The effectiveness of interventions to change six health behaviours: a review of reviews

Chronic diseases, such as cancers, cardiovascular diseases (CVD), diabetes, and respiratory diseases, account for 59% of the 57 million deaths annually and 46% of the global burden of disease [1]. In 2002, the World Health Report [2] identified a number of important lifestyle risk factors for such diseases, including physical inactivity; diet-related factors and obesity; and the use of addictive substances such as tobacco, alcohol and illicit drugs.

These lifestyle factors have significant effects on mortality and morbidity, particularly in industrialised countries. For example, data for the WHO European Region show that physical inactivity is a risk factor for diseases such as cardiovascular diseases, non-insulin- dependent diabetes, hypertension, some forms of cancer, musculoskeletal diseases and psychological disorders. These diseases are estimated to account for nearly 600,000 deaths per year [3]. Similarly, obesity and being overweight are risk factors for diseases such as type 2 diabetes, certain types of cancer and cardiovascular diseases, and affect between 30% and 80% of adults and up to one third of children [4]. Alcohol is also a significant cause of mortality: alcohol-related deaths increased by 15% between 2000 and 2002, and now represent 6.3% of all deaths in the European Region [5].

Sexual risk taking and drug misuse also significantly contribute to ill health and have negative effects on well being among young people. In 28 high income (OECD) nations at least 1.25 million teenagers become pregnant each year; of these, approximately 40% (half a million) will seek to terminate the pregnancy while the other 60% (three quarters of a million) will become teenage mothers [6]. The United States has the highest teenage birth rate in the developed world and the United Kingdom has the highest teenage birth rate in Europe [6]. Worldwide, young people (15-24 years) have the highest rate of sexually transmitted infections (STIs) of any age group. Up to 60% of the new infections and 50% of all people living with HIV globally are in this age group [7].

Most public health and health promotion interventions - whether they focus on the individual, community, whole populations or the environment - seek in some way to change health behaviour by changing health-related knowledge, attitudes and/or structural barriers and facilitators [8]. Social psychological theories such as social cognition theory are commonly used in the development of interventions [9]. Key elements of such theories include knowledge of health risks, perceived self efficacy, goals and motivations and barriers and facilitators [10]. Most health promotion interventions include one or more of the following components: education and knowledge building (around the health issue); motivation and goal setting (e.g. alcohol brief interventions and counselling); and community-based techniques to encourage a change in behaviour or reduce structural or cultural barriers. These interventions can be delivered at three different levels, which we explore below: individual, community and population level interventions. Individually targeted interventions are usually aimed at those with an existing 'risky' behaviour such as smoking or alcohol misuse. Community level interventions focus on particular population groups such as people in a particular workplace or young people in schools. Finally, population level interventions tend to rely on the use of mass media activities, policies or legislation.

All three levels of intervention are aimed at achieving changes in lifestyle, as well as improving knowledge and influencing attitudes towards positive healthy behaviours. However, there is a need to take into account the socio-economic and cultural contexts within which they are located. For instance, an intervention to promote healthy eating within an affluent locality might involve a rather different approach from one undertaken in an area of low income and high unemployment.

In light of the growing concern around the link between 'negative' health behaviours and ill health, we were commissioned by the Public Health section of the UK's National Institute of Health and Clinical Excellence (NICE) under the Behaviour Change Programme Development Group to review the relevant evidence in this area. This paper is an update of the findings of this 'review of reviews' of interventions to change health behaviours [11]. It was one type of evidence used to develop NICE public health programme guidance on behaviour change [12].

Our aim was systematically to collate, evaluate and synthesise review-level findings on the effectiveness of interventions to change unhealthy behaviours or promote healthy behaviours. This synthesis was intended to provide researchers, policy planners, decision-makers and practitioners with an accessible, good quality overview of the evidence in these topic areas. The review focused on six groups of behaviour change interventions:

A subsidiary aim of the review was explore, where possible, the evidence of impact of interventions on health inequalities.

We identified 103 systematic reviews evaluating interventions aimed at changing health behaviour in one or more of the six areas. Some of these reviews covered several behaviours. The reviews included studies which targeted specific individuals or organisations (e.g. through counselling within education) or more generally (e.g. mass media interventions or legislation).

