AUGUST 2011

Exploring the Promise of Management Programs to Improve Health

Suzanne Felt-Lisk and Tricia Higgins

The concept of population health management (PHM)—programs targeted to a defined population that use a variety of individual, organizational, and societal interventions to improve health outcomes—is increas- ingly being looked on by employers, health plans, and others as a promising practice for helping to improve health outcomes and “bend” the cost curve. Improving population health by attacking “the upstream causes of so much of our ill health,” such as poor nutrition, physical inactivity, and substance abuse, is also a core goal of the triple aim for improving health and health care in this nation, articulated by Donald Berwick, administrator of the Centers for Medicare & Services. Despite the concept’s growing prominence and interest, there is little understanding of the prevalence of and evidence base behind PHM programs. This issue brief looks at the state of PHM, highlights the evidence related to PHM pro- grams, examines desirable features, and explores the potential future of PHM in the United States.

The Concept of Population group of stakeholders that provides to help, analysis of key demographic Health Management services aimed at improving popula- and health data on the target population tion health. Although not the only such is important. PHM programs are a set of interven- model, it conveys many concepts shared tions designed to maintain and improve broadly by those involved in PHM. people’s health across the full con- Prevalence of Programs The CCA framework embeds two key tinuum of care—from low-risk, healthy The Mathematica team conducted an points: individuals to high-risk individuals with environmental scan of promising PHM one or more chronic conditions. PHM 1. Population health is person-centered; practices, finding that these programs has elements in common with disease organizational interventions are tai- have increased in popularity in the past management, preventive services, and lored to the individual and community few years. We also found that programs , but differs in both resources are targeted to individuals. combine tools and modalities to engage the scope of services and definition Individuals are evaluated to identify and influence individuals to maintain of target populations. PHM programs 1 their place on a continuum of health or improve their health. The most typically are developed to address the risks, from no or low risk to high risk. recent market survey of purchasers by needs of insured population subgroups Specific interventions, such as health CCA (conducted in 2009) found that 68 for which an employer, health plan, or promotion and wellness, risk manage- percent of those surveyed (mostly large other purchaser bears responsibility. ment, care coordination/advocacy, and employers) purchased population health Populations targeted by PHM are often disease/case management, are targeted improvement services and 84 percent delineated by health benefit source to people based on where they fall on expected to purchase more in the future rather than geography. However, some the continuum of risk/care. (Kelly and Neftzger 2010). proponents argue that because improv- ing population health is a national goal, 2. Operational measures and program a target population can also be identified outcomes help improve the interven- State of the Evidence broadly, as in “all citizens of the United tions and refine the “assessment” Despite the popularity of PHM, the field States,” as well as narrowly, as in “all portion of the program that places is still in its infancy and it is not clear people who call Dr. Jones their doctor” individuals on the continuum of risk/ whether PHM programs can deliver (Berwick et al. 2008). care and determines what interven- better health outcomes. Researchers tions they are eligible to receive. Figure 1 shows a conceptual framework have not yet identified a set of effec- for PHM adopted by the Care Contin- To ensure that a PHM program is tai- tive methods for improving the health uum Alliance (CCA), an industry trade lored to the needs of those it is designed of whole populations, short of care- fully designed, community-wide public Figure 1. FIGURE 2– CONCEPTUAL PHM FRAMEWORK Conceptual PHM Framework

Population Monitoring/Identification ▼

Health Assessment ▼

Risk Stratification

Care Continuum No or Low Risk Moderate Risk High Risk Health Management Interventions

Health Promotion, Health Risk Care Coordination/ Disease/Case

Wellness Management Advocacy Management

▼ ▼ Organizational Interventions Tailored Interventions (Culture/Environment)

Person

Community Resources

Operational Measures

Program Outcomes

Psychosocial Behavior Clinical and Productivity, Financial Outcomes Change Health Status Satisfaction, QOL Outcomes

