Mechanisms of Health Behavior Change in Persons With Chronic Illness or Disability: The Health Action Process Approach (HAPA)

Transcription

1 Rehabilitation Psychology 2011, Vol. 56, No. 3, American Psychological Association /11/$12.00 DOI: /a Mechanisms of Health Behavior Change in Persons With Chronic Illness or Disability: The Health Action Process Approach (HAPA) Ralf Schwarzer Freie Universität Berlin Sonia Lippke Jacobs University Bremen Aleksandra Luszczynska University of Colorado and Warsaw School of Social Sciences & Humanities Objective: The present article presents an overview of theoretical constructs and mechanisms of health behavior change that have been found useful in research on people with chronic illness and disability. A self-regulation framework (Health Action Process Approach) serves as a backdrop, making a distinction between goal setting and goal pursuit. Risk perception, outcome epectancies, and task self-efficacy are seen as predisposing factors in the goal-setting (motivational) phase, whereas planning, action control, and maintenance/recovery self-efficacy are regarded as being influential in the subsequent goal-pursuit (volitional) phase. The first phase leads to forming an intention, and the second to actual behavior change. Such a mediator model serves to eplain social cognitive processes in health behavior change. By adding a second layer, a moderator model is provided in which three stages are distinguished to segment the audience for tailored interventions. Identifying persons as preintenders, intenders, or actors offers an opportunity to match theorybased treatments to specific target groups. Numerous research and assessment eamples, especially within the physical activity domain, serve to illustrate the application of the model to rehabilitation settings and health promotion for people with chronic illness or disability. Conclusions/Implications: The theoretical developments and research evidence for the self-regulation framework eplain the cognitive mechanisms of behavior change and adherence to treatment in the rehabilitation setting. Keywords: rehabilitation, motivation, volition, self-efficacy, intention, risk perception This article describes a health promotion model for persons with chronic illness or disability. The Health Action Process Approach (HAPA) has been found useful to describe, eplain, and predict changes in health behaviors in a variety of settings, in particular in rehabilitation settings. Interventions need to be theory-guided to allow for meaningful interpretations of empirical findings and to draw valid conclusions. Recommendations for successful practice need a theory base. It is therefore important for rehabilitation psychologists to understand different theories and judge their potential for diverse research questions. In the following section, we will provide a brief introduction on lifestyle change in rehabilitation. Then we will describe various theoretical advances in health behavior change. This article was published Online First July 18, Ralf Schwarzer, Department of Psychology, Freie Universität Berlin, Berlin, Germany; Sonia Lippke, Jacobs Center on Lifelong Learning and Institutional Development, Jacobs University Bremen, Bremen, Germany; Aleksandra Luszczynska, Trauma, Health, & Hazards Center, University of Colorado and Warsaw School of Social Sciences & Humanities, Warsaw, Poland. Correspondence concerning this article should be addressed to Ralf Schwarzer, PhD, Department of Psychology, Freie Universität Berlin, Habelschwerdter Allee 45, Berlin, Germany. ralf.schwarzer@ fu-berlin.de Health Behavior Change in Rehabilitation For people with chronic illness or disability, behavior change can be an effective strategy to prevent further morbidity and mortality. When people are enrolled in medical rehabilitation, there is a good chance to help them learn how to improve their health behaviors and maintain them after discharge. For eample, persons with Coronary Artery Disease (CAD) received the following recommendations regarding their health behaviors to improve their prognosis: (1) No smoking; (2) moderate to intensive physical activity, (3) moderate or no alcohol consumption; (4) healthy body weight control; (5) limited intake of saturated fats and trans fatty acids; (6) regular fish consumption; (7) sufficient amounts of fruit and vegetables; (8) Sufficient fiber intake, for eample from grains, legumes, or nuts; (9) reduced salt intake (Iestra et al., 2005). In a systematic review, Iestra et al. (2005) found for all-cause mortality, that positive changes in all behaviors decreased the mortality risk. For eample, physical activity in CAD patients reduced the mortality risk by 25%. The authors also compared their findings with the general population and found comparable effects for four health behaviors. The main challenge is to increase the patients motivation to adhere to health behavior changes. Whether they adhere to an intervention program has a substantial effect on health outcomes. However, some programs fail to show evidence that they increase the level of adherence (Cutrona et al., 2010; Dean, Walters, & Hall, 2010; McLean, Burton, Bradley, & Littlewood, 2010). What 161