We synthesised the results under three research areas:

Full consideration of such a large number of reviews would prove too lengthy for this paper, therefore we have chosen to highlight the main findings rather than provide details of individual interventions discussed within each paper. A fuller version of the original document is available from the authors on request. Table 5 summarises the quality of the included reviews and Table 6 provides a brief overview of the studies, grouped by level of intervention (population, community or individual).

A limitation of a review of reviews approach for knowledge synthesis might also be regarded as one of its strengths. Given the heterogeneity of interventions included within each review and the large number of papers across the six health behaviour areas, it has not been possible to report any particular area in great depth. While some readers might be frustrated by the lack of detail, one of the strengths of this review is its capacity to offer a broad overview of the evidence. We are able to bring to the attention of other researchers, policy makers and commissioners of research the areas where no systematic reviews have been conducted, and suggest potential research questions worthy of investigation at the review level. It was outside the scope of a review of reviews to determine whether this corresponded to gaps in primary data, or whether this reflected the research priorities/questions of systematic review authors.

Given the importance of socio-economic determinants of health and illness, and their impact on morbidity (as well as mortality), it is crucial that interventions designed to improve health take account of those very factors that may work against positive outcomes. As indicated above, this is one of the most significant gaps in the evidence base revealed by our study.

Until the mid 20^th century, infectious diseases were one of the main causes of mortality and morbidity in industrialised nations. Once these were largely brought under control, chronic diseases such as cancer and coronary heart disease assumed prominence. The 1970 s saw the emergence of a 'new' public health, highlighting the social and cultural factors which affect chronic diseases (e.g. lifestyle), and of new disciplines such as health promotion. Today the focus has shifted in line with the epidemiological transition, so there is increasing emphasis on social determinants of health rather than on infectious agents.

A series of reports on health inequalities been published in the UK (for instance, The Black Report [127], the Acheson Report [128] and the Wanless Report [129]). While the Black Report of 1980 focused primarily on material deprivation as a major explanatory factor in health inequalities, more recent research has explored the interplay between social position and the lifecourse in relation to the determinants of health [129‐133, 133]. The more recent change of focus from material deprivation towards a more multi-faceted understanding of determinants of health is important for interventions aimed at effecting changes in behaviour that may have an adverse affect on health. However, what is not clear is whether current health promotion initiatives have absorbed and embedded the 'determinants' approach in order to improve the health of those who might benefit most. It has been argued that the inverse care law is not always considered in relation to local provision of health promoting activities [134]. Indeed, it has been suggested that health promotion interventions frequently increase, rather than decrease, socioeconomic inequalities in health [128], since health promotion messages and interventions have a differential take-up across different social class groups. Those who are more affluent and have a higher level of formal education are more likely to modify their diets, give up smoking and increase levels of physical activity than are the less affluent with lower levels of formal education [135]. For instance, a mass media campaign to increase walking in Scotland ('Fitline') found that it had less appeal for those in the lower socioeconomic groups, despite higher awareness levels among these groups [136]. Furthermore, one-third of those who telephoned 'Fitline' to obtain further information were already regular exercisers, and may not have incurred any additional benefit.

In Tables 8 and 9 we have highlighted areas worthy of further synthesis. Despite the importance of socio-economic status as a determinant of health, few reviews explored socio-economic status or whether interventions had targeted those with the poorest health (e.g. those living in areas of disadvantage). Employment status, occupation, income level, gender, age, education, mobility and ethnicity are only some of the complex factors related to the determinants of health, and these do not include the more recent psychosocial [132] and lifecourse approaches [133]. Indeed, as Wilkinson argues, "The social consequences of people's differing circumstances in terms of stress, self-esteem and social relations may now be one of the most important influences on health" [137], pg 128]. We now understand how individuals become resilient to adverse circumstances (e.g. living in conditions of social deprivation) through, for instance, access to social support and strong social networks.

Studies that have explored the accumulation of risk of ill health over the lifecourse have highlighted the importance of social location and social capital. For instance, Schoon and Bynner [138], in a paper related to risk and resilience in young people, argue that interventions to improve health should aim to bolster incidences of positive adaptations to adverse circumstances rather than simply focusing on trying to put right negative behaviours. Linking into the health promotion agenda, they argue for the importance of primary prevention interventions, which take account of the context of the lives that an intervention seeks to improve, rather than simply taking a 'one size fits all' approach. However, there is little evidence at systematic review level that interventions to effect changes in behaviour are tackling the more complex interplay between health, illness and the wider determinants of health. As we noted above, this does not necessarily mean that the researchers in the primary studies themselves were unaware of a determinants approach. Nevertheless it is a significant finding that the systematic reviews covered here largely failed to take account of the inequalities agenda.

