Appendix B: Final Literature Review with Appendices A-D
Design and Evaluation of Three Administration on Aging (AoA) Programs: Chronic Disease Self-Management Program Evaluation Design (continued)
HHSA290200710071T, Task Order No. 6
Task 2.1 Final Literature Review
Revised February 22, 2011
Agency for Healthcare Research and Quality
U.S. Department of Health and Human Services
10420 Little Patuxent Parkway, Suite 300
Columbia, MD 21044
Phone: 443.367.0088 / Fax: 443.367.0477
Abt Associates, Inc.
1. Background and Purpose
2. Literature Review Methodology
3. Evaluation Methods
3.1 Design Types
3.2 Evaluation Methods—Summary of the Findings
4. Outcomes Studied
4.1 Health Behavior
4.2 Health Status
4.3 Health Care Utilization
4.5 Other Outcomes
4.6 Outcomes—Summary of Findings
5. Program Characteristics: Potential Predictors and Confounding Factors
5.1 Population Targeted
5.2 Mode of Intervention
5.3 The Role of Funding Agencies in CDSMPs
5.4 Program Characteristics: Summary of Findings
6. Summary and Implications for CDSMP Evaluation Design
6.1 Design Implications
6.2 Outcomes of Interest
6.3 Program Characteristics to Consider
Appendix A: Articles Included in Review of the Literature Table
Appendix B: Data Extraction Tables
Appendix C: Additional Extraction Table
Appendix D: Stated Program Modifications
Chronic Disease Self-Management Programs (CDSMPs) are designed to empower adults with chronic disorders to better self-manage their conditions and improve their physical and mental health. The best known and most thoroughly studied CDSMP was developed by Stanford University's Division of Family and Community Medicine in the School of Medicine with funding from the Agency for Healthcare Research and Quality (AHRQ) and the State of California Tobacco-Related Diseases office. The lay-led program emphasizes the individual's role in managing illness and building self-efficacy so he or she can be successful in adopting healthy behaviors. The Administration on Aging (AoA) funds the CDSMP and similar programs through grants to State Units on Aging and Public Health Departments who disburse funding to local public health departments in many states.
The Stanford University CDSMP is an evidence-based disease prevention model designed to help people with chronic diseases better self-manage their conditions, improve their health status, and reduce their need for more costly medical care. The program consists of two and a half hour workshops once a week for six weeks and is generally administered in community settings such as churches, libraries, YW/MCAs, senior centers, public housing projects, community health centers, and cooperative extension programs. The program is also available online. Because the program is not condition-specific, people with different chronic health problems attend together. The workshops are facilitated by two leaders, who are trained and certified by Stanford University, one or both of whom are non-health professionals or lay people with chronic diseases themselves. Workshop topics include: 1) appropriate exercise for maintaining and improving strength, flexibility, and endurance; 2) appropriate use of medications; 3) communicating effectively with health professionals; 4) nutrition; and 5) techniques to deal with problems such as frustration, fatigue, pain, and isolation.
Another well-known and researched CDSMP is the United Kingdom's Expert Patients Programme (EPP), which is based on the Stanford CDSMP model. The EPP is a central component of chronic disease management policy in the United Kingdom (Rogers, et al., 2008), and is expected to target over 100,000 people in England and Wales by 2012 (Richardson, et al., 2008). The EPP is also evidence-based and designed to help people with chronic disease self-manage their conditions, improve their health status, and reduce medical costs. Similar to Stanford's CDSMP, the EPP consists of six weekly workshops conducted in community settings, and is also available as an on-line tool. The topics discussed during the workshops are also similar to those presented in the CDSMP workshops.
In addition to the generic (i.e. non-specific disease) Stanford CDSMP and the British EPP, other disease-specific self-management programs, such as those that are home-based (Jerant, Moore-Hill, & Franks, 2009) or clinic-based (Johnson & Raterink, 2009), have been implemented and evaluated.
The purpose of the evaluation design that will be informed by this literature review is to examine the effectiveness of the Stanford University CDSMP as administered by the AoA at improving participant health status, health behaviors, and reducing healthcare utilization and costs. However, since the Stanford CDSMP is only one of a number of self-management programs, it seems prudent to review the literature more broadly by including evaluations conducted of other self-management programs.
To this end, the IMPAQ/Abt Team (a.k.a. design team) utilized reports from agencies and research organizations to guide a thorough review of the published self-management program literature. In particular, articles were collected and reviewed to gain insight into three primary aspects of evaluation:
- Methods to evaluate self-management programs.
- Key outcomes studied.
- Program characteristics, especially populations targeted by/included in evaluations.
The information extracted from the reviewed articles and summarized in this report is organized by key topics as listed above. The topics are presented by section with relevant findings summarized at the end of each section. Implications and preliminary design recommendations are presented in the final section of this report. The results of this literature review will be used in the development of evaluation design options that are feasible and allow an independent assessment of AoA-funded CDSMP outcomes while remaining relevant to results found in other similar studies. The report concludes with a list of findings for AoA to consider when working with the design team to finalize the evaluation design.
To create guidelines for the CDSMP evaluation design, the design team worked collaboratively with AoA staff to define the key issues and research questions surrounding the evaluation of the CDSMP. Our initial efforts were guided by CDSMP evaluation reports completed in the last five years. We also searched evaluation reports of community-based long-term care programs that are similar to the CDSMP so that we could better understand potential methodological approaches, including the appropriateness of randomized control designs to evaluate the CDSMP.
The design team identified and analyzed literature using specific search criteria/parameters. The criteria/parameters for inclusion in the search included chronic disease, Stanford University chronic disease program, wellness intervention programs, falls prevention programs, diabetes self-management, arthritis self-management, CDSMP, and other related concepts. The primary sources of literature were PubMed/MEDLINE and EBSCOHost, but also include the reference list of the Centers for Disease Control (CDC) meta-analysis and other materials provided by AoA. The search strategy used the National Library of Medicine's Medical Subject Headings (MeSH) key word nomenclature developed for MEDLINE, and was adapted for use in the other databases. Text word searching was also used to supplement MeSH searching. All searches were limited to human subjects, publication dates between 2005-2010, and English language publications (i.e., North American, Canadian British, and Australian). For the initial phase of literature searching, we began in EBSCOHost and PubMed to determine peer-reviewed articles employing the aforementioned keywords. Figure 1 depicts the peer-reviewed articles found strictly through an internet-based keyword search.
|Search Terms||Host||# Results*|
|"chronic disease self management" AND behav* AND Stanford||EBSCO||7|
|chronic AND "self management" AND evaluat*||EBSCO||144|
|chronic AND "self management" AND evaluat* AND behav*||EBSCO||114|
|chronic AND "self management" AND evaluat* AND behav* AND intervention||EBSCO||58|
|"chronic disease self management" AND behav* AND Stanford||PubMed/Medline||7|
After aggregating the preliminary literature database, the design team combed through the articles and selected those that dealt with one or all three major areas of interest. The first area focused on the methodology used to conduct program evaluations including the study design (randomized control design; pretest/posttest), type of data analyzed (primary, secondary), target population (age, disability), selection biases, and ethical issues. The second area of our search focused on outcome variables and the tools used to collect and measure them. Examples of outcome variables included functional outcomes, clinical/health outcomes, self-reported outcomes (e.g., quality of life, well-being), social/behavioral outcomes, and health service utilization. The third area of our search focused on program characteristics such as types of beneficiaries studied and program setting. Additional searches were completed to supplement missing concepts as needed. Results were hand-checked for duplicates and for articles outside of the search parameters, producing a total reference list of 44 peer-reviewed articles.
As resources and references were identified, they were stored in a literature database developed by the design team. This organizational data extraction tool incorporated all empirical information gathered through the literature review, including an on-going list of stakeholders and experts in the field who may be appropriate for participation in the project's Technical Expert Panel. Each row in this tabular summary (Appendix A) presents one published article with columns that note the citation, study design, program name, mode of intervention, data collection strategy, and key findings.
In addition to the Key Article Review table seen in Appendix A, we also completed a data extraction and created a 'one-page snap shot' of the major findings from each of the 25 peer-reviewed CDSMP studies with quantitative data (all design types) in Appendix B. These tables allow us to review outcome measures of interest and develop summary statistics from reviewed articles. To extract additional information relevant to the evaluation design, we created a table (Appendix C) that identifies fidelity to Stanford CDSMP, study randomization, site information, screening procedures, and country or region of study.
All abstracts and reviews were also maintained in an Endnote X4 database for easy extraction and manipulation. The design team also obtained and maintained a hard copy library of all items reviewed.
The studies reviewed by the design team most commonly evaluated CDSMPs and other condition-specific self-management programs through two designs: a non-randomized pre-post test design and randomized controlled trial (RCT) design. While both evaluation designs use a similar approach to data collection (e.g., baseline and follow-up), RCT designs add value by randomizing participants into control or intervention groups, allowing causality to be inferred in the analysis of results. Individual respondents were the unit of analysis for a majority of the empirical articles reviewed. Comparisons between CDSMP programs (i.e. Arthritis self management program versus generic CDSMP) were also analyzed.
