**Eating Vegetables First at Start of Meal and Food Intake among Preschool Children in Japan**

#### **Jiaxi Yang <sup>1</sup> , Yukako Tani <sup>2</sup> , Deirdre K. Tobias 3,4, Manami Ochi <sup>5</sup> and Takeo Fujiwara 2,\***


Received: 13 May 2020; Accepted: 11 June 2020; Published: 12 June 2020

**Abstract:** Eating behavior is an important aspect for dietary quality and long-term health. This study examined associations between eating vegetables first at a meal and food intakes among preschool children in Tokyo, Japan. We used cross-sectional data of 135 preschool children from seven nursery schools in Adachi City, Tokyo, Japan. Caregivers completed a survey on child's eating behaviors and a diet questionnaire. Linear regression was used to examine frequency of eating vegetables first at a meal and food intakes; percent difference and the corresponding 95% confidence interval (95% CI) were presented. Overall, 25.2% of children reported eating vegetables first at a meal every time, 52.6% sometimes, and 22.2% not often or never. In the multivariate analysis, higher vegetable intake remained significant after adjusting for other covariates (compared with the group of eating vegetables first not often or never, the group reported sometimes: 27%, 95% CI: 0–63%; the group reported every time: 93%, 95% CI: 43–159%). No significant difference in intake by frequency categories of eating vegetables first was observed for other food groups, including fruits, meat, fish, cereals, and sweets. Children eating vegetables first at a meal more was associated with higher total intake of vegetables compared with children who did not eat vegetables first, among Japanese preschool children.

**Keywords:** dietary habit; vegetable consumption; food intake; preschool children; Japan; nutrition

### **1. Introduction**

A healthy diet is an essential component for meeting proper nutrition requirement for optimal body growth and body weight during childhood [1–3]. A poor diet can be a result of an imbalanced diet by overeating low-nutrient-dense foods such as refined carbohydrates and sweets and failing to consume other foods with high nutrient density, such as fruits, vegetables, and healthy meats [4]. Having a poor diet by consuming an excessive amount of unhealthy food and an insufficient amount of healthy food can lead to both short-term and long-term negative consequences on children development, such as obesity, nutrition deficiency, and insufficient body growth [5–7]. In addition to its beneficial role on body growth in childhood, a healthy diet pattern formed in childhood will also benefit long-term health if the pattern is maintained in later life [8,9].

Eating behaviors can reflect and even potentially influence overall dietary quality [10,11]. Particularly, choice of foods consumed during the early phase of a meal with respect to energy

intake and intakes of different food groups has been examined in several studies conducted in the United States (U.S.) among people of different age groups. For instance, a randomized cross-over study conducted among adults in the U.S. found that consuming a first course with low-energy-dense salad enhanced satiety and led to reduced energy intake in the subsequent courses [12]. The beneficial role of consuming low-energy-dense foods at the beginning of a meal, particularly vegetables, has also been examined among children. More importantly, studies have revealed that consuming vegetables during the early phase of a meal led to not only reduced meal energy intake but also increased vegetable intake. A study conducted at a daycare center in Pennsylvania by the same research group found that serving low-energy-dense vegetable soup during the early phase of a meal led to reduced intake of energy-dense entrée and increased vegetable consumption at the meal [13]. Another cross-over study examined portion size of vegetables served at the start of meal among preschool children and reported that increasing portion size of vegetables at the start of a meal led to greater vegetable consumption without increasing meal energy intake [14]. Therefore, serving vegetables at the start of a meal and avoiding presence of competing foods that are less healthy may be advocated as an effective strategy to promote higher vegetable intake in preschool children [14].

Lately, there has been a trend of decreasing in vegetable intake and increasing in meat intake among the general Japanese population. The annual survey on citizens' health and nutrition published by the Ministry of Health, Labor and Welfare in Japan suggested that, compared with 10 years ago, the daily intakes of vegetables and fruits among Japanese had been decreasing (277.4 g/day for vegetables with a drop of 18.4 g/day, 110.3 g/day for fruits with a drop of 22.0 g/day), whereas the daily intake of meat had been increasing (80.7 g/day with an increase of 6.7 g/day) [15]. Since one's dietary habit is often formed as early as in childhood, identifying children with insufficient vegetable intake and subsequently developing dietary guidelines and interventions that will likely lead to increased vegetable intake and other healthy food intakes should be considered as a useful strategy to address the current diet problem in Japan [9,16].

A typical Japanese meal set usually consists of a staple food (mostly rice), three side dishes, and a soup served all at once [17]. Therefore, the sequence of food consumed can vary from person to person. As suggested by the earlier studies conducted in the U.S. on consuming vegetables during early phase of a meal, evaluating frequency of eating vegetables first at a meal with respect to various food intakes will be useful to determine if early consumption of vegetables at a meal is informative of assessing vegetable consumption and possibly overall dietary quality among Japanese children.

We used data from a cross-sectional study of preschool children in Japan and evaluated the association between frequency of eating vegetables first at a meal and intakes of different food groups, including vegetables, fruits, meat, fish, cereal, and sweets. We hypothesized that frequently consuming vegetables first at a meal would be associated with higher intake of vegetables and other healthier foods.

#### **2. Materials and Methods**

#### *2.1. Study Population*

We used data from a cross-sectional study of 135 preschool children from Adachi City, Tokyo, Japan in 2017. The study was initiated as a component of a health promotion campaign known as "Eat Vegetable First at Meals", which was launched in Adachi City in 2013. Children in 5-year-old class from seven licensed public nursery schools in Adachi City were invited to participate in the study. Teachers at the nursery schools explained to the children's caregivers about the study and distributed the study questionnaires. The questionnaires included a survey on regular eating behaviors of the participated child and a brief-type diet history questionnaire developed for Japanese preschool children aged 3–6 years (BDHQ3y) [18]. Participants were informed that participation in the study was voluntary and returning the completed questionnaires indicated their consent to participating in the study. The questionnaires were distributed to 165 caregivers, out of which 135 caregivers returned the questionnaires in sealed envelopes via each nursery school (response rate: 81.8%). Use of the

data for this study was approved by the Ethics Committee of Tokyo Medical and Dental University (No. M2016-284).

#### *2.2. Survey on Dietary Behavior*

The dietary behavior survey was filled out by the caregiver of the participated child. It aimed to assess the regular eating and cooking behaviors of both the caregiver and the child. For our study, we were particularly interested in the frequency of eating vegetables first at a meal. In the dietary behavior survey, frequency of eating vegetables first at a meal was assessed in the following question: "how often does your child eat the first bite from vegetables at a given meal?" The caregiver was asked to circle the answer that best applied to his or her child from the following options: "every time", "sometimes", "not often", or "never". Since there were only 4 children who reported "never" for consuming the first bite from vegetables at a meal, we collapsed the groups of children who reported "not often" or "never" into one group in the analysis.

