Discrete strategies to reduce intake of discretionary food choices: a scoping review

On a population level, dietary improvement strategies have had limited success in preventing the surge in overweight and obesity or reducing risk factors for chronic disease. While numerous multi-component studies have examined whole-of-diet strategies, and single component (i.e. discrete) dietary intervention strategies have targeted an increase in core foods (e.g. fruits, vegetables, dairy), there is a paucity of evidence on the effectiveness of dietary intervention strategies targeting a decrease in discretionary choices. The aim of this review was to identify dietary intervention strategies that are potentially relevant to reducing intake of discretionary choices in 2–65 year olds. A scoping review was carried out to map the literature on key discrete dietary intervention strategies that are potentially applicable to reducing discretionary choices, and to identify the targeted health/nutrition effects (e.g. improve nutrient intake, decrease sugar intake, and reduce body weight) of these strategies. Studies conducted in participants aged 2–65 years and published in English by July 20, 2015, were located through electronic searches including the Cochrane Library, Medline, Embase, CINAHL, and Scopus. Three thousand two hundred and eighty three studies were identified from the search, of which 44 met the selection criteria. The dietary intervention strategies included reformulation (n = 13), substitution (n = 5), restriction/elimination (n = 9), supplementation (n = 13), and nutrition education/messages (n = 4). The key findings of the review were: restricting portion size was consistently beneficial for reducing energy intake in the acute setting; reformulating foods from higher fat to lower fat could be useful to reduce saturated fat intake; substituting discretionary choices for high fibre snacks, fruit, or low/no-calorie beverages may be an effective strategy for reducing energy intake; supplementing nutrient dense foods such as nuts and wholegrain cereals supports an improved overall diet quality; and, a combination of permissive and restrictive nutrition messages may effectively modify behavior to reduce discretionary choices intake. Longer-term, well-controlled studies are required to assess the effectiveness of the identified dietary strategies as interventions to reduce discretionary choices intake.


Background
The global prevalence of overweight and obesity, and associated chronic health conditions continues to increase [1]. Dietary recommendations for weight management and chronic disease prevention, including increasing fruit and vegetable intake and decreasing intake of added sugar, saturated fat and salt [2], have failed to be successfully adopted by the majority of the western population [3][4][5].
In Australia, only 6 % of adults consume an adequate intake of fruit and vegetables [6] and only 69 and 36 % of children (aged 5-11 years) consume at least two and three daily serves of fruit and vegetables, respectively [7]. In the UK only 8.5 % of adolescents and 30 % of adults meet the recommended serves of fruit and vegetables [5]. Similarly in the US, only 19 % of the population meet the minimum serves of fruit and vegetables [4]. Discretionary choices are foods or beverages high in saturated fat, added sugars, or salt, such as crisps, sugar sweetened beverages, sweet biscuits, cakes and desserts, sweet and savory pastries and processed meats [8]. In Australia, discretionary choices currently contribute around 36 % of energy intake in 2-13 year olds and 30-40 % of energy intake in those aged ≥14 years [3]. According to US data, 86 % of the population consume more than the recommended limit of discretionary choices [4]. National dietary intake data from the UK also shows mean intake of saturated fat, sodium and added sugars are in excess in all age groups (with the exception of sodium in girls aged 7-10 years) [5]. Research in children has indicated that discretionary choices may displace core foods such as fruit, vegetables, dairy, lean meats, and whole grains [9,10]. Reducing the current intake of discretionary choices will reduce the risk of nutrient deficiencies, obesity and associated chronic disease [8].
While the impetus to reduce discretionary choice intake is clear, the interventions needed to achieve this change are uncertain. The effectiveness of multi-component dietary interventions (e.g. changing whole-of-diet) or discrete dietary intervention strategies targeting an increase in core foods (e.g. fruit and vegetable or dairy intake) have been widely researched; however they have had little success in preventing the surge in overweight and obesity [11][12][13] as well as reducing risk factors for chronic disease [14,15]. In contrast, there is a paucity of evidence on the effectiveness of dietary intervention strategies targeting a decrease in discretionary choices. Increasing our understanding of dietary strategies that are potentially relevant to decreasing discretionary choices intake will inform the design of next generation interventions which are needed to prevent overweight and obesity and/or reduce chronic disease risk factors. Therefore, the aim of this scoping review is to identify dietary intervention strategies that are potentially relevant to reducing intake of discretionary choices in 2-65 year olds. We will explore evidence from interventions targeting discretionary choices and examining dietary intervention strategies that have been applied in the context of core foods but could be applied in the context of reducing discretionary choice intake.