3.1.1 Non-randomized Pre-Post Test Design
Several of the studies reviewed in this report utilized a non-randomized pre-post intervention design with no control group to assess outcome measures at baseline and again at various intervals after the implementation of the self-management program intervention. This type of longitudinal design compares changes in outcomes over time but precludes the inference of causality since the sample is not randomized and does not contain a control group. Rather, participants in the study comprise the intervention group, all of whom are followed before and after the CDSMP.
By nature of the research design, each study tracked participants and administered follow-up questionnaires after the intervention of the CDSMP. While some studies followed up with participants only once after four to six months post-intervention (Bedell, 2008; Gitlin, et al., 2008; Rose, et al., 2008), other studies utilized multiple data collection intervals to evaluate the self-management program—in these cases, data collection occurred at baseline, four to six months post-intervention, and 12 months post-intervention (Barlow, Wright, Turner, & Bancroft, 2005; K. Lorig, et al., 2008; K. Lorig, Ritter, & Jacquez, 2005; K. Lorig, Ritter, & Plant, 2005; K. R. Lorig, Sobel, Ritter, Laurent, & Hobbs, 2001; Sobel, 2002). Two studies (Barlow, et al., 2005; K. Lorig, et al., 2001; Sobel, 2002) tracked the participants up to two years following the onset of intervention. Findings from these studies, as well as from randomized studies, are summarized in Section 3.2.
One pre-post test study reviewed by the design team analyzed secondary, administrative claims data. Ahmed et al. (2006) compared claims data from enrollees in managed care organizations to calculate beneficiary-level changes in cost of care, and utilization of various health care services. In addition, medical claims and medical charts were reviewed to assess selected HEDIS quality indicators (Ahmed & Villagra, 2006).
3.1.2 Randomized Controlled Trial Design
Sixteen of the studies reviewed for this report utilized RCT designs. In this approach, participants are randomized into treatment or control groups. Of the studies reviewed, approximately one-half focused explicitly on the CDSMP (Barlow, Turner, Edwards, & Gilchrist, 2009; Kendall, et al., 2007; K. Lorig, Hurwicz, Sobel, Hobbs, & Ritter, 2005; K. Lorig, Ritter, & González, 2003; K. Lorig, Ritter, Laurent, & Plant, 2006; K. Lorig, et al., 1999; Smeulders, et al., 2009; Swerissen, et al., 2006). Populations included in these trials ranged from persons with chronic disease in general to those with specific chronic diseases such as diabetes, stroke, heart disease, or inflammatory bowel disease. All programs were implemented in community settings, except one that examined the effect of an internet-based CDSMP (K. Lorig, et al., 2006).
Although one study was randomized at the workshop-level (Goeppinger, Armstrong, Schwartz, Ensley, & Brady, 2007), most of the CDSMP trials included in this review randomized participants into treatment and control groups at the individual-level. Randomization occurred after baseline data collection, often using a blinded randomizer strategy. Five of the CDSMP trials used a straightforward wait-list method for control group participants, such that all control group members had the option to enroll in a CDSMP after the study period (usually six-months). Barlow and colleagues (2009) enhanced their treatment and wait-list control group design by adding a second, non-randomized control group consisting of individuals who explicitly reported disinterest in participating in the CDSMP, for which baseline and follow-up data were collected and analyzed.
To ensure that a sufficient number of participants were selected into CDSMP treatment groups, many studies used specific randomization ratios, such as a 3:2 (K. Lorig, et al., 2003), 2:1 (Barlow, et al., 2009), or 6:4 (K. Lorig, et al., 1999) treatment-control ratio. One study investigating outcomes of CDSMP participants across different ethnic groups used a stratified sampling approach before randomizing participants into treatment and control groups (Swerissen, et al., 2006). In this design, eligible prospective participants were stratified based on spoken language and geographic area, and then randomization occurred at a 2:1 treatment-control ratio.
Several RCTs were conducted on programs similar to or adapted from the CDSMP intervention model (Jerant, et al., 2009; Powers, Olsen, Oddone, & Bosworth, 2009; Richardson, et al., 2008). Adaptations were made to accommodate the method of identifying participant eligibility (e.g., self-report rather than physician diagnosis; Richardson et al., 2008) or for use as a disease-specific program, such as hypertension (Powers, et al., 2009).
3.1.3 Data Collection Methods
The most common type of data collection across the non-randomized and RCT studies included in this review were self-administered, self-reported questionnaires that were mailed to participants at baseline and pre-determined follow-up intervals (including one, four, six, and twelve months). Some studies, however, used additional data collection methods, such as telephone interviews (Barlow, et al., 2005; Beckmann, Strassnick, Abel, Hermann, & Oakley, 2007; K. Lorig, Hurwicz, et al., 2005; K. Lorig, Ritter, & Jacquez, 2005) and one-on-one in-person interviews (Jerant, et al., 2009). These interviews gathered more qualitative, in-depth data than the mail surveys, although in a study by Lorig et al. (K. Lorig, Ritter, & Jacquez, 2005), telephone questionnaires were administered to increase response rates at follow-up and addressed the same items as the mailed questionnaires.
One study reviewed by the design team used a qualitative approach to explore the experiences of participants in cancer-specific and generic CDSMP groups (Beckmann, et al., 2007). Telephone-administered, semi-structured interviews were conducted with 25 participants of a cancer-specific CDSMP, and with 10 participants in a general CDSMP.
The estimated attrition rate is an important consideration when designing any evaluation study. For the CDSMP evaluation it is important to review the literature on attrition in similar studies for two primary reasons. First, because the CDSMP is comprised of vulnerable populations, it would not be surprising if study participants have a higher than average dropout rate due to illness and death. High dropout rates could impact longitudinal outcomes due to reduced sample sizes. Analyses conducted with insufficient statistical power may fail to detect pre- and post-program differences. Second, differential attrition rates between control and treatment groups should also be examined in the literature. If studies show that attrition is random, then it is not a problem because respondents would be as likely to drop from the treatment group as they are to drop from the control group. However, if the articles reviewed indicate differential rates of attrition (i.e., consistently greater for one group over the other), then consideration should be given when developing the sampling plan and randomization to account for these differences.
A number of CDSMP studies included in this review measured outcomes at baseline, four or six months post-intervention, and 12-months post-intervention (Barlow, et al., 2009; K. Lorig, et al., 2003; K. Lorig, Ritter, & Jacquez, 2005; K. Lorig, et al., 2006). For non-randomized pre- post-test design studies, attrition ranged from 13 percent for baseline to first follow-up, to 18 percent for baseline to 12 month follow-up (Lorig et al., 2005). For RCTs, attrition ranged from 21 percent for baseline to first follow-up, to 51 percent for baseline to 12-month follow-up (K. Lorig, et al., 2003).
Other RCTs reported data collection at baseline and 6-month post-intervention. Lorig et al. (1999) enrolled 1,140 participants into their study at baseline, and were able to collect follow-up data on 84 percent of treatment and 82 percent of control group participants. Reasons for attrition of those for whom follow-up data was not collected included death (1.2 percent of treatment, .81 percent of control) and illness (3.4 percent of treatment, 7.8 percent of control). Reasons for dropout were reported as unknown for 11.4 percent of treatment and 9.4 percent of control group participants (Lorig et al., 1999). Of the 728 participants enrolled an Australian RCT (Swerissen, et al., 2006), 254 (34.9 percent) withdrew for reasons including changing their minds, family/personal issues, traveling, difficulty completing forms, feeling unwell, feeling as if too much personal information was being requested, busy schedules, death, or due to inaccurate or out-of-date contact information.
The screening process for selecting participants into the study was not always explained. Screening sometimes involved selection of participants with particular disease or condition while others were based on ethnic background and income. The studies describing screening criteria mentioned exclusion by age, language ability (i.e. reading, writing English), eyesight, mental health, or previous participation in a CDSMP.
CDSMP studies were randomized by individual or group. Nine of the RCT studies were randomized individually by participant. The randomization strategy for the remaining studies were randomization by CDSMP group (receiving generic versus disease-specific CDSMP: Goeppinger, Armstrong et al, 2007); by facility and individual (Ersek, Turner, Cain, & Kemp 2008); by provider (PCP) and individual (Powers, Olsen, Oddone, & Bosworth, 2009); and serially (Lorig, Sobel et al, 1999).
3.1.7 Fidelity to Stanford CDSMP
In the data extraction effort (Appendix C), we considered each program's fidelity to the Stanford CDSMP model. In seven of the twenty-five peer reviewed studies with quantitative data, there was no stated modification. Of those, four were pre-post and three were RCTs. Four studies, one of which was online, adhered to the NHS-run Experts Patient Programme version of the CDSMP. While this nationally implemented program follows the CDSMP model, it is slightly adapted to the British public and is always led by two lay leaders. Three of the studies reviewed were modified to older urban African American populations (Rose et al, 2008; Haas, Groupp, et al, 2005; Gitlin, Chernett et al, 2008). Modifications such as language and focus on a reduced-sugar diet were included.