2 162 SCHWARZER, LIPPKE, AND LUSZCZYNSKA makes it more difficult to improve adherence are comorbidities (e.g., depression, aniety) and increased pain levels during eercise, as well as health care-related factors, such as characteristics of the health care providers and organizations. Moreover, patients often face lower preeisting levels of physical activity, helplessness, poor social support, and more perceived barriers to eercise than the general population (McLean et al., 2010). Consequently, more effective interventions to increase adherence are needed. Theory- and evidence-based interventions that are tailored to psychological constructs are more effective (Noar, Benac, & Harris, 2007). Health self-regulation is mandatory, but it is very demanding for someone with chronic illness and disability. Research on health self-regulation eamines how change takes place, and why some individuals change, whereas others do not. It identifies relevant causal factors and their interplay in the change process. This leads to theories that may be generalized to several health behaviors and to several types of illness. For eample, a theory could present particular mechanisms that are valid for motivating orthopedic patients to adopt and maintain physical eercise. This article presents an overview of theoretical constructs and frameworks that have been found useful in the contet of behavior change, in particular for individuals with chronic illness and disability. We will focus on empirical evidence in the contet of rehabilitation settings, confining the discussion to physical activity as one of the main health behaviors, although the principles selected also apply to other health behaviors. Health Behavior Change: The Need for Theory To eplain, predict, and effectively improve the self-regulation of individuals and to optimize treatment, theories of health behavior change are needed (Dunn & Elliot, 2008). In general, such theories are divided into continuum models and stage models. In continuum models, people are positioned along a range that reflects the likelihood of action. Influential predictor variables are identified and combined in one prediction equation. The goal of an intervention is to move the person along this route toward action. Health promotion, then, focuses on increasing all model-inherent variables in all persons, without matching treatments to particular audiences. The Theory of Planned Behavior (TPB; Ajzen, 1991) is such a continuum model that is applied to rehabilitation settings (Dion, Johnston, Rowley, & Pollard, 2008; Eng & Martin Ginis, 2007; Galea & Bray, 2006; Latimer, Martin Ginis, & Arbour, 2006; Yardley & Donovan-Hall, 2007). Other continuum models are the Social-Cognitive Theory (SCT; Bandura, 1986, 2004) and the Protection Motivation Theory (PMT; Rogers, 1975). To overcome the practical limitations of continuum models, the change process has been subdivided by a number of qualitative stages. According to stage theories, health behavior change consists of an ordered set of categories (or stages) into which people can be classified. These categories reflect cognitive or behavioral characteristics, such as the intention to perform a behavior. Following Weinstein, Rothman, and Sutton (1998), the common defining properties of stage models are: (1) Individuals can be classified into different stages by a valid assessment procedure; (2) The stages are ordered, that is, Stage 3 is closer to the criterion behavior than Stages 1 and 2, and Stage 1 is farthest from the criterion behavior. A person in Stage 1 has to move first to Stage 2 before proceeding to Stage 3 and finally adopting the criterion behavior. Finally, (3) individuals in the same stage are more similar than those in different stages; that is, they face the same barriers, but these barriers are different from those in other stages. The main purpose of applying stage models lies in the identification of relatively homogeneous target groups for interventions and the design of stage-matched treatments. The most popular stage theory of health behavior change is the Transtheoretical Model (TTM; J. O. Prochaska & DiClemente, 1983) that proposes five stages of change. It has also been applied to rehabilitation settings (e.g., Gorczynski, Faulkner, Greening, & Cohn, 2010). Both continuum models and stage models have their advantages and disadvantages. Continuum models have been found useful for eplanation and prediction, whereas stage models are often preferred to guide interventions. For health promotion, the continuum models are often too general because all variables involved in such a model need to be addressed in interventions, without considering the special needs of particular subgroups of participants. However, it is possible to integrate both approaches when researchers use a continuum model as a theoretical template and, when it comes to interventions, subdivide the audience into stage groups to allow for stage-matched treatments. Mechanisms of Change: Mediation and Prediction The HAPA (Schwarzer, 2008) has two layers: a continuum layer and a stage layer. We now describe this model in more detail. The traditional continuum models have been criticized mainly because of the so-called intention-behavior gap (referring to the frequent failure of intention to predict behavior). A model that eplicitly includes postintentional factors to overcome this gap is the HAPA. The model suggests a distinction between (1) preintentional motivation processes that lead to a behavioral intention, and (2) postintentional volition processes that lead to the actual health behavior. Within the two phases, different patterns of social cognitive predictors may emerge (see Figure 1). In the initial motivation phase (1), a person develops an intention to act. In this phase, risk perception is seen as a distal antecedent (e.g., I am at risk for cardiovascular disease ). Risk perception in itself is insufficient to enable a person to form an intention. Rather, it sets the Figure 1. The HAPA (Schwarzer, 2008).

