Value-Based Payment in Total Joint Arthroplasty: A Healthcare Market Segmentation Methodology to Improve Valu

The U.S. spends more on healthcare per capita each year than any other nation in the world, yet consistently underperforms other major developed nations in terms of healthcare quality, timeliness, cost, and access to care. Recognizing a need to reform, the U.S. passed the 2010 Patient Protection and Affordable Care Act (ACA) authorizing a Center for Medicare and Medicaid Innovation whose mission includes the development of new payment models that transform the nation from volume- to value-based care. In total joint arthroplasty (TJA), the most common and costly service covered by Medicare, the transformation is manifested in the comprehensive care for joint replacement (CJR). The CJR is a bundled payment initiative that holds hospitals financially accountable for a patient's outcome over a 90 episode of care. Because hospital reimbursement is tied to the total cost of care delivered by multiple independent care providers, hospitals are incentivized to take an active role in coordinating care through the episode to ensure a quality health outcome. The CJR necessitates a new approach to how hospitals manage joint patients from the pre-operative through the post-operative phase of care.;The overall objective of this research is to investigate how a healthcare market segmentation methodology can be implemented by any hospital subject to the CJR to improve healthcare value. The methodology applies machine learning, regression, and process improvement methods, and the analysis behind the methodology uses latent data stored in hospital electronic health records (EHRs). This new methodology has five pillars and addresses four separate but related questions: 1) how can data mining and market segmentation be used to identify unique and distinguishable patient segments (clusters), 2) how do hospitals accurately classify prospective TJA patients such that their assigned segment provides useful information for clinicians or other health professionals, 3) how can clinicians identify cost drivers in TJA that are tied to a patient's cluster, their attributes, or other clinical factors, and 4) how do hospitals identify patient outcome drivers. Using operations research to build a methodology around these four questions will help hospitals adapt to the current transformation in healthcare payment models.;The main body of this dissertation is divided into three sections: 1) clustering and classification of patients, 2) identifying cost and outcome drivers, and 3) assessing interventions that reduce costs and improve healthcare value. In Chapter 3, clustering and classification models applied to EHR data divide a healthcare population into smaller segments for which health interventions can be applied. In Chapter 4, patient segments along with attributes such as gender, procedure code, and comorbidity burden are shown to be predictors of 30 and 90 day readmission and increased supply costs. Additionally, Chapter 4 highlights the impact that a small group of patients with complex bony or ligamentous deformities or infection risk have on a hospital under a value-based payment model. If not managed, this small subset of patients is shown to consume a disproportionate share of the overall implant budget for a hospital. In Chapter 5, two interventions are studied. A rapid recovery protocol focused on individual patients is shown to significantly improve length of stay and discharge to home rate while not impacting readmission rates. Finally, applying a clustering model using attributes related to cost and outcome drivers helps identify a small but high cost segment of the TJA population. A new approach to managing these high cost patients is introduced and modeled using simulation with significant cost savings. This dissertation provides a roadmap for hospitals seeking to improve healthcare value under value-based payment models.