Methods and approach
A scoping review was conducted based on key phases detailed by Arksey et al [16] with the aim of mapping the literature on key discrete dietary intervention strategies that are potentially applicable to reducing discretionary choices, and their targeted health/nutrition effects (e.g. improve nutrient intake, decrease sugar intake and reduce body weight) in 2-65 year olds. According to Daudt et al., a scoping review aims to "map the literature on a particular topic or research area and provide an opportunity to identify key concepts; gaps in the research; and types and sources of evidence to inform practice, policymaking, and research" [17]. The framework for conducting this scoping review was based on key phases detailed by Arksey et al [16]: i) identification of the research question to be addressed; ii) identification of studies relevant to the research question; iii) selection of studies to include in the review; iv) charting of information and data within the included studies; and v) collating, summarizing and reporting results of the review.

Search strategy
The search strategy and procedure were guided by the PRISMA statement [18]. Potential studies were located through electronic searches (Cochrane Library, Ovid [Medline and Embase], EbscoHost [CINAHL], and Scopus). Limits were set to age 2-65 years, English language, and publications released up to and including July 20, 2015 (i.e. the search date). Search terms and MeSH headings in the title, abstract, and index terms were initially identified in Medline and subsequent key words were used for the remaining databases (Appendix 1). An academic librarian assisted with the development of MeSH headings, key words, and conduct of the electronic database search.
Identified studies were assessed for inclusion by the primary author (JAG). Following removal of duplicates, articles were examined for eligibility based initially on titles and abstracts, followed by a full-text appraisal of all remaining articles (Fig. 1). Reasons for full text exclusion were documented.

Study selection
Included participants were children and adults aged 2-65 years who were generally healthy and without chronic disease. Included studies were randomized controlled trials (RCT) or comparator group studies that evaluated strategies to reduce discretionary food intake (or core foods, e.g. dairy products, with an approach that could be applied to discretionary choices) with the aim to improve nutritional intake or health status. Strategies included those pertaining to dietary manipulation with a focus on reformulation (e.g. low fat vs. high fat products), substitution (e.g. replacing biscuits with fruit), restriction/elimination (e.g. reducing portion size), and supplementation (e.g. including specific foods/beverages in the diet). Chronic and acute studies were included: chronic dietary intervention studies were those where participants would consume the food/beverage daily for a number of weeks and dietary intake measured post intervention; acute studies were those where the food/beverage was consumed only once, or once per week over a few weeks, and the impact of subsequent food/calorie intake was measured. Strategies utilizing nutrition education/messages were also included where participants were provided information/strategies/messages on food for any defined period, to alter food intake. Exclusion criteria were studies in pregnant or lactating women or clinical populations that require strictly modified diets (e.g. those with celiac disease); studies related to binge/ disordered/restrained eating; studies assessing results of only the intervention or comparator group either following the RCT or as follow-up post-RCT completion; studies assessing food labelling and media advertising (e.g. through computer games or television adverts); and studies that did not report on nutritional or food intake or body weight.

Data collation and reporting of results
Data was extracted on the study aims, intervention, outcome measurement and main results (see results Tables 1, 2, 3, 4 and 5) and summaries were developed. Unlike a systematic review which synthesizes and weights studies according to level of evidence, a scoping review presents findings in a narrative way. We based our narrative synthesis on recommendations by Arksey et al. [16] and Popay et al. [19]. That is, we firstly gave attention to the numbers of studies reporting on each discrete intervention type (e.g. substitution, reformulation, and elimination), a description of the intervention, and whether it was effective at producing a significant change in the study's main outcome. This allowed us to understand what the most common themes emerging from the identified strategies were, map the strategies back to our aim, and to have a better understanding on their effectiveness and research gaps. Together, these data formed the basis of the scoping review summary.

Review
The initial search retrieved 3283 articles (with removal of duplicates), 3181 were excluded through title and abstract screening (Fig. 1). One hundred and two full text articles were examined; following exclusion, 44 studies were included in the review. Thirteen studies assessed reformulation strategies (Table 1), five studies assessed substitution strategies (Table 2), nine studies assessed restriction/elimination strategies (Table 3) and 13 studies assessed supplementation strategies (Table 4). Four studies assessed nutrition education/messages strategies (Table 5).    Randomized crossover-design. Afternoon yogurt snacks containing: 1) Normal protein yoghurt (5 g protein/170 g serve).