3 SPECIAL SECTION: HEALTH BEHAVIOR CHANGE 163 stage for a contemplation process and further elaboration of thoughts about consequences and competencies. Similarly, positive outcome epectancies (e.g., If I eercise five times per week, I will reduce my cardiovascular risk ) are chiefly seen as being important in the motivation phase, when a person balances the pros and cons of certain behavioral outcomes. Further, one needs to believe in one s capability to perform a desired action (perceived self-efficacy; e.g., I am capable of adhering to my eercise schedule in spite of the temptation to watch TV ). Perceived selfefficacy operates in concert with positive outcome epectancies, both of which contribute substantially to forming an intention. Both beliefs are needed for forming intentions to adopt difficult behaviors, such as regular physical eercise. After forming an intention, the volitional phase (2) is entered. When a person is inclined to adopt a particular health behavior, the good intention has to be transformed into detailed instructions on how to perform the desired action. Once an action has been initiated, it has to be maintained. This is not achieved through a single act of will, but involves self-regulatory skills and strategies. Thus, the postintentional phase should be further broken down into more proimal factors, such as planning, action control, social support, and recovery self-efficacy. Most other social cognition models do not eplicitly address postintentional factors. Social support is one factor reflecting the barriers and resources part of the HAPA model: Support represents a resource, and the lack of it can be a barrier to adopt or maintain health behaviors. Instrumental, emotional, and informational social support can enable the adoption and continuation of behaviors. This was found in studies with chronically ill people, for instance diabetics (Plotnikoff, Lippke, Courneya, Birkett, & Sigal, 2008). It has also been found in a sample of multimorbid older adults that eercising together with a partner has a beneficial effect on adherence (Gellert, Ziegelmann, Warner, & Schwarzer, in press). Including planning and self-efficacy as volitional mediators renders the HAPA into an implicit stage model because it implies the eistence of at least a motivational and a volitional phase. The purpose of such a model is twofold: It allows a better prediction of behavior, and it reflects the assumed causal mechanism of behavior change. Research that is based on this continuum layer of the model, therefore, uses path-analytic methods (e.g., Lippke, Ziegelmann, & Schwarzer, 2005; Luszczynska & Schwarzer, 2003; Renner et al., 2008; Schwarzer et al., 2007; Ziegelmann, Luszczynska, Lippke, & Schwarzer, 2007). Designing Stage-Matched Interventions: Preintenders, Intenders, Actors When it comes to the design of interventions, one can consider turning the implicit stage model into an eplicit one. This is done by identifying individuals who are located either in the motivational stage or in the volitional stage. Then, each group receives a specific treatment that is tailored to this group. Moreover, it is useful and theoretically meaningful to subdivide the volitional group further into those persons who perform and those who only intend to perform. In the postintentional-preactional stage, individuals are labeled intenders, and in the actional stage actors. Thus, a suitable subdivision within the health behavior change process yields three groups: preintenders, intenders, and actors. The term stage in this contet was chosen to allude to the stage theories, but not in the strict definition that includes irreversibility and invariance. The terms phase or mindset may be equally suitable for this distinction. The basic idea is that individuals pass through different mindsets on their way to behavior change. Thus, interventions may be most efficient when tailored to these particular mindsets. For eample, preintenders are supposed to benefit from confrontation with outcome epectancies and some level of risk communication. They need to learn that the new behavior (e.g., becoming physically active) has positive outcomes (e.g., well-being, weight loss, fun) as opposed to the negative outcomes that accompany the current (sedentary) behavior (such as developing an illness or being unattractive). In contrast, intenders should not benefit much from health messages in the form of outcome epectancies because, after setting a goal, they have already moved beyond this mindset. Rather, they should benefit from planning to translate their intentions into action. Finally, actors should be prepared for particular high-risk situations in which lapses are imminent. Interventions should help them if they desire to change something in their routine (e.g., adopting or altering a behavior). A Set of Principles The HAPA is designed as an open architecture that is based on principles rather than on specific testable assumptions. It was developed in 1988 (Schwarzer, 1992) as an attempt to integrate the model of action phases (Heckhausen & Gollwitzer, 1987) with social cognitive theory (Bandura, 1986). It has five major principles that make it distinct from other models and that help to apply the HAPA to research and interventions. Principle 1: Motivation and volition. The first principle suggests that one should divide the health behavior change process into two phases. There is a switch of mindsets when people move from deliberation to action. First is the motivation phase in which people develop their intentions. Afterward, they enter the volition phase. Principle 2: Two volitional phases. In the volition phase, there are two groups of people: those who have not yet translated their intentions into action, and those who have. There are inactive as well as active persons in this phase. In other words, in the volitional phase one finds intenders as well as actors who are characterized by different psychological states. Thus, in addition to health behavior change as a continuous process, one can also create three categories of people with different mindsets, depending on their current location within the course of health behavior change: preintenders, intenders, and actors. Principle 3: Postintentional planning. Intenders who are in the volitional preactional stage are motivated to change, but do not act because they might lack the right skills to translate their intention into action. Planning is a key strategy at this point. It serves as an operative mediator between intentions and behavior. Principle 4: Two kinds of mental simulation. Planning can be divided into action planning and coping planning. Action planning pertains to the when, where, and how of intended action. Coping planning includes the anticipation of barriers and the design of alternative actions that help to attain one s goal despite the impediments.

