Editorial Virtual Special Issue: Health Behavior Change
Changing health behaviour is difficult but do-able. Over the years, Social Science & Medicine has published a number of articles describing attempts to change behaviours. Many attempts have shown little or no success, while a few have been more effective. This Virtual Special Issue highlights articles published since 2003 from a variety of contexts. The articles were selected not because of their success at changing behaviour but because they illustrate the need for a well-considered strategy based on theory. Two review articles are also included and they point out that many programme designers and researchers do not articulate their theory of behaviour change. Instead, the education default is used – teach people with the help of pictures, stories and songs – with the assumption that new knowledge leads to change (e.g. Holford, 1995; Jewkes, Nduna, Levin, Jama, Dunkle, Khuzwayo, Koss, Puren, Wood & Duvvury, 2006). One look in the mirror tells us that knowledge is insufficient. Together, the papers in this virtual special issue on Health Behaviour Change highlight the need to connect interventions to theory. My analysis of the included articles provides a rough grouping based on theoretical principles in order to articulate the change strategy, change agents, and targeted health behaviour.
Important issues arise from this body of literature. First, of concern to social scientists working in developing countries, is whether the new behaviours will be compatible with the culture and adapted to the setting. Some believe that certain practices are common to a setting because they are functional and adaptive; while others believe that cultures are flexible and will not be destroyed by the introduction of new practices. Most agree that change agents should come from the local setting as they fulfill the criteria for effective social supports. A second issue concerns whether change agents, such as non-governmental organizations, should be encouraged to evaluate their programs. Much of the research is based on efficacy trials conducted in highly controlled settings. Fewer effectiveness studies are published, such as evaluations of programs delivered by peer educators or health workers in a community setting. A third issue concerns whether change strategies are to be derived empirically – such as the positive deviance model where successful models in the community are identified and emulated (Lapping, Marsh, Rosenbaum, Swedberg, Sternin, Sternin, & Schroeder, 2002) -- or from theory. Theoretical foundations of change are thankfully in the ascendancy as more researchers become aware of books such as the one edited by Karen Glanz, Barbara Rimer, and Frances Lewis (2002) Health behavior and health education: theory, research and practice. By using a theoretically-derived change strategy, programme designersand researchers allow for generalization of their findings and contribute to a body of evidence on successful strategies. They will also eventually contribute to essential information on the scope of each theory, such as to whom it best applies, and to a comparison of the theories.
Some theories focus on strategies for changing health behaviours, such as communication theory and learning theory. I have noticed that these are commonly used by programme designers in developing countries, but in a token and old-fashioned manner only. They refer adult education to IEC (information, education and communication) and BCC (behaviour change communication). Few are aware of the limitations of these theories and their new, more successful, revisions. One major limitation of the earlier versions of communication and learning theories is that they neglect the power of old habits that still linger alongside the new ones. Old habits may be so over-learned that they are automatically triggered by existing cues. The communication, learning and social influence processes by which people change are still intensively studied by psychologists and educators under controlled conditions. Their findings, described shortly, are beginning to be incorporated and tested in the field.
Other theories focus on what component of behaviour or its determinants should be targeted – the behaviour itself, the surrounding supportive social and physical environment, cognitive precursors, motivators, or values and attitudes. These theories focus on determinants of behaviour rather than change strategies. Arguments continue to rage as to the best component to change. The old KAP (i.e knowledge, attitudes and practice) studies demonstrated that knowledge and attitudes are fairly easy to change, but they do not automatically translate into behaviours (also known as the KAP gap). Dissonance theory turned the KAP strategy on its head by advocating first a change in public behaviour which would then, through a need for consistency, lead to a change in attitudes and values. Motivators are important in order to sustain the behaviour change, but research shows that internal and external motivators vary with cultures (Iyengar & Lepper, 1999). Finally, social support in the form of networks and norms can also help sustain behaviour change.