Leahy KE, 2008 USA
To investigate the effects of reducing the energy density of a popular and familiar entrée-macaroni and cheese-on children's energy intake at lunch. Ages: 2-5 yrs (mean 3.9 yrs) in a university day-care facility Total n completed = 77 Within-subjects crossover design: All children received a standard breakfast and then a manipulated (energy density) entrée (macaroni and cheese, 300 g) 1 day per week for 7 weeks. Included in the meal was broccoli, applesauce, and milk. 1) Higher-energy-density entrée had 2.0 kcal/g. 2) Lower-energy-density entrée had 1.4 kcal/g.
Weighed food before and after eating 1) Decreasing the energy density of the macaroni and cheese by 30 % resulted in a 25 % (72.3 ± 8.3 kcal) decrease in energy consumed from the macaroni and cheese (P < 0.05) and total lunch intake by 18 % (P < 0.0001).
2) Compared with the higher-energydensity macaroni and cheese, children consumed an additional 10.1 ± 4.2 g of the lower-energy-density macaroni and cheese (P < 0.05).

Osterholt KM, 2007 USA
To test how short-term ad libitum intake is affected by variations in the air content of a snack food Ages: 19-45 yrs (mean 27 yrs) Total n completed = 28 Cross-over design with repeated measures within subjects. Subjects consumed a snack on 4 separate afternoons at least 3 days apart (differing in amount of incorporated air). Both snacks had an energy density of 5.7 kcal/g and contained 56 % of energy as fat. Snacks differed slightly in sodium content (less-aerated: 1.0 % of weight; more-aerated: 1.3 % of weight). Subjects were served the same volume of each snack (approximately 1250 ml), but received 55 % less weight and energy when served the more-aerated snack rather than the less-aerated snack.
Weighed food before and after eating 1) Subjects consumed a mean of 70 ± 17 fewer kcal of the more-aerated snack than the less-aerated snack, equivalent to a 21 % decrease in energy intake (P = 0.0003).
2) By volume, consumption of the more-aerated snack was 239 ± 24 ml greater, equivalent to a 73 % increase in the volume consumed (P < 0.0001).
3) NS in subsequent snack intake (data not reported in paper).
BMI body mass index, n number of participants, NS not significant, RTEC ready to eat cereal, SD standard deviation, SE standard error, SEM standard error of the mean, SSB sugar sweetened beverage, yrs years of age, %E percentage of energy 2) Two packets of almonds (28 g; high in protein, total 11.8 g).
3) Control: No snacks and asked to continue habitual eating pattern.

1) 4 d unweighed diet diaries
2) Anthropometry 1) NS in intake of energy, protein, fat or sugar intake between groups after the intervention (P > 0.05).
2) NS between groups in body weight or waist:hip ratio (P > 0.05).
Acute studies Cross-over study. Subjects came to the laboratory to eat lunch once a week for 6 weeks, for a total of six test sessions. Subjects ate a standard breakfast, and then lunch differed in portion size of beverage (same food).
Weighed food before and after eating 1) Subjects consumed more energy from the caloric beverage (regular cola) when served the large portion (151 ± 8 kcal) vs. the small portion (128 ± 4 kcal, P < 0.05).
3) Food intake at lunch did not differ by either type or portion size of the beverage served (P > 0.05).

Rolls BJ, 2010 USA
To investigate the effects on food and energy intakes of varying the portion size and energy density of a vegetable that was added to a meal or substituted for other foods. Ages: 20-45 yrs (mean 27 yrs) Total n completed = 48 in the substitution study Crossover design with repeated measures. Two studies: In both studies, a midday meal of a vegetable, grain, and meat served to participants once a week. Across the meals, the vegetable was increased in portion size (180, 270, or 360 g) and reduced in energy density (0.8 to 0.4 kcal/g). Substitution study: as the vegetable portion was increased, the amounts of the meat and grain decreased equally (i.e. the total amount of food served at the meal did not change).
Weighed food before and after eating 1) Increasing the portion of the vegetable from 180 to 270 g increased vegetable intake in both studies by a mean ± SE of 34 ± 4 g, P < 0.0001, equivalent to~½ serving.
3) Reducing the energy density of the vegetable led to a small decrease in vegetable consumption (9 ± 3 g, P = 0.002). 4) Substitution study: intakes of the meat and grain decreased as the portion of the vegetable increased from 180 to 270 g; significant decrease in energy intake from energy-dense meat and grain as portion sizes decreased (40 ± 10 kcal; P < 0.0001).