4 164 SCHWARZER, LIPPKE, AND LUSZCZYNSKA Principle 5: Phase-specific self-efficacy. Perceived self-efficacy is required throughout the entire process. However, the nature of self-efficacy differs from phase to phase. This is because there are different challenges as people progress from one phase to the net. Goal setting, planning, initiative, action, and maintenance all pose challenges that are not of the same nature. Therefore, one should distinguish between preactional self-efficacy, coping selfefficacy, and recovery self-efficacy. Sometimes the terms task self-efficacy instead of preaction self-efficacy, and maintenance self-efficacy instead of coping and recovery self-efficacy are preferred. In summary, the HAPA has two layers: a continuum layer and a phase (or stage) layer. Depending on the research question, one might choose one or the other. HAPA is designed as a sequence of two continuous self-regulatory processes, a goal-setting phase (motivation) and a goal-pursuit phase (volition). The second phase is subdivided into a preaction phase and an action phase. One can superimpose these three phases on the continuum model as a second layer and regard phase as a moderator. This two-layer architecture allows to switch between the continuum model and the stage model, depending on the given research question. The stage layer is useful for designing stage-matched interventions. For preintenders, one needs risk and resource communication, for eample by addressing the pros and cons of a critical behavior. For intenders, planning treatments are helpful to support those who lack the necessary skills to translate their intentions into behavior. Furthermore, for actors, one needs to stabilize their newly adopted health behaviors by relapse prevention strategies. The HAPA allows both the researcher and the practitioner to make a number of choices. Although it was initially inspired by distinguishing between a motivational and a volitional stage, and later epanded to the distinction between preintenders, intenders, and actors, one need not necessarily group individuals according to such stages. If the purpose is to predict behavior change, one would specify a mediator model that includes postintentional constructs (such as planning and volitional self-efficacy) as proimal predictors of performance (Scholz, Nagy, Göhner, Luszczynska, & Kliegel, 2009). For stage-tailored interventions, however, usually three stage groups would be established. This does not eclude the possibility of generating more than three stages. For eample, for some research questions, one might subdivide the preintenders into precontemplators and contemplators, according to the TTM (J. O. Prochaska & DiClemente, 1993). Or one might opt for a distinction between preintenders, who are either (1) unaware of an issue, (2) aware but unengaged, or (3) still deciding (Weinstein, Lyon, Sandman, & Cuite, 1998). Thus, HAPA is not a puristic stage model, but a versatile theoretical framework that allows for a variety of approaches. Measurement of Theory-Implied Constructs The following section provides general assessment rules, citing eamples from the physical activity domain that can easily be adapted to other behaviors. Behavior Behavior can be assessed either subjectively or objectively: Objective measures could be based on a device, such as a pedometer or accelerometer (tracking steps or movements), a pill-counter, or direct observation of behaviors (attendance rates, observing which products are bought or used for cooking, etc.). Although objective measures are less likely to be biased, they are more demanding to gather. As all social cognitive variables are measured subjectively, the scale correspondence is higher when behavior is also measured subjectively by self-report (Courneya, 1994). Most often, data are collected with questionnaires or in personal interviews. Other subjective measures, such as diary logs, might be more valid and reliable in comparison to typical questionnaires or interviews. However, data are often missing because of noncompliance. A questionnaire regarding physical activity that has been validated with physiological and anthropometric measures (i.e., VO 2 ma and body fat) is the Godin Leisure-Time Eercise Questionnaire (GLTEQ; Godin & Shephard, 1985), or its modified and adapted versions (Lippke, Fleig, Pomp, & Schwarzer, 2010; Plotnikoff et al., 2007). Intention Intention to perform a behavior should be assessed the same way as assessing behavior itself. For physical activity, one could pose the following three items: I intend to perform the following activities at least three [2] days per week for 40 [20] min... (1)... strenuous (heartbeats rapidly, sweating) physical activities ; (2)... moderate (not ehausting, light perspiration) physical activities ; and (3)... mild (minimal effort, no perspiration) physical activity. Answers can be entered on a si-point scale ranging from not at all true (1) to absolutely true (6). Two different scales may be aggregated corresponding to the activity measurement: (1) Intention to Perform Strenuous and Moderate Activities (the correlation of the two items is typically rather low, as the two items have desirable discriminant validity), and (2) Intention to Engage in Strenuous, Moderate, and Mild Activities. Alternatively, intention can also be assessed with an ordinal rating scale in the same way that four domains of behavior are measured: (1) physiotherapeutic eercises (e.g., back training), (2) fitness activities (e.g., using an eercise bike), (3) resistance training (training muscle strength, e.g., on machines), (4) physical activity while commuting (e.g., going by bicycle or walking for longer distances). Intentions should also refer to ehausting activity and physical activity outside of work. Study participants can indicate the intended frequency and duration by ticking one of the following options: not at all (1), less than once per week for 40 [20] min (2), at least once per week for 40 [20] min (3), at least three [1] times per week for 40 [20] min (4), and five times per week for 40 [20] min or more (5). Answers can be categorized in such a way that a dichotomous variable results, namely whether cardiac [orthopedic] patients perform at least the recommended activity three [2] times per week for 40 [20] min (1) or not (0). Planning Action Planning can be assessed with items addressing the when, where, and how of the activity. The items in the rehabilitation study by Lippke, Fleig et al. (2010) were worded: For the month after the rehabilitation, I have already planned... (1)... which physical activity I will perform (e.g., walking), (2)... where I will be