It is generally recognized that no one theory is currently sufficient to explain health behaviour change (Glanz, Rimer, & Lewis, 2002). For this reason, most interventions package several determinants and multiple change strategies. It is less important to test specific theories as they stand than to identify components that are effective. In other words, it is less useful to conclude that one theory is better than other. It is more useful to be able to conclude that specific components of theories X, Y and Z are more effective than other components of these theories. Then, in the end, we hope to have a set of components that are effective under certain conditions and to know why they are effective. In order to contribute to a body of evidence for behaviour change, reports evaluating interventions need to provide specific information. In addition to the CONSORT guidelines (Consolidating Standards of Reporting Trials) outlining essential information to be included when reporting such evaluations, additional behavioural science guidelines outline needed details concerning interventions (Boutron, Moher, Altman, Schulz, & Rivaud, 2008; Davidson, Goldstein, Kaplan, Kaufmann, Knatterud, Orleans et al., 2003; Zwarenstein, Treweek, Gagnier, Altman, Tunis, Haynes, Oxman, & Moher, 2008). Taxonomies of behaviour change techniques (e.g., Abraham & Michie, 2008) attempt to delineate specific strategies and their corresponding theory. As mentioned previously, most interventions combine theories about what is to be changed (e.g., norms) with theories on the strategy or process of change (e.g., communication and diffusion networks), not to mention organizational models of how to combine multi-level interventions (e.g., ecological models). Also, several theories may identify the same construct, such as self-efficacy (social cognitive and planned behaviour) and cues for action (health belief and social learning).
This Virtual Special Issue intends to draw attention to behaviour change interventions with the goal of identifying successful components in specific contexts and the theories they address. In many of the papers that are included, the researchers did not explicitly draw on theories to develop their intervention; when that is the case, I will attempt to identify the theory from their description. In the near future, Social Science & Medicine will be soliciting manuscripts that make explicit their intervention (see Abraham & Michie, 2008; Davidson et al., 2003), the theories underlying its components (e.g., Glanz et al., 2002), and the evaluation (CONSORT) for a regular special issue of the journal.
Several articles included in this virtual special issue reviewed multiple interventions and drew conclusions about what works. Gallant and Maticka-Tyndale (2004) critically analyzed 11 HIV prevention programs delivered to youth in sub-Saharan Africa. Although most articles did not explicitly identify the guiding theory, the authors inferred that most used cognitive theories of behaviour to identify determinants that should be targeted, such as knowledge and attitudes, along with learning theory as a basis for their strategies (e.g., drama, songs and stories, peer role models). In addition, most articles acknowledged the influence of community norms by more or less connecting the programs to people and organizations in the community. Gallant and Maticka-Tyndale conclude that these strategies were successful in changing knowledge and attitudes but not behaviour. In their review Van Empelen, Kok, van Kesteren, van den Borne, Bos and Schaalma (2003) likewise concluded that interventions to prevent HIV among drug users were more effective when delivered at the group and/or community level. These interventions relied on more than one explicitly-articulated theory of determinants and change strategy (role models, social support and network norms, skill building, diffusion). Interventions aimed at the individual may have been ineffective because they were brief attempts to transfer solely knowledge. The article by Cairncross, Shordt, Zacharia, & Govindan, (2005) is not a review, but rather a 9-year follow-up attempt to find out which of many individual and community interventions worked to increase hand-washing and latrine use in south India. Hand-washing after latrine use was higher in females than males, and higher if soap and water were kept beside the latrine (prominently placed cues to action). It was also higher in females who recalled the health education sessions delivered to small groups by community preschool teachers. Hygiene and sanitation behaviours were less strongly related to recall of a variety of other intervention inputs such as individual home visits, video presentations and street dramas that were part of community mobilization activities. These attempts to extract successful interventions provide useful comparisons of different inputs with surprising consistency across different contexts.