#### *2.3. Survery on Food Intake*

Food intakes were assessed using a brief-type diet history questionnaire for Japanese preschool children aged 3–6 years (BDHQ3y), which was developed based on the adult version of a self-administered diet history questionnaire that has been widely used in a range of epidemiologic studies for assessing food intakes in Japanese adults [19]. The caregiver reported the regular food intakes of his or her child during the preceding month by filling out BDHQ3y. The validity of BDHQ3y has been previously tested [18]. Details of BDHQ3y have been described elsewhere [18,20]. Briefly, BDHQ3y is a four-page questionnaire which reflects the typical Japanese dietary pattern, and it includes four sections to assess the food intake frequency: (1) 57 food and nonalcoholic beverage items; (2) daily intakes of rice (the most widely consumed staple food in Japan) and miso soup (widely consumed soup type in Japan); (3) usual cooking methods; and (4) general dietary behaviors. The daily intakes of 66 food items, total energy intake, and nutrient values are then estimated using an ad hoc computer algorithm, which takes into account the age-specific portion size using a specific weighting factor to adjust for the effect of age on the portion size consumed.

We considered the following food groups in our analysis: vegetable, fruit, meat (excluding fish), fish, cereal (including rice, noodles, and bread), and sweets. In the BDHQ3y, vegetable intake was collected based on the consumptions of dark green-leaf vegetables, cabbage, carrots, pumpkins, rooted vegetables, tomatoes, and mushrooms. Fruit intake was collected based on the commonly consumed fruits, except juice and jam made from fruits. Meat intake was collected based on the consumptions of chicken, pork, beef, processed meat, and animal liver. Fish intake was collected based on the intakes from fresh fish, canned fish, dried and salted fish, and food made from fish. Cereal intake was categorized into three sub-groups, and the intake of each group was calculated respectively: rice (including plain white rice, barley, whole grain rice, brown rice, and multigrain rice); noodles (including buckwheat noodles, Japanese wheat noodles, Chinese noodles, fried noodles, instant noodles, and western-style noodles); and bread. Intake on sweets was assessed from the following food sources: western sweets, Japanese sweets, ice cream, chocolate, and other sweet snacks. For each of the food groups, daily intakes of the food items were then summed and a value of total daily intake (g/day) was obtained, respectively. The summed value of the daily intakes from rice, noodles, and bread was reported as the daily intake of cereal. In addition, total energy intake (kcal/day) was also assessed in our study.

For each of the food groups, from the value of daily intake (g/day) estimated based on the nutrient database, we divided the food intake by the total energy intake and then multiplied the value by 1000 to derive the nutrient density (g/1000 kcal per day), so the food intake was represented as a dietary composition (a percentage from daily energy intake) rather than the absolute intake value for each child [4,21]. Nutrient density values were then log-transformed to account for potential non-normality.

#### *2.4. Covariates*

Information on the child's date of birth, sex, height, and weight were filled out by the caregiver in the BDHQ3y questionnaire. Information on the following covariates was additionally collected from the dietary behavior survey: number of people in the household, household economic status (in good standing, normal, indigent), parents' job (self-owned business, full-time, part-time, other), caregiver-rated child's physical health (good, normal, poor), and child's physical activity status, which was assessed by the frequency of conducting physical exercise that was longer than 30 min (almost every day, 5–6 times a week, 3–4 times a week, 1–2 times a week, rarely or never). In addition, the caregiver was asked to recall the average frequency of the child's vegetable consumption (almost at every meal, twice in a day, less than once in a day). We considered the covariates mentioned above as potential confounders for the association between frequency of eating vegetables first at a meal and food intakes and subsequently examined them in the analysis. For any question that was not answered by the caregiver, the missing value was set to the most commonly reported response.

#### *2.5. Statistical Analysis*

The main exposure of interest was frequency of eating vegetables first at a meal. The exposure was evaluated as a categorical variable with the following category: every time, sometimes, and not often or never. The group of children reported as not often or never eating vegetables first at a meal was set as the reference category. We first examined the association between frequency of eating vegetables first at a meal and other demographic or lifestyle-related covariates by conducting a chi-square test for a categorical covariate and analysis of variance for a continuous covariate. For the univariate analysis, the frequency of eating vegetables first at a meal was included as the only predictor in the univariate model. For the multivariate analysis, covariates with a *p*-value less than 0.05 from the chi-square test or the analysis of variance test were considered as significant and were subsequently adjusted in the multivariate model: age (months), physical health status (good, normal, poor), frequency of consuming vegetables (almost at every meal, twice in a day, less than once in a day). For both univariate and multivariate analyses, we used a linear regression model and examined the association between frequency of eating vegetables first at a meal and intake of the food groups. Since the outcome of food intake was on a logarithmic scale, coefficients and standard errors were back-transformed to the original scale with an interpretation of percent difference in the daily nutrient density for a given group compared to the reference group. All analyses were conducted using STATA version 13 (STATA Statistical Software: Release 13. College Station, TX, USA: StataCorp LP).

#### **3. Results**

Characteristics of the overall study sample by the reported frequency of eating vegetables first at a meal are summarized in Table 1. Our study included 135 Japanese preschool children with average age of 6.4 years (SD = 0.3 years) and average body mass index (BMI) of 15.5 kg/m<sup>2</sup> (SD = 1.8 kg/m<sup>2</sup> ). With respect to the frequency of eating vegetables first at a meal reported by the caregiver, 34 (25.2%) participants reported "every time", 71 (52.6%) participants reported "sometimes", and 30 (22.2%) participants reported "not often or never". Compared with the other two groups, the group of children reported as eating vegetables first at a meal every time had slightly higher BMI, a greater proportion of parents who owned self-business or had full-time job, better caregiver-rated physical health, more frequent physical activity, as suggested by a lower proportion of children who rarely or never conducted exercise that was longer than 30 min, and more frequent vegetable consumption (Table 1).


**Table 1.**Population characteristics and characteristics by frequency of eating vegetables first at a meal.

1 *p*-value from chi-square test for categorical covariate and analysis of variance for continuous covariate is presented. 2 SD: standard deviation. 345

 Age is presented in years by dividing age in months by 12. BMI: body mass index. Median is presented for the number of people in the household.

Daily intakes of the major food groups (g/1000 kcal, except for total energy intake) with respect to the frequency of eating vegetables first at a meal are summarized in Table 2. As Table 2 suggests, we observed higher total vegetable intake independent of total energy intake in the groups of children reported as more frequently eating vegetables first at a meal (every time: 147.8 g/1000 kcal, sometimes: 88.7 g/1000 kcal, not often or never: 68.0 g/1000 kcal). Higher intakes of fruits and fish and lower intakes of cereal and sweets were also observed in the group of eating vegetables first every time compared with the other two groups with the lower frequency. We did not observe a difference in total energy intake across the three groups (Table 2).


**Table 2.** Summary of major food group intakes by frequency of eating vegetables first at a meal.

<sup>1</sup> Cereal intake was calculated as the summed value of intakes from rice, noodles, and bread.

We present our main analysis results in Table 3. In the univariate analysis, we observed a significant association between frequently eating vegetables first at a meal and higher total vegetable intake (Table 3). Compared with the group of eating vegetables first at a meal not often or never, we observed 46% (95% CI: 14–88%) higher vegetable intake in the "sometimes" group and 139% (95% CI: 79–219%) higher vegetable intake in the "every time" group. In addition, significantly higher intakes of fruits and fish and lower intake of bread were also observed in the group of eating vegetables first at a meal every time compared with the reference group (Table 3).

