- Theresa M Marteau, director of research1,
- Gareth J Hollands, principal research fellow12,
- Rachel Pechey, Wellcome Trust and Royal Society Sir Henry Dale fellow3,
- James P Reynolds, lecturer in psychology4,
- Susan A Jebb, professor of diet and population health3
1University of Cambridge, Department of Public Health and Primary Care, Cambridge CB2 0SR, UK
2University College London, EPPI-Centre, UCL Social Research Institute, London WC1H 0NR, UK
3University of Oxford, Nuffield Department of Primary Care Health Sciences, Radcliffe Primary Care Building, Woodstock Road, Oxford OX2 6GG, UK
4Aston University, School of Psychology, Birmingham B4 7ET, UK
- Correspondence to: T M Marteau
Diets that are suboptimal—high in saturated fat, free sugar, and salt and low in fibre—are one of the largest contributors to premature death and preventable diseases and to health inequalities.1 Food production, especially livestock, is also a major contributor to environmental harm.2 Reducing the supply and consumption of meat, alcohol, and sugary foods (such as biscuits and confectionary) that contribute to suboptimal diets would improve population health globally, reduce rates of obesity and related diseases such as cardiovascular disease and many cancers,3456789 and could also reduce the health inequalities that stem from their consumption.101112 Limiting these products would also help control the environmental harms associated with their production, processing, transport, and sale.2131415
Reducing consumption enough to improve health equitably and protect the environment will require multiple interventions delivered at scale with the potential to reach everyone. These include price based interventions such as health taxes and carbon pricing1617 and restrictions on price promotions18 and marketing.19 Interventions that change the assortment of products available to consumers (availability interventions) also have the potential to shift consumption at scale, as shown by several recent real world studies (table 1). But this growing evidence has received little systematic analysis by researchers, so remains largely overlooked by policy makers. Defining and characterising effective availability interventions more precisely will provide greater conceptual clarity and strengthen the evidence base. This in turn could increase the use of availability interventions by policy makers and others, to improve population health both equitably and sustainably.
What are availability interventions?
Availability interventions change the variety of options (their number or range) available to consumers in absolute or relative terms. They can be implemented across a broad spectrum of settings in countries (such as limiting the availability of outlets selling higher strength alcohol29), neighbourhoods (such as reducing the density of unhealthier food outlets30), or retail premises, such as a shop, café, restaurant or bar, in private and public sector settings.2731 Our focus is on retail settings (the point of product selection) given that several recent studies have reported large effects on selection and consumption, meriting further discussion.
Progress has been made in defining and conceptualising availability interventions,3233 which can be categorised into two broad types: making changes in relative or absolute availability of different products (fig 1). These types can operate alone or in combination. Relative availability interventions have been most studied; they offer the same number of options to consumers but change the proportions of different subsets of these options—for example, replacing some of the confectionery items on offer with fruit or nuts; sugary drinks with no or low sugar options; alcoholic with non-alcoholic drinks; and meat based with plant based meals. These are particularly well suited to settings where space or the number of options is constrained, such as on supermarket shelves, in a vending machine, or on a limited restaurant menu.
Do they work—and who benefits?
A Cochrane review found that reducing availability resulted in substantial reductions in the selection of targeted foods.34 But it identified only five eligible studies conducted in real world settings. Of these, only three could be included in a meta-analysis, generating a highly uncertain estimate of effect size. The 95% confidence interval ranged from −11% to −60%, encompassing both relatively small and relatively large reductions in selection of target products.
Table 1 summarises four new studies identified from an updated search for this review, using the same search terms and criteria for inclusion. Among the nine real world studies, seven aimed to reduce selection of less healthy foods and drinks, one to increase selection of healthier foods, and one to reduce selection of less sustainable meat based meals. Eight of the studies increased the relative availability of the target foods, only one increased the absolute availability, and none altered both. A consistent direction of effect is evident: reducing the proportion of a target subset of food or drink reduces its selection by consumers, often markedly so. This is noteworthy given the heterogeneity in study settings and interventions. Increasing the proportion of vegetarian meal options in a cafeteria from 25% to 50%, for example, decreased the selection of meat meals by almost eight percentage points (from 81% to 73%).23 Similarly, increasing the proportion of lower energy food options available in cafeterias from 42% to 50% reduced the calories purchased per transaction by almost five percent compared with baseline (from 384 to 366 kcal).27
Although these studies focused on reducing selection of less healthy foods and soft drinks, the findings should apply to other products in principle, including alcoholic drinks. Preliminary results from a study of selection and purchasing of alcoholic and non-alcoholic alternatives from an online supermarket 35 support the findings of an earlier study of hypothetical selection in which participants were asked to select “one drink you would like to drink today.”36 Decreasing the proportion of alcoholic drinks available from 75% to 50% and 25% increased the proportion of non-alcoholic drinks selected (non-alcoholic beers and wines and soft drinks) from 24% to 32% and 45% respectively.35
All these studies were conducted in high income countries. Whether their findings can be applied to low and middle income countries remains unknown and should be evaluated. Differences in baseline availability of products or social norms around different foods and drinks could influence the effect of availability interventions in other settings.
Another remaining uncertainty concerns the link between the size of the intervention (magnitude of change in product availability) and the size of its effect—in other words, whether there is there a dose response effect. This is likely to vary by setting. Relatively small changes in the relative availability of lower energy foods in worksite cafeterias—from about 42% to 50% of the options offered—decreased the total average energy purchased across 19 cafeterias by a modest 4.8% a day.27 In other studies, a larger change such as increasing the proportion of vegetarian meal options from a quarter to a half23 or from one third to two thirds31 had a larger effect, increasing selection of vegetarian options by about 15 and 20 percentage points, respectively.