Communication and Diffusion
Communication is commonly used as a change strategy. Several articles selected for this virtual special issue use communication in the form of media, often as a social marketing strategy to scale-up use of a product or behaviour. The article by Dixon, Scully, Wakefield, White and Crawford (2007) used an experimental analogue of television viewing to examine children's attitude change toward nutritious and junk foods in Australia. Although the correlational study supported the cultivation theory of media communication, namely that junk food advertising creates and normalizes preferences for junk food, the brief intervention interspersing food ads in a regular Simpsons episode was less clear in demonstrating a causal effect. Still, it is imperative that such studies use communication theory to study the impact of advertising on diet (e.g. Finnegan & Viswanath, 2002). Because children are already exposed to ambient messages about junk food and because they are not strong cognitive processors of information, persuasive communication follows the more emotional route. Likewise, social marketing in Tanzania was used to cultivate a market for mosquito nets (Kikumbih, Hanson, Mills, Mponda, and Schellenberg, 2005). Social marketing interventions target a consumer audience after a great deal of formative research on local concerns. According to the Cognitive Elaboration Likelihood model and the two routes to persuasion (Petty & Cacioppo, 1986), the message and messenger need to be altered to fit the targeted audience. People who process information cognitively will be persuaded by sound arguments directed at their concerns. Those who process information superficially will not be persuaded by sound arguments; they resonate to the emotional appeal of the message, the cultural appeal of the medium, and the popular appeal of the sender. Songs and posters advocating bednet repairs used these principles for an audience that was not literate in rural Gambia (Panter-Brick, Clarke, Lomas, Pinder & Lindsay, 2006). Entertainment-education drama series probably need components of both routes to persuasion to cover a wider audience.
Interpersonal communication is also used to diffuse a new product or behaviour (e.g., Oldenburg & Parcel, 2002). Here too, the messenger and the message need to be selected according to the receiver. Diffusion theory identifies receiver categories in terms of how quickly they adopt the new behaviour and suggests that cognitive strategies may work for early, motivational for majority, and problem solving for late adopters. Diffusion theory also considers the nature of the innovator or message sender but they tend to be simply people in authority or opinion leaders. Moser and Mosler (2008) evaluated the process by which new solar technology was adopted by Bolivian villagers to disinfect their drinking water. Opinion leaders influenced both early and middle adopters of the new technology. However, early adopters were more likely to attend promotional activities and fairs, whereas middle adopters were persuaded by home visits of local leaders.
Two other articles studied a diffusion of sorts from Kenyan school children to families and to younger students (Onyango-Ouma, Aagaard-Hansen, and Jensen, 2005) and from Senegalese grandmothers to younger women (Aubel, Toure, & Diagne, 2004). Although children are expected to soak up new information, they have a narrow sphere of influence. Thus the Kenyan students of 9 to 15 years were readily able to diffuse hygiene and sanitation knowledge and practices to younger peers, but less so to adults in their families. In contrast, elders are expected to be opinion leaders of tradition but to resist innovation, i.e., to maintain norms rather than change norms. In the Senegal study, nutrition education concerning pregnancy and breastfeeding was disseminated by grandmothers in the form of stories, songs and problem-solving discussion. One year later, grandmothers had increased their knowledge; mothers in intervention villages evidenced much better health behaviours than those who heard the same messages without grandmother follow-up. These studies on interpersonal communication and diffusion demonstrate the strengths and limitations of strategies to change social norms and thus sustain health behaviour changes.
Social network norms and role models
Two additional articles included in the virtual special issue aimed explicitly to change behaviour through social network norms. Sherman, Sutcliffe, Srirojn, Latkin, and Aramratanna (2009) included communication training along with safe sex and harm reduction to drug users in Chiang-Mai, Thailand, with the intention of having them diffuse the information to their social network members. Thus, the targeted drug users were to function as peer educators, being both communicators and role models. Health behaviours improved in the intervention group, but equal to the comparison group which received cognitive-behaviour skills training. As the authors suggest, there was likely contamination of the diffusion strategy as Chiang Mai youth had networks in both conditions. Mahler, Kulik, Butler, Gerrard, and Gibbons (2008) added norm information on sun protection to a successful video intervention to determine the extra impact of descriptive norms (what peers do) and prescriptive norms (what peers think should be done). They found that Californian youth were more likely to change their behaviour after hearing the combined norm information, and more likely to discuss it with a friend. The idea of changing actual norms, and changing behaviour through social network norms, is gaining ground as a means to sustainability. However, most social scientists would admit that changing social norms is an ambitious project; the first step is to change the behaviour and values of peers and have them pass it on to their friends. Friends tend to socialize each other. This norm-setting sequence also demonstrates the impact of observational learning from role models. Consequently, diffusers of innovative behaviours serve multiple functions in theories of change: they communicate information to change knowledge and attitudes, they give approval concerning new norms, they provide social support during transitions, and they model new behaviours. Peer educators may be particularly successful because they fulfill these functions most effectively.
