Design Choices in Randomized Partnered Evaluations: Veteran-Directed Home and Community Based Services Melissa M Garrido, PhD @GarridoMelissa PE PEPR PReC eC Partnered Evidence-based Policy Resource Center VETERANS HEALTH ADMINISTRATION
PEPR PEPReC eC Overview of PEPReC Partnered Evidence-based Policy Resource Center • HSR&D / QUERI funded resource center • Objective: Provide timely, rigorous data analysis to support the development of high-priority policy, planning, and management initiatives and quantitative program evaluations with strong potential to improve the quality and efficiency of VA healthcare • Director: Austin Frakt, PhD VETERANS HEALTH ADMINISTRATION 2
PEPReC PEPR eC Design Choices in Randomized Program Evaluations Partnered Evidence-based Policy Resource Center • Type of randomized design – Level of randomization? – Timing and distribution of exposure to the intervention? • Choice of allocation technique – Simple or restricted randomization? VETERANS HEALTH ADMINISTRATION 3
Veteran-Directed Home and PEPR PEPReC eC Community Based Services Partnered Evidence-based Policy (VD-HCBS) Resource Center • Goal: Reduce risk of unwanted placement in nursing home or other long-term care facility VETERANS HEALTH ADMINISTRATION 4
PEPReC PEPR eC VD-HCBS Evaluation Goals Partnered Evidence-based Policy Resource Center • Rigorously evaluate effects of VD-HCBS on health outcomes and costs • Work with operations and research partners to produce evidence that will inform GEC’s decisions about the best ways to prevent unnecessary institutionalization of older Veterans VETERANS HEALTH ADMINISTRATION 5
PEPReC PEPR eC Choice: Level of Randomization Individual vs. Cluster Partnered Evidence-based Policy Resource Center • Considerations – Feasibility – Contamination of study groups – Statistical power • Our choice: Cluster (medical center) randomization VETERANS HEALTH ADMINISTRATION 6
Choice: Timing and Distribution of PEPR PEPReC eC Exposure to Intervention Partnered Evidence-based Policy Resource Center Parallel vs. Stepped Wedge Time Time Time Time Time Clusters 1 2 3 4 5 • Considerations 1-7 – Equipoise 8-14 – Feasibility 15-21 22-28 – Statistical power Time Time Time Time Time Clusters 1 2 3 4 5 • Our choice: Stepped wedge 1-7 8-14 15-21 22-28 VETERANS HEALTH ADMINISTRATION 7
VD-HCBS Evaluation: PEPR PEPReC eC Cluster Randomization Partnered Evidence-based Policy Resource Center VAMCs not currently participating in VD-HCBS (n=77) Excluded: • Recently or nearly completed readiness review • Insufficient buy-in from VAMC stakeholders From eligible VAMCs, GEC will identify ~14 sites willing and able to implement VD-HCBS and refer Veterans within the next six months Repeated every 6-10 Randomization months from 2017-2019 Early enrollment Late enrollment VETERANS HEALTH ADMINISTRATION 8
PEPR PEPReC eC VD-HCBS Evaluation: Stepped Wedge Design Partnered Evidence-based Policy Resource Center Every eligible site will participate in VD-HCBS during the evaluation 3/2017 6/2017 9/2017 12/2017 3/2018 6/2018 9/2018 12/2018 3/2019 6/2019 9/2019 12/2019 VAMCs 1-7 8-14 15-21 22-28 29-35 36-42 43-49 50-56 57-63 64-70 71-77 Start times and exact number of sites in each step subject to change VETERANS HEALTH ADMINISTRATION 9
PEPR PEPReC eC Choice: Allocation Technique Simple vs. Restricted Randomization Partnered Evidence-based Policy Resource Center • Considerations – Number of units – Existence of potential confounders – Knowledge of partial or full list of participating units • Our choice: Restricted randomization (covariate constrained randomization) VETERANS HEALTH ADMINISTRATION 10
PEPR PEPReC eC Simple Randomization Partnered Evidence-based Policy Resource Center Early Enrollment Eligible Sites Simple Randomization Late Enrollment VETERANS HEALTH ADMINISTRATION 11
PEPR PEPReC eC Matching Partnered Evidence-based Policy Resource Center Random Assignment within Pairs Matched Pairs Early Enrollment Eligible Sites Matching Late Enrollment VETERANS HEALTH ADMINISTRATION 12
PEPR PEPReC eC Stratification Partnered Evidence-based Policy Resource Center Random Assignment within Strata Strata Early Enrollment Eligible Sites Stratification Late Enrollment VETERANS HEALTH ADMINISTRATION 13
PEPReC PEPR eC Covariate Constrained Randomization Partnered Evidence-based Policy Resource Center Covariate Constrained Randomization Option 1 Option 2 Option 3 Eligible Sites Early Enrollment Early Enrollment Early Enrollment Late Enrollment Late Enrollment Late Enrollment VETERANS HEALTH ADMINISTRATION 14
VD-HCBS Evaluation: PEPReC PEPR eC Covariate Constrained Partnered Evidence-based Policy Randomization Resource Center Site-Specific State or County Patient Case-Mix Patterns of Caring Access to HCBS for Older Patients Size of patient Urban/rural location population VAMC spending on HCBS State participation in CAN scores early participant- VAMC has CLC on directed care Jen Frailty Index campus initiatives scores Market penetration State Medicaid Prospective NOSOS of HCBS spending on HCBS scores VETERANS HEALTH ADMINISTRATION 15
PEPR PEPReC eC Summary Partnered Evidence-based Policy Resource Center • Randomized partnered program evaluations require decisions about unit of randomization, timing of and exposure to intervention, and allocation technique • Decisions need to take into account feasibility, partners’ practical concerns, and ability to generate rigorous evidence VETERANS HEALTH ADMINISTRATION 16
PEPR PEPReC eC Acknowledgements Partnered Evidence-based Policy Resource Center • VD-HCBS Evaluation Team – VHA Office of Geriatrics & Extended Care – Administration for Community Living – PEPReC – Center of Innovation in Long Term Services and Supports – Center for Health Services Research in Primary Care – The Lewin Group – Applied Self Direction • Funding – QUERI: PEC 16-001 / HSR&D: SDR 16-196, CDA 11-201/CDP 12-255 VETERANS HEALTH ADMINISTRATION 17
PEPReC PEPR eC Questions? Partnered Evidence-based Policy Resource Center melissa.garrido@va.gov http://www.isrctn.com/ISRCTN12228144 The views expressed in this presentation do not necessarily reflect the position or policy of the Department of Veterans Affairs or the United States government. VETERANS HEALTH ADMINISTRATION 18
Recommend
More recommend
