A History Embedded Accelerated Failure Time Model to Estimate Nursing Home Length of Stay

Abstract

With its aging population, the United States is under increasing pressure to provide long-term care (LTC) coverage to its citizens and to manage their chronic health conditions. However, the research on LTC transition and utilization modeling remains in its infancy; needless to mention LTC resource allocation and transition pathway optimization. In this paper, we developed a parametric survival model to characterize nursing home (NH) length of stay (LOS), which incorporates information on transition history. In addition, the model addressed issues such as recurrent events and competing risks. To study the effect of covariates on LOS and to ensure the flexibility of the model in evaluating operational-level interventions, we elected to develop an accelerated failure time parametric survival model. We fit the model to care transition data collected from a large cohort of older adults receiving coordinated care in a Midwestern United States urban area. Through our study, we drew the following major conclusions: 1) transition history is a significant factor and a potential predictor of an individual’s LOS in NH; 2) significance of frailty terms indicates that LOS estimates based on data with recurrent transition events can be significantly biased if not accounted for explicitly; and 3) the same clinical covariate can have opposite effects on NH LOS, depending on the destination care setting. Finally, we identified better-suited baseline hazard functions and frailty terms in each survival model from several representative candidates. Findings from our model can aid in operational-level NH care transition and utilization policy development. This paper also serves as the basis for extension into network-wide LTC transition models and utilization simulators. Note to Practitioners —This paper was motivated by the complex problem of characterizing care transitions in an LTC network, based on integrated electronic health records. It exemplifies the beginning of a series of rigors investigations on modeling multisite care transition networks. This line of research has great potential to lead to the development of intelligent decision support systems for LTC capacity planning and transition management, and eventually, empower care organizations to make data-informed management decisions. Furthermore, the empirical aspect of this paper was made possible by the availability of a comprehensive care transition longitudinal data set, at the Regenstrief Institute, constructed from a multiyear follow-up study on an elderly cohort. This cohort was prospectively followed as it moved along a Midwestern United States urban LTC system. This problem is of great interest to population-level LTC policy makers and management of LTC providers. The majority of the existing studies on LTC delivery have relied on cost-effectiveness analysis based on aggregate spending data. Our study, on the other hand, focuses on characterizing time-to-transition event for patients who reside at NH. Such a shift in focus in modeling at an increased level of granularity enhances our ability to evaluate the impact of operational-level interventions on various outcomes of LTC delivery. Finally, the novelty of our study arises from exploring useful transition history patterns to incorporate into the parametric survival model and estimating resource use from censored data. Our research findings highlight the significance of: 1) identifying the most informative time window for incorporation of transition history as a LOS predictor in NH and 2) accounting for bias in transition time estimation arising from an individual tendency to stay longer or shorter in NH through frailty terms incorporated in the parametric survival model.