545108
research-article2014
SJP0010.1177/1403494814545108Unintended effects in settingsMB. Mittelmark
Scandinavian Journal of Public Health, 2014; 42(Suppl 15): 17–24
Original Article
Unintended effects in settings-based health promotion
Maurice B. Mittelmark Department of Health Promotion and Development, Faculty of Psychology, University of Bergen, Norway
Abstract The settings-based approach to health promotion (HP) employs a social ecological (SE) framework to integrate HP into the usual activities of the setting and to increase the setting’s support for healthy choices. The SE approach calls for systems thinking to account for the inextricable relationship between people, their behaviour and their environment. Knowledge about a setting can be used to mobilise people to participate in HP, to optimise success by taking into account the local context, and to anticipate and avoid barriers to success. In other words, the SE approach aims to help HP reach its goals for better health, established in concert with community needs and wishes.Yet, the focus on HP goals may detract attention from how intervention may have unanticipated, and even untoward effects on the setting. There is much evidence from classical ecological research that well-meaning interventions have unintended effects. Biology is so tuned to the possibility that the study of unintended effects is integral to the field. There is some evidence – but much less - that HP also has unexpected, deleterious effects. The evidence is limited because of neglect; the subject of unintended effects is only of peripheral interest in HP. This is a call for a more robust SE approach, in which frameworks used to guide settings-based HP are augmented so as to be concerned with planned effects, and also unplanned effects. What can be done to more responsibly monitor, document and report the full panoply of our effects, including detecting and preventing untoward effects? Key Words: Health promotion settings, social ecology, unintended effects, harm
Introduction Settings-based health promotion has a wide range of possible consequences, a few of which we have planned, but many of which are unplanned, and may be deleterious to the well-being of the setting. This essay addresses the question, what should we do to prevent, monitor, document and report the full panoply of our effects? The term ‘setting’ refers to places where people congregate for work, play and fellowship, such as schools, workplaces, recreational spaces and places of worship. The settings-based approach to health promotion (HP) employs a social ecological (SE) framework to integrate HP into the usual activities of the setting and to increase the setting’s support for healthy choices. The term ‘social ecology’ is a child with many parents. For the present purpose, it helps
us appreciate that human settings have the characteristics of all habitats. There is interaction between inhabitants, and they interact with their physical environment. What happens in one setting may affect what happens in other settings. Settings are not dependent on the particular people who are there at any given time; a school is a school regardless of who studies there or who teaches there. Therefore, the SE approach calls for systems thinking and respects the complex relationship between people, their behaviour and their environment. The complexity of doing interventions in settings is appreciated by HP researchers, and a SE approach to settings-based HP is encouraged. However, those who carry out HP in settings seem to be concerned primarily with how an ecological approach can
Correspondence: Maurice B. Mittelmark, Department of Health Promotion and Development, University of Bergen, P.O. Box 7800, N-5020 Bergen, Norway. E-mail: [email protected] (Accepted 2 July 2014) © 2014 the Nordic Societies of Public Health DOI: 10.1177/1403494814545108
18 MB. Mittelmark enhance the success of their interventions. Ecological sensitivity should also include a keen awareness that almost any intervention in a habitat – a setting - may cause unanticipated, untoward effects. This includes effects experienced by the intervention’s ‘target’, and/ or by other social and physical actors/features of the setting. In field biology, the ecologist is intensely tuned to this possibility, and protection against an intervention’s possible untoward effects is core to her ecological approach. The HP researcher working in human settings seems less alert. This deficit – a seeming lack of appreciation that settings-based HP has the potential to do unanticipated harm - signals the need for a more robust ecological orientation in settings-based research. Health promotion does have the potential to do harm, however unintended. This came as a sharp lesson to me in the 1980s, when I participated in community-based research aimed at preventing heart disease [1]. The Minnesota Heart Health Program (MHHP) was intensively settings-based, with HP in medical clinics and doctor’s offices, schools, food markets, eateries, workplaces, sports clubs and places of worship. The MHHP aimed for health improvements at the individual, the group and the community levels. It included education, structural interventions and policy initiatives, and shared responsibility between the researchers and a broad array of citizens. It was considered the state-of-theart in community-based HP at the time. Our MHHP research team was fortunate to have external scientific advisors, including the preeminent Professor Larry Green, at that time Director of the Center for Health Promotion Research and Development at the University of Texas. During one of his visits midway through the research project, Professor Green posed a blunt question: ‘What harm are you doing with this project?’. His point was that any large-scale community-based HP research project that injected money, people and a ‘new’ agenda into a setting (in this case, whole communities and their many sub-settings) should suspect that it might ‘disturb’ the setting in unplanned ways. We were taken aback by his question. Not once during the planning and implementation of the community-based research project had we seriously contemplated that question. What research questions might we have investigated had we been alert to the possibility of unintended effects?
•• What happens to community welfare organisations that depend on community volunteers, when an exciting new volunteer opportunity is suddenly open? •• What happens to existing health education programmes when media attention and public interest is shifted to the new ‘hot’ heart disease prevention programme?
•• What happens to the community’s existing political and social agenda when a major new agenda item - a well-funded university heart disease prevention project - lands in town?
The settings approach to HP is the instrumental use of places where people gather and spend considerable time for purposes other than HP (e.g. where they eat, shop, work, learn, pray, play, get medical
We did belatedly take the issue to heart, I am happy to report, and started monitoring some possible untoward effects of our work [2]. However, our effort was too little and too late. We never collected the data, nor undertook the analyses, that would have been needed to make a serious study of unintended events and effects in the context of the MHHP. I write above that this was a ‘sharp’ lesson for me, but it was also an embarrassment. I had previously studied behavioural ecology and participated in behavioural ecology research during my period of doctoral study in social psychology. I certainly got a thorough indoctrination into the principles of social and behavioural ecology, but in my zeal as cardiovascular disease prevention researcher, I seemed to have forgotten much of it. I do not think I am alone; my reading of the literature is that although many in HP pay homage to the ideal of a social ecology orientation to HP in settings, we often fall short in our practice and research. We need reminders, often and dogged, that if we are to follow through on a commitment to a social ecology approach to our research, we must make special effort. This paper tries to explain why ecological ideals are hard to implement in a robust way in settings-based HP. It focuses on this issue: settings-based HP has a wide range of possible consequences, a few of which we have planned, but many of which are unplanned, and may be deleterious to the well-being of the setting and/or its inhabitants. What should we do to prevent, monitor, document and report the full panoply of our effects? This is posed as an ethical issue that needs attention. As pithily stated by Leon Eisenberg more than four decades ago, ‘The challenge is our ability to anticipate the second- and thirdorder consequences of interventions in the ecosystem before the event, not merely to rue them afterward.’ [3, p. 123].
