Longitudinal Clustering of High-cost Patients' Spend Trajectories:Delineating Individual Behaviors from Aggregate Trends.

For the past decade, there has been a concerted effort to fulfill the promises of the Triple Aim-improving patient experience and quality while lowering costs by focusing on high-cost patients via Care Management. Despite the well-known fact that high-cost patients make up roughly half of all annual medical expenses, little has been studied about high-cost patients' cost trajectories. This paper focuses on the trajectory patterns for high-cost Medicare patients, which provides another dimension to understanding optimal program intervention. We performed a retrospective observational study employing Longitudinal K-Means Clustering. We discovered there are two major categories based on overall utilization: "persistent" and "episodic". Both cohorts churn to some degree in the post high-cost year. These results highlight that high-cost patients churn, and the existence of high-cost patient sub-cohorts warrant further exploration around the patient profile for each cohort. This finding could influence the current dialogue about the understanding and ability to impact high-cost patients through appropriate intervention and, more importantly, at the right time to attenuate the cost and improve quality of care.