Implementation of a Comprehensive Population Health Management Model

Introduction

The current delivery systems of primary care are unsustainable because they lack resilience to survive in new reimbursement environments where total cost of care, patient experience, and patient-centered outcomes are primary determinants of success. For primary care to remain relevant and viable, practical and transformative solutions must be created that meet 3 main goals: attract and retain new patients, provide consistently high and clearly measurable value, and deliver high-satisfaction services that meet the needs of patients and providers. This quest is imperative to financial viability and sustainability in future health care reimbursement realities. ^1,2 Our institution approached this conundrum by implementing a comprehensive population health management model across its affiliated primary care clinics in multiple geographic areas.

Population health management has received much attention in the medical literature and popular media, but the concept is poorly defined and implementation of a population health framework is not well understood. Herein, we describe the process of implementing a population health–based model within primary care, incorporating the experience of patients and providers of care services.

Background

Mayo Clinic leadership created an Office of Population Health Management to implement a new model of primary care focused on the Institute for Healthcare Improvement's Triple Aim. ³ The Office of Population Health Management strives to create a patient-centered, integrated care delivery model based on aligned incentives and coordinated, collaborative processes that are built on evidence-based prevention and disease management protocols. Our institution's care delivery model is supported by seamless sharing of information between all care delivery sites and community partners, and by wellness and chronic condition management programs that focus on patient engagement and activation. The model is driven by analytics to identify at-risk patients to enable more efficient use of clinical resources. The ultimate goal is support of high-quality outcomes that allow our clinical systems to thrive in an environment of value-based, accountable reimbursement.

Because implementation of the full model requires close alignment with changes in the reimbursement environment, we plan to stage the implementation over several years. We believe this strategy will allow clinics to continue to thrive in the current fee-for-service predominant environments while positioning them to succeed in future value-based reimbursement environments.

Mayo Clinic, a charitable, nonprofit organization, is a large academic medical center with its principal location in Rochester, Minnesota. The Mayo family pioneered the concept of a medical group practice before the turn of the 20th century. Mayo Clinic has grown to become the nation's largest such practice, with campuses in Jacksonville, Florida, and Scottsdale and Phoenix, Arizona. Its mission is achieved through the work of more than 4450 physicians and more than 57,000 allied health care professionals. The organization attends to more than 1 million unique patients each year from all regions of the United States and many foreign countries. Most of patients served (80%) live within a 120-mile radius of the Rochester campus and come from predominantly rural communities in Minnesota, Iowa, and Wisconsin.

The core of Mayo's population health care framework is the Mayo Model of Community Care (MMoCC). As described previously, ⁴ the 10 themes of the model incorporate existing best practices and development of new services in support of the population health framework.

This commentary describes the process used to implement a community health model across geographically distributed clinics and discusses the initial outcomes from the implementation of this new population health care model.

Implementation Methodology

Mayo Clinic's population health framework will be implemented through phases timed in conjunction with changes in payer contracts. Each phase is further divided into an α (limited) phase and a β (full) phase to test the model in a live clinical setting and quickly refine each phase through iterative improvement before broader implementation.

The internal implementation infrastructure included formation of a central α implementation team to support implementation at α sites and to ensure fidelity to the model of care across sites. We also asked each geographic region to develop a β implementation team. The β teams would work with the α teams to learn best practices from the initial implementation and would have responsibility for dissemination of the more mature model within their own regions in the subsequent phase of β implementation.

The α implementation occurred at selected sites in each of our institution's geographic regions (6 sites). Regional leadership chooses an α site on the basis of the desired characteristics provided by the Office of Population Health Management. The office provided centralized resources (the α implementation team) to assist and guide the implementation. This process occurred in collaboration with local site leadership and the regionally based β implementation teams. Centralized resources included a physician and a nursing leader, a health systems engineer, senior project managers, a change management expert, a data analyst, and a media-communications specialist. The central team provided an introduction to the new elements of each phase and described the desired outcomes that the α sites should achieve by the end of the initial α implementation period. The central team then provided support while the local teams developed best practices for implementation, and the α team offered feedback and refinement for the individual thematic elements. Examples of refinements made after testing in α implementation are a step-by-step guide for how to colocate a health care team, a conversation guide for daily huddles conducted by that team, changes in the patient rooming process, and conversation tools to facilitate introduction of the new model to patients.