The settings-based approach to HP
Unintended effects in settings-based health promotion 19 treatment). Interventions are launched as part of or adjacent to the usual activities of the setting. These interventions often call for the active participation of the setting’s inhabitants. A less direct and complementary approach is to change the setting itself, to support healthy choices. An even less direct method is to use social policy to create health-promoting conditions in society generally. The SE approach encourages all these, implemented in a multi-modal, multi-level combination. In HP, the concept of social ecology is conflated with the concept of complexity, while the more fundamental precept of ecology may be the concept of connectedness. Habitats – including human settings - contain interacting biotic and abiotic systems. The systems are more or less open, and a habitat interacts with other habitats. The biotic part is composed of one or more communities (species). Community members act and interact (standing behaviour). Community members fill specific roles – niches – that outlive members. When a habitat is manipulated by an outsider to achieve a planned and worthy goal, unplanned and unexpected effects attend [4]. Some of these effects may be seriously untoward, so monitoring the habitat for unplanned effects is the outsider’s minimal ethical obligation. Better yet is the anticipation and prevention of events that the outsider might otherwise come to rue. Social ecology and HP Even if ecology as a field of biology is almost a century old, the ecological lessons of the past are no guarantee of advancement in our ecological sensitivity and sophistication. What are those lessons for HP researchers? They were crystallised particularly well by Green, Richard and Potvin [5], and the lessons are as valid today as they were when first published almost two decades ago. The attempt to make a change in an ecological system should be preceded by consideration and anticipation of possible unanticipated side effects of intervention. Behaviour is a function of the interaction of individuals and their social and physical environments. Individuals and groups behave differently in different environments, because the environment predisposes, enables and reinforces individual and group behaviour. Interventions must be multi-level and multi-sector because the ecological web is so. Following from this, we can see why ecological approaches to HP are fraught with difficulty. The practical response from many of us is to give a prime place to ecological thinking in our rhetoric and teaching, while retreating to doing settings-based
interventions that are over-simple; calls for an ecological approach ‘have gone largely unheeded’ [6]. Yet, the encouragement to take the ecological path has been steady, and broad guidelines for settingsbased HP have emerged [7-13]. Stokols [8] long ago outlined core assumptions of the social ecology of HP. First, well-being is influenced by our physical and social environments, and by intrapersonal factors. Therefore, HP should reflect the dynamic interplay of these factors and avoid isolated efforts. Second, human environments have many dimensions that affect health and HP should be cognisant of these. They have physical and social components, but also objective and subjective qualities, varying degrees of proximity to inhabitants, and independent and composite attributes (e.g. group size and social climate). Third, the level of analysis and intervention may be at the individual level, but the aggregate social level is also of significance – the family, the work organisation, the community. Fourth, rather than having an oversimplified cause-effect mentality, social ecology calls for systems thinking that takes into account the inextricable inter-relationships of a system’s elements. In fact, Stokols’ [8] core assumptions are unified by systems theory thinking. That is what makes the social ecology approach to HP quite different from what might be called the ‘piecemeal approach’, which employs targeted, isolated interventions to change health behaviour. High-quality SE HP intervention strategies are multi-modal, multi-level and ‘socially valid’: they are grounded in scientific evidence, they are economically feasible, they are likely to reach those they are intended to reach, they are consistent with communities’ priorities and desires, and they are unlikely to cause adverse side effects [8]. Turning a few years later to the question of how to enhance our ability to adhere to the principles of social ecology, Stokols [14] offered design and evaluation guidelines for community HP, helpfully linking guidelines to principles. The principle that well-being is influenced by the environment leads straightforwardly to the practice guideline that we should examine the links between well-being and diverse conditions of the physical and social environment. The principle that intra-personal characteristics and the environment have direct and interactive effects on well-being leads to the practice guideline that we should ‘examine the joint influence of behavioural, dispositional, developmental and demographic factors on people’s exposure and response to environmental hazards and demands’ [14, p. 288]. The principle that the degree of fit between people’s needs and their resources is a key determinant of well-being leads to the guideline that we should
20 MB. Mittelmark assess the degree of person-environment miss-fit and develop intervention to increase fit. The principle that certain behaviours and roles exert pivotal influence on well-being translates to the guideline that we should identify and use ‘leverage points’ in community interventions. Stokols [14] further called for interventions having enduring effects on well-being, and for the use of multiple methods to evaluate the health- and cost-effectiveness of community-based HP. In settings-based HP, the difficulty of adhering to the principles of ecology generally and social ecology in particular, is exacerbated by system connectedness and complexity, but also by managers, funders, and clients, who often expect/demand quick answers to very complex problems. Those who deliver the premises for HP research have the power to override the researcher’s ‘ecological reflexivity’ (awareness) that is in harness with the ecological principles that have been mentioned. When a HP practitioner/researcher is provided with resources to do settings-based HP, it is usually part of a larger agenda with pre-set priorities, assumptions about what is ‘good work’ and deadlines unconnected to the natural rhythm of life in the setting. Hence, even if health promoters working in settings speak of an ‘ecological’ approach, they are seldom robustly ecological in their approach, perhaps because of their own shortcomings, but more likely because of external constraints. Thus, the de facto question addressed in the settings approach to HP is, ‘what can we do to (and with) the person and her setting to achieve our aims for her health’? This is a reasonable question. Health promoters are social change agents, after all, with social change agendas. Poland, Krupa and McCall [15] provide a realistic summary of what a health promoter with a specific agenda can expect a settings approach to accomplish: Knowledge about a setting can be “mobilised in the planning, implementation and evaluation of health promotion interventions”. One can optimise the likelihood of success by taking careful stock of the local context. Knowledge about the setting can be used to “forestall the possibility that a critical oversight could… wash the project up, stall progress…a detailed analysis of the setting… can help practitioners skillfully [sic] anticipate and navigate potentially murky waters…” [15, p. 506].