After adjusting for the relevant covariates (age, physical health status, frequency of consuming vegetables) in the multivariate model, the association between frequently eating vegetables first at a meal and higher intake of vegetables was slightly attenuated, but it still remained statistically significant (Table 3). Compared with the group of children reported as eating vegetables first at a meal not often or never, the "sometimes" group had 27% (95% CI: 0–63%) higher vegetable intake, and the "every time" group had 93% (95% CI: 43–159%) higher vegetable intake. We did not observe significant associations between frequency of eating vegetables first at a meal and food intake for the remaining food groups that we examined, including fruits, meat, fish, cereal, and sweets (Table 3).


**Table 3.** Results on association between frequency of eating vegetables first at a meal and types of food intake <sup>1</sup> .

<sup>1</sup> Food intake was calculated as nutrient density (g/1000 kcal per day) for each food type on the natural log scale (nutrient density was calculated by dividing reported daily food intake (g/day) by total energy intake and then multiplying by 1000). <sup>2</sup> Frequency of eating vegetables first at a meal was included in the univariate model. <sup>3</sup> Multivariate model was adjusted for age (months), physical health status (good, normal, poor), and frequency of consuming vegetables (almost every meal, twice in a day, less than once in a day). <sup>4</sup> Cereal intakes were calculated as the summed value of intakes from rice, noodles, and bread.

#### **4. Discussion**

In our analysis of 135 Japanese preschool children, we found that frequently eating vegetables first at a meal was associated with higher intake of vegetables, and suggestively higher intakes of fruits and fish and lower intake of bread, independent of energy intake. To our knowledge, this is the first study examining frequency of eating vegetables first at a meal and its association with intake of various food groups among Japanese preschool children. The multi-dish style in the Japanese meal culture allowed us to closely examine the role of eating vegetables first at a meal on the intakes of commonly consumed foods among Japanese preschool children.

Our results were consistent with the previous study findings on consuming vegetables during early phase of a meal and greater vegetable consumption. A cross-over study conducted among preschool children in the U.S. found that doubling the portion size of vegetables as the first course led to a subsequent 47% increase in vegetable consumption at a given meal [14]. The same research group conducted other studies examining the role of serving vegetable dishes in the early phase of a meal. They reported similar findings that consuming a vegetable dish early led to increased meal vegetable intake and decreased meal energy intake [12,13]. Based on these study findings, placing vegetable

dishes earlier during the course of a meal can be advocated as a strategy to encourage vegetable intake among children who have insufficient vegetable consumptions. Indeed, serving-vegetable-first has been demonstrated as an effective way to increase vegetable consumption among school children in other settings [22].

In addition to the significant association with higher vegetable intake, we also observed that frequently eating vegetables first at a meal was associated with suggestively higher intakes of fruits and fish, and it was not associated with higher intakes of the food groups that were considered less healthy, such as bread and sweets. In fact, compared with the group of children reported as not often or never eating vegetables first at a meal, the group of children eating vegetables first every time had suggestive lower intakes in bread and sweets (Tables 2 and 3). Further, eating vegetables first more frequently at a given meal did not seem to be associated with increased meal energy intake, which was also consistent with the previous study findings (Table 2) [14]. Therefore, it may be implied that frequently eating vegetables first at a meal was not associated with higher intake of unhealthy food or higher intake of energy. Considering the healthy benefits of eating vegetables, fruits, and fish and current dietary guidelines on limiting intake of refined carbohydrates, our results suggested the possibility of using frequency of eating vegetables first at a meal as a useful tool to assess the overall dietary quality among Japanese preschool children [16,23–31].

Our study provided preliminary evidence that assessing frequency of eating vegetables first at a meal might serve as a convenient and useful method for the policymakers to identify the population of children with generally low vegetable consumption and possibly suboptimal diet quality, and to subsequently develop community interventions or guidelines to improve their diet. Findings from our study also provide useful insights for future interventional studies to further pursue this area of research in order to draw causal conclusions on frequency of eating vegetables first at a meal and increasing total vegetable intake among preschool children in Japan.

There are some limitations in our study. First, with a small sample size (*n* = 135), the statistical power of our study was limited. Therefore, the null associations observed in some food groups may be interpreted as either no association or a possible association but underpowered. In addition, since we collapsed the groups of "not often" and "never" into one group due to the limited sample size, we were unable to separately examine the food intakes for those two groups. Furthermore, a small sample size may limit the generalizability of our results. Therefore, future studies with greater sample size and sufficient statistical power should be conducted to address such limitations. Second, given the cross-sectional nature of the study, our results can only be interpreted as findings of associations. Therefore, we cannot make the causal interpretation that eating vegetables first at a meal will lead to higher vegetable intake. However, our results still suggest that frequently eating vegetables first at the start of a meal is informative of higher total vegetable consumption among Japanese preschool children. Lastly, similar to other nutritional studies, diet was likely to be measured with errors, as the validity and reliability of BDHQ3y filled out by the caregiver may not be high enough to accurately capture the regular dietary pattern of the Japanese preschool children. With respect to the dietary behaviors survey, since it was structured as questions with a reasonable number of choices, misclassification was likely to be low.

#### **5. Conclusions**

In conclusion, our study on 135 Japanese preschool children suggested that compared with children who did not eat vegetables first, eating vegetables first at a meal more was associated with a higher total vegetable intake. Larger-scale studies with a geographically diverse population of preschool children should be conducted to further confirm our findings. Future intervention studies or randomized trials are warranted to further examine the causal role of eating vegetables first at a meal on increasing healthy foods consumptions among Japanese preschool children.

**Author Contributions:** All authors contributed to the study design. T.F., Y.T., and M.O. collected the data. J.Y. conducted data analysis and led the writing of the manuscript. T.F., Y.T., and D.K.T. provided input into data

analysis approach. All authors provided input into interpretation of the study results and have read and approved the final manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was funded by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (JSPS KAKENHI Grant Number 16H03276, 19K14029, and 16K21669).

**Acknowledgments:** The authors thank all the individuals who participated in the study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Dietary Habits in Children with Respiratory Allergies: A Single-Center Polish Pilot Study**

**Eliza Wasilewska 1,\* , Sylwia Małgorzewicz <sup>2</sup> , Marta Gruchała-Niedoszytko <sup>2</sup> , Magdalena Skotnicka <sup>3</sup> and Ewa Jassem <sup>1</sup>**


Received: 13 May 2020; Accepted: 21 May 2020; Published: 23 May 2020

**Abstract:** Background: The rising trend in allergic diseases has developed in parallel with the increasing prevalence of obesity, suggesting a possible association. The links between eating habits and allergies have not been sufficiently clarified. Aim: To evaluate the nutritional status, eating habits, and risk factors of obesity and pulmonary function in children with allergic rhinitis. Materials and methods: We evaluated 106 children with allergic rhinitis (mean age 12.1 ± 3.4 years; M/F 60/46) from the Department of Allergology. Clinical data were collected regarding allergies, physical activity, nutritional status (Bodystat), dietary habits (Food Frequency Questionnaire validated for the Polish population), skin prick test with aeroallergens (Allergopharma), and spirometry (Jaeger). Results: All children suffered from allergic rhinitis; among them, 43 (40.6%) presented symptoms of asthma. There were differences between children with only allergic rhinitis (AR group) and children with both rhinitis and asthma (AA group) in pulmonary function (forced expiratory volume in one second (FEV1) 100 ± 11 vs. 92.1 ± 15.0; *p* < 0.05). A total of 84 children (79%) presented a normal body mass index (BMI) (10–97 percentile), 8 (7.5%) were underweight, and 14 (13.5%) were overweight or obese. There were no differences in body composition between the AR and AA groups. Incorrect eating habits were demonstrated by most of the children, e.g., consumption of three or fewer meals in a day (38%), sweets every day (44%), snacking between meals every day (80%), and eating meals less than 1 h before bedtime (47%). Compared to the AR group, the AA group was more likely to eat more meals a day (*p* = 0.04), snack more often (*p* = 0.04), and eat before sleeping (*p* = 0.005). Multiple regression analysis showed a significant association between high BMI and snacking between meals and low physical activity (adjusted R<sup>2</sup> = 0.97; *p* < 0.05). Conclusions: The risk factors for obesity in children with allergies include snacking and low physical activity. Most children with respiratory allergies, especially those with asthma, reported incorrect eating habits such as snacking and eating before bedtime. A correlation between pulmonary function and body composition or dietary habits was not found.