Reformulation strategies Chronic dietary manipulation studies
Six RCTs assessed the effect of food reformulation over a chronic period ( Table 1). Two of the studies reported positive findings, three studies reported null findings, and two studies reported mixed findings. Gatenby et al. [20] found that in adults, using low fat foods rather than high fat foods ad-libitum over 6 weeks, reduced dietary percentage of energy from fat (8 % absolute) and decreased body weight (700 g difference). A further study in this group [21] utilised the same intervention in females only, but also included another arm of low sugar rather than regular sugar foods. Over 10 weeks, there was no change in energy or sugar intake; however, the percentage of energy from total fat was 4 % lower in the low fat group compared to the low sugar and usual food groups [21]. Two studies reported on altering dairy foods [22,23]: Golley et al. [22] showed that using reduced fat rather than regular fat versions of dairy foods for 12 weeks reduced children's fat and saturated fat consumption but not total energy intake. Gunther et al. [23] showed in young women that manipulating dairy products with either moderate or high fat dairy whilst maintaining isocaloric intake had no effect on body weight.
Two RCTs reported on manipulating beverage consumption. In adolescents, Ebbeling et al. [24] showed that compared to usual beverage consumption, consuming non-caloric beverages for 25 weeks led to a 1 MJ lower energy intake; however, there was no change in BMI between groups. In overweight adults, Raben et al. [25] compared ad-libitum food intake and body weight following 10 weeks of supplemental drinks containing either sucrose or artificial sweetener. It was found that those in the sucrose group tripled their energy intake from sucrose drinks, leading to a >2.5 MJ higher overall total daily energy intake. The sucrose drink group gained 1.6 kg over the 10 weeks compared to a 1 kg reduction in weight in the artificial sweetener group [25].

Acute studies
Acute reformulation strategies were assessed in seven studies (Table 1). In young children, consumption of plain milk (256 kJ/100 mL), sucrose-sweetened chocolate milk (416 kJ/100 mL), or aspartame-sweetened chocolate milk (263 kJ/100 mL) had no significant effect on the consumption of other food items offered at that meal [26]. However, children consuming high-sugar ready-toeat cereal (RTEC) consumed almost twice the amount of refined sugar at the breakfast meal (from RTEC and added sugar: 24.4 g vs. 12.5 g) and consumed a greater quantity of RTEC compared to children consuming the lower sugar RTEC (61 g vs. 35 g) [27].
In adults, provision of snacks enriched with either barley beta-glucan [28] or different macronutrient composition [29,30] had no impact on subsequent food or energy intake. Manipulating the energy density of a meal decreased meal intake in children [31], however, keeping the same energy density but increasing the air content of a snack food reduced energy intake of the snack but not subsequent snack energy intake [32].

Substitution strategies Chronic studies
Two studies reported on substitution as a strategy to improve nutrient/food intake and body weight. Substituting high fibre snacks (10-12 g/d fibre) for 8 weeks in children [33] or substituting cereal bars (high in carbohydrate) or almonds (high in protein) for 12 weeks in adults [34] in place of usual snacks, had no effect on macronutrient intakes [33,34] or body weight [34]  To examine appetite and energy intake following ad-libitum consumption of an afterschool snack of raisins, grapes, potato chips, and chocolate chip cookies in children 8 to 11 yrs. Ages: 8-11 yrs (mean 10 yrs) Total n completed = 26 Within-subjects repeated measures design. Children were given: 1) Grapes (301 g) 2) Raisins (65 g) 3) Potato chips (38 g) 4) Chocolate chip cookies (45 g) to consume (within 15 min) as an afternoon snack.
3) Neither energy density nor volume predicted the effect of the snack on cumulative energy intake (Energy densities of raisins (3.04 kcal/g), chips (5.58 kcal/g), and cookies (4.68 kcal/g) were higher than for grapes (0.69 kcal/g), but grapes and raisins had similar effects on cumulative food intake.
n number of participants, NS not significant, SD standard deviation, SEM standard error of the mean, yrs years of age Within-subject design with repeated measures.
Macaroni and cheese meal and fixed portions of unsweetened applesauce, green beans, and whole-wheat roll served with the entrée (different entrée portions: 100 g, 160 g, 220 g, 280 g, 340 g, 400 g) Weighed food before and after eating 1) The percentage of the macaroni and cheese eaten decreased significantly as portion size increased (i.e. 95 % of 100 g portion eaten down to 64 % of 400 g portion eaten, P < 0.001).
2) Increasing portion size (i.e. 100-400 g) increased children's intake of the macaroni and cheese (~100 g up to~250 g, P < 0.01) 3) Increasing portion size decreased the intake of other foods served with the macaroni and cheese such as fruit and vegetables (~180 g other foods consumed with 100 g macaroni and cheese portion, down to~100 g of other foods consumed with 400 g macaroni and cheese portion, P < 0.0001). 4) There was a 61 % higher energy intake at lunch as portion size increased (P < 0.0001). ( Table 2). Consumption of high fibre snacks in children also had no effect on fibre or whole grain intake; however, intake of total grains, and sweets, was higher [33].