Suboptimal diets are a leading contributor to health inequalities, so it is important to ensure that availability interventions do not widen them. None of the studies in table 1 explored this issue, although related evidence indicates that availability interventions are unlikely to exacerbate existing health inequalities. In one systematic review of interventions to promote healthy eating, none of those classified as “place based”—changing the characteristics of the environment including the availability of some foods—preferentially benefitted the least deprived groups.37 By contrast, a sizeable minority of interventions targeting individuals were more effective in the least deprived groups.
This pattern is replicated in a systematic review of the effect of diet based interventions on people with intellectual disabilities.38 In addition, a mega-analysis that pooled data from four studies conducted online394041 and two laboratory studies with experimental designs4240 (21 360 observations from 7375 participants) assessing the effect of availability interventions on healthier food selection in general populations indicated minimal differences in effect by socioeconomic position.43
Understanding how availability interventions work can help to optimise their effects.3344 Two possible mechanisms that have been investigated are that interventions increase the probability that a preferred option is provided and that they change perceptions of what is a popular option, also known as a social norm. Selection of any option from an assortment is heavily influenced by a person’s pre-existing preference.39 So, adding apples and oranges to a display previously containing only bananas and unhealthy snack bars increases the chances that any individual’s preferred fruit is offered, increasing the chances that fruit (not a snack bar) is selected.40 For maximum effect, the most popular snack bars should be swapped for the most popular fruits.
Social norms have a large influence on what we eat. The choice of stock on display often reflects existing population preferences. The more of a product there is on display—chocolate say, or fruit—the greater the implied popularity of that product, and the greater the chances of its selection.40 Changing the stock can change perceptions, preferences, and ultimately, selection. Other possible mechanisms of action include reducing the effort needed to resist highly palatable energy dense food.4546 Whether the mechanism varies by type of availability intervention is a question for further study.
Realising the potential
The effectiveness of availability interventions as well as their acceptability and feasibility for businesses were shown in several recent studies4731–for example, across 34 stores, increasing the availability of healthier oven cooked chips by adding packs of low fat chips (not previously available) alongside packs of regular chips reduced sales of the regular chips by 23% compared with a 4% reduction in 146 non-interventions stores (inter-rater reliability=0.81 (0.71 to 0.92) using unit sales) with a corresponding increase in sales of the healthier alternative.47 Public acceptability of availability interventions is also high: in a survey of a representative sample of the English adult population (n=4500), 79% of respondents found policies to replace some higher energy products with lower energy alternatives in cafes and restaurants acceptable.48
Availability interventions are ripe for inclusion in local, national, and global policies to improve health and protect the environment at scale given their consistent and often substantial effects on selection and purchasing, with no evidence of adverse effects, particularly on inequalities. As with all interventions, evaluations should be designed in to implementation plans to inform adaptations as evidence of real world effects emerges. Ongoing evaluations are also important to build the evidence base on the size of effect from different magnitudes of change in availability, effects in different contexts and populations, and for different foods and drinks.
There are only a few examples of availability interventions in existing policy. In Scotland, all retail outlets in healthcare buildings are required to ensure that 50% of food and 70% of drinks on sale are not high in fat, sugar, or salt.49 In Portugal, there is now a requirement that cafeterias in public sector settings make available at least one vegetarian dish.50 More recently, the UK government has proposed increasing the availability of alcohol-free and low alcohol products.51
Several factors are likely to contribute to the relative neglect of these interventions. First, they have yet to be systematically evaluated and summarised in a form accessible to policy makers. Second, they don’t fit with the dominant public discourse around personal responsibility for unhealthy behaviour5253—a discourse that is reinforced by some commercial companies to divert attention from (and delay the introduction of) more effective policies that would reduce their profits.545556 Third, availability interventions might be resisted by businesses fearing loss of sales. Although losses have not generally been reported in evaluations,273157 businesses heavily dependent on selling unhealthy and unsustainable foods might lose revenue. Governments are likely to take different approaches to helping businesses transition to providing healthier and more sustainable foods.
Reflecting these and other factors, policies to prevent obesity tend to focus on weaker interventions that target individuals, making high demands on individual agency and sometimes increasing inequalities.37 By contrast, interventions that change the environments shaping decisions (including availability interventions) make fewer demands on individual agency and are less liable to increase inequalities.5859 Summarising the evidence supporting availability interventions in a form useful to policy makers—as we aim to do here—should help highlight the opportunities for including them in strategies promoting health and sustainability.
Evidence for the effectiveness of availability interventions is now mature enough to merit serious consideration by policy makers and others as an effective addition to interventions in public and private sector settings aiming to shift currently unhealthy and unsustainable patterns of consumption towards unmet population health and net zero goals.
Suboptimal diets—particularly the consumption of energy dense foods, meat, and alcohol—are one of the largest contributors to premature death and preventable diseases worldwide, also influencing health inequalities and environmental harm
Interventions that alter the assortment of available food or drink have consistent and often substantial effects on consumer selection and consumption, with no evidence of adverse effects, including increasing health inequalities
This promising evidence would be strengthened by research characterising the nature of these interventions more precisely and estimating the effects achieved by interventions of different magnitude, in different contexts, for different populations, and for different foods and drinks
Availability interventions, currently overlooked by policy makers, merit urgent consideration in public and private sector settings where energy dense foods, meat, or alcohol are on sale given their potential to contribute to population health and net zero goals
Contributors and sources: TMM conceived the idea for this paper. TMM and GJH prepared the first draft of the paper to which all other authors added conceptual ideas and salient evidence. All authors edited the manuscript before approving the final version. TMM is guarantor of the article.
Competing interests: We have read and understood BMJ policy on declaration of interests and have no relevant interests to declare.
Competing interests: All authors have read BMJ competing interests’ policy and declare no relevant interests.