**Keywords:** nutritional status; obesity; dietary habits; allergy; pulmonary function; allergic rhinitis; asthma

#### **1. Introduction**

The incidence of allergic diseases in Poland is increasing concomitantly with improvements in living standards and the adoption of a Western lifestyle. The most common clinical manifestation of hypersensitivity to inhalant allergens is allergic rhinitis (AR), which is one of the strongest factors affecting the quality of life and contributing to missed or unproductive time at work and school. In Europe and the United States, a significant increase in allergic diseases has been observed in recent decades [1,2]. In addition, the multicenter, standardized, randomized Epidemiology of Allergic Disorders in Poland (ECAP) study showed a prevalence of AR among the Polish population of 36% based on self-reported nasal symptoms, and 29% as diagnosed by physicians [3].

Among children with allergies, decreased involvement in outdoor activities and increased problems with concentration, sleep problems, and headaches are seen; moreover, children with AR often also suffer from asthma [4]. It is estimated that up to 40% of people with AR also have asthma, and almost 70% of asthmatics present coexisting AR [5,6]. In the Polish population, the asthma rate was 8% in children and adolescents according to the ECAP study, of which 70% of asthmatics presented with AR, while asthma occurred in 40% of patients with AR [7].

It is known that not only hygiene habits and exposure to allergens, tobacco smoke, and environmental pollution, but also a poor-quality diet, high caloric intake, overweight, and obesity in children and adolescents are important environmental factors that are conducive to the development of allergies [8,9]. Epidemiological and clinical studies suggest a relationship between obesity and allergic rhinitis as well as bronchial asthma [10,11].

In recent years, a significant increase has been noted in the incidence of obesity in children and adolescents in many European countries [12]. Excess body mass was diagnosed in 2% of Polish children in the 1990s, and in 15% of children 20 years later [13–15]. The authors of these studies indicated increased changes in lifestyle and nutritional habits as the causes of increased childhood obesity, i.e., consumption of sweets and unhealthy food; limited consumption of fruits, vegetables, and whole grains; and limited physical activity. In recent decades, fast foods have become a significant component of the diet in Westernized high-income countries, and now also for young people in Poland.

Children with allergic diseases present numerous risk factors for poor nutrition status. There are few studies describing dietary habits and their impact on the nutritional status of people with respiratory allergies. Although allergies are chronic and common diseases, these issues have not yet been clarified. Moreover, early diagnosis of excess body weight in children with allergic diseases, including asthma, seems to be important due to the course and treatment of the disease [16]. Therefore, the aim of this work was to evaluate the pulmonary function, nutritional status, eating habits, and risk factors of obesity in children and adolescents with AR.

#### **2. Methods**

#### *2.1. Study Design*

In this single-center, cross-sectional study, we evaluated, for the first time pediatric patients with symptoms of persistent rhinitis who visited the Department of Allergology of the Medical University in Gda ´nsk, Poland, between 2015 and 2017. The study was performed in compliance with the Code of Ethics of the World Medical Association (Declaration of Helsinki). The study protocol was approved by the Gda ´nsk Medical University Ethics Committee, and written informed consent was obtained from the parents of each patient. The study was supported by local research grant no. ST-554.

#### *2.2. Patients*

Inclusion criteria for the study were as follows: (1) age 7–18 years old, (2) persistent allergic rhinitis (duration at least 6 months in the last 12 months) never diagnosed and never treated with antihistamine drugs, (3) ability to perform spirometry, and (4) signed consent from parents to participate in the study.

Patients were evaluated according to the study protocol by a multidisciplinary team (allergologist, pediatrician, dietician). Children with persistent rhinitis symptoms in the last 12 months who had never been diagnosed and treated with anti-allergic or anti-asthmatic drugs were enrolled in the study (Visit 1, screening). During the next visit (Visit 2), allergy was confirmed by skin prick test, AR was diagnosed according to Allergic Rhinitis and Its Impact on Asthma (ARIA) [6], and asthma according to Global Initiative for Asthma (GINA) guidelines [17]. The medical history and spirometry results indicated newly diagnosed asthma in 43 patients; therefore, patients were divided into two groups: allergic rhinitis (AR group), and allergic rhinitis and asthma (AA group). Anthropometry, bioimpedance assessment, and dietary habits based on Food Frequency Questionnaire (FFQ-6) were collected and compared between the two groups. The scheme of the study is presented in Figure 1.

**Figure 1.** Study design.

Allergy background was confirmed with skin prick test to aeroallergens (*Dermatophagoides pteronyssinus*, *Dermatophagoides farinae*; cat, dog; *Alternaria alternata*, *Cladosporium herbarum*; pollens: grass mix, rye, birch pollen, alder, hazel; Allergopharma, Germany). Children with food allergies and atopic dermatitis were excluded from the study because of the frequent use of elimination diets.

Spirometry with a reversibility test (400 µg salbutamolum) was performed using a MasterScreen Pneumo spirometer, Jaeger Company, Germany. Forced expiratory volume in one second (FEV1), forced vital capacity (FVC) and forced expiratory flow (FEF25–75) were measured in accordance with the procedures recommended by the European Respiratory Society [18] and presented as percentage of predicted value (pv).

#### *2.3. Nutritional Habits*

Data were collected by face-to-face interviews using a researcher-designed standardized questionnaire based on the Food Frequency Questionnaire (FFQ-6) and validated for the Polish population [19]. The FFQ-6 is the most common dietary assessment tool used in large epidemiological studies of diet and health and is validated for the population. The self-administered FFQ-6 asks participants to report the frequency of consumption of approximately 62 line items over a defined period of time (last year). Each line item is defined by a series of foods or beverages. The FFQ-6 includes an assessment of eight food groups (sweets and snacks, dairy products and eggs, grain products, fats, fruits, vegetables and grains, meat products and fish, drinks). Respondents have a choice of six categories of food consumption frequency: (1) never or almost never, (2) once a month or less often, (3) several times a month, (4) several times a week, (5) daily, and (6) several times a day. The FFQ-6 also includes questions on eating habits, i.e., meal intake frequency and snacking between meals.

The following information was obtained: the number of meals in a day, amount of sweets consumed in a week, amount of fast-food eaten in a month, time of last meal before bedtime, and snacking between meals. Fast food was defined as mass-produced food prepared and served very quickly, with poor nutritional quality (hamburgers, takeaways, and carbonated soft drinks).