Acute studies
Three studies reported on the acute effects of substituting lower energy foods/beverages in a main meal or snack (Table 2). In adults, Flood et al. [35] substituted regular calorie with low or no calorie beverages using a variety of portion sizes. It was found that when a larger portion of beverage was served (540 g vs. 260 g), more energy from the beverage was consumed (~100 kJ); however, there was no difference in the overall meal energy intake [35]. Including extra vegetables into a meal while reducing the portion size of meat and grain, increased vegetable intake (~0.75 servings) but also decreased energy intake from the meat and grain component, and therefore the overall meal [36]. In children, an afternoon snack consisting of either raisins or grapes reduced subsequent dinner energy intake by 200-300 kJ compared to a snack of energy dense potato chips or chocolate chip cookies [37].

Restriction/elimination strategies Acute studies
Nine studies reported on restriction/elimination strategies via manipulating portion sizes of entrées, main meals or snacks (Table 3). Increasing the portion size of a salad entrée reduced subsequent main meal intake by 410 kJ in adults [38]; while quadrupling the portion size of macaroni and cheese (energy density of 1.52 kcal/g (6.36 kJ/g)) within a meal (i.e. 100 g vs. 400 g, with an energy density of 1.12 kcal/g vs. 1.27 kcal/g) increased intake of macaroni and cheese by 150 %, and total meal intake by 61 % [39] in children. This supports the findings of the previous study by Fisher et al. [40] that found doubling the portion size (250 g vs. 500 g) of macaroni and cheese (1.42 kcal/g) in a meal increased macaroni and cheese intake by 33 % and total meal intake by 15 % in pre-school children.
In adolescents, serving the same take-away meal in different portion sizes (4 smaller servings presented at a single time point vs. portioned into 4 smaller servings presented at 15-min intervals) had no effect on overall take-away food intake compared to the meal served in one large portion [41]. In younger children, a large portion size of a low calorie snack (applesauce) or high calorie snack (chocolate pudding) increased subsequent meal energy intake compared to a smaller portion size of the same foods [42].

Table 4 Description of included supplementation studies (Continued)
2) To observe if the relative satiating efficiency of protein and carbohydrate changes with preload time interval, specifically by impacting energy adjustment. Ages: 18-34 yrs (mean 23 yrs) Total n completed = 18 males 1) Control test drink (a low-energy apricot and peach fruit drink). 2) Carbohydrate test drink (a higher energy version of the apricot and peach control drink, 97.6 % E carbohydrate).
3) Protein test drink (a higher energy version of the apricot and peach control drink, 50 % protein). 300 ml drinks administered at two time intervals of 120 min and 30 min before the ad libitum test meal. Other meals provided (breakfast, snack, pre-lunch).

Flood JE, 2007 USA
To examine further the effects of consuming different forms of a low-energy-dense soup as a preload on subsequent test meal intake and total energy intake at the meal (soup preload + test meal). Ages: 18-45 yrs (mean 26 yrs) Total n completed = 60 Subjects came to the laboratory for lunch once a week for 5 weeks. Each week, one of four compulsory preload soups containing the same energy density (0.33 kcal/g; 1.4 kJ/g), or no preload, was consumed prior to lunch (pasta and sauce). One and a half (350 ml) of soup was served to women, and two cups (475 ml) of soup was served to men. A test meal was consumed ad-libitum 15 min after the soup was served: 1) Broth and vegetables served separately.

