Patient Self-Management Support Programs: An Evaluation
In the absence of extensive evidence on the most effective ways to design and evaluate self-management support programs, we turn to existing programs to illustrate a wide variety of program designs and evaluation measures. This situation is challenging for providers who are developing Requests for Proposals (RFPs) for programs, negotiating contracts with vendors, or planning or managing their own programs. They need to make decisions about how to structure and evaluate their programs, which features to include, and whether the programs will be useful for meeting their goals in their settings. Even in the absence of extensive evidence, consideration of the following key points can aid decisionmaking.
Support Programs Aim to Change Patient Behavior
Self-management support programs assume a complex sequence of effects (Figure 1). They expect to change patients' behavior by increasing patients' self-efficacy and knowledge. Improved behavior is expected to lead to better disease control, which in turn, should lead to better patient outcomes, improved utilization, and reduced costs. This sequence of assumptions gives self-management support programs multiple objectives and multiple endpoints for evaluation. The pivotal objective, however, is to change people's behavior.
Begin by Considering the Basic Model
An initial step in choosing or building a self-management support program is to decide where in the health care system the program will be positioned—that is, who will manage and administer the program and where will the care be provided—internal or external to the patient's primary care setting. This last distinction frequently has important ramifications for the degree to which the self-management support is integrated with other aspects of the patient's chronic care and thus who the players are, the quantity and kind of data available to support the coaching intervention, and the degree of administrative focus and support.
Program location may depend in part on where in the health care system the purchaser or builder is located. A plan or employer whose eligible population is thinly spread across numerous independent primary care settings might well consider an external model for practical reasons. A medical group or independent delivery system might have more options for organizing the self-management support program either within or outside of the primary care setting.
Provide Coaching in Addition to Patient Education
Other decisions about the program will pertain to factors such as:
- Content of the support.
- Target population of patients.
- Availability of information systems support.
- Protocols for how program staff are to provide the support.
- Staff training.
- In what manner and how often coaches communicate with patients.
Coaching intervention is one of the most important factors in determining a program's success in changing patient behavior. Patient education is necessary but is not sufficient by itself. Rather than being prescriptive or hierarchical, coaching interventions are patient-centered and tailored to the needs and concerns defined by the patient and his or her situation. Coaches therefore must have timely access to information on patients' behaviors, priorities, skills, and needs.
Coaches should have teaching skills and the psychosocial skills necessary to facilitate a patient's change in behavior. While the information needed for the educational interventions may be disease-specific, the core skill set needed for coaching may be applicable for all diseases. Providers and program developers might consider differentiating the self-management support tasks and looking for people with different skills for different tasks.
Training and protocols for the coaches are important program components, especially since many coaching skills are not often taught in professional schools such as nursing schools.
Choose Measures for Evaluating the Success of the Program
The measures used to assess the success of the program should align with the goals of the program. If the pivotal objective of self-management support is to help patients change their behavior and manage their disease, then evaluation should start there. Whether or not patients use their medications is a better indicator of a program's success than whether a physician prescribes medication, since a change in provider behavior usually is not the primary objective of self-management support. In addition, measuring only patient visits to the hospital or the costs of various aspects of patient care would be overly narrow, especially in the short run.
Likewise, the measurement timeframe needs to match the timeframe in which the self-management support objectives can be attained. Some changes happen sooner, some later. For example, patient self-efficacy and behavior must change before lower costs and fewer hospitalizations are realized. To evaluate the success of a program based on hospitalizations and costs before evaluating and improving patient self-efficacy and behavior might result in premature and unnecessarily negative results. Match measurement to the time course in which it is happening, with longer and realistic time allowances for the long-term outcomes.
The best approach is a portfolio of measures that address the different, sequential objectives of self-management support. This approach should include measures of whether patients get better at managing their disease and allow sufficient time to assess if patients' changed behaviors lead to the hoped-for long-term outcomes.
Finally, it is important to carefully consider the potential for bias when selecting measures of patients' changed behavior. Where possible, rely on information that is directly available from an objective source (e.g., pharmacy records) rather than a patient's own report. When it is necessary to rely on patient self-report, try to avoid collecting this information through the coaches (some patients may exaggerate their good behavior in an effort to please their coaches) or only from those who agreed to participate in the program (to avoid selection bias).
The range of choices currently available for the design of self-management support programs reveals the extensive uncertainty that underlies self-management support programming. The research evidence base for program and evaluation design is very limited, and more and better research across most of these areas is sorely needed. Appendix 2 lists a number of the research needs that were highlighted in the interviews that RAND conducted with individuals knowledgeable about self-management support programs. Further research addressing these needs will provide critical guidance to those who are struggling to design optimal self-management support programs, to change patient behavior, and to improve the health of the chronically ill.