#### *2.4. Physical Activity*

The subjects were assigned to four categories depending on their level of physical activity: sedentary lifestyle (up to 2 h per week), low (3–5 h per week), moderate (6–7 h per week), and high (more than 8 h per week). One hour of physical activity corresponded to one hour of classroom attendance (45 min). Subjects were classified as having a sedentary lifestyle if only sometimes present during gym classes or not exercising at all. The low physical activity group attended gym class in school and an additional hour, e.g., swimming. Children identified as high activity trained in some kind of sport.

#### *2.5. Nutritional Status*

Body height was measured by stadiometer and body mass by electronic scale (Tanita Inc., Amsterdam, The Netherlands) by a nurse during the first visit. Body mass index (BMI) was calculated by dividing body mass in kilograms by the square of height in meters. Based on centile charts for sex and age for the Polish population—OLAF/OLA project—percentiles of BMI were specified [20]. According to the OLAF/OLA charts, above the 90th percentile is overweight, above the 97th percentile is obese, and below the 10th percentile is underweight. Body composition values of fat mass (FAT), fat-free mass (LEAN), and water content were measured via the bioimpedance method using a BodyStat 1500 (Bodystat Ltd., Ballafletcher, UK).

#### *2.6. Statistical Analysis*

Differences for somatic traits and between AR and AA groups were evaluated using Student's t-test or using the Mann-Whitney test for asymmetrical distributions. Distributions of values for somatic traits were evaluated using the Kołmogorov-Smirnov test. Differences between qualitative data were compared using the χ 2 test. The association between obesity risk factors and BMI percentile was determined using linear multivariate regression analysis. Differences were considered significant at *p* < 0.05. All analyses were carried out using the Statistica 10.0 software package.

#### **3. Results**

#### *3.1. Patients*

Allergic rhinitis was diagnosed in all 106 patients included in the study; among them, asthma was newly diagnosed in 43 (40.6%). Subjects with only allergic rhinitis were classified into the AR group and those with both allergic rhinitis and atopic asthma into the AA group (see Figure 1). All 106 children (100%) had a positive skin prick test to house dust mite (HDM; *D. farinae* and/or *D. pteronyssimus*), among which 40 (37.7%) were also positive to grass pollen (*n* = 29) and animals (*n* = 11).

Children in the AA group had a positive reversibility test and lower FEV<sup>1</sup> % predicted volume than children in the AR group. The basic characteristics of the study groups and the spirometry parameters are presented in Table 1.

#### *3.2. Eating Habits and Physical Activity*

The results of the eating habits and physical activity assessment are presented in Table 2.


**Table 1.** Clinical and lung function characteristics of patients.

AR: allergic rhinitis; AA: atopic asthma; FEV1: forced expiratory volume in one second; FVC: forced vital capacity; PEF: peak expiratory flow; pv: predicted value; n: number of subjects. \* All children (*n* = 106) had a positive skin prick test.


**Table 2.** Eating habits and physical activity assessment.

AR: allergic rhinitis; AA: atopic asthma; n: number of subjects.

#### 3.2.1. Number of Meals

The majority of children reported eating three (33.4%), four (29.2%), or five (34.1%) meals per day. Children with AA ate more frequently than children with AR (χ <sup>2</sup> = 12.9; *p* = 0.04).

Thirty-five children (34%) did not eat regularly; their meals were at different hours each day. There was no difference in meal regularity between AR and AA groups (χ <sup>2</sup> = 0.26; *p* = 0.60).

#### 3.2.2. Sweets

Almost half of the patients (47; 44.3%) ate sweets every day, in comparison to 12 (11.5%) who consumed sweets only one day per week. There was no difference in the consumption of sweets between AR and AA groups (χ <sup>2</sup> = 2.5; *p* = 0.88).

#### 3.2.3. Snacks

Eight-seven children (80%) snacked (sweet and salty snacks) between meals, 81.4% in the AA group and 79.3% in the AR group. There was no difference in snack consumption, but deeper analysis showed the AA group consumed more salty snacks than AR group (χ <sup>2</sup> = 0.59; *p* = 0.04).

All overweight and obese children (AR and AA) snacked significantly more often between meals (χ <sup>2</sup> = 9.46, *p* = 0.01) than children with normal BMI.

#### 3.2.4. Fast Food

Seventeen patients (16.0%) had never eaten fast food. Most of the children (84%) ate fast food; half of them (*n* = 52) ate it very rarely (once a month) and 8.7% ate it 4–6 times per month. There was no difference in fast food consumption between AR and AA groups (χ <sup>2</sup> = 6.3; *p* = 0.50).

#### 3.2.5. Meals before Bedtime

It was found that children most often consumed their last meal of the day 0.5–2 h before bedtime; a total 12.4% (*n* = 13) did so 2 h before falling asleep, 19.8% (*n* = 21) did so much earlier (from 2.5 to 3 h before bedtime), and 49% (*n* = 52) ate the last meal <1 h before sleeping. Children with AA ate the last meal 1 h before sleep more frequently than those in the AR group (χ <sup>2</sup> = 19.4; *p* = 0.005).

#### *3.3. Physical Activity*

The mean physical activity was 5 h per week. Most children (55%) reported 3–5 h/week physical activity. These children had only physical education (PE) at school and 1 h of additional activities after school (swimming or games). Children with AR reported low (63.4%; *n* = 40) and moderate (19.2%; *n* = 12) physical activity; similarly, children with AA reported low (42.0%; *n* = 8) and moderate (34.8%; *n* = 15) activity. There was no difference in physical activity between the AR and AA groups (χ <sup>2</sup> = 13.1; *p* = 0.15). There was a negative correlation between physical activity level and BMI centile in the whole study population (Spearman's R = –0.19; *p* < 0.05).

#### *3.4. Nutritional Status and Body Composition*

Obesity was diagnosed in six children (6.0%) and eight were overweight (7.5%). In the AA group, obesity was present in 4.7% compared to 6.9% in the AR group (χ <sup>2</sup> = 3.58; *p* = 0.30). The results of body composition measurement are presented in Table 3. There was no difference between the AR and AA groups.

#### *3.5. The Multifactorial Linear Regression Analysis*

Multifactorial linear regression analysis showed an association (independent of age) between BMI percentile and both snacking and physical activity level (see Figure 2 and Table 4).



AR: allergic rhinitis; AA: atopic asthma; BMI: body mass index; LEAN: lean body mass.

**Figure 2.** Association between body mass index (BMI) percentile and risk factors for obesity in all studied groups (adjusted R<sup>2</sup> = 0.97; *p* < 0.05). %F: percentage of body fat; FEV<sup>1</sup> : forced expiratory volume in one second. Legend: red line means < 25th percentile, black line 25th–89th percentile, grey line 90th–97th percentile and yellow line >97th percentile.

**Table 4.** Multivariate regression model predicting BMI value (adjusted R<sup>2</sup> of the model was 0.97, *p* < 0.05).


FEV1: forced expiratory volume in one second; B: Regression coefficient B; Beta: beta standardized regression coefficient.

#### **4. Discussion**

In the present study, we evaluated the nutritional status and dietary habits of Caucasian children with allergic rhinitis alone or with co-existing asthma. Although all of the children presented respiratory allergy symptoms at least 12 months before the study, they were never diagnosed with allergies and had not previously been treated with an antihistamine or anti-asthmatic drugs before.

