with my mentors and collaborators as we develop methods for non-randomized comparative effectiveness research which allow us to predict heterogeneous individual level treatment effects based on a constellation of patient characteristics. During this time, I will participate in the Postdoctoral Medical Informatics Research Training program at Harvard Medical School with a focus on training related to clinical and population health informatics. I will also be an active participant in education and training on topics such as patient/stakeholder engagement through the Brigham and Women's Hospital (BWH) Center for Patient Centered Comparative Effectiveness Research (PCERC). During the R00 phase of the award, I will solicit input from patient and physician members of the Patient and Family Advisory Councils (PFAC) at BWH PCERC regarding choices between alternative lipid lowering therapeutic strategies. I will work with them to identify safety and effectiveness outcomes that are priority concerns for the stakeholder groups they represent and understand what they expect from personalized treatment information. We will apply validated methods for estimating individual level treatment heterogeneity in studies using large, diverse healthcare databases. These studies will be designed to address the patient and provider identified priority questions. There will be continuous input from patients and providers as I work on developing clinical decision support tools designed to communicate relevant evidence, using metrics that healthcare consumers understand. I will use the skills developed through the Medical Informatics Training Program and collaborate with members of the Clinical Informatics Department at BWH to implement and test an evidence-based clinical decision support tool that is compatible with the hospital informatics infrastructure. This tool could be used during the office visit to help physicians and patients discuss the expected risks and benefits that are particular to each patient at a "critical moment", during the office visit when they the treatment decision is being made.