Two studies used the Spanish-language version of the CDSMP, 'Tomando Control de Su Salud,' which is not a direct translation of the CDSMP and has cultural modifications (Lorig, Ritter, Gonzalez et al, 2003; Lorig, Ritter, & Jaquez, 2005). Six studies were modified for a particular disease or condition (i.e. arthritis, stroke, pain, multiple sclerosis, HIV/AIDS) and one study used a CDSMP variant 'Homing in on Health' (Jerant, Moore-Hill, & Franks, 2009). The mode of intervention also varied, with one study utilizing a telephone only intervention while another was online only. One Australian study (Swerissen et al, 2006) was modified by participant's first language (Vietnamese, Chinese, Italian, Greek). Of the twenty five studies reviewed, four took place in England, three took place in Australia, one in the Netherlands, and part of one study took place in Chihuahua, Mexico (K. Lorig, Ritter, & Jacquez, 2005).
A significant number of CDSMP studies (16 of 44 reviewed) have utilized RCT-type designs, randomizing individual CDSMP participants into treatment and control participants post-baseline data collection.
Modest improvements were reported in two of the studies (Powers, et al., 2009; Richardson, et al., 2008), while a study investigating a chronic disease self-management program implemented in the home reported little difference in outcomes between the control and intervention group participants (Jerant, et al., 2009). Further, in a trial investigating the effect of a chronic pain self-management program on a sample of older adults living in retirement communities, Ersek and colleagues (2008) failed to reveal significant intervention effects.
Despite a quarter of the RCT studies found no difference in most health outcomes for treatment and control groups (Ersek, Turner, Cain, & Kemp, 2008; Haas, et al., 2005), in general, the CDSMP evaluation studies that were reviewed revealed improved heath and psychological outcomes for participants, in varying degrees. Some of the variation in evaluation findings might be attributed to factors such as the amount of time between the onset of the program and the follow-up period, or the demographic characteristics of program participants. For example, (K. Lorig, Ritter, & Jacquez, 2005) found significant improvements in health behaviors and a reduction in health care utilization at four months post-intervention; however, improvements in self-efficacy did not reach significance until 12-month follow-up data collection.
Most studies utilized self-reported outcomes as the data of choice for evaluating outcomes; few discussed combining self-reported data with administrative or other data sources (e.g., claims, health assessments administered by clinicians, medical records).
In all reviewed RCTs, participants were asked to self-report changes in a number of different areas surrounding their mental and physical health, as well as changes in how they utilized health care. These measured outcomes varied among studies, but four of the original Stanford CDSMP measures were the most common: health behavior, health status, health care utilization, and self-efficacy. Data were collected at baseline and at pre-determined follow-up intervals. For some indicators, clinical records were used as confirmation of self-reported results. Though psychological well-being indicators such as depression appeared in several studies as outcome measures, no psychometric properties were reported as a method of assuring validity and reliability of the responses.
A set of scales were developed and validated by the Stanford Patient Education Research Center for use in CDSMP studies, including a scale for four health-related behaviors as follows: stretching and strengthening; aerobic exercise; use of cognitive symptom management techniques; and use of techniques to improve communication with health providers (K. Lorig et al., 1996, Rose et al., 2008). Other studies include additional measures of health behavior depending on the condition studied, such as smoking, drinking, and body mass index (as a proxy for diet) for individuals suffering from heart disease, as these behaviors are known to exacerbate the disease (Smeulders, et al., 2009).
Health status in US studies was based on the self-rated health scale used in the National Health Interview Survey, and on a modified version of the Health Assessment Questionnaire (HAQ) physical disabilities scale (K. Lorig, et al., 1996). Indicators measuring health status are all self-reported and included items such as self-rated health, disability, social/role activities/limitations, pain, illness intrusiveness, energy/fatigue, health distress, and shortness of breath (Du & Yuan, 2010; Foster, 2009; Rose, et al., 2008; Sobel, 2002).
Health care utilization indicators used in the reviewed studies generally assessed three types of utilization, and are measured as such during statistical analysis: 1) visits to physicians, including visits to the emergency room that do not result in hospital admission; 2) visits to the hospital during intervention period; and 3) number of nights spent in the hospital. Other studies included more detailed reporting for inpatient and outpatient procedures, readmission, and procedure rates. Program success in health care utilization was assessed by the decrease or reduction in utilization. No specific instruments to measure health care utilization were presented in these studies; rather, participants self-report on questionnaires and the responses are checked against the participant's clinical records. Some studies show that reductions in health care utilization can also reduce health care expenditures (K. Lorig, et al., 1999; Swerissen, et al., 2006); these are further discussed below.
Self-efficacy is the theoretical foundation of the CDSMP. Du and Yuan (2010) regard self-efficacy as the "most important evaluation indicator of self-management outcomes, composed of efficacy expectations and outcome expectations" (Du & Yuan, 2010). Efficacy is measured as the perception of an individual as to whether s/he can perform a certain behavior (efficacy expectations) and whether a specific behavior will cause a certain outcome (outcome expectation). While self-efficacy is used as an outcome measure for most studies reviewed by the design team, it is also often used as a predictor of future health outcomes. The self-efficacy measures developed by Lorig et al. at Stanford University were used most often in the studies reviewed for this report (Barlow, et al., 2009; Du & Yuan, 2010; K. Lorig, et al., 2001).
Self-efficacy measures are based on two scales that were developed and designed specifically for the CDSMP. The scales measure a participant's perceived adaptability to manage the disease and manage the symptoms, and are rated by respondents on a scale that ranges from 1-7 with a one indicating low self-efficacy and a seven indicating high self-efficacy. The sum of the two scales provides the overall self-efficacy score (10 to 70 points). Barlow and colleagues (Barlow, et al., 2009) reported results on general self-management self-efficacy and multiple sclerosis self-efficacy using the Liverpool Self-Efficacy Scale. Results indicated an improvement in both types of self-efficacy at four months, but not at the 12-month follow-up.
Some studies that measure self-efficacy report the correlations between self-efficacy and outcome measures. Lorig and colleagues (2001) reported that higher levels of self-efficacy are associated with lower outpatient service utilization. In support of these findings, a more recent study showed that lower reported levels of self-efficacy were associated with higher levels of depression (Barlow, et al., 2009). However, caution should be taken when interpreting the results of correlation analyses; since data on both measures were collected at the same time, causation cannot be determined. Based on these findings, the evaluation contractor should consider including self-efficacy as an outcome measure, controlling for levels of depression.
Self-efficacy theory, developed by Albert Bandura, states that high-level self-efficacy is a prerequisite to realizing self-management goals, as well as critical in determining whether a person will maintain or improve their health status (Bandura, 1989 as cited in Du & Yuan, 2010). For this reason, a participant's belief in his or her ability to manage the condition can also act as a predictor of health outcomes. For example, Lorig and colleagues (K. Lorig, et al., 2006) found that change in self-efficacy at six-months was associated with health status at one year.
4.5.1 Quality of Life
Quality of life (QOL) was measured in studies using common indicators and accepted scales such as the Flanagan Quality of Life Scale (Bedell, 2008), the pain and fatigue scale adapted by Lorig et al. (Goeppinger, et al., 2007), and the Quality Adjusted Life Years (QALY) profile (Kennedy, et al., 2007; Richardson, et al., 2008). Though many indicators for QOL overlap with those for health status, they provide a multidimensional view of self-management effectiveness. Du and Yuan stated in their self-management outcomes evaluation study that QOL and health status reflect effectiveness and effects of self-management programs because "QOL refers to subjective feels towards life, while health status emphasizes an objective condition of living status" (2010).
4.5.2 Health Outcomes in CDSMP versus Condition-Specific Programs
In 2005, Chodosh, Morton et al. used empirical data from a systematic review of 53 studies (780 were screened) to quantitatively evaluate RCTs of chronic disease self-management programs for older adults. The authors attempted to answer two research questions:
- Do self-management programs result in improved outcomes for specific chronic diseases of high prevalence among older adults (namely diabetes, hypertension, and osteoarthritis)?
- For effective interventions, are there specific components – i.e., characteristics such as tailoring programs to specific health conditions, group setting, feedback, psychological emphasis, and medical care—that are most responsible for these observed effects?
While analyses showed the use of feedback in diabetes programs, group settings in hypertension programs, and tailoring in osteoarthritis programs to be effective, these results did not retain statistical significance when examined across conditions (Chodosh, et al., 2005). Instead, the study suggests that the attributes of a condition-specific program are responsible for these positive results. Consequently, this study suggests that condition-specific tailoring will allow for improved health outcomes by focusing on the interventions most effective for each condition rather than more generally.
