Poor dietary intake is the single most important preventable behavioural risk factor affecting health globally^{( 1 )}. In high-income countries the burden of disease due to poor dietary intake is at least 14 % of the total disease burden, which is about double that due to cigarette smoking^{( 1 , 2 )}. There is evidence that a range of nutrition strategies can effectively reduce diet-related disease, including individual-focused ‘downstream’ approaches and more ‘upstream’ population-based approaches. Downstream interventions often focus on adverse health behaviours through direct dietary interventions and require active participation by individuals⁽ Reference Lytle and Fulkerson ^{3 )}; upstream interventions focus on the wider circumstances that produce the adverse health behaviours. Intervening on the wider circumstances can be achieved through implementing policies that improve the social or physical food environment; for example, improving the availability and quality of healthy food choices, decreasing the cost of healthy food choices or changing the socio-cultural norms around food through public awareness campaigns⁽ Reference Raine ^{4 )}. Regulatory actions from governments combined with joint efforts from industry and society are often required to implement these policies in order to create healthier food systems⁽ Reference Roberto, Swinburn and Hawkes ^{5 )}. There is increasing evidence that upstream population approaches are more cost-effective and can provide larger savings to the health-care system than downstream individual approaches⁽ Reference Sacks, Veerman and Moodie ^{6 – 8 )}. However, despite this, there has been limited progress towards public policy action in many countries⁽ Reference Johnson, Cheadle and Podrabsky ^{9 ,} Reference Swinburn and Wood ^{10 )}.

Public policy is defined as ‘a program of action (or inaction) of a government to achieve specific goals’⁽ Reference Clavier and De Leeuw ^{11 )}; it is the bedrock of the political process and is essential in directing investment⁽ Reference Palmer and Short ^{12 )}. Public policy making is rarely a linear process and is informed by interactions between institutions, politicians, advisers, interest groups, the public service plus a range of other participants⁽ Reference Bridgman and Davis ^{13 )}. Current public policy to address nutrition issues in high-income, democratic countries is inconsistent⁽ Reference Swinburn and Wood ^{10 ,} Reference Vogel ^{14 )}. Numerous hypotheses have been given for this inconsistency including the power of vested interests, a lack of evidence and that tackling nutrition issues is too expensive and complex⁽ Reference Swinburn and Wood ^{10 ,} Reference Vogel ^{14 ,} Reference Corvalán, Reyes and Garmendia ^{15 )}. To understand why nutrition is not addressed more fully in public policy, the policy process needs to be better elucidated in terms of how it is influenced and by whom. This does not mean a focus on the technical content and design of policies; rather it is the analysis of specific aspects of the policy process. These include how ideas, knowledge, interests, power and institutions influence the policy decision-making process⁽ Reference Ham and Hill ^{16 )}. In analysing the policy process, stakeholders may be able to use the findings to more effectively leverage for policy change.

Any discussion of the policy-making process needs to consider the nature of power. Power can be defined as control over outcomes, and this can occur through control over resources and/or control over actors⁽ Reference Dahl ^{17 )}. The role of power in policy making has been explored by political scientists since the 1950s, with Lasswell and Kaplan noting ‘the political process is the shaping, distribution, and exercise of power’⁽ Reference Lasswell and Kaplan ^{18 )}. Public policies, it is argued, reflect the values of those with the greatest influence and are imbued with how those groups perceive the world to be, or ought to be⁽ Reference Altensetter ^{19 )}. Some believe that health policy making, like policy in other sectors, rests on the accumulation and use of power by those involved in the policy-making process⁽ Reference Lewis ^{20 )}. One way to increase one’s power in the policy-making process is through effective advocacy. Advocacy is defined as ‘active interventions by organisations on behalf of the collective interests they represent, that has the explicit goal of influencing public policy or the decisions of any institutional elite’⁽ Reference Onyx, Dalton and Melville ^{21 )}.