Towards a more robust approach Thus, a settings approach helps us to achieve our aims. However, if we take a robust ecological approach, additional issues arise. What is the standing pattern of behaviour in the setting prior to
intervention? How can we develop an intervention that is compatible with and protective of the preexisting ecology of the setting? Once we intervene, how can we monitor our impact to document not only planned, but also unplanned (and especially untoward) events and effects of the intervention on the natural state of the setting? How can we react quickly and positively to ameliorate untoward effects, if any are observed? What happens to and in the setting long after our project is complete? Were health promoters working in settings to add these questions to the de facto question, they would move tellingly in the direction of a robust ecological approach. One might guess that these issues are at least noted in ecological frameworks for HP. However, even the freshest frameworks focus on the means to achieve the HP ‘end’, and do not attend to the ‘ancillary’ questions raised in the paragraph above [13]. In the rare and welcome instances when unintended effects have been considered, the focus has usually been on the intervention aim, and how unintended effects may detract from intended effects [16]. When the pre-existing ecology of a setting is considered, the concern is often about how the ‘core business’ of a setting might be in conflict with a project’s aim, how the core business might suggest ways (incentives) to assure the cooperation of the setting [17], or how the intervention might produce effects opposite to those wished for [18]. Yet some studies have had a wider ecological reach. Guttman and Salmon [19] observed how public health communication campaigns might inadvertently stigmatise vulnerable groups and expand social gaps. Van Hal et al. [20] present case studies showing how empowerment interventions intended to include people in society may actually have excluding consequences. Burgess, Fu and van Ryn [21] review literature showing that some mothers who are exposed to health education about the harm to themselves and their children because of their smoking, unfortunately experience lower selfworth and increased guilt and sadness. On the brighter side, looking for unintended effects may provide reassurance. Williamson and colleagues [22] searched for and found little evidence of unintended effects of smoke-free ordinances. A few studies are so attuned to the issue of possible unintended effects that their research models explicitly included the possibility of such effects. Figure 1 is from intervention work by Krølner and colleagues [23] that aimed to improve health via increased fruit and vegetable consumption among 13-year-olds, showing hypothesised relationships between intended outcomes (increased intake), side
Unintended effects in settings-based health promotion 21
Figure 1. Hypothesised causal relationships between the Boost Intervention Programme (aimed to increase fruit and vegetable consumption among 13-year-olds), outcomes, side effects and unintended adverse outcomes [23]. Reproduced with permission.
effects (academic achievement) and unintended adverse outcomes (e.g. bullying and eating-related teasing). This attention to the possibility of unintended outcomes is rare, but not unique [17,24-26]. A practical example from the health behaviour research arena is the study of Horigian et al. [27], of a family therapy intervention for the treatment of adolescents with drug abuse. The investigators assembled a team of family therapy experts, medical safety officers and ethicists, who developed principles for defining, and then monitored possible adverse events that participants and their families might experience during the course of the study. Such events might be consequences of the intervention or have other roots, and the investigators were sensitive to the difficulty of assigning causes to the adverse events. As it happened, more than 50% of participating adolescents experienced at least one adverse event, for example arrest, ejection from school or home, and/or violence/abuse,
suicidal behaviour and hospitalization. The takehome lesson for settings-based HP was that the investigtors anticipated adverse events and monitored, observed and reported what happened. Behavioural intervention trials and HP interventions in settings have stark dissimilarities, but one feature they have in common is the rarity with which adverse events are anticipated. The key lesson for HP from the Horigian et al. [27] study, even if it was not a settings study, is the desirability and feasibility of including in the research protocol a real concern for the lives and experiences of participants, including the possibility that a wellmeaning intervention might have untoward consequences. The salience of this concern for HP in settings is poignantly illustrated in the work of Pinhas, et al. [28], in which some children exposed to school-based ‘healthy weights’ initiatives experienced adverse events. They describe the case of:
22 MB. Mittelmark ‘a 14-year-old …student with excellent grades who was perfectionistic and worried tremendously about her marks. […]she was assigned a school project on eating disorders as part of a health and nutrition section of her physical education program. She reviewed common weight loss behaviours found in patients with eating disorders for her project. In doing so, she began to worry about her own weight and whether she was too fat. She became preoccupied with eating disordered thoughts and behaviours that she had documented for her project and found herself trying them out. She began restricting the amount of energy dense foods she consumed and decreasing her total portion sizes. She did not engage in bingeing, purging,or over-exercising as a means to lose weight. Approximately 6 months later she presented to the hospital emergency department as a result of not eating solid foods for 5 days’ [28, pp. 110-11].
Yet another example is the work of Moore et al. [29] in which settings-based HP – smoke-free workplace laws for bars – had the unanticipated outcome that some women who smoked outside experienced threats to their physical safety. How might we promote more explicit attention to the ecology of a HP setting, as in the examples above, leading researchers to concern about the ‘core business’ of the setting in its own right, as well as concern about how intervention might interfere with project goals? One answer is to enhance existing HP planning and evaluation frameworks to include the concept of unintended harm [30], including not only harm to the aims of the intervention project, but also harm to the integrity of the setting. This would provide the means and incentive to theorise about, monitor, prevent and report adverse events and processes, alongside the health promoting events and processes that we aim to achieve. A practical step in that direction is the work of Allen-Scott, Hatfield and McIntyre [30], who are in the process of constructing an empirically founded typology of unintended harm and its underlying factors. The typology articulates five classes of unintended harm and five classes of underlying factors, meaning factors under the ostensible control of the researcher, or at least awareness of which may lead to harm avoidance in the design and implmentation of a HP intervention. Beginning with underlying factors, Allen-Scott, Hatfield and McIntyre [30] identify limited or poor-quality evidence as potentially contributing to harm, citing the example from their empirical material of the lack of sufficient evidence on long-term physical and psychosocial effects of obesity reduction interventions. A second underlying factor supported by their empirical database is the boomerang effect (prevention of one extreme leads to anther extreme), as in the study
already referred to by Pinhas et al. [28], in which an intervention aiming to prevent body weight problems triggered dietary restraint to a dangerous degree. Lack of community engagement is third factor underlying potential harm according to evidence in Pinhas et al.’s [28] database, with minimal or ‘tokenistic’ community involvement associated with stigma, descrimination and environmental contamination. Ignoring root causes is the fourth underlying factor in the harm typology, referring, for example, to the possible stigmatisation of people whose poor health is in theory modifiable by an intervention’s lifestyle change focus, with the intervention ignoring more fundamental causes such as living and social conditions. The fifth underlying factor leading to unintended harm is the fallacy of one size fits all (my label), referring to how a successful intervention developed in one context/culture may cause harm in another context/culture. Allen-Scott, Hatfield and McIntyre [30] focus particular attention on the case of interventions developed in highincome countries that cause unexpected harm in low-income countries. The typology’s underlying factors correlate more or less to five categories of unintended harm: physical, psychosocial, economic, cultural and environmental. Not unexpectedly, the strongest evidence for links between underlying factors and types of harm in Allen-Scott, Hatfield and McIntyre’s [30] empirical material is for the link between the boomerang effects and poor-quality evidence, on the one hand, and individual-level physical and psychosocial harm on the other hand. Harm that would be most manifest at group, community and societal levels – economic, cultural, environmental – is much less studied. This is consistent with the contention raised earlier that when we are concerned with the ecology of unintended effects, it is mostly a concern that our planned intervention effect might be jeopardised. As HP interventions in settings rarely if ever have the main aim to affect economic, cultural or environmental conditions, it is not a surprise that variables in these arenas are scarcely scrutinised. Conclusion This circles back to the core question of this essay: Settings-based health promotion has a wide range of possible consequences, a few of which we have planned, but many of which are unplanned, and may be deleterious to the well-being of the setting.What should we do to prevent, monitor, document and report the full panoply of our effects?