One study reviewed by the design team used a qualitative approach to explore the experiences of participants in cancer-specific and generic CDSMP groups (Beckmann et al., 2007). Qualitative analyses of responses revealed themes related to decreased sense of isolation, increased motivation to improve one's health and wellbeing, and improved sense of control and achievement. Findings from this study suggest that a disease-specific model may be more beneficial as cancer participants found the information to be more relevant to their condition, and experienced more bonding with other participants, compared to the generic CDSMP.
4.5.3 Cost Effectiveness
An important outcome measure relates to the cost effectiveness of self-management programs. As defined in the reviewed articles, cost effectiveness is demonstrated when there is a reduction in overall cost of health care utilization and no deterioration of health outcomes for a CDSMP participant. Though reduction in cost for some studies were generated from the savings of reduced ER visits, hospital visits, and overnight stays (K. Lorig et al., 2001; K. Lorig et al., 1999), two studies used Quality Adjusted Life Years as an indicator of health gains (Kennedy et al., 2007; Richardson et al., 2008). Data on resource use were combined with unit cost data to provide estimates of overall costs per patient, including the estimated cost of the self-management program per patient.
Kennedy, Reeves, et al. (2007) studied the cost effectiveness of the generic Expert Patients Programme (EPP) in the United Kingdom by estimating the cost of the program per participant, comparing the reduction of heath care utilization among intervention and control groups. The authors found that, though not statistically significant, the reduction in resource use would off-set the cost of participants for the program. Additionally, the study utilized the net benefit approach, which includes the "societal perspective (including the costs to patients) with effects assessed in terms of health gains, measured in terms of QALYs".
Though some studies found that tailoring self-management programs to the specific disease can improve outcomes more than a generic CDSMP, the generic CDSMP may be more cost effective. As stated in Goeppinger, Armstrong, et al. (2007), "population density, community resources, participant levels of comorbidity, and professional and participant preferences" influence the preference for a generic versus condition-specific self management program.
The literature reviewed by the design team demonstrates that overwhelmingly investigators agree that self-efficacy is a critical concept to measure when evaluating a CDSMP program; however, how it is used analytically is inconsistent across studies. Scales developed by Lorig and colleagues (1999) have been widely used to measure this outcome.
Measures of health status and health behaviors are also common, with studies looking specifically at self-rated health, degree of pain and discomfort, role limitations, time spent engaging in exercise and other indicators through the use of validated scales. One study reviewed included the validation of one particular measurement developed specifically for self-management program outcome evaluation—the Health Education Impact Questionnaire (HEIQ), which was shown to have strong psychometric properties including validity (Osborne, Elsworth, & Whitfield, 2007).
Health care utilization has been examined as an outcome, primarily through the use of self-reported visits to physicians and emergency rooms, and number and duration of hospital stays. One study relied on analysis of claims data from participants enrolled in a managed care organization (Ahmed & Villagra, 2006). In addition, cost-effectiveness of self-management programs has been assessed to demonstrate potential reductions in health care utilization (Kennedy et al., 2007; Richardson et al., 2008).
While some studies found that tailoring self-management programs to the specific disease can improve outcomes more than a generic CDSMP, the cost effectiveness of the generic CDSMP may prove to be most beneficial to both patients and Federal programs funding CDSMPs. The Goeppinger, Armstrong, et al. (2007) study investigated the cost effectiveness of an arthritis-specific program versus that of a generic CDSMP and found that CDSMP is in fact more cost effective when used for populations with arthritis and multiple comorbid conditions (vs. the arthritis-specific program).
Throughout our review and extraction of information from the literature, we attended to program characteristics that have the potential to influence participant outcomes, such as race or ethnic differences, the location at which the intervention occurred, or the mode of program transmission (e.g., community group meetings, on-line). This information will be used to determine potential implementation-related variables to include in the AoA CDSMP evaluation design, so that the study will have the ability to detect any differences in outcomes associated with these factors.
The vast majority of the articles synthesized for this review reported on studies that evaluated the Stanford University CDSMP (Gitlin, et al., 2008; K. Lorig, Hurwicz, et al., 2005; K. Lorig, Ritter, & Plant, 2005) or the British version of the CDSMP, the EPP (Kennedy, et al., 2007; K. Lorig, et al., 2008). Although the Stanford CDSMP and its off-shoot, the EPP, were designed to be inclusive of a variety of chronic diseases, recent studies have examined the effectiveness of these and other similar programs on specific disorders (Goeppinger, et al., 2007; Haas, et al., 2005; K. Lorig, Ritter, & Plant, 2005). In addition, studies on the efficacy of the CDSMP for Spanish-speaking recipients have been reported (K. Lorig, et al., 2003; K. Lorig, Ritter, & Jacquez, 2005). In the following sections we present an overview of demographic variables that were reported in the reviewed literature that may be important to consider in the CDSMP evaluation design.
5.1.1 Chronic Conditions
The Arthritis Self-Management Program (ASMP) was the first CDSMP implemented and was designed to assist a disease-specific population cope with one chronic condition. The Stanford University CDSMP, modeled after the ASMP, expanded eligibility to include participants with varied chronic conditions. These more generic CDSMPs have been found in multiple studies to have significant positive results on self-efficacy, health status, health behaviors, and healthcare utilization (Beckmann, et al., 2007; K. Lorig, et al., 2001; K. Lorig, et al., 2008; K. Lorig, et al., 2003; K. Lorig, Ritter, & Jacquez, 2005; K. Lorig, Ritter, & Plant, 2005; K. Lorig, et al., 1999; Steward et al., 1999).
Although there are many generic CDSMPs today, including the United Kingdom EPP, there are also numerous self-management programs that target one specific disability or disease (Armstrong & Powell, 2008; Battersby, et al., 2009; Beckmann, et al., 2007; Bedell, 2008; Cummings & Turner, 2009; Ersek, et al., 2008; Goeppinger, et al., 2007; Haas, et al., 2005; Kendall, et al., 2007; Rogers, Kennedy, Nelson, & Robinson, 2005; Smeulders, et al., 2009) or two (Chodosh, et al., 2005; Powers, et al., 2009).
Considerable research has been conducted to compare the efficacy of condition-specific programs to generic CDSMPs. As cited earlier, Goeppinger and colleagues (Goeppinger, et al., 2007) compared the cost effectiveness of a generic CDSMP with an Arthritis Self-Help Course, and found positive benefits for arthritis patients from both programs. The results of this study suggest that generic CDSMPs may be a more cost-effective approach to self-management. However, Lorig, Ritter, and Plant (K. Lorig, Ritter, & Plant, 2005) found improvement in more outcomes measured among arthritis patients who participated the Arthritis Self-Management Program than in those who participated in the generic CDSMP, particularly at four-months. In contrast to the Goeppinger study, these findings indicate that disease-specific programs should be considered first as long as there are sufficient resources and interested participants.
5.1.2 Age groups
CDSMPs were originally designed to empower seniors to better self-manage their chronic conditions and improve their physical and mental health. The mean age of a majority of the studies reviewed was 60 years and older (go to summary table in Appendix B). However, many of these programs include adults under the age of sixty. For example, Goeppinger et al. (2007) included adults 18 and over in a comparative study of a small arthritis education program against a traditional CDSMP. Gordon's review of the CDSMP literature (Gordon, 2008) shows that many U.S. CDSMP programs are inclusive of middle-aged adults (i.e., 40 years old and older). Although many of the CDSMP reviews that included adults under the age of sixty were conducted in the United States (Goeppinger, et al., 2007; K. Lorig, et al., 2001; K. Lorig, et al., 2003; K. Lorig, et al., 2006; K. Lorig, et al., 1999), several were conducted in the United Kingdom and Austria where they typically include adults 18 years and older (Barlow, et al., 2005; Battersby, et al., 2009; Kennedy, et al., 2007).
Age is reported in most studies to describe the sample; however, only one report (Nolte, Elsworth, Sinclair, & Osborne, 2007) analyzed the effect of age differences on outcome measures. This review of Australian CDSMPs revealed that younger participants, particularly younger women, reported benefits of the self-management program on most outcomes measured while few older adults showed improvement.
5.1.3 Race and Ethnicity
Relative to the number of studies that have been undertaken to evaluate the CDSMP and other similar self-management programs, very few have explored the benefits of these programs among ethnic groups. In fact, overwhelmingly, the samples included in the studies reviewed for this report were predominately white and female. However, a study among Hispanic-Americans in Northern California showed continued significant improvement in all health behaviors and health status outcomes at 12-month follow-up as compared to baseline. In addition, there was significant improvement in emergency room visits that persisted through the 12-month follow-up, although no significant differences were found in the number of physician visits, or hospital days (K. Lorig, et al., 2003). Similarly, in a study that examined the Spanish-language version of the CDSMP along the Texas, New Mexico, and Mexico border (K. Lorig, Ritter, & Jacquez, 2005), participants showed improvements in health behaviors, health status, and self-efficacy at both four- and 12-month follow-up interviews.