Source: Outcomes Guidelines Report Volume 5. (2010). Washington, DC: Care Continuum Alliance. Reprinted with permission. QOL = quality of life. Conceptual PHM Framework In addition to a revised version of the 2009 released PHM Framework, a conceptual population health management framework health campaigns.is included inThis this suggests report, (Figure that 2, above).literature The intent on worksiteof this model health is to promotion, complement the tiondetailed and PHM disease Program management. Process Within when large organizations, or a nation, defined broadly as using a health risk all these areas it will be important to choose toFramework pursue a byPHM identifying program, the general componentsassessment and (HRA) stakeholders with feedbackof population and health. Likelook the at thedetailed quality PHM of Program the research and strong monitoringProcess Framework, and evaluation the Conceptual will PHMfollow-up Model depicts (which the identification, varied widely). assessment Health and stratificationto better understand of program how participants. this research be criticalThe to core guide of the its modelevolution (central over blue box) promotion—throughincludes the continuum the of worksitecare, as well or as healthapplies management to PHM. interventions. The time to personmaximize is central results. in the model, and is surroundedother venues—is by various aoverlapping key part of sources PHM, of as influence on the management of his or her noted previously. The task force con- Evidence suggests that not all PHM Many ofhealth. the individual This can elementsinclude, but of isPHM not limited cluded to, organizational that these interventions interventions, generally provider interventions programs and are family equal: and a recentcommunity survey programsresources. have been Operational separately measures imple- are representedwere effective under program in influencing outcomes, tobacco as are the core outcomeshows that domains many fromlarge theemployers—the PHM mented and studied, as shown in Table and alcohol use, seatbelt nonuse, dietary main customers of PHM programs—do 1, but moreProgram has to Process be done Framework. to determine Finally, the cycle of program improvement based on process learningsnot think andtheir outcomes vendors isare very effec- prominently depicted by the large curved fatgreen intake, arrows. blood pressure, cholesterol, how these elements work together. The summary health risk estimates, worker tive, whereas a smaller portion do find challenge that PHM brings is to fit these absenteeism, and health care service them effective. For example, 66 percent individual pieces together into a patient- use (Soler et al. 2010). However, most responded that their vendors are not at centered, evidence-based whole that is of the studies were not well structured all effective or only slightly effective in more effective than individual elements and most of the participants are likely influencing members to make healthy have proved to date. to be the “worried well,” or those who lifestyle decisions; 51 percent gave the same negative response regarding 19 There are also broader, related areas in are motivatedCARE CONTINUUM to change their ALLIANCE behavior, reducing confidence in this conclusion. vendors’ effectiveness at encouraging which research is being applied. For members to comply with preventive care example, the Task Force on Community Comprehensive research is also being conducted in the fields of care coordina- guidelines (Kelley and Neftzger 2010). Preventive Services recently reviewed The difficulty of effectively changing 2 Table 1. PHM PROGRAM COMPONENTS Program Component Advantages Disadvantages Means of Targeting Interventions to Population’s Needs Health risk appraisals (HRAs) HRAs have demonstrated health benefits when Percentage of target population that completes combined with programs to address identified HRAs is typically low, leading to underestimation needs (Shekelle et al. 2003; Soler et al. 2010) of problems (National Business Coalition on Health [NBCH] and eValue8 Health Care 2010) Claims/lab/electronic health record These sources provide additional information, There is a time lag for claims data; a variety of (EHR) data used to identify appro- compared with using HRAs alone methods are used, but the best approach is unknown priate level of intervention Data on racial, cultural, language, and Using these data is considered a promising The best approach is not known; this method socioeconomic factors used to indicate practice by NBCH is not widely used at present best ways to reach subpopulations Policy and Program Design Tools Coverage and cost-sharing policy/ Cost-sharing for medications is associated with rates Rewarding the use of high-value services is value-based insurance design of treatment and adherence (Goldman et al. 2007); increasingly popular, but the research base for logically, coverage should support lifestyle changes effectiveness is weak except for medication cost- (e.g., tobacco and alcohol cessation) as such sharing (Choudhry et al. 2010) changes are usually a goal of the overall program Opt-out program design More of the relevant members are included Might not lessen the challenge of engaging in the program members meaningfully Incentives Incentives to individuals for “Incentives for participants” was one of the top- The research base is weak; offering incentives is participation in PHM program ranked changes that program operators believed worth it only if the program is successful for those could improve results, according to a recent sur- who were enticed to participate vey (Disease Management Association of America [DMAA] 2010) Incentives, including cash, gift cards, and mer- chandise, are common Provider incentives Aligning incentives with goals of better health Annual, bonus-type pay-for-performance pro- care is widely accepted as helpful; the most prom- grams have not been shown to be very effective ising types of incentives appear to be (1) use of a (Christianson et al. 2007; Rosenthal et al. 2007) shared savings model, when possible (with larger groups); (2) additional payment for desired care coordination or medical home services; and (3) incentives to support EHR adoption and use (as a foundation for quality improvement and PHM) Provider Support Feedback to providers on gaps in Many primary care providers will follow up on The cost of producing and distributing this infor- the care of specific patients patients identified by health plans as potentially mation could be significant, particularly at the needing services (such as chronically ill patients start of such an effort when data might not be as who need routine monitoring tests) (Felt-Lisk clean as anticipated. The overall effectiveness is et al. 2009) not proven; the cost benefit is not established Consumer Engagement and Interventions Heavy overall program emphasis Large majority of program operators cite con- Costs will increase with heavy emphasis on this on this component sumer engagement and interventions as a critical component; the cost benefit is not established program component (DMAA 2010) Common program components that Inclusion of these program components in The research base is generally weak are included in NCQA’s Wellness NCQA standards reflects a general consensus and Health Promotion accreditation that programs should include them program standards: • Reminders to patients for preven- tive services • Educational resources and self- management tools • Linking enrollees to other resources • Health coaching Social networking A recent study of veterans showed strong effec- A 2004 literature review did not show tiveness of monthly telephone calls between evidence that online peer support is effective matched diabetic patients (Heisler et al. 2010) (Eysenbach et al. 2004) Blue Cross Blue Shield Association has begun a major new program (Blue365 Online Community) and was a 2010 finalist in NBCH’s Health Innova- tion awards