Unintended effects in settings-based health promotion 23 We must first develop more fully our social ecological sensitivity. This will lead us to pose questions like these, in addition to the de facto research question of a settings-based HP study: •• What is the standing pattern of behaviour in the setting prior to intervention? •• How can we develop an intervention that is compatible with and protective of the pre-existing ecology of the setting? •• Once we intervene, how can we monitor our impact to document not only planned, but also unplanned (and especially untoward) events and effects of the intervention on the natural state of the setting? •• How can we react quickly and positively to ameliorate untoward effects, if any are observed? •• What happens to and in the setting long after our project is complete? To routinely introduce questions like these into the planning, implementation and evaluation of settingsbased HP, our intervention frameworks and models need augmentation, for example by the harm typology of Allen-Scott, Hatfield and McIntyre [30], or by emerging variants. Succeeding in this, we will have moved in the direction of a more robust ecological orientation in settings-based HP. Acknowledgement This paper is based in part on a presentation by the author at the 7th Nordic Health Promotion Research Conference, Tønsberg, Norway, 17-19 June, 2013. Conflict of interest The author declares that there is no conflict of interest. Funding This research received no specific grant from any funding agency in the public, commercial, or not-forprofit sectors. References [1] Mittelmark MB, Luepker RV, Jacobs DR, et al. Communitywide prevention of cardiovascular disease: education strategies of the Minnesota Heart Health Program. Prev Med 1986;15:1–17. [2] Weisbrod RR, Pirie PL and Bracht NF. Impact of a community health promotion program on existing organizations: the Minnesota Heart Health Program. Soc Sci Med 1992;34:639–48. [3] Eisenberg L. The human nature of human nature. Science 1972;176:123–8. [4] Pratt PD and Center TD. Biocontrol without borders: the unintended spread of introduced weed biological control agents. BioControl 2012;57:319–29. [5] Green LW, Richard L and Potvin L. Ecological foundation of health promotion. Am J Health Promot 1996;10: 270–81.
[6] Golden SD and Earp JAL. Social ecological approaches to individuals and their contexts: Twenty years of Health Education and Behavior health promotion interventions. Health Educ Behav 2012;39:364–72. [7] McLeroy KR, Bibeau D, Steckler A, et al. An ecological perspective on health promotion programs. Health Educ Quart 1988;15:351–77. [8] Stokols D. Establishing and maintaining healthy environments: toward a social ecology of health promotion. Am Psychol 1992;47:6–22. [9] Richard L, Gauvin L and Raine K. Ecological models revisited: Their use and evolution in health promotion over two decades. Ann Rev Public Health 2011;32:307–26. [10] Poland B and Dooris M. A green and healthy future: the settings approach to building health, equity and sustainability. Crit Public Health 2010;2083:281–98. [11] Whitelaw S, Baxendale A, Bryce C, et al. ‘Settings’ based health promotion: a review. Health Promot Int 2001;16:339–53. [12] Dooris M. Healthy settings: challenges to generating evidence of effectiveness. Health Promot Int 2005;21:55–65. [13] Shareck M, Frohlich KL and Poland B. Reducing social equalities in health through settings-related interventions – a conceptual framework. Glob Health Promot 2013;20: 39–52. [14] Stokols D. Translating social ecology theory into guidelines for community health promotion. Am J Health Promot 1996;10:282–98. [15] Poland B, Krupa G and McCall D. Settings for health promotion: An analytic framework to guide intervention design and implementation. Health Promot Prac 2009;10:505–16. [16] Rychetnik L, Frommer M, Hawe P, et al. Criteria for evaluating evidence on public health interventions. J Epidemiol Comm Health 2002;56:119–27. [17] Kokko S, Green LW and Kannas L. A review of settingsbased health promotion with applications to sports clubs. Health Promot Int. 2014; 29(3): 494–509. [18] Sallis JF, Calfas KJ, Alcaraz JE, et al. Potential mediators of change in a physical activity promotion course for university students: project GRAD. Ann Behav Med 1999;21:149–58. [19] Guttman N and Salmon CT. Guilt, fear, stigma and knowledge gaps: ethical issues in public health communication interventions. Bioethics 2004;18:531–52. [20] van Hal LBE, Meershoek A, Nijhuis F, et al. The ‘empowered client’ in vocational rehabilitation: the excluding impact of inclusive strategies. Health Care Anal 2012;20:213–30. [21] Burgess DJ, Fu SS and van Ryn M. Potential unintended consequences of tobacco-control policies on mothers who smoke. A review of the literature. Am J Prev Med 2009;37:S151-8. [22] Williamson AA, Fox BJ, Creswell PD, et al. An observational study of the secondary effects of a local smoke-free ordinance. Prev Chronic Dis 2011;8:A83. http://www.cdc.gov/ pcd/issues/2011/jul/10_0123.htm. Accessed 7 January 2014. [23] Krølner R, Jørgensen TS, Aarestrup AK, et al. The BOOST study: design of a school- and community-based randomised trial to promote fruit and vegetable consumption among teenagers. BMC Public Health 2012. Epub DOI:10.1186/1471–2458–12–191. [24] Dunton GF, Intille SS, Wolch J, et al. Investigating the impact of a smart growth community on the contexts of children’s physical activity using Ecological Momentary Assessment. Health & Place 2012;18:76–84. [25] Busza J, Walker D, Hairston A, et al. Community-based approaches for prevention of others to child transmission in resource-poor settings. A social ecological review. J Int AIDS Soc 2012;15(Suppl 2). DOI: 10.7448/IAS.15.4.17373. [26] Moore L, de Silva-Sanigorski A and Moore SN. A socio-ecological perspective on behavioural interven tions to influence food choice in schools: alternative,
24 MB. Mittelmark complementary and synergistic? Public Health Nutr 2013;1000– 5. Epub 4 March 2013. DOI: 10.1017/S1368980012005605. [27] Horigian VE, Robbins MS, Dominguez R, et al. Principles for defining adverse effects in behavioral intervention research: lessons from a family-focused adolescent drug abuse trial. Clin Trials 2010;7:58–68. [28] Pinhas L, Walker GM, Walker KS, et al. Trading health for a healthy weight: the uncharted side of healthy weights initiatives. Eat Disord 2013;21:109–16.