Another study evaluated a CDSMP program that used an adapted version of the CDSMP training manual, developed with rural, African American older adults (Rose, et al., 2008). The study found improvement in health behaviors and health status among the sample of rural older African Americans; however, several outcomes measured did not persist to the six-month follow-up interview.
5.1.4 Beneficiary Types
The majority of studies reviewed for this report recruited samples through public advertisement and word of mouth, and did not document beneficiary type. Only one article reviewed identified a Medicaid population (Rosenman, et al., 2006), and none reported specifically on Medicare recipients. However, a few of the investigations selected samples from specific beneficiary populations such as managed care organizations (MCO) and provider networks (Ahmed & Villagra, 2006; Jerant, et al., 2009; K. R. Lorig, et al., 2001; Sobel, 2002). Ahmed and Villagra (Ahmed & Villagra, 2006) evaluated the impact of a comprehensive Diabetes Disease Management Program (DDMP) on cost and quality across ten US urban markets, and concluded that the DDMP improves quality of care and reduces overall medical costs in MCOs. In addition, patients from a Veteran Affairs Medical Clinic were recruited for participation in a hypertension self-management program (Damush, et al., 2010; Powers, et al., 2009) to test program spillover effects for patients with diabetes and high cholesterol. Although the results were mixed, demonstrating a significant positive effect for self-managing diabetes but not cholesterol, the authors conclude that disease-specific self-management programs may have spillover effects on patients' comorbid conditions.
5.1.5 Dual Eligibles & Medicaid funding in the CDSMP context
Vulnerable populations, such as those receiving Medicaid services, are underserved by evidence-based medicine programs (Counsell, 2010). This section explores dual eligibles and the importance of Medicaid funding in the CDSMP context through a review of the grey literature. While there is evidence of potential cost-savings and benefit to this group, it may not be possible in the short term to include dual eligibles in the evaluation design. Only one of the articles from the literature review provides information about the Medicaid population (Rosenman, et al, 2006). However, an online search of CDSMP programs and mention of the Medicaid population, found that states such as New York are beginning while others (Maryland) intend to include the population in later years of the program (NACDD, 2010). With their state data indicating that 5% of Medicaid chronic care population accounts for 50% of the Medicaid health care expenses (Goehring, 2010), Washington State now offers reimbursement for diabetes SMP and aims to provide CDSMP reimbursement for Medicaid.
Based on high health care expenditures for Medicaid participants who also receive Medicare (dual eligibles), it is likely that inclusion of this population would result in cost savings to Medicare. Dual eligibles include 9 million low-income elderly and disabled Medicare beneficiaries who qualify for coverage based on their low income. Medicaid covers important services and co-pays that Medicare limits or does not cover, such as long-term care. Dual eligibles account for 18% of Medicaid enrollees but 46% of Medicaid spending. The management of chronic conditions in this group is likely to result in substantial savings. The dual eligible population is also growing. Medicaid coverage rates for the community among the over 65 population increased from 7.6 percent in 1987 to 14.1 percent in 1996.
While it is unclear what proportion of CDSMPs are Medicaid beneficiaries, there may be more data on this population in the near future. ARRA funded CDSMPS require the State Medicaid Agency to be involved in the development and implementation of the program (AoA, 2010). The new CDSMP programs are required to give special attention to serving low-income, minority and limited English speaking older adults, including Medicaid eligible individuals. In Rhode Island, Medicaid brought CDSMP to the State in 2006 in conjunction with the Department of Health and the Department of Elderly Affairs (Arora et al, 2008). Existing disease-management programs similar to CDSMP may develop specific programs for dual eligibles. In Florida, dual eligibles can participate in a LifeMasters program, or a Medicare Disease Management Demonstration, providing services to certain chronically ill beneficiaries (Edlin, 2006). LifeMasters tests whether disease management in the traditional fee-for-service (FFS) program leads to improved outcomes and lower total costs to the Medicare program, and has demonstrated substantial savings for the Medicaid population (Business Wire, 2003).
According to a recent CDC brief, a few states are moving toward Medicaid reimbursement for CDSMP (Gordon & Galloway, 2010; www.healthyagingprograms.com). While this has been occurring on a relatively small scale to date, the brief reports one state has Medicaid clinics specializing in asthma and diabetes and patients receive referrals to CDSMP programs. Another strategy has been to train Medicaid managers to run CDSMP programs within their clinics. A Partners in Care Foundation conference in 2010 argued from a Social Enterprise Reimbursement Model that once Medicaid accepts CDSMPs as a reimbursable benefit, they can cover the benefit under the Medicaid Waiver program. The state of Washington amended their Aged/Disabled Waiver to include provision of CDSMPs and California is pursuing a similar strategy. There are a number of insurers reimbursing CDSMPs, including Kaiser Permanente, who serves as the U.S.'s largest non-profit health plan; the NHS providing the world's largest publicly funded CDSMP program, and the LA Care Health Plan, who offer the nation's largest public health plan. Through Oregon's "Living Well" program, CDSMP coaches refer Medicaid Fee-for-Service clients to programs in their area.
An ARHQ User Guide on implementing Medicaid disease management programs (Arora et al, 2008) sheds light on how inclusion of the Medicaid population into CDSMP programs may mean adapting the program to meet their needs. Communication and social supports are important issues for individuals in the Medicaid program. Public distrust of Medicaid and other public programs creates barriers to contacting members. During a health intervention in Indiana, it was discovered that members were not opening mailings from the program because the envelopes had the same logo as the Medicaid program. After Indiana changed the envelopes to have a program specific logo, members began to open them. In addition to differences in recruiting Medicaid members, states must anticipate that members may have low literacy levels. It is vital that program materials target an appropriate reading level and be made available in prominent languages. Informal focus groups with select Medicaid members may facilitate communication with members and illuminate "perceived barriers" to communication.
Despites benefits to the Medicaid population, it may not be possible include them in the evaluation design. However, this issue will be explored further during upcoming site visits to CDSMPs. Obtaining Medicaid data and linking it to Medicaid claims data is both difficult and sometimes extremely complicated to compare across states, and this creates a barrier to inclusion of the Medicaid population. Depending on the research question and variables of interest, it may also be unreliable source of data. Because Medicaid is funded through federal and state funds, the benefits vary greatly and there are disparities in access to health care as well. This issue will be explored further in the evaluation design report.
The traditional CDSMP and EPP consist of two and a half hour workshops once a week for six weeks and are generally administered in community setting such as churches, libraries, YW/MCAs, senior centers, public housing projects, community health centers, and cooperative extension programs. Only one study reviewed reported on the distribution of the location of the workshops, but did not report on the effectiveness of holding the workshops in one location over another (Haas, et al., 2005).
Several self-management programs are designed to accommodate homebound adults or adults who for other reasons prefer not to attend group sessions in the community. On-line versions of the CDSMP (K. Lorig, et al., 2008; K. Lorig, et al., 2006) and EPP (K. Lorig, et al., 2008) are available, and have demonstrated positive benefits such as decrease in symptoms and health care utilization, and improvement in health behaviors, self-efficacy, and satisfaction in the health care system. The on-line version of the CDSMP may be a viable option for persons with chronic disability who prefer not to attend group sessions and who are computer savvy.
A study of an in-home self-management program failed to demonstrate lasting positive results (Jerant, et al., 2009). The authors concluded that, despite leading to improvements in self-efficacy compared to those in CDSMP studies, the in-home program had limited sustained effects and probably was not a cost-effectiveness option from the health system perspective.
The purpose of this literature review is to inform an evaluation design for AoA funded CDSMPS. However, the studies identified in the literature focused almost exclusively on non-AoA funded CDSMPs which may differ from AoA funded CDSMPs because of differences in program setting. In order to bridge that gap, we conducted a review of the grey literature to understand CDSMPs in AoA-funded settings and, in particular, to determine whether and how AoA-funded CDSMP programs are different from other CDSMP programs. In a structural sense, CDSMPs in AoA-funded settings benefit from the National Aging Network, which includes 56 State Units on Aging, 629 Area Agencies on Aging, 244 tribal organizations, some 20,000 local community service organizations, 500,000 volunteers, and a wide variety of national organizations (Tilly, 2010). Therefore, it is reasonable to define AoA's role as providing a tremendous amount of resources to CDSMPs and linking CDSMPs to a larger service delivery system. The following section will examine 1) how the AoA describes their role; 2) non-AoA funded CDSMPs; and 3) other setting factors and reporting structures, in order to inform the CDSMP evaluation design.
5.3.1 CDSMPs in AoA-funded Settings
AoA began funding evidence-based programs in 2003, working with a variety of federal agencies, the Aging Network and other partners. Since 2006, this grant funding has resulted in the delivery of evidence-based programs to 25.000 seniors in over 1,200 community‐based sites across 27 states (AoA Fact Sheet, http://www.healthyagingprograms.org). AoA administers the grant program, and has a contract with the National Council on Aging (NCOA) to provide Technical Assistance to CDSMP grantees. Technical Assistance includes web‐based training, on‐site visits, target teleconferences, peer‐to‐peer mentoring, strategies and models for developing statewide CDSMP distribution systems, and strategies to sustain programs beyond the grant cycle. AoA plans to incorporate successful CDSMPs into its Aging Network's array of services and to provide ongoing support of these programs through technical assistance. In addition, many states are using OAA Title III-D funds to support CDSMPs (CITATION is personal contact via email with AoA, 01/04/11).