Rolls BJ, 2010 USA
To investigate the effects on food and energy intakes of varying the portion size and energy density of a vegetable that was added to a meal or substituted for other foods. Ages: 20-45 yrs (mean 27 yrs) Total n completed = 49 in the addition study.
Crossover design with repeated measures. Two studies. In both studies, a midday meal of a vegetable, grain, and meat was served to participants once a week. Across the meals, the vegetable was increased in portion size (180, 270, or 360 g) and reduced in energy density (from 0.8 to 0.4 kcal/g). Addition study: as the vegetable portion was increased, the amounts of the meat and grain were not changed (i.e. total amount of food served at the meal was increased).
Weighed food before and after eating 1) Increasing the portion of the vegetable from 180 to 270 g increased vegetable intake in both studies by a mean ± SE of 34 ± 4 g, equivalent to about half a serving. 2) Doubling the portion of the vegetable (180 to 360 g) increased vegetable intake by 60 ± 5 g (49 ± 4 %).
3) Addition study: Intake of the meat and grain did not differ as the vegetable portion size was increased; there was no significant change in energy intake from the meat and grain and no significant difference in total energy intake at the meal.
BMI body mass index, n number of participants, NS not significant, RCT randomized controlled trial, RTEC ready to eat cereals, SE standard error, SEM standard error of the mean, yrs years of age, %E percentage of energy In children, halving the item size of candies [43] or cookies [44] led to a reduction in intake of 60 kcal and 68 kcal, respectively. In adults, consuming portion controlled 100 kcal snack packs (19 g-26 g) of crisps, crackers, pretzels or cookies led to a weekly caloric deficit of 841 kcal/week compared to larger, usual snack packs (187 g-369 g) [45]; while consuming smaller portions of a sandwich (6, 8, 10, or 12 in.) also reduced energy intake [46].

Supplementation strategies Chronic studies
Seven chronic studies reported on the effect of supplementing the diet with specific foods/beverages on micronutrient intake or body weight, in which the majority of studies reported null results (Table 4). Supplementing 43 g/d of almonds for 4 weeks in adults with type 2 diabetes [47] or walnuts for 12 weeks [48] did not affect micronutrient intake [47], body weight [47,48] or body composition [48]. Six months of almond consumption (range of intakes: 42-71 g/d) significantly increased micronutrient intakes and decreased intakes of sodium and sugars by more than 10 % [49].
Consumption of a grain bar high in carbohydrate prior to the evening meal for 8 weeks decreased body weight compared to consumption of a peanut bar high in protein (mean difference 1.5 kg), yet there was no difference between groups in reported energy or protein intake [50]. Consuming RTEC in amounts ranging from 33-60 g/d did not consistently improve dietary intake or anthropometric measures [51][52][53], and these studies did not report whether and what types of foods RTEC supplemented, except in the study by Kirk et al. [51] who supplemented bread and toast. The study by Kirk et al. [51] also reported a greater decrease in the percent contribution of biscuits and cakes to mean daily energy intake compared with the control group who received no advice to consume RTEC for 12 weeks (-6.0 %E vs -1.4 %E); one or two servings of RTEC for 12 weeks did not affect body weight in children [53]; but when RTEC and milk was consumed in place of the usual evening snack (food type not reported), an increase in evening energy intake (+500 kJ) was found [52].

Acute studies
Six studies reported on supplementation using food or beverage pre-loads (low calorie food or beverage) before or within a meal, where most studies reported a reduction in meal energy intake (Table 4). In adults, a snack of prunes before a meal reduced subsequent energy intake at the meal compared to a snack of white bread, however the reduced energy intake did not remain over a 24-h period [54]. A 500 ml water pre-load significantly reduced meal energy intake by 13 % [55] and 8 % [56] in older adults; however a 375 ml or 500 ml water pre-load in younger women and men did not alter meal and energy intake [56]. Comparatively, consumption of an apricot and peach drink high in protein led to a reduction in subsequent test meal food intake in young adults compared to the same apricot and peach drink that was higher in carbohydrate, or a low energy control version [57].
In adults, consumption of 1.5-2 cups of soup in a variety of flavours (with the same energy density) before a meal, reduced meal energy intake by 20 %; however the type of soup had no significant effect on test meal intake or total meal energy intake [58]. In the same study described earlier, Rolls et al. [36] also showed that including extra vegetables into a meal without reducing the portion size of the meat and grain component increased vegetable intake (~0.75 servings); however the addition of more vegetables did not significantly affect meal energy intake.