[29] Moore RS, Annechino RM and Lee JP. Unintended consequences of smoke-free bar policies for low-SES women in three California counties. Am J Prev Med 2009;37: S138-43. [30] Allen-Scott LK, Hatfield JM and McIntyre L. A scoping review of unintended harms associated with public health interventions. Towards a typology and an understanding of underlying factors. Int J Public Health. Epub 1 January 2014. DOI 10.1007/s00038–013–0526–6.
research-article2014
SJP0010.1177/1403494814545108Unintended effects in settingsMB. Mittelmark
Scandinavian Journal of Public Health, 2014; 42(Suppl 15): 17–24
Original Article
Unintended effects in settings-based health promotion
Maurice B. Mittelmark Department of Health Promotion and Development, Faculty of Psychology, University of Bergen, Norway
Abstract The settings-based approach to health promotion (HP) employs a social ecological (SE) framework to integrate HP into the usual activities of the setting and to increase the setting’s support for healthy choices. The SE approach calls for systems thinking to account for the inextricable relationship between people, their behaviour and their environment. Knowledge about a setting can be used to mobilise people to participate in HP, to optimise success by taking into account the local context, and to anticipate and avoid barriers to success. In other words, the SE approach aims to help HP reach its goals for better health, established in concert with community needs and wishes.Yet, the focus on HP goals may detract attention from how intervention may have unanticipated, and even untoward effects on the setting. There is much evidence from classical ecological research that well-meaning interventions have unintended effects. Biology is so tuned to the possibility that the study of unintended effects is integral to the field. There is some evidence – but much less - that HP also has unexpected, deleterious effects. The evidence is limited because of neglect; the subject of unintended effects is only of peripheral interest in HP. This is a call for a more robust SE approach, in which frameworks used to guide settings-based HP are augmented so as to be concerned with planned effects, and also unplanned effects. What can be done to more responsibly monitor, document and report the full panoply of our effects, including detecting and preventing untoward effects? Key Words: Health promotion settings, social ecology, unintended effects, harm
Introduction Settings-based health promotion has a wide range of possible consequences, a few of which we have planned, but many of which are unplanned, and may be deleterious to the well-being of the setting. This essay addresses the question, what should we do to prevent, monitor, document and report the full panoply of our effects? The term ‘setting’ refers to places where people congregate for work, play and fellowship, such as schools, workplaces, recreational spaces and places of worship. The settings-based approach to health promotion (HP) employs a social ecological (SE) framework to integrate HP into the usual activities of the setting and to increase the setting’s support for healthy choices. The term ‘social ecology’ is a child with many parents. For the present purpose, it helps
us appreciate that human settings have the characteristics of all habitats. There is interaction between inhabitants, and they interact with their physical environment. What happens in one setting may affect what happens in other settings. Settings are not dependent on the particular people who are there at any given time; a school is a school regardless of who studies there or who teaches there. Therefore, the SE approach calls for systems thinking and respects the complex relationship between people, their behaviour and their environment. The complexity of doing interventions in settings is appreciated by HP researchers, and a SE approach to settings-based HP is encouraged. However, those who carry out HP in settings seem to be concerned primarily with how an ecological approach can
Correspondence: Maurice B. Mittelmark, Department of Health Promotion and Development, University of Bergen, P.O. Box 7800, N-5020 Bergen, Norway. E-mail: [email protected] (Accepted 2 July 2014) © 2014 the Nordic Societies of Public Health DOI: 10.1177/1403494814545108
18 MB. Mittelmark enhance the success of their interventions. Ecological sensitivity should also include a keen awareness that almost any intervention in a habitat – a setting - may cause unanticipated, untoward effects. This includes effects experienced by the intervention’s ‘target’, and/ or by other social and physical actors/features of the setting. In field biology, the ecologist is intensely tuned to this possibility, and protection against an intervention’s possible untoward effects is core to her ecological approach. The HP researcher working in human settings seems less alert. This deficit – a seeming lack of appreciation that settings-based HP has the potential to do unanticipated harm - signals the need for a more robust ecological orientation in settings-based research. Health promotion does have the potential to do harm, however unintended. This came as a sharp lesson to me in the 1980s, when I participated in community-based research aimed at preventing heart disease [1]. The Minnesota Heart Health Program (MHHP) was intensively settings-based, with HP in medical clinics and doctor’s offices, schools, food markets, eateries, workplaces, sports clubs and places of worship. The MHHP aimed for health improvements at the individual, the group and the community levels. It included education, structural interventions and policy initiatives, and shared responsibility between the researchers and a broad array of citizens. It was considered the state-of-theart in community-based HP at the time. Our MHHP research team was fortunate to have external scientific advisors, including the preeminent Professor Larry Green, at that time Director of the Center for Health Promotion Research and Development at the University of Texas. During one of his visits midway through the research project, Professor Green posed a blunt question: ‘What harm are you doing with this project?’. His point was that any large-scale community-based HP research project that injected money, people and a ‘new’ agenda into a setting (in this case, whole communities and their many sub-settings) should suspect that it might ‘disturb’ the setting in unplanned ways. We were taken aback by his question. Not once during the planning and implementation of the community-based research project had we seriously contemplated that question. What research questions might we have investigated had we been alert to the possibility of unintended effects?
•• What happens to community welfare organisations that depend on community volunteers, when an exciting new volunteer opportunity is suddenly open? •• What happens to existing health education programmes when media attention and public interest is shifted to the new ‘hot’ heart disease prevention programme?
•• What happens to the community’s existing political and social agenda when a major new agenda item - a well-funded university heart disease prevention project - lands in town?
The settings approach to HP is the instrumental use of places where people gather and spend considerable time for purposes other than HP (e.g. where they eat, shop, work, learn, pray, play, get medical
We did belatedly take the issue to heart, I am happy to report, and started monitoring some possible untoward effects of our work [2]. However, our effort was too little and too late. We never collected the data, nor undertook the analyses, that would have been needed to make a serious study of unintended events and effects in the context of the MHHP. I write above that this was a ‘sharp’ lesson for me, but it was also an embarrassment. I had previously studied behavioural ecology and participated in behavioural ecology research during my period of doctoral study in social psychology. I certainly got a thorough indoctrination into the principles of social and behavioural ecology, but in my zeal as cardiovascular disease prevention researcher, I seemed to have forgotten much of it. I do not think I am alone; my reading of the literature is that although many in HP pay homage to the ideal of a social ecology orientation to HP in settings, we often fall short in our practice and research. We need reminders, often and dogged, that if we are to follow through on a commitment to a social ecology approach to our research, we must make special effort. This paper tries to explain why ecological ideals are hard to implement in a robust way in settings-based HP. It focuses on this issue: settings-based HP has a wide range of possible consequences, a few of which we have planned, but many of which are unplanned, and may be deleterious to the well-being of the setting and/or its inhabitants. What should we do to prevent, monitor, document and report the full panoply of our effects? This is posed as an ethical issue that needs attention. As pithily stated by Leon Eisenberg more than four decades ago, ‘The challenge is our ability to anticipate the second- and thirdorder consequences of interventions in the ecosystem before the event, not merely to rue them afterward.’ [3, p. 123].