Social Learning and Social Cognitive Change
Learning theory is frequently espoused by behaviour change programs whenever models are provided. However, there is much more to learning theory that could easily be incorporated. One aspect is providing self-reinforcement, rehearsal of the behaviour, and well-placed cues to remind the person to enact the behaviour. Social learning theory became more cognitive when researchers realized that people were not passive imitators of role models or passive responders to situational cues. They selectively attend to certain models and often arrange their own cues to remind themselves to act. Moreover, as a result of rehearsal, reinforcement and constructive feedback, people acquire a skill and gain confidence in their ability to overcome obstacles to perform the behaviour (Baranowski, Perry, & Parcel, 2002). Social learning and cognitive theory do not specify what will change (except for self-efficacy) as much as identify strategies for change (e.g., observational learning, mastery learning).
One learning-based intervention included in this virtual special issue was conducted with 12- to 13-year-old low-income British adolescents in an effort to improve nutrition (Hyland, Stacy, Adamson, & Moynihan, 2006). In after-school Food Clubs, students learned the skills for making inexpensive nutritious meals. They took the meals home to their families as a way of diffusing the healthy food message. Qualitative methods used to evaluate the program showed that skills and confidence improved but diets changed only if parents made the effort to buy foods and let their children use the kitchen. Another learning-based intervention (Jason, Pokorny, & Adams, 2008) examined the addition of fines for under-age purchase, possession or use of tobacco in the United States. Fines are a clear punishment and aversive for both financial and social reasons. Here they were both a disincentive (knowing ahead that you might get fined if caught with cigarettes) and a punishment (losing money when you pay the fine). The authors found that over 4 years, the threat of fines increased the rate of smoking abstinence among US high school students above the standard practice of controlling sales through stores.
The Information-Motivation-Behaviour skills model also incorporates cognitive and skill acquisition for behaviour change. It was used to prevent HIV among truck drivers in south India by increasing knowledge about transmission and condom-use skills (Cornman, Schmiege, Bryan, Benziger, & Fisher, 2007). The motivation construct is a mixture of attitudes and intentions, which are largely cognitive rather than intrinsic or extrinsic motivation. However, skills were rehearsed and found to influence the use of condoms with wives but not non-marital partners 10 months later. The comparison group here was an information-only intervention. Similarly, the Theory of Planned Behaviour (e.g. Montano & Kasprzyk, 2002) outlines determinants of behaviour that should be changed, such as behavioural beliefs, subjective norms and perceived behavioural control. This theory has been applied by Hill, Abraham, & Wright (2007) in an article in this virtual special issue to increase exercise in high school students in Britain The Theory of Planned Behaviour does not specify a strategy for change, so attempts to change cognitive determinants are normally brief written or verbal communications. In the study by Hill, Abraham and Wright (2007) people were trained to prompt cognitively their behaviour by specifying when and where they would act or were motivated to recall the information through quizzes. Self-reported exercise was higher in all information groups compared to a no-information control group. Although most studies using this theory include the original four cognitions (namely beliefs, norms, behaviour control and intentions), Fishbein's revision of the theory which adds past habits, skill acquisition and environmental constraints is rarely tested (Fishbein & Yzer, 2003). The revision comes closer to including variables found to be effective in many contexts.
The articles included in this Virtual Special Issue on Health Behaviour Change are instructive in several ways. They inform us about the importance of theories to guide the development of interventions – theories about the determinants of behaviour and theories about strategies for behaviour change. They demonstrate that often theories are used in combination because it is more important to identify effective components of theories than to test the value of a single theory. Furthermore, they show how individual studies can be integrated to derive conclusions generalizable to many contexts. Finding comparable results in different contexts using different measures of the same constructs strengthens our conclusions. The articles also remind us that theoretically driven interventions can also be "culturally compelling" (Panter-Brick et al., 2006). While the message may be about new behaviours, the medium and format of the message should fit the audience (according to Communication theories), cues to action need to be properly placed (according to Social Learning theory), and the change agent needs to have local respect (according to Diffusion of Innovation theory). To be culturally compelling may also mean that we need to work with norm setters, helping to move influential community organizations through the stages of change (McCoy, Malow, Edwards, Thurland, & Rosenberg, 2007).