Behaviour change: general approaches

The PDG was influenced by a number of different theories, concepts and accounts of behaviour and behaviour change, drawn from the social and behavioural sciences. These include: resilience, coping, self-efficacy, planned behaviour, structure and agency, 'habitus' and social capital. (Ajzen 1991, 2001; Antonovsky 1985, 1987; Bandura 1997; Bourdieu 1977, 1986; Conner and Sparks 2005; Giddens 1979, 1982, 1984; Lazarus 1976, 1985; Lazarus and Folkman 1984; Morgan and Swann 2004; Putnam 2000; for more details, see appendix A.)

The PDG discussed efforts to use policy and legislation to change behaviour (although relatively little formal evidence on legislation was identified). Such measures tend to work through a combination of awareness-raising, compulsion and enforcement, providing legislative or environmental 'structure' to the decisions people make about their behaviour. It was noted that legislation can appear to be a simple and powerful tool, and the evidence suggests that introducing legislation, in conjunction with other interventions, can be effective at the individual, community and population levels. However, it also suggests that it can be subject to contingencies and side effects, including criminalisation, compensating or displaced behaviour, and lack of public support (Gostin 2000; Haw et al. 2006; WHO 2005).

The PDG observed that people's health behaviours may change, depending on their social and material circumstances and their time of life. It was also noted that many other factors (such as place of birth, parental income, education and employment opportunities, or the impact of prejudice and discrimination) can have both direct and indirect effects on health, and on people's ability to change, leading to a cumulative effect over the life course (Graham and Power 2004; Kuh et al. 1997). The PDG considered the concept of the life course and evidence was sought on the potential benefits of intervening at key life stages or transition points. Explicit, formal evidence (at the level searched) was scarce.

The PDG further noted that the knowledge and evidence from different disciplines are very different in the concepts they use, the assumptions they make about cause and explanation, and (sometimes) the methods that they favour. Consequently, combining knowledge and evidence from different levels – such as the social and the individual – is extremely difficult. To ensure that as broad a range as possible of knowledge and evidence was taken into account, the PDG adopted a pluralistic approach that acknowledged the value of different forms of evidence and research methods.

The psychological literature is extensive and provides a number of general models of health behaviour and behaviour change. However, the research literature evaluating the relevance and use of these models is inconsistent. For example, it includes multiple adaptations of particular models, poor study designs and studies that fail to take account of all the confounding factors. Having considered some of the more commonly used models of health behaviour, the PDG concluded that the evidence did not support any particular model (although some have more evidence of effectiveness than others). For this reason, it believes training should focus on generic competencies and skills, rather than on specific models. These include the ability to:

• critically evaluate the evidence for different approaches to behaviour change

• design valid and reliable interventions and programmes, that take account of the social, environmental and economic context of behaviours

• Identify and use clear and appropriate outcome measures to assess changes in behaviour

• employ a range of behaviour change methods and approaches, according to the best available evidence

• regularly review the allocation of resources to interventions and programmes in light of current evidence.

For the purposes of this guidance, human behaviour is defined as: 'the product of individual or collective human actions, seen within and influenced by their structural, social and economic context'. These actions produce observable social, cultural and economic patterns which limit – or enable – what individuals can do. The recommendations in this guidance span the individual, social and group processes involved in human behaviour.

The PDG considered the psychological models showing relationships between knowledge, attitudes and behaviour, according to the various definitions outlined in the identified literature. The PDG noted that for some actions the links between intentions and behaviour can be described precisely. However, simple models do not capture more complex or population-level dynamics.

Although the evidence on psychological models was found to be limited, a number of concepts drawn from the psychological literature are helpful when planning work on behaviour change with individuals. When used in conjunction with recommendations here on planning and social context, these concepts could be used to structure and inform interventions. They include:

• outcome expectancies (helping people to develop accurate knowledge about the health consequences of their behaviours)

• personal relevance (emphasising the personal salience of health behaviours)

• positive attitude (promoting positive feelings towards the outcomes of behaviour change)

• self-efficacy (enhancing people's belief in their ability to change)

• descriptive norms (promoting the visibility of positive health behaviours in people's reference groups – that is, the groups they compare themselves to, or aspire to)

• subjective norms (enhancing social approval for positive health behaviours in significant others and reference groups)

• personal and moral norms (promoting personal and moral commitments to behaviour change)

• intention formation and concrete plans (helping people to form plans and goals for changing behaviours, over time and in specific contexts)

• behavioural contracts (asking people to share their plans and goals with others)

• relapse prevention (helping people develop skills to cope with difficult situations and conflicting goals).