5 SPECIAL SECTION: HEALTH BEHAVIOR CHANGE 165 physically active (e.g., in the park), (3)... on which days of the week I will be physically active, and (4)... for how long I will be physically active. Coping planning, on the other hand, can be measured with the item stem I have made a detailed plan regarding... and the items (1)... what to do if something interferes with my plans, (2)... how to cope with possible setbacks, (3)... what to do in difficult situations in order to act according to my intentions, (4)... which good opportunities for action to take, and (5)... when I have to pay etra attention to prevent lapses. In a study with cardiac patients, the items proved high measurement qualities (Sniehotta, Schwarzer, Scholz, & Schüz, 2005). Action Control While planning is a prospective strategy, that is, behavioral plans are made before the situation is encountered, action control is a concurrent self-regulatory strategy, where the ongoing behavior is continuously evaluated with regard to a behavioral standard. Action control can be assessed with a 6-item scale comprising three facets of the action control process (Sniehotta et al., 2005): self-monitoring (Items a and b), awareness of standards (Items c and d), and self-regulatory effort (Items e and f): (1) I consistently monitored myself whether I eercised frequently enough, (2) I consistently monitored when, where, and how long I eercise, (3) I have always been aware of my prescribed training program, (4) I often had my eercise intention on my mind, (5) I really tried hard to eercise regularly, and (6) I took care to train as much as I indented to. Self-Efficacy Perceived motivational and volitional self-efficacy can be composed of items such as the following. Motivational selfefficacy refers to the goal-setting phase and can be measured with the stem I am certain... followed by the two items... that I can be physically active on a regular basis, even if I have to mobilize myself, and... that I can be physically active on a regular basis, even if it is difficult (correlation of the two items r.79; Lippke, Fleig et al., 2010). Volitional selfefficacy refers to the goal-pursuit phase. It can be subdivided into maintenance self-efficacy and recovery self-efficacy. Maintenance self-efficacy has been measured with the stem I am capable of continuous physical eercise on a regular basis... followed by the two items... even if it takes some time until it becomes routine, and... even if I need several tries until I am successful (correlation of the two items r.82; Lippke, Fleig et al., 2010). Items on recovery self-efficacy can be worded I am confident that I can resume a physically active lifestyle, even if I have relapsed several times, I am confident that I am able to resume my regular eercises after failures to pull myself together, or I am confident that I can resume my physical activity, even when feeling weak after an illness. In three rehabilitation studies, the scales consisting of three to si items ehibited high reliability and validity in orthopedic and cardiac rehabilitation samples (Schwarzer, Luszczynska, Ziegelmann, Scholz, & Lippke, 2008). Risk Perception Risk perception can be measured by items such as: How high do you rate the likelihood that you will ever get one of the following diseases, or that you will relapse to them? (a) cardiovascular disease (e.g., heart attack, stroke), (b) diseases of the musculoskeletal system (e.g., osteoarthritis, herniated vertebral disk)? Any other health risk can be added, especially if relevant to the individual sample included in the study (Fleig, Lippke, Pomp, & Schwarzer, 2011; Schwarzer et al., 2007). Outcome Epectancies Positive outcome epectancies (pros) and negative outcome epectancies (cons) can be assessed with the stem If I engage in physical activity at least three [two] times per week for 40 [20] min... plus the following items. Pros are measured with five items, for eample,... then I feel better afterward,... then I meet friendly people, or... then my elasticity would increase. Cons were assessed by five items such as... then every time would cost me a lot of money,... then I would be financially burned, or... then I would have to invest a lot (e.g., into the organization). Social Support Received social support can be from different sources, such as family and friends. In a rehabilitation sample, social support regarding physical activity was assessed with 10 items: My family/ friends... (1)... have encouraged me to perform my planned activities, (2)... have reminded me to engage in physical activity, (3)... have helped me to organize my physical activity, (4)... have managed the household for me so that I could engage in physical activity, (5)... have engaged in physical activity with me (Jackson, Lippke, & Ziegelmann, 2010). Stage Algorithms The assessment of stages should be measured without using a specific time frame, in accordance with Sutton (2000). Study participants are asked Please think about the month before the rehabilitation started: Did you engage in physical activity at least 3 [2] days per week for 40 [20] min or more? Additionally, the study participants read the instruction If you have had an accident or some health conditions restraining you from performing your typical activities, please think about the month before that incident. Possible answers were yes or no. Another question was: For the month after the rehabilitation: Do you intend to perform physical activities 3 [2] times per week for 40 [20] min or more? with the answers yes or no. Again, for the stage items, the following information was eplicitly provided: When thinking about physical activities, please consider: (1) You are performing the activities during your recreational time or while commuting. (2) You are not achieving them during your working time. (3) You are doing them because you intend to do physical activities. (4) The activities have to be at least somewhat ehausting. Participants are asked to answer with either yes or no. Those individuals who indicate that they were active before the rehabilitation are then

