Increasing vegetable intakes: rationale and systematic review of published interventions

Purpose While the health benefits of a high fruit and vegetable consumption are well known and considerable work has attempted to improve intakes, increasing evidence also recognises a distinction between fruit and vegetables, both in their impacts on health and in consumption patterns. Increasing work suggests health benefits from a high consumption specifically of vegetables, yet intakes remain low, and barriers to increasing intakes are prevalent making intervention difficult. A systematic review was undertaken to identify from the published literature all studies reporting an intervention to increase intakes of vegetables as a distinct food group. Methods Databases—PubMed, PsychInfo and Medline—were searched over all years of records until April 2015 using pre-specified terms. Results Our searches identified 77 studies, detailing 140 interventions, of which 133 (81 %) interventions were conducted in children. Interventions aimed to use or change hedonic factors, such as taste, liking and familiarity (n = 72), use or change environmental factors (n = 39), use or change cognitive factors (n = 19), or a combination of strategies (n = 10). Increased vegetable acceptance, selection and/or consumption were reported to some degree in 116 (83 %) interventions, but the majority of effects seem small and inconsistent. Conclusions Greater percent success is currently found from environmental, educational and multi-component interventions, but publication bias is likely, and long-term effects and cost-effectiveness are rarely considered. A focus on long-term benefits and sustained behaviour change is required. Certain population groups are also noticeably absent from the current list of tried interventions.


Introduction Health benefits of high fruit and vegetable intakes
The health benefits of a high consumption of fruits and vegetables are well known [1]. Associations with all-cause

Abstract
Purpose While the health benefits of a high fruit and vegetable consumption are well known and considerable work has attempted to improve intakes, increasing evidence also recognises a distinction between fruit and vegetables, both in their impacts on health and in consumption patterns. Increasing work suggests health benefits from a high consumption specifically of vegetables, yet intakes remain low, and barriers to increasing intakes are prevalent making intervention difficult. A systematic review was undertaken to identify from the published literature all studies reporting an intervention to increase intakes of vegetables as a distinct food group. Methods Databases-PubMed, PsychInfo and Medlinewere searched over all years of records until April 2015 using pre-specified terms. mortality [2,3] and mortality from cardiovascular disease [2,3], including coronary heart disease [4] and stroke [5,6], are well evidenced. Associations also suggest a reduced risk of hypertension [7], osteoporosis [8], body weight and adiposity [9,10], dementia and cognitive decline [11,12], and some cancers [13][14][15], although the evidence for cancers is less consistent [1,3]. Intervention studies increasing the consumption of fruits and vegetables also demonstrate improved microvascular function [16], improved microvascular function and inflammatory status [17], improved profiles in inflammatory and oxidative stress [18], improved immune response [19], and improved weight maintenance [20].