Advocating for food and nutrition policy and government support to protect and promote health has been identified internationally as a core function for the public health nutrition workforce⁽ Reference Hughes ^{22 )}. However, research has shown that the current workforce is largely disengaged from the advocacy component of the policy-making process and is unprepared for this type of work⁽ Reference Hughes ^{23 )}. Nutrition practitioners and researchers often embrace policy advocacy with the belief that a scientific approach is all that is required; that is, providing high-level scientific evidence is enough to influence policy decisions. However, expecting a linear connection between scientific results and policy making reveals a misunderstanding of the political nature of policy environments⁽ Reference Walt and Gilson ^{24 ,} Reference Bambra ^{25 )}. While evidence is an important ingredient, it is only one component in the inherently political policy-making process. Assuming that policy making is a linear, rational process ignores the broader influences of political ideologies, institutions and powerful vested interests in shaping and selecting the ‘knowledge’ provided in the formulation of public policy⁽ Reference Bernier and Clavier ^{26 ,} Reference Bryant ^{27 )}. By relying solely on scientific evidence as a platform for policy change and not acknowledging or understanding the political dimension of policy making, public health nutrition practitioners and researchers are unlikely to fully appreciate why desired policy outcomes fail to emerge and therefore why nutrition is often not addressed at a broad population level beyond the individual.

Conversely, the systematic study of the policy-making process has occurred in the field of political science since the 1950s, where many theories have been developed to explain the various components, including politics, process and power⁽ Reference Sabatier ^{28 )}. Applying political science policy process theories (PPT) retrospectively allows a greater understanding of why particular policy decisions were made, which can help advocates understand the complexities of the policy-making process as well as identify key actors and leverage points for power and influence in the process. An understanding of PPT can also aid in predicting the outcomes of ongoing or future initiatives⁽ Reference Lapping, Frongillo and Studdert ^{29 ,} Reference Blanchet and Fox ^{30 )}. Three commonly used PPT that have been validated empirically in multiple countries to explain and predict the policy-making process include the Advocacy Coalition Framework⁽ Reference Jenkins-Smith and Sabatier ^{31 )}, the Punctuated-Equilibrium Theory⁽ Reference Baumgartner and Jones ^{32 )} and the Multiples Streams Theory⁽ Reference Kingdon ^{33 )} (see Table 1 for a summary)⁽ Reference Sabatier ^{28 )}. These PPT have different components; they are, however, all theories of policy change which focus on agenda setting and policy adoption. Despite this raft of knowledge, very few public health experts utilise these empirically based PPT when analysing or trying to influence the policy-making process or the content of policies⁽ Reference Breton and De Leeuw ^{34 –} Reference Gilson and Raphaely ^{36 )}. Publications in the public health policy area tend to focus on describing policy content and evaluating policy impact and outcomes, paying little attention to the policy-making process⁽ Reference Bernier and Clavier ^{26 ,} Reference Macintyre ^{37 ,} Reference Catford ^{38 )}.

Table 1 Summary of influential theories of the policy-making process

A review of peer-reviewed papers reporting on policy process or content in 2010 found only 18 % applied a PPT⁽ Reference Breton and De Leeuw ^{34 )}. A systematic literature review of the health policy literature in low- and middle-income countries between 1994 and 2007 found that health policy analysis was in its infancy with the majority of studies being descriptive and very few utilising PPT⁽ Reference Gilson and Raphaely ^{36 )}. While these reviews have been conducted across health promotion and public health generally, to date, no systematic review of public health nutrition policy papers has been undertaken.

While reviews of health promotion and public health literature can provide general insights, public health nutrition policy is a particularly unique field. Unlike cigarettes or pathogens, the healthfulness of particular foods can be open to individual interpretation. Adding to this, choosing food is a necessary, but complex process. This complexity is due to food choices being deeply embedded in culture and influenced by many factors internal and external to the person, including biological, social, psychological, economic and environmental⁽ Reference Drewnowski and Darmon ^{39 –} Reference Burns, Bentley and Thornton ^{42 )}. Macro-level environmental factors, including social norms, economic price structures, food marketing, agricultural policies, and food production and distribution, play an indirect role but have a substantial and powerful effect on what people eat⁽ Reference Story, Kaphingst and Robinson-O’Brien ^{43 )}. Successfully impacting on these factors will often require a structural or political response. The complexity of the issue is further exacerbated by the many different voices of interest groups. These include not only those with obvious vested (social or financial) interests such as public health groups and food industry groups, but also ‘alternative’ health practitioners as well as celebrities and even actors advocating dietary recommendations. To advance our understanding of the unique field of public health nutrition policy, we conducted a systematic review of peer-reviewed literature to determine whether PPT is utilised to guide analysis in the literature.