5.3.2 CDSMPs in non-AoA funded Settings
Another way to understand AoA's role in CDSMP programs is to compare to non-AoA funded programs. However, few of the articles reviewed for the literature review report the funding sources of the program, with the exception of international programs and those run by Kaiser Permanente. Still, it is possible to understand AoA's role by comparing AoA-funded programs to programs run in the United Kingdom (UK) by the National Health Service (NHS). An evaluation by Rogers et al (2006) considered the program within the NHS setting. The study discusses how CDSMP programs were difficult to adapt to the way NHS normally provides services, with a patient addressing a single condition with a health professional. The NHS-run CDSMP programs have conformity in who runs the program, being run through Primary Care Trusts (PCTs) by two individuals with personal experience of a chronic condition. However, AoA-funded CDSMPs must demonstrate fidelity to the Stanford model, which requires only one lay leader and another leader of varying background. In practice, some AoA-funded CDSMPs have lay leaders and health professionals or other volunteers. Because both the delivery and content of the NHS-funded CDSMP was prescribed with limited flexibility, it created tension and difficulty in meeting the needs of local communities (Rogers et al, 2006). Given the flexibility and diversity in AoA-funded CDSMPs, this may be less of an issue.
Similar to NHS-funding CDSMPs, Australian courses use a highly structured course protocol, but instead of lay leaders, use trained health professionals (Kendall et al, 2007). On a national level, the Australian Commonwealth Department of Health and Ageing implemented the Sharing Health Care Initiative, which involved several large demonstration projects across a variety of settings (Commonwealth Government Department, 2005). As self-management courses are now being applied in a variety of settings and implemented at the government policy level, there is an urgent need to understand and document the impact of self-management courses across settings. Numerous controlled trials have been conducted across disease groups and have been summarized in meta-analytic and narrative reviews. These studies, however, suggest that self-management courses might not be suitable for all types of chronic conditions and population subgroups.
5.3.3 Other Setting Factors & Reporting Structures
There is current research under way to explore how setting factors influence CDSMPs (K. Lorig, personal communication, Sept 7, 2010). A number of setting factors may influence AoA's role in particular and CDSMPs in general. Two factors that have a larger influence are the funding sources and the organization that administers the program. A third factor would be data or information collected by the CDSMP and what is required by funding source(s). Funding mechanisms for a CDSMP program are likely to influence AoA's role, with sometimes four or five organizations funding a CDSMP program, from both public and private entities, as well as interested community organizations. When sources of funding change, it is likely that AoA's role in the CDSMP will also change. A program that no longer receives funding from AoA may continue the partnership or may develop a relationship with new funding agencies. Stronger CDSMP programs that are able to become self-sustaining and acquire new funding may also be unique "best practice" programs.
The agency completing the training may also influence AoA's role, especially if there is a prior relationship with AoA. The agency that runs the CDSMP program will hold a license that makes them legally responsible and it may be a different agency than who funds the program. The organizations supervising and setting up the program will vary and range from small private organizations with state contracts to run the program to large organizations that have a history of running CDSMP programs (e.g. Kaiser Permanente). The leadership's relationship to AoA may also play a role. For example, a leader with prior experience working in the Aging Network will have a better sense of outside resources and services. The types of leaders running the programs are health professionals and non-health professionals; staff or volunteers; and leaders who are similar to program participants or different from program participants. The location of the site is another key setting issue. For example, how and where the program is advertised and whether participation is open or closed. The programs can take place in a public or private setting, ranging from churches and public libraries to senior centers. Some non-AoA funded CDSMPs are run in health facilities, such as primary physician offices. The same CDSMP program may take place in multiple venues.
Another way to compare AoA's role across diverse settings is to compare two programs that are similar in structure and resources. One method is to compare the per participant program cost. The resources allocated to a particular CDSMP are likely to impact the benefit to participants. The cost per participant of a CDSMP program varies from $70 to $200 depending on the program and the actual 2-year savings are between $390 and $520 (Lorig et al, 2001). An evaluation of a 7-week CDSMP Kaiser program reported an average cost per person to be around $200 and the savings to near a 1:4 cost-to-savings ratio (Lorig et al., 2001). Another evaluation calculated the cost per participant for a 6 month program to be $70 and the savings in health care expenditures to be around $750 per participant, more than 10 times the cost of program (Lorig et al, 1999). The cost per participant in NHS programs is roughly 3 to 6 times greater than CDSMPs in the U.S. A study of cost effectiveness of a nationally implemented CDSMP in the NHS estimated costs at $489 per participant and a cost savings between $721 and $827 (Kennedy et al, 2007).
While the reports reviewed of Australian CDSMPs did not mentioned per participant cost, it was noted that the program was free for participants and they received reimbursement for travel. The per participant cost may vary for disease specific SMPs. Meta-analytic reviews (Chodosh et al, 2005) suggest that generic self-management courses might not be suitable for all types of chronic conditions and population subgroups. In the evaluation of an arthritis SMP, savings were found to be 4-5 times greater than the cost of the program. Per participant cost is one of the ways in which the CDSMP implementation sites will differ from one to another which may impact the outcomes.
Finally, reporting structures also vary by CDSMP program. Several CDSMP programs, such as the San Jose Trust in California, have established CDSMP programs and report data to AoA but receive no AoA funding. Even in the case of a CDSMP program with no AoA funding, AoA may have a role in how the program is run, through an established relationship. Finally, fidelity to Stanford CDSMP Program is also likely to influence AoA's role, as they are most familiar with the Stanford model and licensing. Only seven of the CDSMPs in the literature review are based solely on the Stanford model. There is some support for the idea that disease specific SMP may be more effective than generic CDSMP programs (Lorig, Ritter, & Plant, 2005).
The flexibility that AoA affords its CDSMP programs may be instrumental in their success. Given the diversity of CDSMP programs, AoA may have less of an influence than CDSMP programs run in England or Australia, but this flexibility may be key in meeting the needs of a diverse older population. Given the substantial heterogeneity in the settings of AoA funded CDSMP programs, it may be advisable to select the evaluation sites in such a way to include that variation (i.e., nationally representative sites).
Many studies reviewed for this report describe specific inclusion criteria related to certain chronic conditions, such as arthritis, cancer, or heart disease, while others stipulate the presence of any chronic disease as sufficient for participation. In general, findings revealed significant program effects for both the disease-specific (Barlow, et al., 2005; Beckmann, et al., 2007; Goeppinger, et al., 2007) and generic CDSMPs (K. Lorig, et al., 2001; K. Lorig, et al., 2003; K. Lorig, Ritter, & Jacquez, 2005; K. Lorig, Ritter, & Plant, 2005; K.R. Lorig, et al., 2001; Rose, et al., 2008).
When the two types of programs were compared head-to-head, there were seemingly contradictive findings of which program type emerged to lead to better outcomes. For example, one study found positive benefits for arthritis patients who participated in an arthritis-specific program and a generic CDSMP (Goeppinger, et al., 2007) while another study found more positive benefits for arthritis patients who participated in the disease-specific program over the generic program (K. Lorig, Ritter, & Plant, 2005).
Overall, the studies reviewed for this report included demographic characteristics as descriptive variables rather than as predictors or to stratify the sample. As noted above, only one report reviewed examined age as a predictor of benefits of the self-management program (Nolte, et al., 2007). This study revealed that younger participants, particularly younger women, reported benefits of the self-management program on most outcomes measured while few older adults showed improvement.
Most of the US based studies targeted populations of adults aged 40 and older while those in the UK targeted adults 18 years and older. The two key studies which demonstrated few or no significant differences between the CDSMP treatment versus control groups were conducted on populations of adults 65 years and older (Ersek, et al., 2008; Haas, et al., 2005). Thus, the effectiveness of CDSMPs on older adults does not seem to be established yet.
In addition to comparing disease-specific to generic programs, analyses can be conducted to compare programs that determine eligibility via self-diagnoses versus physician diagnosis. Our review of the literature revealed that in general, a critical difference between the CDSMP and the EPP is that patients who participate in the CDSMP generally have a physician-diagnosed chronic condition. In the United Kingdom, where the EPP is based on social inclusion and patient empowerment, access to the program is not based on medical diagnoses but on self-defined long-term conditions (Kennedy, et al., 2007)
The majority of the studies reviewed reported program effects on samples that were predominately white and female; however, a few studies explicitly targeted Hispanic or African-American populations (Gitlin, et al., 2008; K. Lorig, et al., 2003; Powers, et al., 2009). Similar to the results of the CDSMP among white samples, studies targeting Hispanic and African American participants revealed significant program effects. Rose and colleagues (Rose, et al., 2008) noted that the CDSMP was feasible and well-received with the older African American adults but also noted that they may benefit from a "booster" session to assure lasting programmatic effects in an older adult population.
