Understanding the groups of care transition strategies used by U.S. hospitals: an application of factor analytic and latent class methods

Abstract Background After activation of the Hospital Readmission Reduction Program (HRRP) in 2012, hospitals nationwide experimented broadly with the implementation of Transitional Care (TC) strategies to reduce hospital readmissions. Although numerous evidence-based TC models exist, they are often adapted to local contexts, rendering large-scale evaluation difficult. Little systematic evidence exists about prevailing implementation patterns of TC strategies among hospitals, nor which strategies in which combinations are most effective at improving patient outcomes. We aimed to identify and define combinations of TC strategies, or groups of transitional care activities, implemented among a large and diverse cohort of U.S. hospitals, with the ultimate goal of evaluating their comparative effectiveness. Methods We collected implementation data for 13 TC strategies through a nationwide, web-based survey of representatives from short-term acute-care and critical access hospitals (N = 370) and obtained Medicare claims data for patients discharged from participating hospitals. TC strategies were grouped separately through factor analysis and latent class analysis. Results We observed 348 variations in how hospitals implemented 13 TC strategies, highlighting the diversity of hospitals’ TC strategy implementation. Factor analysis resulted in five overlapping groups of TC strategies, including those characterized by 1) medication reconciliation, 2) shared decision making, 3) identifying high risk patients, 4) care plan, and 5) cross-setting information exchange. We determined that the groups suggested by factor analysis results provided a more logical grouping. Further, groups of TC strategies based on factor analysis performed better than the ones based on latent class analysis in detecting differences in 30-day readmission trends. Conclusions U.S. hospitals uniquely combine TC strategies in ways that require further evaluation. Factor analysis provides a logical method for grouping such strategies for comparative effectiveness analysis when the groups are dependent. Our findings provide hospitals and health systems 1) information about what groups of TC strategies are commonly being implemented by hospitals, 2) strengths associated with the factor analysis approach for classifying these groups, and ultimately, 3) information upon which comparative effectiveness trials can be designed. Our results further reveal promising targets for comparative effectiveness analyses, including groups incorporating cross-setting information exchange.