Consideration of fruits and vegetables as different food groups
However, while fruits and vegetables share health benefits as a result of the provision and interaction of a number of bioactive compounds, including vitamins, minerals, antioxidants, carotenoids and flavenoids [21,22], the specific bioactive compounds in fruits and in vegetables can vary greatly [21][22][23][24][25]. Their contribution to other dietary features also vary. Fruits typically contain greater dietary sugars, with potential negative impacts on both health and on public willingness to consume them [22,26]. Vegetables, by comparison, can contain more protein and fibre [22] and are more often processed prior to consumption. This processing can both increase and decrease micronutrient bioavailability and activity, again impacting on health benefits [22,25,[27][28][29][30][31]. Many studies that separate fruits and vegetables find different effects of the different food classes on health outcomes [2,12,[32][33][34][35]. These differences between fruits and vegetables argue for the consideration of fruits and vegetables, in terms of health, as separate and different food types.
While differing in their potential health benefits, fruits and vegetables also taste very different, are generally of a different texture and are typically consumed in different manners [36][37][38][39]. Fruit is generally sweet, is typically softer in texture, is more often consumed raw, is more frequently consumed and is generally considered more acceptable as a snack, as a drink or as dessert [36][37][38][39]. Vegetables, by comparison, can taste bitter, are generally harder in texture, are more often cooked, are more typically consumed and considered more acceptable as part of a meal [36][37][38][39], and thus are also more often consumed with other foods as opposed to alone [40]. These different consumption patterns suggest that fruit and vegetable consumption may be differentially determined. Glasson et al. [41] directly compared the determinants of fruit consumption and vegetable consumption in an Australian population, to find fruit consumption to be largely prevented by cost, food preferences, quality, availability and wastage concerns, while vegetable consumption was more frequently prevented by food preferences, lack of time, cost and taste. Differential determinants again argue for the consideration of fruits and vegetables, as separate and different food types, and suggest the potential need for different intervention strategies for increasing fruit and for increasing vegetable consumption.
Furthermore, while population levels of both fruit and vegetable consumption remain below World Health Organization recommendations across the world [42,43], interventions to increase fruit and vegetable intakes more often target fruit, and typically report greater success for fruit consumption compared to that for vegetables, for both children and adults [44][45][46]. These findings suggest not only a need for different intervention strategies for increasing fruit and increasing vegetable consumption, but a real need for strategies that achieve successful increases in vegetablespecific consumption. This paper focuses specifically on vegetable consumption.

Vegetable-specific consumption Health benefits of high vegetable-specific intakes
Various studies demonstrate health benefits from a high consumption specifically of vegetables (i.e. from vegetables alone, as opposed to in combination with fruits, as occurs when considering fruits and vegetables together). Observational studies demonstrate reduced risk of cardiovascular disease [2], type II diabetes [34], non-gallstonerelated acute pancreatitis [33], various cancers [2,[47][48][49][50] and cognitive decline [32]. Meta-analyses of observational studies demonstrate associations between a higher vegetable consumption and reduced risk of stroke [6], dementia and cognitive decline [12], and from various cancers [14,15,35,51,52], although again the evidence for cancers is inconsistent. Meta-analyses of prospective studies find no benefits for breast cancer risk [53], gastric cancer risk [54], pancreatic cancer risk [55] and bladder cancer risk [56]. Meta-analyses also report no benefits of overall vegetable consumption for type II diabetes [57][58][59], although doseresponse meta-analyses also suggest benefit up to 2-3 servings/day and a threshold beyond this where type II diabetes risk does not reduce further [60].
Specific vegetable groups or types of vegetables have also been associated with improved health outcomes. Intakes of dark green leafy vegetables have been associated with reduced risk for type II diabetes [57][58][59], reduced risk for a number of cancers [48,49,61] and with reduced depression [62]. High intakes of cruciferous vegetables have been associated with reduced risk from various cancers [63][64][65][66][67][68][69][70]. Intakes of beta-carotene-rich vegetables, yellow-and red-pigmented vegetables, and fruiting vegetables have also been associated with reduced risk from various cancers [48][49][50]61], and root vegetable consumption has been associated with reduced type II diabetes risk [60].
Much of this evidence, however, stems from limited studies, and the body of evidence is far from conclusive [64,66]. Prospective and cross-sectional studies are easily criticised for potential confounding, case control comparisons may suffer from bias towards differences between groups due to comparisons between cases and health conscious (and consequently) healthy controls [71], and study designs do not allow determination of causality. Considerable further work is required before conclusions can be drawn. The majority of studies investigating effects of vegetable consumption, furthermore, do not investigate vegetable consumption independent of fruit consumption or other aspects of the diet. While fruits and vegetables are frequently consumed together, associations may reflect not just associations with vegetables, but associations with produce consumption in general, or with a healthier diet/ lifestyle [1,72]. A recent systematic review by Fulton et al. [72] reports impacts for fruit and vegetable interventions not only from micronutrient intakes but also from changes to the whole dietary profile. Lifestyle factors are frequently included in studies as confounders, but it is often difficult to control for all potential confounders, and adjustment for other dietary aspects, particularly fruit consumption, is less common. Associations will also depend on the definition of vegetables used, and the inclusion or not of certain vegetables in certain categories. Potatoes, for example, are sometimes included among vegetables, sometimes included as 'white' vegetables, and sometimes not considered at all [21,22].
Thus, for improved health, increasing intakes of vegetables are required. For intakes of vegetables to be increased, strategies and interventions are needed. These interventions should be based on in-depth understanding of the underlying determinants of low vegetable consumption.