The most important finding of the study is that the majority of children with respiratory allergies reported incorrect eating habits and low physical activity, with 7.5% being overweight and 6.0% being obese. In the study population, excess body weight was significantly associated with snacking between meals and low physical activity.

#### *4.1. Nutritional Status*

Unexpectedly, the prevalence of overweight and obesity among allergic children was similar to the population of healthy children in Poland [13,14,21,22]. This aligned with data from the International Obesity Task Force (IOTF) showing that approximately 10% of children worldwide are overweight [12].

Although AR is a common disease, most authors focus on children with food allergies or asthma. These studies have suggested that adiposity indicators are associated with asthma, asthma severity, and atopy [23,24]. It is obvious that a positive energy balance is associated with changes in immune system functioning, including chronic inflammation, which is clearly an unfavorable phenomenon [22]. Overweight and obese children with allergic diseases have metabolic derangements, and obesity may have an impact on inflammation and clinical symptoms in asthma. The cause of impact of the obesity on asthma risk is still unknown. Potential etiologies include airway smooth muscle dysfunction from thoracic restriction, obesity-related circulating inflammation priming the lung, and obesity-related comorbidities mediating asthma symptom development. Studies suggest that obesity in children with asthma appears to be associated with greater airflow obstruction and a mildly diminished response to inhaled corticosteroids [25]. Additionally, anti-allergic and anti-asthmatic medications may be risk factors for obesity and physiological factors associated with puberty, also intensifying the tendency to gain weight in adolescents [23]. In our study, we did not take into account the effects of medicines because all of the children were newly diagnosed with respiratory allergies and had not been treated with an antihistamine or anti-asthmatic drugs. This may be one of the reasons for the relatively small number of children with obesity observed in our study.

Recent prospective evidence supports the notion that increased body weight precedes asthma development, but there is an ongoing debate as to whether obesity directly increases this risk or whether patients first experience asthma and then become overweight or obese, possibly because of respiratory constraints and reduced physical activity [26].

There are only a few studies on nutritional status and allergic rhinitis. A cross-sectional study of obesity indicators and AR in 8165 participants from the 2005–2006 National Health and Nutrition Examination Survey (NHANES) showed that overweight and obesity were associated with increased risk of AR in adults, but no such evidence was found among children [27].

Interestingly, in our study, although children with co-existing asthma were younger than and not as tall as the children with only AR, they had similar weight. However, there were no statistical differences between the number of overweight and obese children and body composition (FAT, LEAN) in the two groups. This is interesting because other authors have reported more than 50% of children with excessive body weight among children with asthma [28]. Spirometry parameters also did not correlate to BMI, body fat, and lean body mass content in the whole study group, although pulmonary function tests were lower in asthmatics. There were no differences in terms of family burden between allergy, asthma, obesity, exposure to tobacco smoke, and pet allergens.

#### *4.2. Dietary Habits*

Our study showed that incorrect eating habits were reported by most of the children with allergies, such as frequent consumption of fast foods and sweets, snacking between meals, and eating meals less than 1 h before bedtime.

Many studies have confirmed that fast-food consumption is linked to childhood obesity [29,30]. The multicenter International Study of Asthma and Allergies in Children (ISAAC) showed that fast food consumption is high in childhood (6–7 years), increases in adolescence (13–14 years), and is associated with higher BMI [31]. In our study, 35% of the children reported fast food consumption at least several times a month. This result is similar to the ISAAC results, showing that 27% of children and 52% of adolescents reported more than weekly fast food consumption [31]. We did not find an association between dietary habits and pulmonary function. There were also no differences in fast food consumption between children with asthma and those with only rhinitis. These results are different from those reported by other authors, suggesting that fast food consumption may contribute to the increasing prevalence of asthma, rhinoconjunctivitis, and eczema in adolescents and children [31]. Other results from case-control [32–34] and cross-sectional [35–40] studies indicate that consumption of fast foods is significantly related to current asthma and allergic rhinitis (pollen fever). Wang et al. suggested that the amount of processed foods eaten correlates with the frequency and severity of asthma [29].

Another important finding from this study was that approximately 80% of children with respiratory allergies snacked between meals every day. Moreover, although all children with excess body mass consumed more snacks compared to normal-weight patients and reported low physical activity, asthmatics consumed snacks more frequently (χ <sup>2</sup> = 0.59; *p* = 0.04) and were more likely to eat their last meal of the day 1 h before sleeping (χ <sup>2</sup> = 19.4; *p* = 0.001). Similar results were seen in the PANACEA study, which showed that among a population of 700 Greek children 10–12 years old with a 23.7% prevalence of asthma symptoms, almost half the children reported salty snack consumption ≥1 times/week [41]. In the cited study, consumption of salty snacks >3 times/week (vs. never/rarely) was associated with a 4.8 times higher likelihood of having asthma symptoms, irrespective of potential confounders. The authors noted that the association of salty snack eating and asthma symptoms was more prominent in children who watched television or played video games >2 h/day [41].

Unlike other researchers, we studied the times of meals consumed and found, interestingly, that almost half of the children ate in the last hour before bedtime. This incorrect habit was more common in children with asthma symptoms. There are well-known factors that affect and exacerbate inflammation in the lower respiratory tract in asthmatics, such as infection or gastroesophageal reflux. Eating immediately before bed might have contributed to the formation of gastroesophageal reflux and bronchial hyperreactivity in the studied group of children with AR. This is also interesting because children with AR differed compared to asthmatics in lung function (FEV1%pv), but not in nutritional status or other eating habits except for snacking and meals consumed less than 1 h before bedtime. Unfortunately, we did not study the symptoms of gastroesophageal reflux, and we, therefore, cannot form any specific conclusions.

Braithwaite et al. [31] postulated some possible mechanisms to explain the relationship between asthma and allergic disease and the consumption of fast food, which may involve higher concentrations of saturated fatty acids, trans fatty acids, sodium, carbohydrates, and sugar, as well as preservatives that may modulate immune reactions. Consumption of processed foods reduces the consumption of foods that are rich in protective nutrients, such as fruits and vegetables. A reduced intake of fruits and vegetables, which have antioxidative and anti-inflammatory properties, is likely to have an unfavorable impact on asthma prevalence/management [42]. Additionally, indications are that a diet poor in antioxidants is a key factor influencing the development of allergic diseases; a Western lifestyle and processed food consumption can also cause reduced exposure to microbial products and a changed microbiome, which are thus possible causes of the increase in allergic disease [43].

#### *4.3. Physical Activity*

In our study, children from both groups (AR and AA) in over 50% reported low physical activity. It is well known that a lack of exercise increases the risk of obesity. This was confirmed by research in the Phase 3 ISAAC trial, in which television viewing (5+ h/day vs. <1 h/day, *p* < 0.001) (the group with low physical activity) was statistically significantly associated with higher BMI in comparison to vigorous physical activity (3+ h/week vs. never, *p* < 0.001) (the group with high physical activity) in adolescents. The authors also suggested that current behaviors are more important than other factors such as birth weight, breastfeeding, current maternal or paternal smoking in early childhood in the development of obesity [24].

A few studies have shown a relationship between physical activity and allergy. In the ISAAC study, associations were found between vigorous physical activity and a sedentary lifestyle for 13-year-olds with allergic rhinoconjunctivitis. Mitchell et al. showed that several hours of TV viewing was associated with symptoms of current asthma in adolescents [24]. Similarly, studies indicate that physical activity could be protective against the development of asthma [44]. On the other hand, Byberg et al. found no association between physical activity and allergic rhinoconjunctivitis [45].

