Abstract

Non-communicable disease prevention strategies usually target the four major risk factors of poor diet, tobacco, alcohol and physical inactivity. Yet, the most effective approaches remain disputed. However, increasing evidence supports the concept of an effectiveness hierarchy. Thus, ‘downstream’ preventive activities targeting individuals (such as 1:1 personal advice, health education, ‘nudge’ or primary prevention medications) consistently achieve a smaller population health impact than interventions aimed further ‘upstream’ (for instance, smoke-free legislation, alcohol minimum pricing or regulations eliminating dietary transfats). These comprehensive, policy-based interventions reach all parts of the population and do not depend on a sustained ‘agentic’ individual response. They thus tend to be more effective, more rapid, more equitable and also cost-saving. This effectiveness hierarchy is self-evident to many professionals working in public health. Previously neglected in the wider world, this effectiveness hierarchy now needs to be acknowledged by policy makers.

Background

This perspective briefly summarizes the growing evidence for a public health ‘effectiveness hierarchy’, and examines the policy implications for future preventive health strategies.

The global burden of disease and disability is now mainly caused by non-communicable diseases (NCDs), notably heart disease, stroke, dementia, diabetes and cancer.^1,2

NCD prevention strategies are now prioritizing four major risk factors: tobacco, poor diet, physical inactivity and alcohol.^1–3 However, there is debate about the most effective approaches to prevention. Many countries have prioritized ‘downstream’ approaches targeting individuals (such as screening to detect high risk patients, personal advice, primary prevention medications and ‘nudge’).³ These highly visible strategies are politically less challenging than ‘upstream’ population-wide policy interventions (such as legislation, regulation, taxation or subsidies.^1–3) However, the growing effectiveness evidence clearly points ‘upstream’.⁴

(The apocryphal story illustrating the ‘upstream/downstream’ metaphor is summarized in Box 1.)

Effectiveness hierarchy: neglected paradigm or self-evident truth?

Most public health practitioners intuitively accept an effectiveness hierarchy as an obvious, self-evident truth.^5–8 Thus Frieden proposed a five tier ‘Health Impact Pyramid’ in 2010.⁹ This explicitly suggests that ‘upstream’ interventions addressing socioeconomic determinants of health have the greatest potential population impact, and counselling and health education the least.⁹ Some enlightened clinicians have also highlighted the potentially large role of policies to prevent cardiovascular disease^8,10 or cancer,¹¹ and advocated using ‘the longest lever possible’.¹²

However, in the wider world, most politicians and policy makers, pundits and ordinary people appear unaware or indifferent to this effectiveness hierarchy. They thus risk over-looking the best strategies for maximizing the future health of their families and friends.^4,7

We therefore now propose to briefly review the evidence for this ‘effectiveness hierarchy’.

Evidence supporting the effectiveness hierarchy

The evidence supporting a hierarchy of effectiveness now appears relatively extensive for all four of the major NCD risk factors: tobacco, diet, physical inactivity and alcohol.^1,2 This evidence is briefly summarized below.

Tobacco control

Systematic reviews of tobacco control strategies consistently suggest that larger scale, comprehensive population-based approaches are more effective than individual approaches or local community strategies.^13–15 The relative power of these different tobacco control interventions has been usefully summarized and quantified by a variety of scales including the US Tobacco Control Index and the European Tobacco Control Scale (TCS).¹⁶ The TCS illustrates a clear effectiveness hierarchy for tobacco control interventions. From a total of 100 points, 30 points are allocated to tobacco price, 22 points to comprehensive smoke-free legislation (but only 10 points if limited to workplaces), 10 points for health warnings on cigarette packs, six points for nationwide cessation services for individuals and only two points for individual patients accessing telephone quitline advice¹⁶ (Fig. 1).

Fig. 1

Tobacco Control Score (TCS): estimated effects of different policy options.

Fig. 1

Tobacco Control Score (TCS): estimated effects of different policy options.