Determinants of vegetable-specific consumption
Various research has been undertaken to understand the associations with, and reasons for, vegetable consumption, independent of fruit consumption. In young children, the bitter and undesirable taste of vegetables is often provided as a major barrier to vegetable consumption [73][74][75], and food neophobia (the reluctance to eat, or the avoidance of, novel foods [76]) particularly, can interfere with young children's acceptance of vegetables [76][77][78]. This neophobia typically results in the rejection of bitter tasting foods and foods that do not ''look right'' [76], of which vegetables are good examples.
As children age, taste, appearance and liking continue to be important [79], but low vegetable consumption is frequently also associated with various characteristics of the family environment. These factors include low parental education and socio-economic status [80][81][82], low vegetable consumption by parents and caregivers [83][84][85], low availability and negative perceptions of vegetables in the home [86] and a family environment that is unsupportive of vegetable consumption [83,85,87]. Vegetable consumption is higher, for example, in families where vegetables are disguised or sauces are used to mask undesirable tastes and appearances [79,83,85], where vegetables are more often incorporated into composite foods as opposed to consumed alone to dilute negative tastes and appearances [79], where meals are home cooked to accommodate individual preferences [83], and where games are played to encourage vegetable consumption [85]. The expression of neophobic behaviour towards vegetables also appears to be mitigated by high parental education and high socio-economic status [77,88] and again by a positive and supportive environment [89,90].
Taste, appearance, liking and the surrounding environment continue to be important as children become adolescents, but individual cognitions also gain increasing importance. Low vegetable consumption in adolescents has again been associated with low parental education and socio-economic status [91], low vegetable consumption by the parents [92], low availability and a family environment that is unsupportive of vegetable consumption [91,93]. Vegetable consumption in adolescents, however, has also been associated with an awareness of the importance of vegetables for health, and a willingness and ability to ask for vegetables from parents [93].
In adults, higher vegetable consumption has been associated with higher liking for the taste of vegetables [41,94,95], higher appreciation of health and the value of a healthy diet [94], greater nutritional and culinary knowledge [96,97], and with several related food habits and eating practices [98,99], including usual consumption of meals as opposed to snacks [94,100], increased time and willingness to prepare and cook home-cooked meals [41,94,95,97], and a low consumption of fast food [94]. The transfer of childhood eating habits and food preferences into adulthood is well known, and adult vegetable intake is often related to childhood experiences [94]. Neophobic tendencies also typically last well into adulthood, and typically correlate negatively with liking for and frequency of vegetable consumption in adulthood [101][102][103]. The individual preferences of one family member can also have impacts on the rest of the family, with most family units opting to cook only one meal of acceptability to all family members [104,105]. Given the importance of adult consumption for children, many of the determinants of adult consumption will also impact on child consumption.
Alongside individual preferences, higher vegetable consumption in adults is also related to increased availability [98,106,107] and reduced cost [41,95,97,108], and low consumption is largely associated with lower socio-economic status [109,110], lower income [44,107], living in a more deprived area or lower income neighbourhood (an indirect measure of socio-economic status) [44] and lower education [109,111].
Research thus, suggests a variety of reasons behind low vegetable consumption, ranging from taste and pleasure, to individual cognitions and health beliefs, and to aspects beyond the individual including society and the environment. Many of these reasons have been targeted by interventions.

Strategies to increase vegetable-specific consumption: systematic review
Various reports of strategies to increase vegetable-specific consumption are available. A comprehensive collection of these interventions, and an evaluation of success, however, is currently lacking. The aim of this work was to systematically review the published literature to identify all published interventions aiming to increase vegetable-specific consumption.

