Characterizing High-Cost, High-Need Patient Populations and Response to Care Management Programs at a NetworkModel Medicare Advantage Health Maintenance Organization

There will be two aims. Aim 1 will describe and categorize high-need, high-cost populations. The investigators will use CareMore inpatient, outpatient, and pharmacy claims from years 2010-2014. The investigators will construct mortality and readmission risk category and they will utilize traditional claims-based risk models developed on out of sample data to categorize patient risk level. For the high health expenditure category, they will examine inpatient, outpatient, and pharmacy data to identify groups of patients who drive disproportionate health spending. Finally, for utilization the investigators will use claims and outpatient visit data (to the extent any electronic health record (EHR) data becomes available) to identify patients with disproportionate visit intensity. The approach to describe patients within these groups will be to categorize patients into deciles and utilize multivariate regression models to identify demographic, clinical, socioeconomic, and geographic characteristics associated with presence in a top decile along each metric. For example, the investigators will describe individual and clusters of diagnoses that are associated with high spending. They will also examine patients who meet multiple criteria and describe associated characteristics. Aim 2 will examine the impact of CareMore's end stage renal disease (ESRD) care model. They will use CareMore inpatient, outpatient, skilled nursing facilities (SNF), inpatient rehabilitation facilities (IRF), other facilities, pharmacy, and post-acute care claims and Center for Medicare and Medicaid (CMS) Fee-For-Service (FFS) data from matched geographies. They will also attempt to obtain the healthcare effectiveness data information set (HEDIS) data from health plan at patient and physician level, CMS stars measure data, available electronic medical record (EMR) data, and physician characteristics data to the extent CareMore can provide. They will perform a retrospective analysis of claims data. The investigators will use a primary comparison group of ESRD patients in Medicare FFS, particularly in localities where no ESRD special need plans (SNP) is offered to mitigate some effects of selection. The team will also explore a second comparison group of CareMore traditional MA plan enrollees who develop ESRD. They will first examine patient, physician, and plan or site characteristics associated with variation in utilization of the specialized services including enhanced nutritional counseling, personalized care plans, dialysis treatment evaluation, and supplemental medical evaluations -- using logistic regression. Second, they will perform a descriptive analysis of the association between use of these services and cost and quality outcomes.