6 166 SCHWARZER, LIPPKE, AND LUSZCZYNSKA categorized as Actors. Those who answer that they were not active, but intended to perform the recommended goal activity, are called Intenders. And those answering that they were not active and did not intend to perform the recommended goal activities are assumed to be Preintenders. An alternative stage algorithm can be similar to the one presented in Table 1, which was used in the study by Lippke et al. (in press). Persons indicating 1 and 2 on the list above would be categorized as Preintenders, those answering 3 as Intenders, and 4 and 5 as Actors. The algorithm was designed as a rating scale, with (precontemplating) preintenders on the very left and (maintaining) actors on the very right, for two reasons: (1) This format is more similar to the assessment of other social cognitive variables, such as intention and planning, and, by this, measurement correspondence is achieved. (2) The format is less space consuming (Lippke, Ziegelmann, Schwarzer, & Velicer, 2009). The algorithm has been successfully adapted to other behaviors that are relevant for individuals with chronic illness: In a study by Lippke et al. (in press), the adapted algorithm was successfully used to assess the stages regarding foot care, nutrition, smoking, alcohol, and healthy drinking in two samples of persons who had diabetes. The measurement qualities of a stage algorithm cannot be evaluated in terms of Cronbach s alpha, but in regard to Sensitivity (agreement between classification as being active by the stage algorithm, and performing the goal activity according to a different measure) and Specificity (agreement between classification as being nonactive, and the nonperformance of the goal activity). The algorithm presented in Table 1 (Lippke, Ziegelmann et al., 2009) had high sensitivity (70%) and specificity (80%) in comparison to previous studies (Nigg, 2005; Plotnikoff et al., 2007). The algorithm described above, specifically designed for rehabilitation patients, has demonstrated very high sensitivity, but comparably lower specificity. So far, it remains open whether this depends on the stage algorithm or the rehabilitation contets. Empirical Evidence How Does the Model Work? Mediating Effects Empirical evidence on HAPA assumptions range from single case studies in therapy and counseling (Fiore, 2007) to large-scale investigations in rehabilitation settings (e.g., Schwarzer et al., 2008). Eperimental manipulation (Lippke, Ziegelmann, & Schwarzer, 2004; Luszczynska, Gregajtys, & Abraham, 2007; Sniehotta, Scholz, & Schwarzer, 2006; Scholz, Knoll, Sniehotta, & Schwarzer, 2006; Ziegelmann, Lippke, & Schwarzer, 2006) and eperimental causal-chain-studies (Reuter, Ziegelmann, Lippke, & Schwarzer, 2009) provide further evidence. In addition, metaanalyses included aspects of the HAPA (e.g., planning: Gollwitzer & Sheeran, 2006; self-efficacy, outcome epectancies and risk perception: Milne, Sheeran, & Orbell, 2000; behavior change techniques: Michie, Abraham, Whittington, McAteer, & Gupta, 2009; stages, self-efficacy, social support, intention, risk perception: Noar et al., 2007). In the following, findings are presented on the five HAPA principles that may have implications for further research and theory. Principle 1: Motivation and volition. The two phases of behavior change were found to be empirically different, for instance in studies by Lippke et al. (2004, 2005) and Lippke, Schwarzer et al. (2010). In an orthopedic sample, the three HAPA stages were assessed, and it was investigated whether qualitatively different stages of health behavior change and stage-specific mindsets would be evident. Significant differences between motivational and volitional individuals were supported by paired comparisons and nonlinear trends (Lippke et al., 2005). In particular, preintenders reported significantly less self-efficacy, intention, and planning than intenders. In addition, in longitudinal stage-specific prediction patterns, it was important to consider stages: Preintenders with high risk perception were much more likely to develop higher intentions over time than preintenders with low risk perception. However, this effect of risk perception was not found in intenders (Lippke et al., 2005). In an eperimental trial with orthopedic patients, a volitional intervention proved to be matched only to volitional study participants. Accordingly, planning was not beneficial for persons in the motivational stages (Lippke et al., 2004). This was replicated in an Internet study that also demonstrated the specific mechanisms: The planning intervention facilitated behavior change by means of intention and planning, but only in the case of intenders, not in preintenders (Lippke, Schwarzer et al., 2010). Thus, preintenders first need to develop an intention. Accomplishing this depends mainly on a motivational intervention that targets goal setting. Principle 2: Two volitional phases, three stages. The study outlined above with orthopedic participants (Lippke et al., 2005) investigated the third HAPA stage in addition. This action stage was found to be significantly different from the other ones. When eploring the effects of a planning intervention on actors, they seemed to benefit, but not as much as those volitional participants who had not been active before (i.e., intenders). Thus, it is imperative to distinguish the intention stage from the action stage. Intenders benefit mainly from planning when adopting a new behavior. On the contrary, actors rather need relapse prevention Table 1 Stage Assessment Please think about the month before the rehabilitation started. Did you engage in physical activity at least 3 days per week for 40 min or more? Please choose the statement that describes you best. No, and I do not intend to start No, but I am considering it No, but I seriously intend to start Yes, but only for a brief period of time Yes, and for a long period of time Note. This stage assessment can be adapted to any other goal behavior.