Beneficiary type was left largely vague by most studies. Notable exceptions are Rosenman et al.'s (2006) review that specified Medicaid beneficiaries, and studies that reported on managed care organization beneficiaries (Ahmed & Villagra, 2006; Jerant, et al., 2009; K. Lorig, et al., 2001). And finally, nearly all articles reviewed dealt with self-management programs in community settings. Armstrong et al. (2008) and Lorig et al. (2006) described positive outcomes of an internet-based self-managed program.
The purpose of this literature review was to inform the design of an evaluation study to assess the efficacy of the Stanford University CDSMP in AoA funded sites. The data gathered through this review has provided logistic and methodological insight that will help the design team to narrow the focus of the design and strengthen the potential for significant, relevant findings regarding participant improvement in health status and health behaviors, and in reducing healthcare utilization and costs.
Overall, the studies reviewed for this report provide evidence supporting the utility of the CDSMP and similar self-management programs in improving self-efficacy, health status, and health behaviors. In addition, while relatively few studies investigated the effects on health care utilization, those that did found significant reductions in physician visits and hospital stay duration, suggesting that savings to health care financing programs such as Medicare may be possible.
Among the RCTs included in this review, the most common approach for evaluating the CDSMP was through a post-baseline, blinded randomization strategy, in which participants were recruited, completed baseline data collection, and were randomized into treatment and control groups. To avoid ethical implications of withholding what is considered to be a beneficial intervention, control group participants were placed on waiting lists, and received the intervention after the study period had ended. Some studies collected follow-up data on wait-listed participants, to further examine changes in outcomes from baseline to post-intervention timeframes (K. Lorig, et al., 2001). Given the ethical concerns, we recommend that AoA evaluate the CDSMP using a RCT-type design with a waitlist control group and blind randomization.
The most common data collection intervals used in these studies were four to six months post-intervention, and 12 months post-intervention; many of the RCTs reviewed collected data at least two times, and some extended the data collection period to two years post-intervention (K. Lorig, et al., 2001; Powers, et al., 2009). In these designs, average attrition rates ranged from 21 to 51 percent, depending upon the data collection interval, with common reasons for drop-out including illness, disinterest, and incorrect contact information. To maintain adequate sample for statistical power during the national CDSMP evaluation, we recommend that AoA adopt a recruitment/sampling strategy that accounts for attrition and utilizes these and other RCT studies to appropriately estimate and control for sample attrition.
Based on the review of literature of both randomized and non-randomized trials, positive outcomes were reported in CDSMP studies regardless of design method, suggesting that an evaluation conducted by the AoA could find similar positive health outcomes in a longitudinal study. However, the reviewed studies vary enough in health outcome findings and in the length of follow-up data collection to suggest that the national CDSMP evaluation take care to include at least a 12-month post-intervention period to determine the sustainability of the outcomes measured.
Data collection methodologies in most CDSMP studies focused almost exclusively on participant self-report, via mail survey questionnaires or telephone surveys. Instruments for assessing health status were developed for use in these studies (e.g., the modified Health Assessment Questionnaire physical disabilities scale (K. Lorig et al., 1996), the Health Education Impact Questionnaire (Osborne, Elsworth, & Whitfield, 2007), the Lorig Health Behavior Scale ((Osborne, et al., 2007)) and found to have good psychometric properties. These instruments and data collection methodology have made important contributions to our understanding of participant-centered program outcomes; however, in order to more fully understand the affect of these programs on health services utilization and public/private payor health care expenditures, additional datasets should be explored for the national evaluation. We recommend that AoA's evaluation contractor use validated instruments for participant self-report, supplemented by administrative claims and/or uniformly available health assessment data, to allow study of broader sets of outcome measures.
Four of the original Stanford CDSMP measures were most commonly used in the studies reviewed: health behavior, health status, health care utilization and self-efficacy. The original studies developed and validated instruments and scales for assessing these metrics, and additional measures have been developed and validated since, as described above. There is evidence that CDSMP improves ratings of self-reported health and increases self-efficacy and health-related QOL, but these findings vary by study and depending upon follow-up period. Some studies report decreases in health services utilization (i.e. inpatient hospital use) and potentially reduced costs for CDSMP participants; however, these studies have several limitations including the reliance on participant self-report.
Based upon these reviewed studies, as well as upon AoA's stated purposes for designing a national evaluation of the CDSMP, we recommend that the following outcomes be compared for CDSMP treatment and control participants, using a mixed methods approach to data collection and analysis:
- Health status (e.g., functioning, fatigue, social/role activity limitations), measured using both validated self-report instruments and independent clinical assessments.
- Health behaviors (e.g., communication with physician) using validated self-report instruments.
- Health care utilization (e.g., physician visits, inpatient hospital stays, emergency department use).
- Health care expenditures (e.g., costs to Medicare, Medicaid, other payors and out-of-pocket individual participants).
Many studies reviewed captured and analyzed participant demographic variables such as age, sex, and race/ethnicity to determine program effects. Though most studies reviewed sampled white and female CDSMP participants, several studies specifically targeted Hispanic participants (two studies) and rural African American older adults (one study). Similarly, most studies report age as a sample descriptor, rather than age in relation to outcomes. Given AoA's interest in understanding program effectiveness (including cost effectiveness) among elders, we recommend that the national evaluation specifically target elder (defined as 60-plus, per AoA guidance) CDSMP participants with broad cultural representation in order to demonstrate program effectiveness.
Findings are mixed with regard to whether disease-specific or generic CDSMP programs result in better outcomes for participants. Given these mixed results, we recommend that AoA's national evaluation contractor consider sampling both types of programs for study.
Finally, given that most reviewed studies focused on health outcomes rather than cost or utilization outcomes, types of beneficiaries by payor source were not a targeted group for sampling or for analysis. Again, given AoA's interest in understanding cost effectiveness for elders, we recommend that the national evaluation specifically target Medicare (and perhaps Medicaid) beneficiaries for study.
*References marked with an asterisk indicate studies included in Key Article Review Table (Appendix A).
American Recovery and Reinvestment Act: Communities Putting Prevention to Work, Chronic Disease Self-Management Program. (2010). (January 2010). Retrieved from http://www.aoa.gov/AoA_programs/HPW/ARRA/index.aspx
Arora, R., Boehm, J., Chimento, L., Moldawer, L., & Tsien, C. (2008). Designing and Implementing Medicaid Disease and Care Management Programs: A User's Guide. Retrieved 22 September 2010, from Agency for Healthcare Research and Quality: http://www.ahrq.gov/professionals/systems/long-term-care/resources/hcbs/medicaidmgmt/medicaidmgmt.pdf (1.59 MB)
Arora, R., Boehm, J. et al. (2008) Designing and Implementing Medicaid Disease and Care Management Programs: A User's Guide. AHRQ Publication No. 07(08)-0063, March 2008, Agency for Healthcare Research and Quality, Rockville, MD. http://www.ahrq.gov/professionals/systems/long-term-care/resources/hcbs/medicaidmgmt/index.html
Arthritis Integrated Dissemination State Descriptions. Retrieved from http://www.chronicdisease.org/files/public/ArthritisIntegratedDisseminat....
*Barlow, J., Turner, A., Edwards, R., & Gilchrist, M. (2009). A randomised controlled trial of lay-led self-management for people with multiple sclerosis. [Article]. Patient Education & Counseling 77(1), 81-89.
*Barlow, J., Wright, C. C., Turner, A. P., & Bancroft, G. V. (2005). A 12-month follow-up study of self-management training for people with chronic disease: are changes maintained over time? British Journal Of Health Psychology 10(Pt 4), 589-599.
*Battersby, M., Hoffmann, S., Cadilhac, D., Osborne, R., Lalor, E., & Lindley, R. (2009). 'Getting your life back on track after stroke': a Phase II multi-centered, single-blind, randomized, controlled trial of the Stroke Self-Management Program vs. the Stanford Chronic Condition Self-Management Program or standard care in. [Article]. International Journal of Stroke 4(2), 137-144.
*Beckmann, K., Strassnick, K., Abel, L., Hermann, J., & Oakley, B. (2007). Is a Chronic Disease Self-Management Program Beneficial to People Affected by Cancer? [Article]. Australian Journal of Primary Health 13(1), 36-44.
Business Wire. (2003). State of Florida and LifeMasters Extend Medicaid Disease-Management Program that Improves Health, Lowers Costs. FindArticles.com. Retrieved 15 Sept, 2010 from http://findarticles.com/p/articles/mi_m0EIN/is_2003_Nov_12/ai_n27674865/
*Chodosh, J., Morton, S. C., Mojica, W., Maglione, M., Suttorp, M. J., Hilton, L., et al. (2005). Chronic Disease Self-Management Programs for Older Adults. [Article]. Annals of Internal Medicine 143(6), I32-I32.