Our results do not contradict the association between pulmonary function and physical activity but show a correlation between physical activity level and BMI percentile in the whole study population (Spearman's R = –0.19; *p* < 0.05).

Our study provides two very important issues in the study of respiratory affections such as allergy and asthma. These are the fact that this is the first study on nutrition carried out in newly diagnosed AR teenagers, before any medication that could mislead any result and the fact that this is another study from a few existent about nutrition in respiratory allergy. The limitation of our study is the relatively small group of patients; therefore, a more accurate analysis was not possible, for example based on age or sex. This study is not generalizable to the Polish population because it was performed in a clinical sample of children. Although the relationship between incorrect dietary habits, low physical activity, and obesity in children with respiratory allergies is supported by our findings, no conclusions about causality can be made due to the cross-sectional design.

Further studies with large groups are necessary to determine the relationship between respiratory allergy, body weight, and diet.

#### **5. Conclusions**

The risk factors of obesity in allergic children were found to be snacking and low physical activity. Most children with respiratory allergies, especially asthmatics, reported incorrect eating habits such as snacking and eating before bedtime. A correlation between pulmonary function and body composition or dietary habits was not found.

Our study also indicated that in groups of children with respiratory allergies, there is a need for correction of diet and lifestyle. We suggest that early dietary correction may be helpful for children with allergic rhinitis and a high risk of asthma.

**Author Contributions:** E.W.: S.M., M.S. designed the study. E.W., S.M. collected and analysed the data. M.S., M.G.-N. and E.J. carried out independent internal peer review of the data. All authors agreed the final version of the manuscript submitted for publication.

**Funding:** The study was supported by local research grant no ST-554.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

1. McNeill, G.; Tagiyeva, N.; Aucott, L.; Russell, G.; Helms, P.J. Changes in the prevalence of asthma, eczema and hay fever in pre-pubertal children: A 40-year perspective. *Paediatr. Perinatal. Epidemiol.* **2009**, *23*, 506–512. [CrossRef] [PubMed]


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **A Serious Game Approach to Improve Food Behavior in Families—A Pilot Study**

#### **Sigrid Skouw, Anja Suldrup and Annemarie Olsen \***

Food Design and Consumer Behavior Section, Department of Food Science, University of Copenhagen, Rolighedsvej 26, 1958 Frederiksberg C, Denmark; ssn@food.ku.dk (S.S.); clg487@alumni.ku.dk (A.S.)

**\*** Correspondence: ano@food.ku.dk; Tel.: +45-3533-1018

Received: 15 April 2020; Accepted: 9 May 2020; Published: 14 May 2020

**Abstract:** The objective of this pilot study was to investigate the effect of a specially developed serious game to improve food behavior in families with children aged 5–13 years using mixed methods. Fourteen families were randomized into a game-group and a non-game-group and divided into age groups (game-children (GC), game-parents (GP), non-game-children (nGC), and non-game-parents (nGP)). The families completed a baseline test, a three-week intervention period with or without a game element, and a follow-up test. Qualitative results showed a positive change in food behavior in all families. Quantitative results mainly showed an effect in food neophobia as a decrease was seen in all groups; however, it was only significant (*p* < 0.05) in three groups (GP, nGC, nGP). No changes were seen in willingness to taste, and only limited changes in liking and number of words used to describe the stimuli. In conclusion, qualitative results showed positive change in the children's food behavior in most families, indicating a positive effect of performing tastings and tasks together as a family—regardless of the presence of a game element. However, this was not as clear in the quantitative data, indicating that current quantitative tools are less suited to measure complex concepts like willingness to taste.

**Keywords:** serious game; gamification; eating behavior; food neophobia; willingness to taste

#### **1. Background**

Low intake of fruit and vegetables (F&V) was according to WHO among the top 10 leading risk factor causes of death in middle- and high-income countries and among 6 diet-related risks of disability-adjusted life years in 2004 [1]. Surveys from 2005 [2] and 2014 [3] showed F&V intake among European 11-year-old children to be below the recommended levels of 400 g/day [4].

Issanchou and Nicklaus [5] put together a conceptual framework showing a number of different concepts determining children's food choice, one of these being experience and social influence from parents and peers. Genetics will affect children's sensory perceptions, and parenting style will further be determining preferences, choices, and intake [5]. This has also been shown in experimental research, like a recent review of different strategies to change children's eating behavior [6]. Parental control and using rewards/instrumental feeding was shown to largely impact eating behavior both positively and negatively. Examples of such strategies are availability of food in the household, restriction of the amount of food a child is allowed to eat, and use of rewards to get children to eat particular foods [6].

A report on vegetable consumption in Denmark showed intake to be limited to only a few types of vegetables such as carrot, onion, and tomato [7]. The most limiting factors of vegetable purchase in Danish families were found to be the lack of ideas on how to use and to get children to eat different and new vegetables [7]. Children's limited food choices are also a challenge in other countries. For instance, a survey from Uruguay found similar results of low variety of vegetable intake and low liking, and some vegetables were never offered to the children due to either parents not eating them

themselves or not knowing how to prepare them [8]. To meet the national recommendations for fruit and vegetable intake and to prevent picky eating and food neophobia (reluctance to eat new foods [9]), these limiting factors should be addressed. Encouraging families to approach novel or disliked F&V in a more explorative manner may reduce these limitations, e.g., through sensory exploration and involvement [6], and increase in F&V intake.

The use of game elements to change eating behavior has gained more attention over the last two decades [10–12]. Games created with the intention of developing skills and knowledge are classified as *serious games*. Serious games were initially defined by Abt [13] as games that " . . . *have an explicit and carefully thought-out educational purpose and are not intended to be played primarily for amusement. This does not mean that serious games are not, or should not be, entertaining*". Playing games is usually associated with fun social interactions of a competitive nature and is driven by both intrinsic and extrinsic motivation [14,15], providing a hands-on approach. The latter has been found more effective in increasing vegetable consumption in children compared to educational programs [6]. Thus, specially designed games might be useful tools for motivating and encouraging exploration of a variety of foods, including that of F&V, and to further promote a change in eating behavior.

Games have demonstrated potential for increasing children's F&V consumption [16,17], while studies on the effect of games on adults are scarce and show only little or no effect [18,19]. Investigation into the effect of games on families does not exist to the knowledge of the authors of this study, constituting a gap in knowledge. This gap is particularly interesting as efforts made to change eating behavior have been found to be more efficient when directed at the family level rather than at the individual level [20], since parental food habits is one of the most important determinants of children's food choice and behavior [21].

Thus, the aim of this pilot study was to investigate if a specially developed serious game could improve food behavior in relation to fruit and vegetables in families with children aged 5 to 13 years. Food behavior was investigated through measures related to the game content: food neophobia and willingness to taste, food vocabulary used to describe F&V, and qualitative measures.