Method
The objective of the review was to identify from the published literature all studies reporting an intervention to increase vegetable intakes, where vegetables were considered as a separate and distinct food group, and the intervention focused specifically on increasing intakes of this food group. Three databases: PubMed, PsychInfo, and Medline, were searched over all years of records for all studies with the terms 'vegetable' and 'vegetables' in the 'title'. These search criteria were used to limit the search results to studies with a focus on vegetables. No other search criteria and no limits were used. All titles were screened for relevance and then all abstracts. Two review authors independently conducted all searches, screened all titles and screened all abstracts (KMA, AH or HH). Studies were included in the review if they involved an intervention designed primarily to increase vegetable intakes as a specific and distinct food group, and if they intended to change behaviour-vegetable selection, purchasing or vegetable consumption. Studies were not included if they did not include an intervention, if the intervention targeted fruit and vegetable intakes [112], if the intervention targeted vegetables and other foods, e.g. vegetables and wholegrains [113,114]; if the intervention involved changing consumption as opposed to increasing consumption [115], or if they did not include a measure of behaviour, but instead only measured correlates of behaviour such as intentions, attitudes, and knowledge [116,117]. Studies measuring tasting were included where tasting was voluntary, where amount tasted was voluntary and where tasting/amount was measured, but studies where tasting was compulsory and/or prespecified, e.g. to make hedonic judgements, were not included [118,119]. Studies were included regardless of the use or not of a comparison for an intervention, or the type of comparison used. A study using a vegetable-specific intervention that is compared with a fruit-specific intervention, for example, is included [120] (all other criteria were also met). Relevant articles were also searched for other suitable studies. Searches of conference abstracts, book chapters, etc., were not undertaken, thus studies are only included if reported in full articles. Details from all studies were subsequently tabulated by one review author (KMA) and checked by an additional review author (AH or HH). All tables are provided in the "Results" section. No other data were extracted. Due to the early nature of the research area, the limited number of studies available per intervention type, and high heterogeneity between study methodologies, risk of bias was not assessed, and no attempt was made to combine studies, e.g. through meta-analysis. The review was undertaken using PRISMA guidelines and a PRISMA diagram illustrating the outcomes of the review process is given in Fig. 1.

Results
Searches were most recently conducted on 28 April 2015. The results of the searches are given in the PRISMA diagram in Fig. 1. A total of 77 studies were identified, reporting the impacts of 140 interventions. Details of all studies are presented by intervention type in Tables 1, 2, 3, 4, 5 and 6. Interventions have been broadly classified as those focussing on hedonic determinants of vegetable intake, such as taste, familiarity and liking-Tables 1, 2 and 3, those focussing on environmental determinants of vegetable intake- Table 4, those focussing on cognitive determinants- Table 5, and those using a combination of approaches- Table 6. Studies reporting two or more different interventions are included separately in separate tables, where appropriate. Within each table, studies are ordered by age of target audience. Of the interventions identified, 113 (81 %) interventions focus on improving intakes in children. Early intervention will maximize health benefits [45], and eating habits in childhood are likely to extend into adulthood [75,94].