Nutrition education/messages strategies
Four discrete intervention studies were identified that assessed nutrition education/messages strategies to improve nutritional intake and/or decrease discretionary choices ( Table 5). The study by Sichieri et al. [59] encouraged simple messages "to consume water instead of sugar-sweetened beverages" in 9-12 year olds over 7 months. Overall, there was a significant reduction in Table 5 Description of included nutrition education/messages studies (Continued) 1) Social norm: 'Students eat less junk food than you might realise. Most students limit how much junk food they are eating to 1 or less than 1 serving/d (based on a 2012 study). 2) Health condition: 'Reducing junk food intake is good for your health. Limiting junk food to 1 or less than 1 serving a day is part of a healthy diet (based on a 2012 study). 3) Control: Message emphasised the importance of preparing in Ladvance for exams.
BMI body mass index, CI confidence interval, n number of participants, NS not significant, RCT randomized controlled trial, SE standard error, yrs years of age carbonated beverage intake (~390 ml/d) compared with the control group who received no information on substitution. However, this did not equate to a reduction in weight or BMI, potentially due to slight increases in fruit juice intake in both groups (35-84 ml/week). In the study by Alinia et al. [60] free access to a fruit basket for 5 months at various workplaces increased the consumption of fruit by nearly 1 serving/d vs control; intake of dietary fibre increased by 3 g/d and added sugar decreased by 2 teaspoons/d in the intervention group vs baseline. In contrast, provision of a fruit tuck shop in primary and junior schools in the UK had no effect on increasing fruit intake over the 9 month period [61]. However, in the schools with a 'no food' or 'fruit only' policy, the fruit tuck shop intervention had a greater impact than in schools with no restrictions, whereby there was 0.37 portions greater consumption of fruit in schools with a fruit only policy compared with 0.14 portions with a no food policy and -0.13 where there were no restrictions [61].
The final study assessed positive and negative nutrition messages on food intake [62]. Young adults who read messages about the health effects of junk food (i.e. "reducing junk food intake is good for your health") or social expectations (i.e. "students eat less junk food than you might realise") consumed fewer high calorie snack foods compared with those who read messages unrelated to junk food consumption; however, there was no effect on fruit and vegetable intakes between groups [62].