The settings-based approach to HP
Unintended effects in settings-based health promotion 19 treatment). Interventions are launched as part of or adjacent to the usual activities of the setting. These interventions often call for the active participation of the setting’s inhabitants. A less direct and complementary approach is to change the setting itself, to support healthy choices. An even less direct method is to use social policy to create health-promoting conditions in society generally. The SE approach encourages all these, implemented in a multi-modal, multi-level combination. In HP, the concept of social ecology is conflated with the concept of complexity, while the more fundamental precept of ecology may be the concept of connectedness. Habitats – including human settings - contain interacting biotic and abiotic systems. The systems are more or less open, and a habitat interacts with other habitats. The biotic part is composed of one or more communities (species). Community members act and interact (standing behaviour). Community members fill specific roles – niches – that outlive members. When a habitat is manipulated by an outsider to achieve a planned and worthy goal, unplanned and unexpected effects attend [4]. Some of these effects may be seriously untoward, so monitoring the habitat for unplanned effects is the outsider’s minimal ethical obligation. Better yet is the anticipation and prevention of events that the outsider might otherwise come to rue. Social ecology and HP Even if ecology as a field of biology is almost a century old, the ecological lessons of the past are no guarantee of advancement in our ecological sensitivity and sophistication. What are those lessons for HP researchers? They were crystallised particularly well by Green, Richard and Potvin [5], and the lessons are as valid today as they were when first published almost two decades ago. The attempt to make a change in an ecological system should be preceded by consideration and anticipation of possible unanticipated side effects of intervention. Behaviour is a function of the interaction of individuals and their social and physical environments. Individuals and groups behave differently in different environments, because the environment predisposes, enables and reinforces individual and group behaviour. Interventions must be multi-level and multi-sector because the ecological web is so. Following from this, we can see why ecological approaches to HP are fraught with difficulty. The practical response from many of us is to give a prime place to ecological thinking in our rhetoric and teaching, while retreating to doing settings-based
interventions that are over-simple; calls for an ecological approach ‘have gone largely unheeded’ [6]. Yet, the encouragement to take the ecological path has been steady, and broad guidelines for settingsbased HP have emerged [7-13]. Stokols [8] long ago outlined core assumptions of the social ecology of HP. First, well-being is influenced by our physical and social environments, and by intrapersonal factors. Therefore, HP should reflect the dynamic interplay of these factors and avoid isolated efforts. Second, human environments have many dimensions that affect health and HP should be cognisant of these. They have physical and social components, but also objective and subjective qualities, varying degrees of proximity to inhabitants, and independent and composite attributes (e.g. group size and social climate). Third, the level of analysis and intervention may be at the individual level, but the aggregate social level is also of significance – the family, the work organisation, the community. Fourth, rather than having an oversimplified cause-effect mentality, social ecology calls for systems thinking that takes into account the inextricable inter-relationships of a system’s elements. In fact, Stokols’ [8] core assumptions are unified by systems theory thinking. That is what makes the social ecology approach to HP quite different from what might be called the ‘piecemeal approach’, which employs targeted, isolated interventions to change health behaviour. High-quality SE HP intervention strategies are multi-modal, multi-level and ‘socially valid’: they are grounded in scientific evidence, they are economically feasible, they are likely to reach those they are intended to reach, they are consistent with communities’ priorities and desires, and they are unlikely to cause adverse side effects [8]. Turning a few years later to the question of how to enhance our ability to adhere to the principles of social ecology, Stokols [14] offered design and evaluation guidelines for community HP, helpfully linking guidelines to principles. The principle that well-being is influenced by the environment leads straightforwardly to the practice guideline that we should examine the links between well-being and diverse conditions of the physical and social environment. The principle that intra-personal characteristics and the environment have direct and interactive effects on well-being leads to the practice guideline that we should ‘examine the joint influence of behavioural, dispositional, developmental and demographic factors on people’s exposure and response to environmental hazards and demands’ [14, p. 288]. The principle that the degree of fit between people’s needs and their resources is a key determinant of well-being leads to the guideline that we should
20 MB. Mittelmark assess the degree of person-environment miss-fit and develop intervention to increase fit. The principle that certain behaviours and roles exert pivotal influence on well-being translates to the guideline that we should identify and use ‘leverage points’ in community interventions. Stokols [14] further called for interventions having enduring effects on well-being, and for the use of multiple methods to evaluate the health- and cost-effectiveness of community-based HP. In settings-based HP, the difficulty of adhering to the principles of ecology generally and social ecology in particular, is exacerbated by system connectedness and complexity, but also by managers, funders, and clients, who often expect/demand quick answers to very complex problems. Those who deliver the premises for HP research have the power to override the researcher’s ‘ecological reflexivity’ (awareness) that is in harness with the ecological principles that have been mentioned. When a HP practitioner/researcher is provided with resources to do settings-based HP, it is usually part of a larger agenda with pre-set priorities, assumptions about what is ‘good work’ and deadlines unconnected to the natural rhythm of life in the setting. Hence, even if health promoters working in settings speak of an ‘ecological’ approach, they are seldom robustly ecological in their approach, perhaps because of their own shortcomings, but more likely because of external constraints. Thus, the de facto question addressed in the settings approach to HP is, ‘what can we do to (and with) the person and her setting to achieve our aims for her health’? This is a reasonable question. Health promoters are social change agents, after all, with social change agendas. Poland, Krupa and McCall [15] provide a realistic summary of what a health promoter with a specific agenda can expect a settings approach to accomplish: Knowledge about a setting can be “mobilised in the planning, implementation and evaluation of health promotion interventions”. One can optimise the likelihood of success by taking careful stock of the local context. Knowledge about the setting can be used to “forestall the possibility that a critical oversight could… wash the project up, stall progress…a detailed analysis of the setting… can help practitioners skillfully [sic] anticipate and navigate potentially murky waters…” [15, p. 506].
