Training Physician Data Scientists: The Duke LHS Training Program at 1 year Nick Wysham, MD AAMC Learning Health Systems Kickoff February 12, 2015 Outline • • • • • • • • Background Curriculum Trainees Projects Challenges/ opportunities Future Directions Summary Acknowledgments Greene et al. Ann Intern Med, 2012;157:207-10. Evolution in Application of Evidence to Care It is time for EBM 2.0! • Evidence is derived from formal research plus systematic observation of clinical practice • Need environment where the use of real-time data • Drives discovery and continuous improvement • Becomes a routine part of patient care • Ultimately improves both individual and population health • Contemporary clinical practice demands efficiency, focuses on value, and incorporates electronic tools • Data data everywhere • Physicians have minimal exposure to informatics, data interpretation and practice based assessment • Continuous quality improvement models expanding LHSTP Curriculum Goals • Teach LHS Concepts • • • • • • Data use Clinical informatics Data quality Statistics PDSA-cycles EHRs and IT solutions to collect and review data • Infrastructure for trainees to learn concepts • Practical projects to apply concepts • Align with health system priorities Curricular Goals Acquisition of analytic skills Boot camp Familiarity with DUHS Data Introduction to core LHS concepts DEDUCE Training Sessions on research and statistical methods Introduction to Performance Services Understanding of regulatory boundaries for QI work Project Development/Delivery Working with DUHS operational leaders to determine systemConnections with key informants wide priorities for projects in DUHS who can further access Using analytic skills and data to systems-based data understand problems and possible solutions Applying project results to drive iterative improvement and outcomes assessments Topics and Lecturers • Framing a Clinical Question: – Greg Samsa, PhD statistics and bioinformatics • RedCap and eCRF: – Ursula Rogers, database analyst • Data Storytelling: – RJ Andrews • Chart Review and Data Management – Greg Samsa, PhD • Longitudinal Data – Greg Samsa, PhD • QI in Healthcare – Jonathan Bae, MD, Hospitalist and Chief of Quality for Department of Medicine • Overview of the Logic in Statistical Methods in QI and LHS – Alex Cho, MD, MBA and Amy Abernethy, MD, PhD • QI Part II – George Cheely, MD and Joe Boggan, MD • NIH Collaboratory – Rachel Richesson, RN, PhD • Project Charters • LHSTP Bootcamp – Amy Abernethy, MD, PhD • LHS Frames of Reference – Erich Huang, MD, PhD, Department of Surgery and Bioinformatics and Biostatistics Trainees • Cohort 1 – 3 medicine residents – 3 fellows (HOx2, and PCCM) • Cohort 2 – 1 medicine resident – 3 fellows (Rheum, Nephrology, HO) – 2 Surgery residents – 1 anesthesia resident – 2 neonatology fellows LHS Projects • Cohort 1 – Appropriateness of DVT prophylaxis – Dexemedetomidine for ICU sedation • Cohort 2 – Maternal Mg and neonatal bowel perforations for severely pre-term infants – Nomogram for predicting outcomes in the era of enhanced recovery colorectal surgery – Streamlining emergent initiation of chronic dialysis Example: Appropriateness of ICU Sedation • BIG question! • Findings: – Health System is outlier with respects to dex expenditure and ICU LOS – Institutional sedation practice is not guideline consistent – Necessary data is not captured • RAAS, Pain, Delirium scores – missing, inaccurate? – Pain scores not validated Example: ICU Sedation • Deliverables: – Presentation to DUHS leadership: CMO • Need for reliable metrics – Interface with performance services data analysts » Utilization » Appropriateness – Nursing education » Assessments » Practice standards • Need for system-wide sedation protocol – Segued into system-wide initiative • With LHS trainees in leadership roles Trainee Feedback 9% • Web-based Survey: – 11/15 response rate 27% 64% – 9/11 endorsed “topics were of interest” – 6/9 would recommend LHSTP to peers Positive Negative Neutral Trainee Comments • Initial expectations: – Centered on acquiring QI and statistical skills • Project selection: – Desire to progress towards implementation • Desired content: – Statistics – Unifying concepts • Impact on career: – “How hard "Big data" or "data mining" really are. Data and and insight are not synonymous terms, and that has become markedly apparent. In addition the willingness to think through data structure and data architecture in advance of a major IT system are critically important. Within EPIC there is some subtle but important nuances that clearly did not have a clinical input and has made finding data on the back end more challenging. Overall though there is huge promise to this and would be helpful moving forward.” LHSTP Challenges • Evolving curriculum – No road map! – Need to integrate topics • Diverse expertise/experience – Targeting offerings to learners • Time – Coordinating schedules with clinically active learners/instructors • Learning from feedback – Goal near 100% “recommend to others” Opportunities • Fortuitous Collaborations – Ex. nephrology with surgical oncology/bioinformatics • EHR tool to screen for risk to progress to ESRD – Trainees with hospital leadership – EHR-based VAE surveillance project • Enhanced presence in university • Potential for GME concentration/certificate Acknowledgements: Funding • Appreciation/support for AAMC pioneer award – Kicked off visibility for program • Led to presentation to CMO – Led to significant institutional support • Office of the Chancellor for Health Affairs, Duke University • Duke University School of Medicine, Department of Medicine • Duke Institute for Healthcare Innovation (DIHI) • DCRI Center for Learning Health Care (CLHC) Future directions • Statistical and database support for trainees • Optimize trainee selection – Ensure trainees can commit time and effort • Scheduling challenges – Smaller Cohorts? • Continued curriculum development – Unifying themes, cohesiveness – Can this curriculum to translated to other centers? • Carry forward selected projects towards implementation – Rapid data feedback cycle – Complete the circle Summary • A program in Learning Health for clinical trainees is feasible – Challenges inherent to starting educational program • Successful methods include multidisciplinary expertise and hands-on projects • Such a program can form the nucleus of diverse expertise with common goals – Necessary for implementation of LHS ideal More Thanks! • Leaders – Amy Abernethy, MD, PhD – Aimee Zaas, MD, MHS • Lecturers • Project Mentors • My fellow “trailblazers” – – – – – Lynn Howie, MD, MA Aaron Mitchell, MD Kevin Shah, MD, MBA Krish Patel, MD Angela Lowenstern, MD • 2014-15 Cohort – Jennifer Hauck, MD – Leslie Pineda, MD – Anastasiya Chystsiakova, MD – Jeffrey Yang, MD – Mohammed Adam, MD – Blake Cameron, MD – Melissa Wells, MD – John Yeatts, MD