Coordinated attempts to promote or support behaviour change can take a number of forms. These activities can also be delivered at a number of levels, ranging from local, one-to-one interactions with individuals to national campaigns. Many terms are used to describe these activities and sometimes these are used interchangeably (see glossary). Broadly, interventions can be divided into 4 main categories:

• policy – such as legislation, workplace policies or voluntary agreements with industry

• education or communication – such as one-to-one advice, group teaching or media campaigns

• technologies – such as the use of seat belts, breathalysers or child proof containers for toxic products

• resources – such as leisure centre entry, free condoms or free nicotine replacement therapy.

This guidance adopts the NICE definitions for public health interventions and programmes, unless another specific term has been used in the literature (such as 'campaign' to refer to a media initiative). See 'The public health guidance development process: an overview for stakeholders including public health practitioners, policy makers and the public' (details in appendix E).

Whether an intervention or programme is delivered to individuals, in community or family settings, or at population level, the effects are rarely restricted to 1 level. For example, a brief primary care intervention aimed at reducing alcohol consumption among individuals could have an impact:

• on the individual's behaviour (for example, level of alcohol consumption, individual health outcomes, or incidence of domestic violence)

• on the local community (for example, local alcohol sales, alcohol-related crime or accident and emergency [A&E] events)

• at population level (for example, national alcohol sales and consumption, national statistics on alcohol-related crime and A&E events, or demographic patterns of liver cirrhosis).

The PDG noted that it is important to specify 3 things with respect to any intervention that aims to change behaviour. First, be as specific as possible about its content. Second, spell out what is done, to whom, in what social and economic context, and in what way. Third, make it clear which underlying theories will help make explicit the key causal links between actions and outcomes (Davidson et al. 2003; Pawson 2006; Weiss 1995). The PDG noted that the evidence is often very weak in these respects.

It is important for those planning health improvement interventions to be clear about the behaviours that need to be changed, any relevant contextual changes that also need to be made, and the level at which the intervention will be delivered (individual, community or population). The following questions should be used as a guide:

• Whose health are you seeking to improve (target population/s)?

• What behaviour are you seeking to change (behavioural target)?

• What contextual factors need to be taken into account (what are the barriers to and opportunities for change and what are the strengths/potential of the people you are working with)?

• How will you know if you have succeeded in changing behaviour (what are your intended outcomes and outcome measures)?

• Which social factors may directly affect the behaviour, and can they be tackled?

• What assumptions have been made about the theoretical links between the intervention and outcome?

A range of resources provide access to good quality, up-to-date evidence on the effectiveness of interventions and programmes aimed at changing behaviour. These include: NICE public health guidance, research and review databases (for example, the Database of Abstracts of Reviews of Effectiveness, the Cochrane Library, Medline, and the Social Science Citation Index), and current texts on behaviour change (for example, Conner and Norman 2005). When drawing up plans to change people's behaviour, enough time needs to be set aside to consult these resources to establish which interventions and programmes will be most appropriate.

Time and resources should be set aside for evaluation. The size and nature of the intervention, its aims and objectives and the underlying theory of change used should determine the form of evaluation (see below).

Attempts to change behaviour have not always led to universal improvements in the population's health. For example, different groups (measured by age, socioeconomic position, ethnicity or gender) react differently to incentives and disincentives, or 'fear' messages. Effective interventions target specific groups and are tailored to meet their needs. This is particularly important where health equity is 1 of the goals. Service user views may be helpful when planning interventions.

The cultural acceptability and value of different forms of behaviour varies according to age, ethnicity, gender and socioeconomic position. It is important not to stereotype or stigmatise groups or individuals because of these variations. This can be avoided by working closely with communities over time, by tackling prejudice and discrimination in professional practice, and by using needs assessments to gather local and cultural information to ensure interventions are tailored appropriately.

Changing behaviour may not be a priority for the individuals being targeted. People do not necessarily make their own long-term health a priority and may want to focus on other, more immediate needs and goals (for example, relieving stress, or complying with peer pressure).

Some damaging and, therefore, apparently negative health behaviours may provide positive psychological, social or physical benefits for individuals in certain social and cultural contexts. For example, smoking cigarettes may provide 'time out' for people in difficult circumstances. Effective interventions take account of the social, cultural and economic acceptability of the intervention and the target group's attitudes toward the behaviour. They recognise diversity in the values people use to guide their lives and behaviour.

Interventions may have unintended and negative consequences. When planning an intervention, it is often helpful to conduct a prospective health and equity impact assessment.

No single method can be universally applied to influence all behaviour and all people. Universal interventions do not invariably have uniform effects, and may be more effective among some population groups, or in some settings, than others.