Poor diet

A similar effectiveness hierarchy is becoming apparent in diverse interventions to improve diet. Our BMJ Analysis summarized the growing evidence for reducing dietary salt consumption in populations.¹⁷ A gradient was clearly apparent. Advice to individuals or social marketing was generally weak. The UK programme of sustained pressure for industry reformulation reinforced by media messaging was more powerful. Furthermore, comprehensive ‘upstream’ strategies including regulation and marketing control (as in Finland and Japan) have been yet more powerful, achieving even greater reductions in daily salt consumption^17,18 (Fig. 2).

Fig. 2

Dietary salt reduction: estimated effects of different policy options.

Fig. 2

Dietary salt reduction: estimated effects of different policy options.

Similar hierarchies of effectiveness are apparent for interventions to reduce the dietary intake of industrial transfats, saturated fats and sugars, and also for interventions to increase the consumption of fresh fruit and vegetables. For instance, the biggest reductions in industrial transfat intake have been seen in Denmark (and soon perhaps in the USA), as the consequence of progressive and comprehensive policy interventions culminating in legislation^19,20 (Fig. 3).

Fig. 3

Dietary industrial transfat reduction: estimated effects of different policy options.

Fig. 3

Dietary industrial transfat reduction: estimated effects of different policy options.

Likewise, the largest country-wide reductions in saturated fat intake were achieved in Finland (reflecting comprehensive and sustained strategies to progressively reduce the production and consumption of animal and dairy fats), and in Mauritius (following a regulation banning palm oil imports). Mean blood cholesterol levels subsequently fell dramatically, by 1.0 and 0.8 mmol/l, respectively.^21,22

Physical activity/inactivity

Recent NICE reviews have summarized the growing evidence demonstrating a similar effectiveness hierarchy in individuals and in populations (Fig. 4). First prize goes to Cuba’s 37% increase in physically active adults because of active travel, this being a rapid and unintended consequence of the major economic crisis commencing in 1989.²³ As GDP plummeted by 80% and fuel became scarce, the government needed to dramatically reduce private transport, promote public transport and also distribute a million bicycles.²³

Fig. 4

Interventions increasing physical activity: estimated effects of different policy options.

Fig. 4

Interventions increasing physical activity: estimated effects of different policy options.

Between 1972 and 2002, Finland implemented comprehensive national policies promoting walking, cycling and leisure activities. These policies substantially increased the proportion of physically active adults, by ~27% in women and by some 11% in men.²⁴ In contrast, more limited transport and exercise policies further ‘downstream’, typically achieve smaller increases in physical activity, such as media campaigns, or time-limited active travel schemes promoting walking and cycling. In general, least has been achieved by targeting individuals for advice, information leaflets or exercise prescription.^25,26 Furthermore, those modest benefits typically then diminish over time^26,27 (Fig. 4).

Alcohol control and cost-effectiveness

Anderson and colleagues recently reviewed the growing evidence for reducing alcohol consumption in populations.²⁸ This provided considerable additional support for a hierarchy of effectiveness and, crucially, also of cost-effectiveness. Thus, brief interventions advising individuals cost approximately $2 700 per disability adjusted life year (DALY) saved. Interventions targeted further ‘upstream’ become increasingly cost-effective, with a 50% excise tax increase costing much less, only $330 per DALY saved.²⁸

Wider cost-effectiveness: an emerging hierarchy of interventions

Health economists with NICE have identified increasing evidence of a similar hierarchy in the effectiveness and cost-effectiveness of diverse preventive interventions. Most are now easily compared using the standard NICE metric: cost per QALY. Owen and colleagues recently reviewed the cost-effectiveness estimates for some 200 public health interventions, most being relatively ‘downstream’.²⁹ Thus, preventive interventions in individuals requiring statin medication often cost many thousands of pounds per QALY, while individual advice on behaviour changes, (for instance, exercise on prescription or mass media campaigns promoting healthy eating) usually only cost a few hundred pounds per QALY.²⁹

In contrast, ‘upstream’ population-wide policy interventions are generally cost-saving, such as regulations to control tobacco or alcohol, or to reduce dietary salt or transfats; likewise subsidies to promote fresh fruit.³⁰ Reassuringly similar results come from economic analyses in the UK, USA and Australia.^31,32 The systematic review by Masters et al.³¹ likewise recently observed an effectiveness hierarchy in over 50 diverse public health interventions. Interventions implemented locally typically gained a return on investment averaging £4 for every pound spent. ‘Upstream’ nation-wide public health interventions typically demonstrated even larger benefits with a median return on investment of £27 for each pound invested.³¹

Some interventions may even be cost-positive, i.e. revenue raising. Taxes on soda, and tobacco, e.g. could actually serve to fund targeted interventions for those individuals in most need, and thus potentially improve health equity.