7 SPECIAL SECTION: HEALTH BEHAVIOR CHANGE 167 that helps them to recover from setbacks, and, in case of a relapse, they need strategies for coping with them effectively. Principle 3: Postintentional processes. As described above, planning was mainly beneficial in the postintentional phase, that is, for the volitional individuals (Lippke, Schwarzer et al., 2010). As such, planning serves as an operative mediator between intentions and behavior. This was also shown in structural equation analyses (e.g., Schwarzer et al., 2008, 2007). Empirical tests on whether planning mediates the relation between intention and behavior have confirmed this assumption in the general population (Scholz, Schüz, Ziegelmann, Lippke, & Schwarzer, 2008) as well as in rehabilitation (Reuter et al., 2009). That planning mediates between intention and behavior was also demonstrated in orthopedic outpatients (Lippke et al., 2005). Principle 4: Action planning and coping planning. Comple planning interventions promote sustained behavior change after rehabilitation. Planning can be subdivided into action planning and coping planning, as has been evaluated in different samples of persons with chronic illness and disability. Sniehotta, Schwarzer et al. (2005) eamined the factorial structure of the two planning aspects. When addressing action planning and coping planning separately in interventions, different effects were found: Those persons in cardiac rehabilitation became more active when both kinds of planning were addressed in the intervention, as opposed to a mere action planning intervention. In another research sample, at 6 months after cardiac rehabilitation, participants involved in a planning intervention consumed 13% less saturated fat than controls (Luszczynska, Scholz, & Sutton, 2007). In orthopedic rehabilitation, a delayed effect of coping planning on behavior demonstrated that coping planning seems to be of higher importance for long-term maintenance: Whereas behavior adoption 2 weeks after rehabilitation was only correlated with action planning, coping planning came into play only 4 and 26 weeks later (Ziegelmann et al., 2006). In another study, action planning transpired to be the mediator between intention and coping planning, and coping planning mediated between action planning and behavior. This was true for different age groups in rehabilitation participants with orthopedic diagnoses (Ziegelmann & Lippke, 2007). The discriminant validity between action and coping planning has been confirmed in an eperiment by Wiedemann, Lippke, Reuter, and Schüz (2011). Principle 5: Phase-specific self-efficacy. Various studies proved that self-efficacy is required throughout the entire process of behavior adoption (Luszczynska, Gregajtys, & Abraham, 2007) and maintenance (e.g., Lippke et al., 2005). There is also evidence from rehabilitation samples on the changing nature of self-efficacy as people pass through different stages of change (Schwarzer et al., 2008; Ziegelmann & Lippke, 2007). In cardiac rehabilitation, three phase-specific kinds of selfefficacy were distinguished. They demonstrated discriminant validity and were analyzed regarding their predictions of intentions and behavior. Those individuals in the maintenance phase (actors who remained actors) were more likely to perform physical activities if they reported more maintenance self-efficacy than all others. Study participants resuming their physical eercise after a health-related break were more successful if they had higher recovery self-efficacy in comparison to those who were active without a break (Scholz, Sniehotta, & Schwarzer, 2005). Individuals in cardiac rehabilitation who have strong coping self-efficacy are likely to maintain the recommended rehabilitation eercise at follow-up, whereas this type of self-efficacy does not help people who have relapsed. Those who had a setback in adhering to recommended eercise, but harbor strong recovery self-efficacy beliefs, were more likely to regain control after a relapse (Luszczynska & Sutton, 2006). For Whom Does the Model Work? Moderator Effects The previous discussion about mediation addressed the question: How does it work? To further understand the mechanisms of health behavior change, we also have to ask: For whom does it work? Thus, we need to identify mediator effects as well as moderator effects. The HAPA, for eample, as a parsimonious mediator model, includes stage as a moderator. That indicates that a prediction model within one stage group operates in a different way than a prediction model within an adjacent stage group. This is similar to the assumption that one set of social cognitive variables can move people from Stage a to b, whereas a different set of variables can move people from Stage b to c. Although action planning has been found to mediate the intention-behavior relation (Gollwitzer & Sheeran, 2006), some studies have failed to find such mediation effects (Norman & Conner, 2005). This suggests that the relationships between intentions, planning, and behavior might also depend on other factors. For eample, the degree to which planning mediates between intentions and behavior has been found to be higher in older than in younger people (Renner, Spivak, Kwon, & Schwarzer, 2007; Scholz, Sniehotta, Burkert, & Schwarzer, 2007). This represents a case of moderated mediation. Perceived self-efficacy is one potential moderator for the degree to which planning has an effect on subsequent behaviors. It is epected to moderate the planningbehavior relation because people harboring self-doubts might fail to act upon their plans. For persons with a high level of selfefficacy, planning might be more likely to facilitate goal achievement. Self-efficacious people feel more confident about translating their plans into actual behavior. In other words, whether planning interventions (independent variable) actually affect behavior (dependent variable) might depend on the individual s level of selfefficacy (moderator). In a study on physical activity, longitudinal data from an online survey were used to eamine similar interrelationships (Lippke, Wiedemann, Ziegelmann, Reuter, & Schwarzer, 2009). Only those persons who had a sufficiently high level of eercise self-efficacy acted upon their plans. Conversely, participants who were harboring self-doubts failed to act upon their plans (see also Gutiérrez-Doña, Lippke, Renner, Kwon, & Schwarzer, 2009; Richert et al., 2010). Mediator models work well in some groups, but not in others. By comparing men and women, younger and older people, and those from different cultures, we identify relevant moderators (Renner et al., 2007; Ziegelmann et al., 2006). When a mediator model has strong interrelations within one category of participants, but weak associations within a different category, then this is a case of moderated mediation. The amount to which the mediator translates the effect of the independent variable on the dependent variable depends on the levels of a moderator variable. Such moderators can be age, chronic illness, disability, and so forth, but also psychological variables that are closely related to the constructs used in health behavior models. Temporal stability