Methods

The systematic review was conducted based on the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement and a review protocol used by Breton and De Leeuw⁽ Reference Breton and De Leeuw ^{34 )} that examined the use of PPT in health promotion, refined to focus on food and nutrition policy. This review included English-language nutrition policy literature published between 1986 and December 2014 (Fig. 1). The start date was chosen as this was the year the Ottawa Charter^{( 44 )}, which recommended a focus on healthy public policy as an effective strategy of health promotion, was released.

Fig. 1 PRISMA 2009 flow diagram

Five electronic databases (PubMed, Informit, CINAHL, PsycINFO and Scopus) were searched using the following keywords: ‘(politic* OR advoc* OR coalition OR influenc*) AND (nutri* OR food OR obesity) AND (polic*) AND (government)’. To limit settings-based approaches, on the two occasions when more than 500 papers were retrieved from a search, the terms ‘AND NOT agriculture*’ were included. Reference lists of relevant articles were also searched for eligible papers. Multiple papers referring to the same study contributed only once to the review.

The studies from the initial search (n 1932) were reviewed against the inclusion criteria (see Table 2). Papers were excluded if they were not primarily concerned with the food or nutrition policy-making process or policy content. In addition, papers were excluded if they recommended developing a new policy but provided no detail on how that should occur or if they only examined the views of the community about general policy directions. The application of the criteria resulted in 147 eligible papers.

Table 2 Inclusion and exclusion criteria for the identification of journal articles reporting on policy research

All 147 papers were fully read by the first author and assessed as to whether insights from any empirically tested PPT had guided the policy research or theoretical reflections they reported, consistent with the method in Breton and De Leeuw⁽ Reference Breton and De Leeuw ^{34 )}. When there was uncertainty as to whether a paper fit the inclusion criteria, it was reviewed by a second independent reviewer. Disagreements between reviewers were resolved by discussion until consensus was reached. A second screening was undertaken to exclude all papers concerning low-income and/or partial or non-democratic countries as defined by the Democracy Index 2014 (n 19), as it was felt these settings were too different from the health policy environments and systems of government of prime interest/relevance in the present study^{( 45 )}.

The content of papers was scanned for topic areas, year of publication, use of PPT, country of focus and level of government, then results were tallied and depicted graphically (Microsoft^® Excel version 2010). A data visualisation technique known as ebb-and-flow analysis was conducted to examine graphically the thematic changes over time. This enables readers to identify emerging, dominant, enduring and transient themes within a field of literature, and in the case of the present review, highlight the evolution of a field of research as new areas of inquiry are developed and gain traction over time⁽ Reference Havre, Hetzler and Whitney ^{46 )}.

The ebb-and-flow analysis was created through a process of thematically coding articles to create a database that enumerated the number of articles published per year for each theme identified (that is, for example, obesity, childhood obesity, food security). This was transformed into a normalised data set (percentage of a theme’s publications compared with the total volume of publications for a given time period; that is, for example, every year, 2 years, 5 years) and graphed in Microsoft Excel version 2010 using the ‘100 % Stacked Area’ chart function. A line graph representing the volume of publications per time period was overlain to provide additional context to the data.

Findings

After the full-text review, sixty-three studies were selected for inclusion. The studies focused on a variety of countries, settings and nutrition policy issues. Most studies examined single cases of nutrition policy in one country although eleven studies included multiple country comparisons or data. The number of nutrition policy papers per annum increased over time, with a greater concentration occurring post 2007 (see Fig. 2).