Secondary benefits of prevention

Many of the same interventions designed to prevent unhealthy behaviours can also serve to ‘treat’ those who partake in behaviours which undermine their health. Taxes may prevention initiation of cigarettes, but also may increase/improve quit attempts for price sensitive smokers. Likewise, banning the sale of tobacco products in pharmacies also decreases overall tobacco retailer density, thus discouraging smoking in adolescents and potentially decreasing relapse in smokers trying to quit.^33,34 Furthermore, evidence to support this effectiveness hierarchy paradigm is also steadily emerging in other arenas including climate change, road safety, crime prevention and social policy⁴ (Box 2).

Discussion

Main findings of this study

Increasing evidence supports the concept of an effectiveness hierarchy. Thus, ‘downstream’ preventive activities targeting individuals (such as 1:1 personal advice, health education, ‘nudge’ or primary prevention medications) consistently achieve a smaller health impact than interventions aimed further ‘upstream’ (for instance, smoke-free legislation, alcohol minimum pricing or regulations eliminating dietary transfats). These comprehensive, policy-based interventions reach all parts of the population and do not depend on a sustained ‘agentic’ individual response. They thus tend to be more effective, more equitable and also cost-saving.

The proposed concept of an effectiveness hierarchy raises several important issues. These include the underlying theoretical and ethical frameworks, equity and durability. Also crucial are the challenges of overcoming vested interests, wider political feasibility and operationalizing the policy evidence into interventions which are then effectively implemented.

What is already known on this topic

Geoffrey Rose famously demonstrated the simple mathematical principle that a small improvement in the whole population (for instance dietary salt reduction to modestly lower average blood pressure) consistently generates larger net benefits than more intensive interventions just targeting those fewer individuals at higher risk (for instance using medications for hypertension).³⁵ (Individual approaches typically have higher delta, but on far fewer people and with a higher per-person expense).

Dahlgren and Whitehead’s³⁶ ‘layers of influence’ rainbow model then helped to better conceptualize the ‘upstream’ political, economic, cultural and environmental influencing factors through to ‘downstream’ factors acting directly on communities and individuals.

More recently, McLaren and colleagues have usefully observed that all preventive interventions sit on a ‘structural/agentic continuum’: starting from fiscal or legislative actions which change the environment (‘structural’ interventions) through to information leaflets and advice which are completely ‘agentic’, being totally dependent on an individual’s active response.³⁷

The latter point being recently emphasized by Adams et al.³⁸

The Nuffield Bioethics review suggested that public health interventions should be proportional to the hazard.³⁷ Their useful Ladder of Interventions specified actions escalating from ‘doing nothing’ (for a minimal risk), through progressive restrictions of an individual’s freedom of choice and culminating in the ‘elimination of choice’ (for instance Danish regulations to eradicate industrial transfats from food).¹⁹

Equity effects

Individual preventive approaches depend mainly on agency—a person’s active response.^35,36 Such interventions therefore tend to favour affluent and educated groups, hence potentially increasing inequalities.^35,36 Conversely, ‘structural’ population-wide approaches which make the environment healthier generally benefit all individuals.^35,36 They may thus sometimes narrow the inequalities gap.^36–40,42

Durability and sustainability issues

Interventions attempting behaviour change in individuals typically diminish over time: smokers relapse, statin adherence decreases, healthy diets drift and joggers give up.^25–27 In contrast, the benefits of legislation tend to be durable (reflecting the persistent effect of an intervention, removing the need to reapply it), and also self-sustaining Thus, once smoking is eliminated from bars, or arsenic or industrial transfats eliminated from food, these health hazards are seldom permitted to return.

