Statistics Core

点击翻译按钮获取中文摘要

PROJECT SUMMARY ED-LEAD is proposing an embedded pragmatic clinical trial of three independent yet potentially synergistic interventions all targeted at improving the care of Persons Living with Dementia (PLWD) and their care partners. The three interventions – emergency care redesign, nurse-led telephonic care, and community paramedicine – all focus on PLWD who present to the Emergency Department (ED) for care and the need for careful attention to care transitions in the triadic encounter between the PLWD, care partner, and healthcare team. These interventions will share patient-level outcomes that will benefit from a joint analysis. The proposed randomization structure will be based on a multifactorial design, where EDs will be randomized to any combination of the three interventions. This design will generate substantial quantities of data that will need to be evaluated to assess implementation fidelity of each intervention and assess a range of intervention-specific and universal outcomes. The Statistical Analysis Core (SAC) will provide biostatistical expertise for the overall project. The SAC’s key function is to develop the modeling framework that will enable the study team to evaluate each intervention individually and in combination with others. The factorial design is a key element of the joint study that will allow the investigators to explore how the interventions might work together to have a greater impact than any single intervention. Traditional methods of statistical inference typically require very large sample sizes to perform complex factorial experiments. Furthermore, unreasonable assumptions regarding the absence of interaction effects are sometimes required in analyzing a factorial design. We have developed a Bayesian modeling approach that will enable us to present the results in a way that will allow health care providers, health care systems, and health policy makers to assess the individual and joint impacts of these three very different interventions never evaluated simultaneously. With the analytic framework serving as a foundation, the SAC will support data management related to patient-level health utilization data, training and intervention fidelity, non-CMS, intervention-specific patient-level outcome measurement, and intervention- specific implementation outcome measurement. The SAC will oversee the randomization process to ensure that the 80 ED sites participating in the study are distributed across the eight arms of the factorial design in a way that minimizes imbalance of key site-level characteristics, such as location and size. The SAC will perform statistical analyses and data exploration using appropriate statistical and computing methodologies, and assist in interpreting and presenting results.