Spending and Utilization Indicators in the New Jersey Medicaid ACO Demonstration Project.

Derek DeLia, Rizie Kumar, Jose Nova, Kristen Lloyd, David Goldin

Publication Date: 11/01/2018

This report provides a first assessment of spending and utilization indicators related to the New Jersey Medicaid Accountable Care Organization (ACO) Demonstration Project (NJ P.L. 2011, c.114). It documents quarterly trends in total costs of care (TCOC), emergency department visits, total inpatient admissions, preventable admissions, readmissions, and post-discharge follow-up visits during the pre-Demonstration period (2012-Q1 through 2015-Q2) and the first year of the Demonstration (2015-Q3 through 2016-Q2). As described below and in other reports, Demonstration Year 1 was predominantly a transitional year, as the ACOs were just beginning to receive data feeds from the state and were contending with an uncertain funding landscape. Thus, the information in this report reflects mostly baseline conditions that the ACOs did not expect to alter in a significant way in Year 1. Moreover, this report is part of a sequence of evaluation documents and should not be viewed in isolation from the rest of the sequence.

Overall, trends in the spending and utilization indicators among the three certified ACOs moved roughly in line with trends in a statistically similar comparison group. The analysis provides some additional detail about Camden residents who were members of United and Horizon health plans, which had executed shared savings arrangements with the Camden ACO. Among Camden’s United members, there was an apparent decrease in total inpatient admissions and preventable admissions around the time of the execution of the shared savings arrangement (which occurred before the start of the statewide Demonstration). Among Camden’s Horizon members, there was a shallow decline in readmissions, which took place throughout the study period and not coincident with the timing of the shared savings arrangement. These observed trends, however, are statistically “noisy” with no consistent indication of statistical significance.