Implementation can also become more effective over time, as illustrated by seatbelt legislation or smoke-free environments: political resistance fades, younger cohorts inherit the new ‘social norms’ and public support actually increases.^4,6,38

Some interventions are sustainable over time without the need for additional resources or upkeep. However, others like water fluoridation will require upkeep. However that cost is modest when compared with treating individuals with dental decay.

Political feasibility and ‘upstream’ approaches

Translating evidence in to policy is neither simple nor linear. Decision making by planners and policy makers is complex, reflecting many factors often considered more powerful than scientific evidence.^44–46 Many hurdles exist, because the political and scientific arenas differ in terms of their perspectives, aims, values and practices. Neither can legislation offer a ‘quick fix’. In liberal democracies, successful laws generally only follow extensive public debate and growing support.^18,33

Happily, there is a long history of governments legislating to protect their citizens’ health. These regulatory successes include clean water, sanitation, air pollution, immunization, seatbelts and smoke-free statutes.⁴⁷ Such legislative public health proposals are typically supported by the political centre and left; but often initially opposed by the libertarian right and commercial interests.⁴⁷

Framing is also crucially important. Thus, smoke-free policies were framed to protect non-smokers, where public smoking infringed on their rights. Likewise strategies framed to protect youth, e.g. from being exposed to marketing of tobacco products or junk food.

There are also many political levers to use, from taxes or zoning (to prevent tobacco and alcohol retailers from locating near schools), to licensing (alcohol retailers, tobacco retailers), or frank prohibition of signage or product usage.

Applying the logic of an effectiveness hierarchy would clearly favour regulation and fiscal interventions at the highest levels—national, regional and global. The Framework Convention for Tobacco Control (FCTC) represents a notable global success now signed by over 160 countries.⁴³ Similar approaches have therefore been subsequently suggested to control other harmful products such as alcohol, dietary transfats or sugars.^48,49 The recent UN High Level Meeting on NCD prevention and control was thus potentially powerful, with the WHO subsequently agreeing a ‘25 × 25’ target: a 25% reduction in NCD deaths by the year 2025.^1,2

However, many hurdles remain. The FCTC was only achieved in 2005, 5 decades after clear scientific evidence of tobacco harm, and only after 2 decades of sustained advocacy and activism.⁴¹ In reality, any such concerted global actions will be predictably and energetically opposed and obstructed by commercial vested interests.^48–54 As suppliers of unhealthy commodities, ‘Big Tobacco’, ‘Big Alcohol’ and ‘Big Food’ companies obviously prioritize profit, not public health.^48–56 Furthermore, these ‘disease promoting industries’ use remarkably similar tactics intended to delay, dilute or demolish effective regulation.^48–56 For instance, claiming that ‘nanny state’ policies erode individual autonomy or harm economic growth.^47,49

Similarly obstructive commercial lobbying has recently been seen at the US federal level, frustrating public health regulation to combat child obesity.^48–50 In contrast, over 25 individual states and US metropolitan areas have successfully implemented ‘soda taxes’ on sugary drinks, most recently Berkeley, San Francisco, Philadelphia, Oakland, Boulder and Chicago.⁵⁴ Similarly, over 30 cities and states have now enacted protective legislation on smoke-free public spaces.^55,56,57

Similarly in Europe, public health progress has often been similarly sabotaged at the European Union level. For instance, the recent Tobacco Products Directive was systematically weakened by sustained lobbying, and the EU proposal for front-of-pack food labelling to effectively inform consumers was derailed by massive food industry opposition.^55,56 Happily, however, individual European member countries have been politically more agile, (much like leading US cities and states^57,58). They have successfully achieved effective regulation and taxation to control tobacco,^15,16 alcohol,⁵¹ transfats,^19,20 salt^17,18 and sugar.^53,54

The failure of ‘nudge’ and partnership approaches

In stark contrast to healthy policies, non-regulatory voluntary agreements and ‘partnerships’ with industry have consistently proven weak or ineffective, not least by demoting upstream issues. For example, ‘responsibility deals’ which depend on ‘nudge’ and non-specific ‘pledges’ from industry have received increasingly devastating criticism in the UK, Europe and the USA^59–64.