Towards a more robust approach Thus, a settings approach helps us to achieve our aims. However, if we take a robust ecological approach, additional issues arise. What is the standing pattern of behaviour in the setting prior to
intervention? How can we develop an intervention that is compatible with and protective of the preexisting ecology of the setting? Once we intervene, how can we monitor our impact to document not only planned, but also unplanned (and especially untoward) events and effects of the intervention on the natural state of the setting? How can we react quickly and positively to ameliorate untoward effects, if any are observed? What happens to and in the setting long after our project is complete? Were health promoters working in settings to add these questions to the de facto question, they would move tellingly in the direction of a robust ecological approach. One might guess that these issues are at least noted in ecological frameworks for HP. However, even the freshest frameworks focus on the means to achieve the HP ‘end’, and do not attend to the ‘ancillary’ questions raised in the paragraph above [13]. In the rare and welcome instances when unintended effects have been considered, the focus has usually been on the intervention aim, and how unintended effects may detract from intended effects [16]. When the pre-existing ecology of a setting is considered, the concern is often about how the ‘core business’ of a setting might be in conflict with a project’s aim, how the core business might suggest ways (incentives) to assure the cooperation of the setting [17], or how the intervention might produce effects opposite to those wished for [18]. Yet some studies have had a wider ecological reach. Guttman and Salmon [19] observed how public health communication campaigns might inadvertently stigmatise vulnerable groups and expand social gaps. Van Hal et al. [20] present case studies showing how empowerment interventions intended to include people in society may actually have excluding consequences. Burgess, Fu and van Ryn [21] review literature showing that some mothers who are exposed to health education about the harm to themselves and their children because of their smoking, unfortunately experience lower selfworth and increased guilt and sadness. On the brighter side, looking for unintended effects may provide reassurance. Williamson and colleagues [22] searched for and found little evidence of unintended effects of smoke-free ordinances. A few studies are so attuned to the issue of possible unintended effects that their research models explicitly included the possibility of such effects. Figure 1 is from intervention work by Krølner and colleagues [23] that aimed to improve health via increased fruit and vegetable consumption among 13-year-olds, showing hypothesised relationships between intended outcomes (increased intake), side
Unintended effects in settings-based health promotion 21
Figure 1. Hypothesised causal relationships between the Boost Intervention Programme (aimed to increase fruit and vegetable consumption among 13-year-olds), outcomes, side effects and unintended adverse outcomes [23]. Reproduced with permission.
effects (academic achievement) and unintended adverse outcomes (e.g. bullying and eating-related teasing). This attention to the possibility of unintended outcomes is rare, but not unique [17,24-26]. A practical example from the health behaviour research arena is the study of Horigian et al. [27], of a family therapy intervention for the treatment of adolescents with drug abuse. The investigators assembled a team of family therapy experts, medical safety officers and ethicists, who developed principles for defining, and then monitored possible adverse events that participants and their families might experience during the course of the study. Such events might be consequences of the intervention or have other roots, and the investigators were sensitive to the difficulty of assigning causes to the adverse events. As it happened, more than 50% of participating adolescents experienced at least one adverse event, for example arrest, ejection from school or home, and/or violence/abuse,
suicidal behaviour and hospitalization. The takehome lesson for settings-based HP was that the investigtors anticipated adverse events and monitored, observed and reported what happened. Behavioural intervention trials and HP interventions in settings have stark dissimilarities, but one feature they have in common is the rarity with which adverse events are anticipated. The key lesson for HP from the Horigian et al. [27] study, even if it was not a settings study, is the desirability and feasibility of including in the research protocol a real concern for the lives and experiences of participants, including the possibility that a wellmeaning intervention might have untoward consequences. The salience of this concern for HP in settings is poignantly illustrated in the work of Pinhas, et al. [28], in which some children exposed to school-based ‘healthy weights’ initiatives experienced adverse events. They describe the case of:
22 MB. Mittelmark ‘a 14-year-old …student with excellent grades who was perfectionistic and worried tremendously about her marks. […]she was assigned a school project on eating disorders as part of a health and nutrition section of her physical education program. She reviewed common weight loss behaviours found in patients with eating disorders for her project. In doing so, she began to worry about her own weight and whether she was too fat. She became preoccupied with eating disordered thoughts and behaviours that she had documented for her project and found herself trying them out. She began restricting the amount of energy dense foods she consumed and decreasing her total portion sizes. She did not engage in bingeing, purging,or over-exercising as a means to lose weight. Approximately 6 months later she presented to the hospital emergency department as a result of not eating solid foods for 5 days’ [28, pp. 110-11].
Yet another example is the work of Moore et al. [29] in which settings-based HP – smoke-free workplace laws for bars – had the unanticipated outcome that some women who smoked outside experienced threats to their physical safety. How might we promote more explicit attention to the ecology of a HP setting, as in the examples above, leading researchers to concern about the ‘core business’ of the setting in its own right, as well as concern about how intervention might interfere with project goals? One answer is to enhance existing HP planning and evaluation frameworks to include the concept of unintended harm [30], including not only harm to the aims of the intervention project, but also harm to the integrity of the setting. This would provide the means and incentive to theorise about, monitor, prevent and report adverse events and processes, alongside the health promoting events and processes that we aim to achieve. A practical step in that direction is the work of Allen-Scott, Hatfield and McIntyre [30], who are in the process of constructing an empirically founded typology of unintended harm and its underlying factors. The typology articulates five classes of unintended harm and five classes of underlying factors, meaning factors under the ostensible control of the researcher, or at least awareness of which may lead to harm avoidance in the design and implmentation of a HP intervention. Beginning with underlying factors, Allen-Scott, Hatfield and McIntyre [30] identify limited or poor-quality evidence as potentially contributing to harm, citing the example from their empirical material of the lack of sufficient evidence on long-term physical and psychosocial effects of obesity reduction interventions. A second underlying factor supported by their empirical database is the boomerang effect (prevention of one extreme leads to anther extreme), as in the study
already referred to by Pinhas et al. [28], in which an intervention aiming to prevent body weight problems triggered dietary restraint to a dangerous degree. Lack of community engagement is third factor underlying potential harm according to evidence in Pinhas et al.’s [28] database, with minimal or ‘tokenistic’ community involvement associated with stigma, descrimination and environmental contamination. Ignoring root causes is the fourth underlying factor in the harm typology, referring, for example, to the possible stigmatisation of people whose poor health is in theory modifiable by an intervention’s lifestyle change focus, with the intervention ignoring more fundamental causes such as living and social conditions. The fifth underlying factor leading to unintended harm is the fallacy of one size fits all (my label), referring to how a successful intervention developed in one context/culture may cause harm in another context/culture. Allen-Scott, Hatfield and McIntyre [30] focus particular attention on the case of interventions developed in highincome countries that cause unexpected harm in low-income countries. The typology’s underlying factors correlate more or less to five categories of unintended harm: physical, psychosocial, economic, cultural and environmental. Not unexpectedly, the strongest evidence for links between underlying factors and types of harm in Allen-Scott, Hatfield and McIntyre’s [30] empirical material is for the link between the boomerang effects and poor-quality evidence, on the one hand, and individual-level physical and psychosocial harm on the other hand. Harm that would be most manifest at group, community and societal levels – economic, cultural, environmental – is much less studied. This is consistent with the contention raised earlier that when we are concerned with the ecology of unintended effects, it is mostly a concern that our planned intervention effect might be jeopardised. As HP interventions in settings rarely if ever have the main aim to affect economic, cultural or environmental conditions, it is not a surprise that variables in these arenas are scarcely scrutinised. Conclusion This circles back to the core question of this essay: Settings-based health promotion has a wide range of possible consequences, a few of which we have planned, but many of which are unplanned, and may be deleterious to the well-being of the setting.What should we do to prevent, monitor, document and report the full panoply of our effects?