Interventions aiming to change or use hedonic factors
Eleven interventions focus on changing or using the taste or familiarity of a vegetable/vegetable product on a single occasion ( Table 1). Six of these interventions suggest that the addition of a liked taste or flavour in the form of salt [121,123] and in the form of a flavoured dip [122] or condiments [125] can increase vegetable consumption on a single occasion. Guidelines regarding salt intake and the possible impact of increasing preferences for salty flavours must also be considered, but these studies demonstrate a potentially useful strategy. The addition of fat to a vegetable product did not result in increased intakes [121], but increased intakes were found following the use of familiar as opposed to novel vegetables products [124].
Fifty-two interventions focus on increasing liking and familiarity with repeated experience. These interventions use learning techniques, including repeated exposure (n = 23), ensuring that exposure is positive via pairing with liked flavours (n = 14), pairing with beneficial nutrients (n = 6), pairing with external reinforcement (n = 7), the use of positive models (n = 1) and the use of reinforcement plus models (n = 1) ( Table 2). Many of these interventions demonstrate success by improving liking and/or consumption: 16 of 23 (70 %) using repeated exposure; 12 of 14 (86 %) using pairing with liked flavours; four of six (67 %) using pairing with nutrients; all seven (100 %) using pairing with reinforcement; and the one (100 %) using reinforcement plus modelling. Effects, however, are far from robust or consistent (i.e. effects are often found in one measure, but not in , are often small, and tend to be limited to the specific vegetable used during exposure. Conditions within studies, furthermore, are often confounded, making mechanisms difficult to elucidate. In many studies that purport to investigate exposure, the exposure is in combination with other food components [129], modelling [75] or rewards in the form of praise or other positive interactions [139], thus effects may in fact occur partly due to conditioning. In many studies that purport to measure conditioning, exposure is not controlled for [136,151]. Many of these interventions furthermore also involve children's parents, and so may have benefits not just by allowing tasting and experience for the child, but also by improving parental perceptions of vegetables, improving attitudes towards vegetables in the home, and improving parental education and knowledge [75]. Interestingly, some of the interventions included in Table 2 report parental opinions of the intervention [75,129], but as far as we can tell, none specifically tested parental knowledge or education as a result of the intervention for their children. While largely successful, particularly over considerable exposures, exposure, however, is a relatively time-consuming practice that results in small changes, and typically only for the vegetable to which children have been exposed. Nine interventions have extended the use of exposure to consider exposure to vegetables via picture books and stories (Table 3). These procedures appear beneficial, although few studies have currently tested these ideas, and effects again appear small or unreliable, and typically only apply to the vegetable to which the child has been exposed. Little evidence suggests that effects generalise to other vegetables, so neither taste or visual exposure appears to encourage consumption of a variety of vegetables. Repeated exposure to many vegetables may result in increased consumption of many vegetables, and some studies are beginning to demonstrate these effects [156,157], but generalisation of exposure to non-exposed vegetables has not yet been demonstrated reliably either through the use of taste or visual stimuli. The potential for exposure to multiple as opposed to single vegetables at one time point, however, may be greater using visual as opposed to taste stimuli.

Interventions based on changing the environment
An alternative approach to encourage vegetable consumption focuses on changing the environment and increasing consumption through increasing the provision of vegetables, or improving the manner in which provision is implemented (Table 4). Thirty-nine interventions were found using these strategies. All of these, with the exception of three interventions (one increasing variety [158] and two improving presentation [141,164]) resulted in increased selection and/or consumption of vegetables in children and adults, through the increased provision of vegetables (n = 20), through the increased provision of a variety of vegetables (n = 7), through improved presentation (n = 5), through changing the location of vegetables (n = 1), through changing the order in which vegetables and other foods are served (n = 1) and through changing the serving order, while also increasing availability (n = 2). Increased consumption as a result of increased provision is unsurprising, but concerns have been raised regarding increased energy intakes as a result, and increased potential for food wastage. An absence of effects on overall energy intakes is reported in some studies [161], and concerns are mitigated if vegetables are substituted for other foods in the meal as opposed to simply added [161]. Suggestions to reduce potential food wastage include the use of family style serving dishes for individual meals [159,161] or allowing differential selection, but again the costeffectiveness of interventions that can increase waste will be questioned. Strategies that improve the presentation of vegetables may offer a valuable alternative. These interventions typically change the salience or likely appeal of vegetables [141,164,166], and have again demonstrated success, but relatively few studies are currently available. Exact mechanisms are again unclear-attractive labels may rely partly on modelling, effects due to serving order may rely partly on hunger and exposure, but the relative ease and low cost of these interventions add to their value.

