|
Abstract:
|
In cluster randomized trials (CRTs), groups of individuals (centers) are randomly allocated to an intervention or control arm and we are often interested in the intervention effect at the center-level; the primary focus is the center-level effect of the intervention, and centers rather than individuals are the entities of interest. Because symbolic data analysis builds on the notion that statistical inferences are commonly required at the group-level rather than at the individual-level, it is natural to think about CRTs within the symbolic data framework. The recently developed symbolic two-step method provides an alternative to the single-step multi-level model in the design and analysis of CRTs evaluating interventions within care delivery research. This method applies the symbolic data analysis framework to adjust for patient-level factors when estimating and testing effects of center-level factors on both the average center-level outcome and its variation. Estimation and testing of center-level effects on center outcome variation of the patient-adjusted outcomes is the innovation of the method. Herein, we recommend when to apply the method in the design and analysis of CRTs.
|