Limitations of this study

This brief review has many limitations. Firstly, the scientific evidence still remains relatively sparse, and might be selectively quoted. However, recently emerging results are supportive, notably recent systematic reviews of policies to reduce the dietary intake of salt, and transfats,^18,20 and the economic return on investment of diverse public health interventions.³¹ A more comprehensive programme of primary research and systematic reviews should therefore now formally test the effectiveness hierarchy paradigm in a wider range of specific areas.⁶⁵ Looking beyond the classical four NCD risk factors will also be important and potentially exciting (Box 2).⁴

Secondly, demanding the medical ‘gold-standard’ of randomized trials is easy when assessing patient therapies, but seldom feasible for evaluating upstream national policy interventions. Thus most policy effect sizes represent estimates based on analyses of natural experiments interrupted time-series or observational cohorts.^66,67 Furthermore, these effect sizes have been quantified using a variety of outcome metrics. This heterogeneity limits comparisons of effectiveness and highlights the value of promoting a standard methodology, such as costs per QALY or DALY.^26,29

Thirdly, needing to acknowledge the potential for unintended consequences of population wide interventions. Thus, if population wide policies are enduring, sustainable and wide reaching, they should be fully evaluated for unintended effects prior to general implementation. For example, the 1920s US prohibition of alcohol fuelling crime, or media campaigns on obesity resulting in shaming, bullying, stigma or negative mental health consequences.

Fourthly, considering any policy intervention in isolation is slightly artificial. Preventive interventions in the messy real world are delivered in a wider social, cultural and economic context which may be supportive, neutral or obstructive. Thus, in tobacco control, combinations of interventions within comprehensive strategies can produce additive or even synergistic benefits.⁶⁸ Therefore ‘upstream’ interventions banning advertising typically create a more favourable environment which will then support the individual advised by their doctor to stop smoking. A ‘systems approach’ involving a comprehensive, multi-level, multi-sectorial strategy may thus achieve ‘the maximum possible sustained public health gains’.^{6–9,18,20,66,68,69} Furthermore, benefits which can often occur surprisingly rapidly, within months rather than decades.⁷⁰

The public and policy implications of a preventive effectiveness hierarchy

As the emerging evidence strengthens, so the implications become increasingly clear.

1. Future strategies proposed for NCD prevention should strive to prioritize ‘upstream’ approaches using regulation, taxation, subsidies and comprehensive approaches. Policy makers and politicians will therefore need to champion these ‘upstream’ interventions at the highest international levels (including the World Bank, United Nations, World Health Organization, World Trade Organization and Codex Alimentarius).
2. Interventions further ‘downstream’ and closer to the individual generally become progressively weaker and more expensive. They therefore merit greater scepticism from planners and policy makers.
3. High income countries can fudge the choice between population-wide and individual approaches to prevention. They can afford both, and might therefore portray these strategies as ‘complementary’. However, low and middle income countries face limited budgets and starker choices. They cannot quite so easily dismiss the emerging evidence on effectiveness and cost-effectiveness summarized here.
4. Opposition by commercial vested interests might be anticipated. Because corporations will routinely resist any proposed public health regulation which might threaten their profits. However, the health of the public tends to triumph, eventually.⁴⁷

In conclusion, most policy makers and planners are facing growing disease burdens and shrinking healthcare budgets. Thus when considering future prevention strategies, they will increasingly need to prioritize those upstream policies which would most benefit their entire populations. They can no longer afford to neglect the public health effectiveness hierarchy.

Supplementary data

Supplementary data are available at Journal of Public Health online.

Acknowledgements

We thank many friends for their constructive suggestions and support, notably Martin McKee, Martin O’Flaherty, Margaret Whitehead and colleagues in the University of Liverpool; and also anonymous referees at JPH and a prior journal.

Contributors and sources

This article summarizes analysis and prolonged reflection, refined by extensive discussions with the colleagues thanked above. The evidence has been progressively gathered by a combination of scoping reviews and more detailed searches. A series of well resourced systematic reviews are now proposed as the logical next stage in the process.

Conflicts of interest

None.

References

© The Author 2017. Published by Oxford University Press on behalf of Faculty of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com