Research evidence clearly indicates that illness tends to cluster within lower socio-economic groups [139, 140]. This means that it is even more crucial that those conducting systematic reviews (as well as those designing interventions) make health inequalities a central concern. As a recent WHO report states: "Together, the structural determinants and conditions of daily life constitute the social determinants of health and are responsible for a major part of health inequities between and within countries" [130].

The Cochrane Collaboration has a Cochrane Health Equity Field which aims to encourage reviewers to 'include explicit descriptions of the effect of the interventions not only on the whole population but to describe their effect upon the disadvantaged and/or their ability to reduce socioeconomic inequalities in health and to promote their use to the wider community.' To aid reviewers in this task, they have developed an 'Equity Tool' which is available from their website http://​equity.​cochrane.​org/​en/​index.​html. We would urge reviewers to access this tool and include data on inequities in future publications.

Another notable finding is that there were many more reviews evaluating individual-level interventions than those at community or population level, which again may reflect practical/methodological concerns. What we were not able to ascertain was whether the lack of such reviews reflected a lack of primary population based intervention studies. However, as the success of mass media advertising campaigns in marketing consumer goods clearly demonstrates, population-level interventions certainly hold promise, despite not lending themselves to evaluation by traditional means, such as the randomised controlled trial.

Whatever the level or target of the intervention, one of the most significant challenges is to bring about a change in behaviour that is sustained over time. In the smoking cessation field this is dealt with through relapse prevention schemes. However, sustained change is a difficulty also faced by healthy eating and physical activity programmes, where the intervention may be highly effective for the duration of contact with participants but then become less effective and even lead to a reversal of positive change in behaviour over time.

Interventions aiming to achieve long-term, sustained behaviour change will require a different approach to evaluation. Currently, the majority of studies have a relatively short period of follow-up, with the longest usually no more than two years. In order to determine long-term effectiveness, there is a need to develop longitudinal studies that can run alongside the intervention and revisit the participants at several time points, charting the challenges to sustaining the healthy behaviours and learning from those who have successfully maintained new, healthful habits. This would enable policy makers and those delivering the interventions to gain a deeper understanding of the strengths and weaknesses of the intervention, with a view to improving the effectiveness of future interventions. Participatory and ethnographic approaches may be particularly suited to this form of evaluation.

Despite the limitations of a review of reviews, we have gathered together a wide body of evidence which illustrates that there are many interventions that have effectively achieved behaviour change across a range of health behaviours. The interventions that appear to be most successful include workplace interventions to support smoking cessation, physical activity and healthy eating; school based interventions across the health behaviours; individual-level interventions drawing on physician advice to promote healthy eating, smoking cessation and responsible/safe levels of alcohol use; and counselling for tobacco and alcohol use. Pregnancy may be a point in the lifecourse when women are especially amenable to making health improving changes (particularly in smoking cessation, and physical activity).

Mass media interventions are relatively effective in addressing the interconnectedness of knowledge, attitudes and behaviour within health promoting social marketing. Similarly, population-level legislative interventions (such as smoking bans and age limits for alcohol use) tend to be a part of much wider, more comprehensive campaigns to secure public support for the underlying health promoting message. Clearly, interventions at the individual and community levels should recognise the importance of changing knowledge and attitudes along with promoting healthy behaviour. Although many of the interventions currently include educational components, few explicitly attempt to evaluate their success.

Systematic reviews should explore data from the primary studies showing how effectiveness varies in relation to social and economic difference (including wider determinants of health, such as gender, ethnicity and geographic location). Further research is needed to synthesise primary studies of the effectiveness of interventions across all six health behaviours to target those of lower socio-economic status who may be at highest risk of ill-health and who may engage in a range of unhealthy behaviours. For this reason, it is likely that health promoting interventions might benefit from a more multi-faceted approach: the effectiveness of the specific interventions reviewed here is likely to be enhanced through complementary supportive action. The inter-relationship between activities, such as alcohol use and smoking, may lend itself to a two-pronged intervention.

While there is a good body of high quality evidence of effectiveness of interventions to promote healthy behaviours, longer term follow-up is required in order to determine whether positive change is sustained and for how long.