Interventions based on changing or using cognitive factors
Nineteen interventions were found that used information, education or other cognitive techniques (Table 5). These interventions are largely aimed at older audiences (those where cognitive factors have a greater impact on vegetable consumption and non-consumption), and used a range of techniques from providing information and education on nutrition (n = 6) or nutrition-related skills (n = 2), providing education plus a demonstration (n = 1) or gardening experience (n = 2), providing tailored information (n = 2), providing information on social norms (n = 1), invoking choice (n = 4) and invoking a memory (n = 1). With the exception of one intervention that aimed to educate [75], and two interventions that utilised choice [174], all of the studies using these types of strategy reported success to some degree, but multiple measures of impact were often taken, and success is not necessarily reported for all measures. The cost-effectiveness of these types of interventions is, however, also often questioned. Educational interventions can be costly, particularly those involving classes or courses to be delivered by a professional, but the long-term benefit of these interventions can also be difficult to assess. Knowledge accumulates over time and experience, and it can be difficult for individuals to pinpoint the exact source/ sources of beneficial education. Greater liking and intake of green beans, and greater intake of novel V, postintervention in I vs C, but no effects for artichoke Weaning with V but not with fruits, may promote V acceptance in children  The findings support the effectiveness of using a reward with a repeated exposure strategy. In particular, such rewards can facilitate the tastings necessary to change liking Increased selection of V in I vs C, so increased consumption. Students selecting beans consumed the same in I and C, but students selecting carrots also consumed more in I vs C Placing photographs in cafeteria lunch trays incurs minimal costs, but was associated with an increase in V consumption within the range of those found in more expensive interventions

Multi-component interventions
Multi-component interventions involve a combination of strategies ( Table 6). Ten of these interventions were found. Again all the published reports evaluating these interventions report success, but again multiple measures are often taken, which demonstrate varying degrees of benefit. These types of intervention can also be time-consuming and costly to implement. Success is furthermore not easily attributable to the combination of many strategies as opposed to the use of any single one.