Counsell, S. R. (2010). Team Developed Care Plan and Ongoing Care Management by Social Workers and Nurse Practitioners Results in Better Outcomes and Fewer Emergency Department Visits for Low-Income Seniors.
*Damush, T. M., Jackson, G. L., Powers, B. J., Bosworth, H. B., Cheng, E., Anderson, J., et al. (2010). Implementing Evidence-Based Patient Self-Management Programs in the Veterans Health Administration: Perspectives on Delivery System Design Considerations. [Article]. JGIM: Journal of General Internal Medicine 25, 68-71.
Agency for Healthcare Research and Quality. (2010) Designing and Implementing Medicaid Disease and Care Management Programs. Retrieved from http://www.ahrq.gov/professionals/systems/long-term-care/resources/hcbs/medicaidmgmt/medicaidmgmt4.html.
Edlin, M. (2006). Opening Doors on Disease Management. American's Health Insurance Plans (AHIP). Retrieved 15 Sept 2010 from http://www.ahip.org/content/default.aspx?bc=31|130|136|17340|17341
*Ersek, M., Turner, J. A., Cain, K. C., & Kemp, C. A. (2008). Results of a randomized controlled trial to examine the efficacy of a chronic pain self-management group for older adults [ISRCTN11899548]. [Article]. Pain (03043959), 138(1), 29-40.
*Gitlin, L. N., Chernett, N. L., Harris, L. F., Palmer, D., Hopkins, P., & Dennis, M. P. (2008). Harvest Health: Translation of the Chronic Disease Self-Management Program for Older African Americans in a Senior Setting. [Article]. Gerontologist 48(5), 698-705.
Goehring, C. (2010). Fact Sheet: Chronic Disease Self-Management Program. Retrieved from www.adsa.dshs.wa.gov.
*Goeppinger, J., Armstrong, B., Schwartz, T., Ensley, D., & Brady, T. J. (2007). Self-management education for persons with arthritis: Managing comorbidity and eliminating health disparities. Arthritis And Rheumatism 57(6), 1081-1088.
*Gordon, C. a. T. G. (2008). Review of Findings on Chronic Disease Self-Management Program (CDSMP) Outcomes: Physical, Emotional & Health-Related Quality of Life, Healthcare Utilization and Costs 1-10.
*Haas, M., Groupp, E., Muench, J., Kraemer, D., Brummel-Smith, K., Sharma, R., et al. (2005). Chronic disease self-management program for low back pain in the elderly. Journal Of Manipulative And Physiological Therapeutics 28(4), 228-237.
*Heisler, M. (2010). Different models to mobilize peer support to improve diabetes self-management and clinical outcomes: evidence, logistics, evaluation considerations and needs for future research. [Article]. Family Practice 27(suppl), i23-i32.
*Jerant, A., Moore-Hill, M., & Franks, P. (2009). Home-based, peer-led chronic illness self-management training: findings from a 1-year randomized controlled trial. Annals Of Family Medicine 7(4), 319-327.
*Johnson, P., & Raterink, G. (2009). Implementation of a diabetes clinic-in-a-clinic project in a family practice setting: using the plan, do, study, act model. [Article]. Journal of Clinical Nursing 18(14), 2096-2103.
*Kendall, E., Catalano, T., Kuipers, P., Posner, N., Buys, N., & Charker, J. (2007). Recovery following stroke: The role of self-management education. [Article]. Social Science & Medicine 64(3), 735-746.
*Kennedy, A., Reeves, D., Bower, P., Lee, V., Middleton, E., Richardson, G., et al. (2007). The effectiveness and cost effectiveness of a national lay-led self care support programme for patients with long-term conditions: a pragmatic randomised controlled trial. Journal Of Epidemiology And Community Health 61(3), 254-261.
*Lorig, K., Hurwicz, M.-L., Sobel, D., Hobbs, M., & Ritter, P. L. (2005). A national dissemination of an evidence-based self-management program: a process evaluation study. [Article]. Patient Education & Counseling 59(1), 69-79.
*Lorig, K., Ritter, P., Stewart, A. L., Sobel, D. S., Brown, B. W., Jr., Bandura, A., et al. (2001). Chronic disease self-management program: 2-year health status and health care utilization outcomes. Medical Care 39(11), 1217-1223.
*Lorig, K., Ritter, P. L., Dost, A., Plant, K., Laurent, D. D., & McNeil, I. (2008). The Expert Patients Programme online, a 1-year study of an Internet-based self-management programme for people with long-term conditions. Chronic Illness 4(4), 247-256.
*Lorig, K., Ritter, P. L., & Plant, K. (2005). A disease-specific self-help program compared with a generalized chronic disease self-help program for arthritis patients. Arthritis And Rheumatism 53(6), 950-957.
*Lorig, K., Sobel, D. S., Stewart, A. L., Brown, B. W., Jr., Bandura, A., Ritter, P., et al. (1999). Evidence suggesting that a chronic disease self-management program can improve health status while reducing hospitalization: a randomized trial. Medical Care 37(1), 5-14.
Lorig, K., Stewart, A., Ritter, P., Gonzalez, V., Laurent, D., & Lynch, J. (1996). Outcome measures for health education and other health care interventions. Thousand Oaks, CA: Sage Publications, Inc.
National Evaluation of the Sharing Health Care Initiative: Final Technical Report. (2005). Retrieved from http://www.health.gov.au/internet/main/publishing.nsf/Content/B71197DF35...$File/FINAL%20SHCI%20Technical%20Report.pdf
*Nolte, S., Elsworth, G. R., Sinclair, A. J., & Osborne, R. H. (2007). The extent and breadth of benefits from participating in chronic disease self-management courses: A national patient-reported outcomes survey. [Article]. Patient Education & Counseling 65(3), 351-360.
*Osborne, R. H., Elsworth, G. R., & Whitfield, K. (2007). The Health Education Impact Questionnaire (heiQ): An outcomes and evaluation measure for patient education and self-management interventions for people with chronic conditions. [Article]. Patient Education & Counseling 66(2), 192-201.
*Powers, B. J., Olsen, M. K., Oddone, E. Z., & Bosworth, H. B. (2009). The Effect of a Hypertension Self-Management Intervention on Diabetes and Cholesterol Control. [Article]. American Journal of Medicine 122(7), 639-646.
*Richardson, G., Kennedy, A., Reeves, D., Bower, P., Lee, V., Middleton, E., et al. (2008). Cost effectiveness of the Expert Patients Programme (EPP) for patients with chronic conditions. [Article]. Journal of Epidemiology & Community Health 62(4), 361-367.
*Rogers, A., Kennedy, A., Bower, P., Gardner, C., Gately, C., Lee, V., et al. (2008). The United Kingdom Expert Patients Programme: results and implications from a national evaluation. The Medical Journal Of Australia 189(10 Suppl), S21-S24.
*Rogers, A., Kennedy, A., Nelson, E., & Robinson, A. (2005). Uncovering the Limits of Patient-Centeredness: Implementing a Self-Management Trial for Chronic Illness. [Article]. Qualitative Health Research 15(2), 224-239.
*Rose, M. A., Arenson, C., Harrod, P., Salkey, R., Santana, A., & Diamond, J. (2008). Evaluation of the Chronic Disease Self-Management Program With Low-Income, Urban, African American Older Adults. [Article]. Journal of Community Health Nursing 25(4), 193-202.
*Rosenman, M. B., Holmes, A. M., Ackermann, R. T., Murray, M. D., Doebbeling, C. C., Katz, B., et al. (2006). The Indiana Chronic Disease Management Program. [Article]. Milbank Quarterly 84(1), 135-163.
*Sequist, T. D., von Glahn, T., Li, A., Rogers, W. H., & Safran, D. G. (2009). Statewide evaluation of measuring physician delivery of self-management support in chronic disease care. Journal Of General Internal Medicine 24(8), 939-945.
*Smeulders, E. S. T. F., van Haastregt, J. C. M., Ambergen, T., Janssen-Boyne, J. J. J., van Eijk, J. T. M., & Kempen, G. I. J. M. (2009). The impact of a self-management group programme on health behaviour and healthcare utilization among congestive heart failure patients. European Journal Of Heart Failure: Journal Of The Working Group On Heart Failure Of The European Society Of Cardiology 11(6), 609-616.
*Swerissen, H., Belfrage, J., Weeks, A., Jordan, L., Walker, C., Furler, J., et al. (2006). A randomised control trial of a self-management program for people with a chronic illness from Vietnamese, Chinese, Italian and Greek backgrounds. Patient Education And Counseling 64(1-3), 360-368.
*Young, A. S., Chaney, E., Shoai, R., Bonner, L., Cohen, A. N., Doebbeling, B., et al. (2007). Information Technology to Support Improved Care For Chronic Illness. [Article]. JGIM: Journal of General Internal Medicine 22, 425-430.
Page originally created May 2011