Discussion
This scoping review assessed the impact of discrete dietary manipulation strategies that were potentially applicable to reducing discretionary choices in adults and children. Although no definitively effective single discrete strategy was identified, there were a number that show potential, including reducing portion size, reformulation of fat (from higher to lower saturated fat), substituting high fibre snacks or low/no-caloric beverages into the diet, supplementing healthy nutrient dense foods such as nuts and wholegrain cereals, using a combination of permissive and restrictive education messages, and incorporating cheap and/or complimentary healthier foods in the workplace. The findings of this review can inform future research, in particular adapting these strategies in the development of interventions to reduce excess intake of discretionary choices.
A limited number of chronic dietary manipulation studies addressed the use of reformulation as a strategy to improve diet quality. While a change from regular fat foods to reduced fat foods appears to be a useful strategy in terms of lowering fat intake [20,21], the impact on overall energy intake and bodyweight are not clear [23]. Consumption of non-caloric beverages appears to reduce overall energy intake in adolescents [24] and adults [25], however reductions in body weight tend to occur only in adults.
Short-term consumption of sucrose compared to artificially sweetened milk did not alter subsequent food intake in children [26], and consuming snacks with different macronutrient profiles [28][29][30] or consuming a more-aerated snack [32] had no effect on subsequent meal energy intake. Nevertheless, since a more aerated snack did result in less energy consumed from the snack, and reformulating energy dense meals for lower density versions reduced energy intake at that meal [31], manipulating food density has potential for reducing discretionary choices and should be investigated further.
There were mixed results regarding substitution strategies. Chronic studies substituting high fibre snacks [33], cereal bars or almonds [34] for usual snacks did not improve micronutrient intake, and high fibre snacks in children actually increased intake of sweets. Interestingly, in the high fiber snack study by Brachula et al. [33] participants in the intervention group did not usually consume snacks. Thus incorporating two eating occasions of a high fiber snack to their usual routine, with no change in total grain but an increase in whole grain intake, indicates that the children were likely displacing refined grains and hence discretionary choices, and therefore undertaking a positive behavior change.
Smaller portions of non-caloric beverages reduced meal energy intake [35], and increasing vegetable portions within a meal increased vegetable consumption and decreased meat and grain intake [36]. Reducing energy density of foods and beverages appears to be an appropriate strategy, at least in the short term, for reducing energy intake. This approach requires investigation in contexts when the meal consists heavily of discretionary choices, such as fast food outlets, to review the impact on reducing energy intake and improving nutrient intake. Studies assessing low/no-caloric beverages or raisins as a substitute for high calorie beverages and crisps/cookies, respectively, suggest these may be effective strategies for displacing discretionary choices. The long-term impact of such strategies remains to be evaluated.
Most restriction/elimination studies were effective in reducing short-term (within-meal) energy intake. These studies typically reduced the portion size of specific meal components or increased the portion of a low-density meal (e.g. a salad), which subsequently reduced energy density and energy intake [38][39][40]42]. The theory behind this is that lower energy density foods might displace intake of some of the higher energy-density foods at a meal or at a subsequent meal, such as discretionary choices, through its satiating effect [38,63]. However, one study showed that reducing the energy content of various entrées led to an increase in consumption of discretionary foods, and while this resulted in a decrease in daily energy intake, ratings of hunger were increased [64]. Increased hunger could translate into increased consumption of discretionary choices. Large portions of low energy dense foods, which provide a feeling of fullness, may be a more effective strategy to moderate energy intake and decrease discretionary choices at a meal or between meals. Halving item sizes of candies [43] or cookies [44], or reducing portion sizes of discretionary choice snacks [45], or a sandwich [46] was also an effective strategy for reducing energy intake. Although these are effective in an acute setting, the longer-term effects of restricting portion or snack item sizes should be investigated in the context of reducing subsequent discretionary choices and energy intake.
Results were mixed for the effect of chronic supplementation strategies on nutrient profile and body weight [47][48][49][50][51][52][53]. Given the nutrient dense profile of nuts and wholegrain foods, supplementing the diet with these foods compared to discretionary choices would support overall diet quality. The net benefits of supplementing the diet with RTEC are less clear since many are high in added sugar and salt.
Acute supplementation studies generally have a positive effect on reducing subsequent meal energy intake and could be a strategy to reduce discretionary choices after a main meal. A water pre-load could be a useful strategy to decrease energy intake in older adults who are overweight or obese, however its effect on replacing or reducing higher calorie foods in the diet requires further investigation, so too does its effect in non-obese younger and older adults. Including low calorie, nonenergy dense meals prior to a discretionary choices main meal has a beneficial impact on subsequent meal energy intake.
Free access to fruit is a positive strategy to increase fruit and fibre intake, and reduce added sugar intake in adults [60], and therefore potentially discretionary choices; however, the impact of this in children was ineffective and requires nutrition policies within the school to produce any positive effect [61]. Interestingly, the effect in adults occurred without any guidance or nutrition education and only with the provision of a free fruit basket at the office workplace, yet this was only one study and therefore requires further investigation. Nevertheless, differences in outcomes/effects between these studies may be the result of older adults having a better understanding on the benefits of fruit intake compared to younger children; possibly lower accessibility of fruit at a tuck shop where students are required to get the fruit themselves compared to having it in the classroom; or potential problems in the validity and reliability of self-reported fruit intake between adults and children.
Only a limited number of studies utilised nutrition messages to change nutritional intake. Permissive messages to increase water consumption are beneficial to reducing sugar sweetened beverage intake [59] and restrictive messages about the impact of junk food on health and its social acceptance, also appears to reduce discretionary food intake [62]. However, the longer-term implications of these messages are unknown.
Strengths and limitations to this study are acknowledged. This review is the first to comprehensively investigate the impact of discrete dietary strategies that were potentially applicable to reducing discretionary choices, on discretionary choices intake and/or the strategies targeted health/nutrition effects. The review captured a wide range of studies as the inclusion criteria was broad and was not limited by age; and the search was extensive by way of multiple databases. Although we relied on published content and did not contact authors or search grey literature, the purpose of a scoping review is to be broader than a systematic review so that key findings can be later assessed in a systematic process. Limitations include the small number of studies that utilized similar dietary strategies, thus drawing firm conclusions on study outcomes/effects was difficult; and although not a key feature of scoping review [16] we did not assess study bias, however this review is transparent and has not placed undue emphasis on one study relative to another. As this study did not assess intake of discretionary choices per se for all identified strategies, the impact on this outcome was not always able to be determined; further research is required to investigate how the alternative targeted health/nutrition effects for these strategies are affected by and/or affect intake of discretionary choices. Another limitation is that only one author reviewed the potentially relevant studies; however, initial discussion around inclusion/ exclusion criteria and final studies extracted involved all authors, thereby minimizing any error in included studies. Finally, as there was heterogeneity between study populations, it remains unclear whether some strategies are more feasible in particular population groups.

Conclusions
No single discrete strategy was identified that definitively reduced discretionary choices in adults or children. However, restriction/elimination strategies (specifically reducing portion size) were consistently beneficial for reducing energy intake, at least in the acute setting. Reformulation of fat (from higher to lower fat) could be useful to reduce saturated fat intake, and substituting high fibre snacks, fruit, or low/no-caloric beverages for discretionary choices may be effective. Supplementation strategies where nutrient dense foods such as nuts and wholegrain cereals are consumed in place of discretionary choices support an improved overall diet quality. Regarding education strategies, a combination of permissive and restrictive messages may reach a large audience to effectively modify behavior, while incorporation of cheaper and/or freely accessible healthier foods in the workplace may be a further option to reduce discretionary choices and support diet quality in adults. Longerterm, well-controlled and larger studies are required to confirm the effectiveness of the proposed strategies and assess their impact in multi-component interventions.