3 patients’ behavior has also been evident components listed in Table 1, which risk, because (1) any success through from the lack of success of two popula- describes what we know and do not yet prevention of chronic disease would be tion-based disease management pro- know about the advantages and disad- a powerful and crucial outcome, and (2) grams within Medicare—the Medicare vantages of each of these elements of such success would be unlikely to be Health Support program, which ran from PHM programs. measured directly and related cost sav- 2005 to 2008, and the LifeMasters Sup- ings might take many years to accrue. ported SelfCare Demonstration (Crom- The Future of PHM well et al. 2008; Esposito et al. 2008). Presently it is unclear whether any sys- Programs tematic infrastructure—in the form of financed, rigorous research to improve If improving population health contin- Current Thinking About our understanding of which models ues to be a national goal, will it occur Desirable Program Features work—is being put into place. through organized PHM programs? If To identify potentially desirable features so, how many PHM programs will there It is easy to think that PHM could of a PHM program, recognizing that a be and who will run them? Many large provide benefits in improving the set of desirable features has not been employers and public and private payers nation’s health and reducing health care proven, we looked at promising prac- responsible for large populations have expenses compared with a health care tices (identified by researchers through already established programs, but there environment without PHM. How- a systematic process), new accreditation is no obvious mechanism for involv- ever, without defined and coordinated standards, and research on typical pro- ing the many individuals employed by research to better understand if and how gram components (Goetzel et al. 2007; smaller companies and organizations. these programs can work, the promise NCQA 2011).2 Desirable features may Because of nascent evidence and the of PHM seems unlikely to be realized. include the following: lack of a proven model, it is likely that the short-term future of PHM will Endnotes 1. Integrating the PHM program into involve multiple models that coexist benefit design and financial incen- and reach a relatively small percentage 1 The environmental scan consisted of con- tives, including compensation of the U.S. population. Good first steps tacting key organizations and individuals practices when the sponsoring orga- should include defining and coordinat- who provided and pointed us to key pub- nization is also an employer ing implementation of these competing lished and gray literature. In particular, we thank the staff of the National Committee 2. Using a combination of initiatives PHM models with strong research to understand their relative effects and for Quality Assurance (NCQA), America’s and tools that address the full con- Health Insurance Plans, and CCA for their whether they can improve health and tinuum of care assistance. We supplemented this scan perhaps even reduce health care costs with targeted online research, focusing on 3. Providing incentives for eligible over time. If PHM models are rigor- literature reviews already completed by individuals to participate based on a ously monitored and evaluated over the other authors. clear understanding of what they find longer term, the most beneficial models 2 meaningful could be spread more comprehensively The criteria for accreditation can be viewed by opening the report card for any and evaluated further. PHM coordina- 4. Tailoring programs to individual accredited organization and focusing on the needs—for example, allowing partici- tion and research could occur within format of the results rather than the results pants to obtain services through mul- the Medicare or Medicaid programs, for themselves (http://reportcard.ncqa.org/ tiple modalities, such as the internet, example, through the Center for Medi- WHP/External/). telephone, and printed materials care & Medicaid Innovation. Further- more, government and/or foundation References 5. Fulfilling NCQA’s Wellness and sponsorship of partnerships between Health Promotion accreditation research organizations and large For the full list of references, go to standards, which represent recent employers could support high-quality www.mathematica-mpr.com/health/ consensus among experts on the evaluations alongside PHM interven- phm_brief_ref_7_11.asp. components of a strong program tions. Careful attention will have to be paid to tracking these programs’ ability For more information, contact Sue Felt-Lisk In keeping with these promising prac- to influence interim markers of health at [email protected]. tices, PHM programs often include the

Princeton, NJ • Ann Arbor, MI • Cambridge, MA • Chicago, IL • Oakland, CA • Washington, DC

Visit our website at www.mathematica-mpr.com Mathematica® is a registered trademark of Mathematica Policy Research, Inc.