An intervention aimed at changing 1 behaviour may inadvertently lead to other changes. For example, someone who gives up smoking may start eating more food to compensate, leading to other health risks.

Motivated individuals actively seeking to make changes in their behaviour require a different approach from those who are unmotivated. The latter may need more information about the benefits of change, as well a realistic plan of action. Equally, different methods may be required at different times and to reach different people. This guidance identifies the broad principles.

Enabling individuals and communities to develop more control (or enhancing their perception of control) over their lives can act as a buffer against the effects of disadvantage, facilitating positive behaviour change.

A range of cognitive, social and environmental resources can help to boost the resilience of people living in difficult circumstances. These resources can help promote their health and protect them against illness and other negative outcomes. They include a positive attitude to health (leading to positive, health-related behaviours), coping skills and 'social capital', the relationships of trust and reciprocity built up through, for example, friendship, family and faith networks.

Action taken earlier, rather than later, in an individual's life can sometimes be more effective at preventing health-damaging behaviours. Consequently, interventions that focus on children and young people (and usually, their carers too) are important. However, interventions with other population groups can be highly effective and cost effective. An example is action to prevent falls among older people.

All interventions need to be developed and evaluated in stages, using an established approach such as the Medical Research Council's framework for the development and evaluation of complex interventions (Campbell et al. 2000; see also Campbell et al. 2007; Flay 1986; Nutbeam 1998). Such an approach will help ensure interventions are based on the best available evidence of feasibility, acceptability, safety, effectiveness, efficiency or equity.

As well as focusing on individual factors, it is important that policy makers and commissioners take steps to address the social, environmental, economic and legislative factors that affect people's ability to change their behaviour.

A large number of mechanisms could be used to influence behaviour but the amount of evidence varies. Generally, there is far more evidence on activities aimed at individuals than on policies and other activities aimed at tackling the wider determinants of health. The evidence on efficacy and equity is also variable. The PDG could not review all the possibilities, but noted that the following mechanisms were successful in some circumstances:

• legislation and taxation

• mass media campaigns

• social marketing

• community programmes

• point of sale promotions.

Population-level interventions have the greatest potential, however, if supported by government and implemented effectively. (Legislation making it compulsory to wear seatbelts in the front seats of cars is an example of a highly effective, population-level intervention.)

Epidemiological theory suggests that even small degrees of change, over time, can result in significant improvements in population-level health (Rose 1985). Population-level interventions could be an effective and cost-effective way of changing behaviour.

The PDG noted that a wide range of policies and the actions of a range of government and non-governmental organisations impact directly and indirectly on health. (Relevant policies and actions include those related to taxation, the licensing laws and the benefits system.) This could be explicitly acknowledged by carrying out routine health impact assessments on how a policy, law or system affects people's health-related behaviour. It could also be acknowledged through partnership and cross-government working.

The level of skills, knowledge and the competencies required by those providing health-related interventions will differ, according to their specific role. However, some are central to most public health activity. These include: knowledge of the full range of difference approaches to behaviour change, competence in planning and evaluation, understanding the principles of non-discriminatory practice; and the ability to use evidence from research and practice.

The PDG noted that the capacity of the public health workforce requires assessment. An education and training strategy to support the development needs of those involved in helping to change people's behaviour (within both NHS and non-NHS settings) could improve effectiveness. National training standards to reflect the skills and competencies described in the recommendations would support their implementation.

The distinction between monitoring and evaluation is important. Monitoring involves routinely collecting information on a day-to-day basis and using shared information resources and statistics to keep local and national health activity under surveillance. It is part of quality and safety assurance. Evaluation, on the other hand, is the formal assessment of the process and impact of a programme or intervention. Where an intervention is employed that has already been rigorously evaluated (for example, in NICE public health guidance) and demonstrated to be effective in equivalent conditions, then monitoring, rather than a full evaluation, is likely to be sufficient.

Complex public health interventions can be systematically evaluated, based on the relevant theory and evidence, if they use a well-planned, 'staged' approach to evaluation.

Formal outcome and process evaluation can be challenging, but it is an important way of assessing efforts to change behaviour. An effective evaluation is based on clearly defined outcome measures – at individual, community and population levels, as appropriate. Qualitative research looking at the experience, meaning and value of changes to individuals may also be appropriate. Methods and outcome measures are identified during the planning phase. In addition, effective interventions specify their 'programme theory' (or reason why particular actions are expected to have particular outcomes). They also use a framework of 'action – reason – outcome' to guide evaluation (Campbell et al. 2000; Campbell et al. 2007; Flay 1986; Nutbeam 1998; Pawson 2006; Weiss 1995).