Data were gathered that reflected the effect of the MMoCC (MMoCC2) implementation on the practice. Metrics centered on the Triple Aim and included (1) performance on 2 ambulatory quality measures (mammography completion and diabetes mellitus outcomes), (2) patient satisfaction, and (3) production of relative value units. The latter was selected as a cost and financial metric because at this stage of the project, the main concern expressed by the practice was potential loss of revenue.

To ensure fidelity to the overall model despite diffusion across diverse sites, the central implementation team developed and oversaw a certification process. The certification process consisted of several self-reported measures, as well as objective assessments of quality outcomes and team development through a validated team development questionnaire. ⁵

All α sites passed the certification process. Our assessment showed that implementation of the MMoCC in the α sites had a positive or neutral effect on health care quality and patient experience. The effect on revenue appeared to be neutral to slightly positive. We then sought and received approval from practice oversight committees to proceed with β implementation of MMoCC2 to the other practices.

This later implementation was monitored monthly with use of scorecards and status reports. Scorecards tracked progress at each site on program element implementation. Status reports included high points of progress, barriers, and other useful information and contained a simple visual overall assessment of green (progressing well, on target to meet plan), yellow (some issues need addressing; moderate risk of not meeting plan), or red (barriers exist, causing high risk of not meeting plan), depending on project status.

Quarterly review with regional leadership allowed the identification of problems and barriers that impeded progress. Problems were addressed at the appropriate level of intervention. Issues relating to data or infrastructure support were addressed at the enterprise level; issues of staffing, management, or process were addressed at a regional or site level.

Results and Lessons Learned

A major lesson learned came from our self-imposed aggressive time line. To meet it, we needed to begin implementation at β sites before all learnings from α implementation were analyzed and the model updated to reflect these learnings and best practices. Considerable confusion resulted among some care teams when original and updated models became intermingled. Our implementation time line had to be delayed 4 months to allow sufficient time to repackage the model.

The original reason for grouping elements into phased packages was to ensure that the suggested changes were compatible with one another. As with many implementations that depend on availability of clinical applications, the recommended processes for implementation were dependent on a clinical software application that was not yet ready. This situation caused a trickle-down effect of inability to begin a new process because the resource that would allow access to the software was not available. Some elements of the new model were put on hold until the software (ie, disease and prevention registries) became available.

The differing effect of executive leadership vs frontline leadership was highlighted during the implementation process. Although we had secured the support of central executive leadership, the majority of the day-to-day decisions, as well as resource allocation, is done at regional and site levels. In regions where the local and frontline executives supported the model, our progress was greater than in those regions where support did not exist.

The implementation process also brought to light important cultural barriers that are likely to be universal in health care organizations. The most acute barriers were those that participants reported the most frequently or we believed to be the most likely barriers to implementation. First, and probably most important, is the slow progress of the move away from fee-for-service payment models, at least in our geographic service area. This payment model recognizes the work of physicians as the only work that has reimbursement value. Because rational persons are not likely to do anything that goes against their own financial self-interest, we cannot expect physicians to begin delegating care to other team members if that delegation would lead directly to income reduction. Second, and probably just as important, is the needed change in the perception that population health models simply add more work to already overburdened primary care clinics and services. The goal of population health must be clear—to shift care to more appropriate pathways, which ultimately should move the institution toward the Triple Aim and lower the cost curve. Although the accomplishment of this goal means that primary care physicians will be expected to focus more on patients with complex and multiple chronic conditions, it also means that the routine care currently provided to healthy patients will either move to other care team members or to nonvisit care delivery.

Our institutional experience shows that barriers to implementation of a population health model are substantial and should not be taken lightly. However, our initial data also show that this new model provides the staff with a sense of new meaning and purpose in their delivery of primary care to their patient population.

Conclusions

Consensus is building that implementation of a population health framework is a positive response to the economic and professional challenges facing primary and community care. Although the definition of population health is not exact and an ideal model has yet to be identified, many health care delivery institutions are moving forward with framework implementation in anticipation of pending payment reform. Our findings suggest that the considerable organizational, economic, and cultural barriers to implementation will require concerted use of energy and resources to overcome them. With recent focus on lowering health care costs and improving care quality, our initial experience suggests that devoting these efforts will be worthwhile.