Unintended effects in settings-based health promotion 23 We must first develop more fully our social ecological sensitivity. This will lead us to pose questions like these, in addition to the de facto research question of a settings-based HP study: •• What is the standing pattern of behaviour in the setting prior to intervention? •• How can we develop an intervention that is compatible with and protective of the pre-existing ecology of the setting? •• Once we intervene, how can we monitor our impact to document not only planned, but also unplanned (and especially untoward) events and effects of the intervention on the natural state of the setting? •• How can we react quickly and positively to ameliorate untoward effects, if any are observed? •• What happens to and in the setting long after our project is complete? To routinely introduce questions like these into the planning, implementation and evaluation of settingsbased HP, our intervention frameworks and models need augmentation, for example by the harm typology of Allen-Scott, Hatfield and McIntyre [30], or by emerging variants. Succeeding in this, we will have moved in the direction of a more robust ecological orientation in settings-based HP. Acknowledgement This paper is based in part on a presentation by the author at the 7th Nordic Health Promotion Research Conference, Tønsberg, Norway, 17-19 June, 2013. Conflict of interest The author declares that there is no conflict of interest. Funding This research received no specific grant from any funding agency in the public, commercial, or not-forprofit sectors. References [1] Mittelmark MB, Luepker RV, Jacobs DR, et al. Communitywide prevention of cardiovascular disease: education strategies of the Minnesota Heart Health Program. Prev Med 1986;15:1–17. [2] Weisbrod RR, Pirie PL and Bracht NF. Impact of a community health promotion program on existing organizations: the Minnesota Heart Health Program. Soc Sci Med 1992;34:639–48. [3] Eisenberg L. The human nature of human nature. Science 1972;176:123–8. [4] Pratt PD and Center TD. Biocontrol without borders: the unintended spread of introduced weed biological control agents. BioControl 2012;57:319–29. [5] Green LW, Richard L and Potvin L. Ecological foundation of health promotion. Am J Health Promot 1996;10: 270–81.
[6] Golden SD and Earp JAL. Social ecological approaches to individuals and their contexts: Twenty years of Health Education and Behavior health promotion interventions. Health Educ Behav 2012;39:364–72. [7] McLeroy KR, Bibeau D, Steckler A, et al. An ecological perspective on health promotion programs. Health Educ Quart 1988;15:351–77. [8] Stokols D. Establishing and maintaining healthy environments: toward a social ecology of health promotion. Am Psychol 1992;47:6–22. [9] Richard L, Gauvin L and Raine K. Ecological models revisited: Their use and evolution in health promotion over two decades. Ann Rev Public Health 2011;32:307–26. [10] Poland B and Dooris M. A green and healthy future: the settings approach to building health, equity and sustainability. Crit Public Health 2010;2083:281–98. [11] Whitelaw S, Baxendale A, Bryce C, et al. ‘Settings’ based health promotion: a review. Health Promot Int 2001;16:339–53. [12] Dooris M. Healthy settings: challenges to generating evidence of effectiveness. Health Promot Int 2005;21:55–65. [13] Shareck M, Frohlich KL and Poland B. Reducing social equalities in health through settings-related interventions – a conceptual framework. Glob Health Promot 2013;20: 39–52. [14] Stokols D. Translating social ecology theory into guidelines for community health promotion. Am J Health Promot 1996;10:282–98. [15] Poland B, Krupa G and McCall D. Settings for health promotion: An analytic framework to guide intervention design and implementation. Health Promot Prac 2009;10:505–16. [16] Rychetnik L, Frommer M, Hawe P, et al. Criteria for evaluating evidence on public health interventions. J Epidemiol Comm Health 2002;56:119–27. [17] Kokko S, Green LW and Kannas L. A review of settingsbased health promotion with applications to sports clubs. Health Promot Int. 2014; 29(3): 494–509. [18] Sallis JF, Calfas KJ, Alcaraz JE, et al. Potential mediators of change in a physical activity promotion course for university students: project GRAD. Ann Behav Med 1999;21:149–58. [19] Guttman N and Salmon CT. Guilt, fear, stigma and knowledge gaps: ethical issues in public health communication interventions. Bioethics 2004;18:531–52. [20] van Hal LBE, Meershoek A, Nijhuis F, et al. The ‘empowered client’ in vocational rehabilitation: the excluding impact of inclusive strategies. Health Care Anal 2012;20:213–30. [21] Burgess DJ, Fu SS and van Ryn M. Potential unintended consequences of tobacco-control policies on mothers who smoke. A review of the literature. Am J Prev Med 2009;37:S151-8. [22] Williamson AA, Fox BJ, Creswell PD, et al. An observational study of the secondary effects of a local smoke-free ordinance. Prev Chronic Dis 2011;8:A83. http://www.cdc.gov/ pcd/issues/2011/jul/10_0123.htm. Accessed 7 January 2014. [23] Krølner R, Jørgensen TS, Aarestrup AK, et al. The BOOST study: design of a school- and community-based randomised trial to promote fruit and vegetable consumption among teenagers. BMC Public Health 2012. Epub DOI:10.1186/1471–2458–12–191. [24] Dunton GF, Intille SS, Wolch J, et al. Investigating the impact of a smart growth community on the contexts of children’s physical activity using Ecological Momentary Assessment. Health & Place 2012;18:76–84. [25] Busza J, Walker D, Hairston A, et al. Community-based approaches for prevention of others to child transmission in resource-poor settings. A social ecological review. J Int AIDS Soc 2012;15(Suppl 2). DOI: 10.7448/IAS.15.4.17373. [26] Moore L, de Silva-Sanigorski A and Moore SN. A socio-ecological perspective on behavioural interven tions to influence food choice in schools: alternative,
24 MB. Mittelmark complementary and synergistic? Public Health Nutr 2013;1000– 5. Epub 4 March 2013. DOI: 10.1017/S1368980012005605. [27] Horigian VE, Robbins MS, Dominguez R, et al. Principles for defining adverse effects in behavioral intervention research: lessons from a family-focused adolescent drug abuse trial. Clin Trials 2010;7:58–68. [28] Pinhas L, Walker GM, Walker KS, et al. Trading health for a healthy weight: the uncharted side of healthy weights initiatives. Eat Disord 2013;21:109–16.
[29] Moore RS, Annechino RM and Lee JP. Unintended consequences of smoke-free bar policies for low-SES women in three California counties. Am J Prev Med 2009;37: S138-43. [30] Allen-Scott LK, Hatfield JM and McIntyre L. A scoping review of unintended harms associated with public health interventions. Towards a typology and an understanding of underlying factors. Int J Public Health. Epub 1 January 2014. DOI 10.1007/s00038–013–0526–6.
![[Health promotion: Mental health promotion].](/assets/img/hold.png)