Fig. 2 Frequency of eligible nutrition policy articles (Jan 1986–Dec 2015)

The majority of papers had a geographical focus on the USA (n 31, 49 % of all papers); this accounted for more than double the papers from the next most common geographical focus, Australia (n 12, 19 %). The remaining countries or regions represented included the UK, Japan, New Zealand, Canada, Norway, Finland and the European Union. A large number of papers (n 29, 46 %) focused their attention on national nutrition policy. State-focused nutrition policy, then local- and global-focused nutrition policy were the next most common (respectively: n 14, 22 %; n 11, 17 %; n 8, 13 %).

The topics covered in the nutrition policy papers were wide-ranging (see Fig. 3). Thirty-eight per cent of the papers focused on obesity, divided into general obesity (n 15, 24 %) and childhood obesity (n 9, 14 %). Other more common topics were interest group influence (n 14, 22 %), local government food policy (n 6, 9 %) and the process of developing nutrition policy (n 6, 9 %). The popularity of these topics has changed over time as can be seen in the ebb-and-flow analysis (Fig. 4). Earlier papers focused more on general policy content and implementation as well as school nutrition policy, whereas more recently there has been a stronger focus on obesity and the influence of interest groups in policy making. The ebb-and-flow analysis also illustrates the growing field of nutrition policy research and captures the fragmentation of the field as a wide variety of nutrition issues have become more salient over time.

Fig. 3 Nutrition policy topics (F&V, fruit and vegetables; SSB, sugar-sweetened beverages; reg, regulations; TV, television; govt, government; movt, movement)

Fig. 4 Ebb-and-flow analysis of nutrition policy topics (SSB, sugar-sweetened beverages)

Use of political science policy process theories

There has been an increase in the use of PPT in the reviewed nutrition policy papers in recent years (Fig. 5). However, this increase is fairly small with only 14 % (n 9) of all the nutrition policy papers reviewed using empirically tested PPT. In 2014 there appeared to be a reversal of the slight upward trend, with none of the nutrition policy papers utilising PPT to analyse the described policy process.

Fig. 5 Comparison of nutrition policy papers published () and those that use political science policy process theory ()

While papers were reviewed for all empirically tested PPT, the only PPT utilised included Multiple Streams Theory (five papers)⁽ Reference Caraher, Carey and Mcconell ^{47 –} Reference Yeatman ^{51 )}, the Advocacy Coalition Framework (three papers)⁽ Reference Johnson, Cheadle and Podrabsky ^{9 ,} Reference Hobbs, Ricketts and Dodds ^{52 ,} Reference Johnson, Payne and Mcneese ^{53 )} and one study used both Multiple Streams Theory and the Advocacy Coalition Framework⁽ Reference Gilson Sistrom ^{54 )}. Five papers referred to PPT in their introduction or discussion but it was unclear whether they utilised theory in analysing their data⁽ Reference Allender, Gleeson and Crammond ^{55 –} Reference Dodson, Fleming and Boehmer ^{59 )}. Theory was used to retrospectively analyse a case study/studies in all but one of these papers. Gilson Sistrom⁽ Reference Gilson Sistrom ^{54 )} utilised PPT prospectively, with the application of Multiple Streams Theory and the Advocacy Coalition Framework to guide advocacy around regulating ‘junk food’ in schools.

Seven papers (13 %) referred to theories other than PPT. The theories used included: Framing Theory⁽ Reference Shelley ^{60 )}, which looks at one component of the policy process; and the Health Belief Model⁽ Reference Freudenberg ^{61 )} and Stages of Change (Transtheoretical Model)⁽ Reference Swinburn and Wood ^{10 )}, both of which are theories related to individual behaviour change rather than policy or system change. Three authors developed their own theories⁽ Reference Field, Gauld and Lawrence ^{62 –} Reference Miller and Harkins ^{64 )} and one paper used Walt and Gilson’s Health Policy Triangle which is a framework that identifies the elements and relationships of the policy process. This framework is a useful starting point for health policy analysts; however, the Health Policy Triangle does not explain or predict behaviour and outcomes as do the PPT referred to earlier⁽ Reference Breton, Richard and Gagnon ^{66 )}.