Discussion
While a variety of successful strategies for increasing vegetable intakes have been tried, evaluated and published, evaluation periods are typically short, effect sizes can be small, and those studies that use longer follow-up periods often report reductions in effect size as follow-up periods are extended. These findings are unsurprising and have persuaded many researchers to recommend repeated interventions or a combination of interventions with the hope of improving long-term benefits. However, cost-effectiveness is rarely considered, yet cost-effectiveness becomes an increasing concern in long-lasting and multi-component interventions. Further work is clearly still required. A greater number and variety of intervention evaluations would increase the evidence base, and more reliably inform future policies. Longer-term follow-ups for interventions are imperative, and consideration of more sustainable behaviours or the more sustainable elements of behaviour, such as habit formation or behavioural norm changes, would be of value. While the review highlights strategies of benefit furthermore, the review also identified noticeable absences. Based on the search strategies and current literature, very few interventions were identified specifically for adolescents or older individuals. The eating attitudes, practices and intakes of these groups are known to differ from those of other members of the population, and the simple generalisation of successful strategies from other population groups may not occur. Adolescence is a period of rapid development, from physical, cognitive and social perspectives, and changes to eating practices and dietary intake during this period are well reported [194]. Studies in this group on barriers to consumption identify constraints largely similar to those in younger children, but also identify an increased recognition of cognitive factors. Strategies then that involve education may be particularly beneficial. Older individuals similarly will experience changes in physical and cognitive abilities, many of which will have an impact on eating practices and food intake [195][196][197]. Barriers specifically to vegetable consumption in this group have not been identified as far as we are aware, but barriers to fruit and vegetable consumption are similar to those for other adults [195,196], although the impact of demographic characteristics and environmental factors tend to be exacerbated. Changes to living circumstances for example, will impact negatively on existing impacts as a result of availability, cost and cooking abilities [195][196][197].
Vegetable consumption is also known to be low in individuals of low education and of low socio-economic status [198], and these factors are specifically highlighted as barriers to increasing consumption, yet few of the interventions published to date focus on or even include individuals with these demographics. There are some exceptions-the intervention by Clarke et al. [184] focuses specifically on individuals using community pantries, and many of the studies in the developing world focus not only on increasing vegetable intakes but also on sustainable vegetable provision and improved food security [185,191], but more work is clearly needed in relation to socio-economic disparities. Interventions that improve fruit and vegetable intakes are available [199]. Increased efforts to reduce socio-economic disparities, however, are often requested [200][201][202][203], and concerns that intervention success is most easily achieved in those of little need of benefit are difficult to allay.
Consideration of the barriers to vegetable consumption suggests that many of the strategies that have shown success so far in certain groups may be beneficial for other groups. Almost all individuals will arguably benefit from increasing vegetable intakes, and the strategies found to be successful in one population group may easily transfer to another. Exposure type strategies to increase liking, for example, have shown effects for fruit consumption in older adults as well as children [204], although vegetables were not specifically investigated here. We recommend careful consideration of barriers however, and caution against a 'one size fits all' approach. While interventions may be successful across individual and population groups, testing is clearly required. At present, there is a real lack of comparisons between interventions-i.e. interventions have not been compared, e.g. in the same age group or population, with the exception of comparisons of differing exposure and conditioning strategies in young children. This lack of comparison may reflect the early nature of the field, but even where multi-component interventions have been successful, identification of the successful component/components is rarely undertaken. With a view to lasting impact and cost-effectiveness, comparison of interventions, or the identification of more effective intervention components would clearly be of value.
Several types of broader population-based interventions have also not yet been considered specifically for vegetable consumption. Strategies such as pricing and marketing interventions, improved product provision, government subsidies, and population-wide awareness and education campaigns [205][206][207] specifically for vegetables do not yet exist or have not yet been evaluated as far as we are aware. In some countries, WHO recommendations have been separated for fruit and vegetables. Dutch consumers are asked to consume 2 + 2 (2 portions of fruit and 2 portions of vegetables per day), and Australian consumers are asked to aim for 2 + 5 (2 portions of fruit and 5 portion of vegetables per day), but these types of recommendations rely heavily on an individual's ability to identify and categorise fruits and vegetables, and limited work suggests that consumers find this difficult [41,104].

Limitations of the review
While our review has identified a number and variety of interventions, we have only considered the published literature, and our search strategy is likely to be biased towards articles published in English and away from related grey literature, such as lay publications and conference proceedings. Due to the early nature of the research area, publication bias is also highly likely. There is a noticeable absence of publications that report failures, or that demonstrate cost-inefficiency or other negative impacts of interventions. We also chose not to attempt to combine interventions. Due to the early nature of the research area, the limited number of studies available per intervention type, and high heterogeneity between study methodologies, we considered combination to be inappropriate.

Future directions
There is an urgent need for the development and evaluation of interventions to target all population groups. Interventions are particularly required for certain population groups, including adolescents, older adults and those of low socioeconomic status. Interventions for groups 'at risk' of disease may also be beneficial, given the often increased success of interventions in these individuals [1]. Assessments of the long-term benefits, sustainability and cost-effectiveness of interventions are also clearly required. While many interventions report success, effect sizes are typically small, long-term follow-up is rarely undertaken and studies that do report follow-up often fail to find sustained benefits. Interventions with a focus on long-term benefits and sustained behaviour change are required, as is increased work understanding the principles underlying behaviour change and behaviour change maintenance. Comparisons between interventions, to identify those of greatest likely benefit, would also be of interest.

Conclusion
In conclusion, increasing evidence suggests health benefits from the increased consumption specifically of vegetables, yet barriers to increasing intakes are prevalent, and while successful interventions have been published, the true value of these, both in cost-efficiency and sustainability are yet to be determined. Considerable further work is needed in developing new and adapting existing interventions for all population groups, and in evaluating benefit and cost-efficiency over the longer term.