Provider-Payer Partnerships as an Engine for Continuous Quality Improvement

In Arizona, a regional behavioral health authority (RBHA) is contracted by the state to serve as the point of accountability for behavioral health service design and delivery for a given region. The RBHA is financed via Medicaid and other funds, and it contracts with multiple providers to deliver a complete continuum of care. Cenpatico Integrated Care, a partnership between Centene Corporation and Banner–University Family Care, has been the RBHA for southern Arizona since October 2015.

The crisis response center (CRC), located in Tucson on the campus of Banner–University of Arizona Medical Center South, was built in 2011 with Pima County bond funds to reduce the numbers of persons with behavioral health needs in jails and emergency rooms. The CRC has been managed by Connections Health Solutions since 2014 and provides 24/7 urgent care, 23-hour observation, and brief inpatient care to approximately 12,000 adults and 2,400 youths annually.

In January 2016, CRC and RBHA leadership began a series of regular conference calls to discuss and problem-solve system issues that led to crisis center utilization. By spring, the group had developed a process for working as a collaborative QI team, using data sharing and analysis to generate rapid-change cycles for the purpose of improvement (Figure 1).

Comparing fiscal years 2016 and 2017, there was a 30% decrease in 72-hour return visits to the observation units, dropping from 1.6% to 1.1% of youths (p<.03) and from 3.3% to 2.2% of adults (p<.02) (72-hour return visits are defined as the percentage of patients discharged from the observation unit who return to the observation unit within 72 hours, as described previously [3]). There was a 50% decrease in the percentage of adults discharged from the inpatient unit who were readmitted to the inpatient unit within 30 days, from 5.5% to 2.8% (p<.01).

This project demonstrates how crisis providers and payers can collaborate to improve care at both the individual and system level. Crisis utilization data provided a mechanism for more up-to-date analysis of trends, and regular collaborative QI meetings facilitated the rapid development of multiple QI cycles. Outcomes were improved for individuals in the high-utilizer group, as indicated by decreased crisis visits. In addition, the decrease in readmissions and decrease in total number of high utilizers suggests improvements in the system as a whole, as people were better able to get their needs met without accessing crisis services. Furthermore, these outcomes were achieved via more efficient use of existing resources rather than through increases in staffing or cost.

The project also underscores the importance of the composition of the team in QI initiatives (4). It is critical to include both leaders who can act as decision makers and individuals who have first-hand experience working in the environment that is the target of improvement efforts.

Using a formal data-driven QI process—instead of a traditional research approach—has been essential for our success. To appreciate why that is true, it is important to understand the different methods and indications for each method (5).

Both research and QI have value in improving services and systems, but they must be understood as distinct processes. Research attempts to add to the general body of knowledge regarding evidence-based care, and it is by necessity slow and methodical. Research methods isolate a particular intervention, control all other variables if possible, and test whether that intervention achieved the desired result. To do so, researchers test hypotheses one at a time in controlled settings with well-defined subsets of patients in order to minimize confounding effects.

In contrast, QI is rapid and iterative. It focuses on applying what is already known to solve or improve complex real-world problems rather than on expanding the general body of knowledge. Many interventions may be tried in rapid succession or at the same time, in uncontrolled natural processes. Results are measured quickly to see whether improvement has occurred, and modifications are immediately applied and reanalyzed, Furthermore, there is no control group; interventions are applied to everyone, and recognition of confounding variabilities is welcomed, rather than “screened out.” However, like research, QI is thoroughly data driven and follows formal rules and processes.

In this project, we employed QI rather than a research approach in order to deploy multiple interventions rapidly and simultaneously. For example, in the school-based crisis intervention, there was no comparison group, many other improvements were made at the same time, and we did not control for differences in why certain schools call 911 more than others. Rather, the value of the QI approach was that it provided a process for studying the data so that one could discover that 911 calls from schools were a problem in the first place. Then the QI partnership team worked together to rapidly implement and test potential solutions. The details of the solution that ultimately worked may not be generalizable to other communities, where, for example, increasing school-based counseling services may be a better solution. The point is to have a process to discover problems, act on them quickly, test the results, modify as needed, and maintain what works.

Although this project involves a single crisis provider and one payer, we believe the most important element is the collaborative QI process itself. Based on the success of this project, similar protocols are now being developed between the RBHA and local emergency departments and between the CRC and other payers. Working with others to apply QI methods to achieve a common goal provides value to any organization in any community.