8 168 SCHWARZER, LIPPKE, AND LUSZCZYNSKA of intention, for eample, may be a moderator (Conner, 2008). Older age and stronger baseline self-efficacy levels may moderate the effects of self-efficacy interventions among individuals with degenerative spine diseases enrolled in a rehabilitation program (Luszczynska, Gregajtys, & Abraham, 2007). Moderated mediation is also possible, with psychosocial variables, such as intention or planning. For eample, the intention behavior link is mediated by planning, and this mediator effect can be moderated by level of intention (Wiedemann, Schüz, Sniehotta, Scholz, & Schwarzer, 2009). Here we have a special case in which the independent variable (intention) of a mediator model serves the function of a moderator in addition. In other words, only in highly motivated persons does the intention operate via planning on the improvement of adherence, whereas in poorly motivated persons no such mediator effect is visible. The best way to demonstrate the mechanisms of health behavior change is the eperimental manipulation of those variables that are supposed to produce behaviors or to move people from one stage to another (Michie, Rothman, & Sheeran, 2007; Reuter, Ziegelmann, Wiedemann, & Lippke, 2008). Various eperimental studies have shown that self-efficacy interventions do make a difference, which attests to the fact that self-efficacy is indeed an operative construct that facilitates volitional processes, such as effort and persistence (Luszczynska, Tryburcy, & Schwarzer, 2007). Self-efficacy beliefs, in turn, are mobilized by family support during rehabilitation (Luszczynska & Cieslak, 2009). Outlook: Better Tailoring of Interventions for Persons With Chronic Illness or Disability Table 2 Intervention Matri for HAPA-Based Stage-Specific Treatments Stage group Preintender Intender Actor Motivational constructs Self-efficacy (motivational) Risk perception Outcome epectancies Goal setting Volitional constructs Action planning Coping planning Social support Self-efficacy (maintenance) Self-efficacy (recovery) Action control Based on their systematic review, McLean et al. (2010) conclude that interventions may be most effective when they (1) have motivational-cognitive behavioral approaches (as theorized by the HAPA, improving motivational factors), (2) help patients to manage barriers (as theorized by the HAPA, improving volitional factors as coping planning and maintenance self-efficacy), and (3) tackle issues related to health care providers and organizations. So far, the principles of the HAPA have proven to be convincing for the general population as well as for persons participating in orthopedic or cardiac rehabilitation. However, we know much less about the main differences between persons with various chronic illnesses and disabilities or their special needs compared to the regular population. In studies investigating theories other than the SCT, TPB, PMT, and TTM, only few differences between various chronic conditions and between individuals with and without diabetes were found (e.g., Plotnikoff et al., 2008, 2007). However, is this also true for the HAPA? This warrants further study, as only more empirical findings on such similarities and differences can help to tailor interventions effectively. Evidence from previous intervention studies shows results for people with various chronic illnesses and disabilities. Especially planning interventions and action control programs proved to be successful (Lippke et al., 2004; Luszczynska, 2006; Sniehotta et al., 2006; Sniehotta, Schwarzer et al., 2005; Ziegelmann et al., 2006). These interventions are mainly matched to intenders and actors adopting a new behavior or a behavior change. However, also preintenders and maintainers need effective interventions. From noninstitutionalized samples, there is evidence that interventions specifically targeting intenders or maintainers are promising (Reuter et al., 2008). Such programs need to be further developed and tested in people with different chronic illnesses and disabilities. The key characteristic of theory-based interventions such as one based on HAPA is to first assess the needs of the recipients, for eample, their stage of change. Then, the stage-specific variables are to be addressed. In an intervention with cardiac and orthopedic participants, the following contents were included (see Table 2). The stage-specific evaluation of effects is imperative: If different stage-specific interventions are applied to different stages, it is essential not only to search for changes in behavior or progression from inactive to active stages, but also to look for stage transitions between all stages. Alternatively, increases in the stage-specific variables (such as goal setting or motivational self-efficacy in preintenders) are relevant outcomes as well. People with chronic illness and disability have a higher likelihood of ehibiting multiple behavior risk factors, and they are at risk for premature death compared to those with only one behavioral risk factor or none. Furthermore, they also account for a disproportionate percentage of health care costs (J. J. Prochaska, Spring, & Nigg, 2008). Thus, it is important to understand not only single behaviors (such as physical activity or smoking), but also multiple health behavior mechanisms. Only if more is known about multiple behavior mechanisms can we help people to adopt and maintain as many recommended health behaviors as possible and, eventually, to stay healthy (Flay & Petraitis, 1994; Morabia & Costanza, 2010). Furthermore, we need a more elaborated modeling of multiple health behaviors to design effective treatments for people with chronic illness or disability. References Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, Bandura, A. (1986). Social foundations of thought and action. Englewood Cliffs, NJ: Prentice Hall. Bandura, A. (2004). Health promotion by social cognitive means. Health Education & Behavior, 31, Conner, M. (2008). Initiation and maintenance of health behaviors. Applied Psychology, 57, Courneya, K. S. (1994). Predicting repeated behavior from intention: The