#### **2. Material and Methods**

#### *2.1. Recruitment and Randomization*

Sixteen families were recruited through social media and a newsletter shared by the project Taste for Life (a research and communication collaboration of scientists in Denmark with focus on taste, www.taste-for-life.org) to participate in the pilot study. Inclusion criteria were no F&V allergies and address in or around the area of Copenhagen for logistic reasons. Most participating families consisted of two children and two parents. Some families contained one or two children in the target age group and one child outside of the target age group, who participated in the game but not in the tests. The families were randomly assigned to either a game group or a non-game. Two families (one game and one non-game family) dropped out of the study before the baseline test; one for unknown reasons and another due to illness. The game-group and non-game-group each contained seven families at the beginning of the intervention. The study complied with the Helsinki declaration. After reviewing the study protocol, the study was found not to require ethical approval (j.nr. 19007287). The data collection and handling plan was approved by the institutional GDPR office (j.nr.: 514-0120/19-5000). Parents gave written, informed consent on behalf of themselves and their children, and children agreed to participate and for data to be used for scientific publications.

#### *2.2. Intervention Material*

Developing a game, which unites fitting motivators, a fitting social situation, and mere exposure to novel or disliked foods through sensory interactions, have the potential to be a successful strategy to encourage food exploration and possibly change eating behavior in families.

A serious game was developed for the purpose of this study and was called *The Kingdom of Taste*. The game is played by up to five players and is composed of:


Six different F&V are to be used during the game. The F&V are cut into appropriate pieces, one for each player, and placed in the six cups and covered with the lids. The lids were included to add an additional element of surprise and excitement for the players, as they would not see which F&V they were to taste, before landing on a taste task. This could potentially change the level of arousal and the participant's optimal complexity of foods, as described by Dember and Earl [22], before uncovering and thereby affect willingness to taste the hidden F&V.

The game is typically played by 3–5 players (1–3 children and 2 parents from the participating families) and takes 30–60 min. The board game is centered around a story of a chef who has forgotten to purchase F&V for a dinner party at the castle. The chef asks the players to help him collect as many F&V (point cards) as possible on their way from the village to the castle. To collect F&V, the players have to solve different tasks present on the game board. The tasks fall within three categories: (1) descriptive tasks where F&V are to be described with regard to flavor, appearance, and associations; (2) taste tasks where the players has the opportunity of tasting up to six different and unknown F&V; and (3) creativity tasks related to preparation, cooking techniques, and construction of meals. The tasks are represented on the game board as three distinctive zones as shown in Figure 1. The tasks are created to increase familiarity of a large variety of F&V, both through descriptive tasks and through tastings. Mere exposure to the F&V through pictures, words, and tastings could potentially increase affection of these [23]. Creative meal planning is a part of the game in the last zone and as the game is finalized by each player composing a three-course meal with his/her collected F&V card (points). Sparking exploration and interest in meal composition could inspire players to bring this creativity to the kitchen and further affect food behavior. The game aimed at obtaining a suitable level of difficulty for the target group, in accordance with the Theory of Flow, which describes how the relationship between skill level and posed challenges needs to be balanced to achieve a state of flow; i.e., when the challenge a person is faced with is not too difficult nor too easy to solve [24,25]. Before the pilot test, the game was tested by five families with children aged 4 to 9 years and one school class with students aged 11 to 12 years. The families received all necessary materials (except F&V, which they were to provide themselves with the possibility of receiving compensation for their purchases) and a questionnaire with questions regarding the game elements, age group, entertainment, etc., to be filled out after having tested the game. The game was modified according to this feedback.

The game was used as intervention material for the game-families.

The non-game-families were provided with a representative selection of the three categories of tasks present on the board game, including tastings of F&V, but without the game context. The non-game material was comprised of a sheet of task instructions, 30 food cards (with names of the F&V and no pictures), one parental instruction with examples/suggestions on how to solve the tasks, and containers without lids for taste samples. Lids were not included in order to limit any game element that could create additional excitement during a task.

**Figure 1.** The Kingdom of Taste and an overview of the three zones and their distinctive tasks.

#### *2.3. Study Design*

The study timeline consisted of a baseline test, a three-week intervention period, and a follow-up test. The families were instructed to perform their designated assignments at home once a week during the three-week intervention period; i.e., the game-families were to play the serious game, and the non-game-families were to perform similar tasks without the game element a total of three times. F&V for the home assignments were delivered to all families (both game and non-game families) once a week during the intervention period, at their home address.

Table 1 provides an overview of the taste samples used at the home assignments. The F&V for taste samples were chosen based on the theory of *Zone of Proximal Development* [26], as this theory has recently been used to explain flavor preference development in children [27]. The theory of the Zone of Proximal Development is originally a model with three levels (Zone of Actual Development, Zone of Proximal Development, and Zone of Insurmountable Difficulty) describing children's cognitive development as a result of social interaction between individuals with different skill levels [26]. In a food context, The Zone of Actual Development represents foods that are familiar, liked, and considered to be safe to eat for the child on his/her own, whereas the Zone of Proximal Development represents foods that are considered exiting, unknown, and associated with some degree of uncertainty, which the child is only able or willing to taste under adult guidance and support [27]. F&V choices aimed at having two F&V belonging to the Zone of Actual Development (upper two rows in Table 1) and four F&V in the Zone of Proximal Development (lower four rows in Table 1).


**Table 1.** Overview of the taste samples used for the home assignments.

At the end of the follow-up session, all families received a 500 DKK (75 USD) gift card of their choice as a thank-you-gift for their participation in the study, and the non-game-families further received a *Kingdom of Taste* game.

#### *2.4. Questionnaire Design and Test Protocol*

Baseline and follow-up tests were performed at the university with a maximum of two families present in the same room at a time placed in far ends of the room, never mixing families from different intervention groups. Each test took approximately one hour.

Mixed methods were applied for the data collection of this study in order to obtain a fuller insight into the intervention effects, as a combination of quantitative and qualitative methods can provide insights that may be missed when only using one of them [28]. The choice of using mixed methods was further based on the expectation that food behavior is a complex concept, which is possibly difficult to measure by current quantitative tools. The questionnaire developed for the baseline test consisted of three quantitative parts: (1) A Danish version of the six-item version of the original food neophobia scale (FNS) [29], first used in [30]; (2) a box for describing the presented F&V (a measure of food vocabulary, single words only); and (3) willingness to taste the presented F&V (yes/no), including liking measured on a 7-point hedonic smiley scale and stating familiarity and frequency of consumption to provide an idea of the level of preliminary knowledge about the F&V used in the tests. The questionnaire was to be filled out individually, though the younger children had the opportunity to receive assistance either from one of the two first authors of this article or their parents.

The follow-up test consisted of the same quantitative tasks as the baseline test to measure change during the intervention period, but further included qualitative questions for the parents to answer. The qualitative questionnaire consisted of open-ended questions regarding observed changes in food behavior at home during the intervention period; changes in willingness to taste, how they discussed F&V, and changes in the children's engagement with F&V.

The F&V used at the baseline and the follow-up test were the same; banana, carrot, broccoli, papaya, prune, and caper berry, which were chosen based on the same considerations as the taste samples for the home assignments. The order of the F&V was randomized and varied between all the families at the two test sessions. The F&V were placed on two plates, one plate with taste samples and another plate with the F&V in its true shape (uncut, except for the broccoli and papaya which were cut in half), for the participants to use as a visual tool when answering the questions. The F&V were presented one at a time. Additional materials present were pens, paper plates, napkins, crispbread, cups, and water.

#### *2.5. Data Analysis*

All statistical data analyses were conducted using R-studio statistical free software (version 1.1.456, Boston, United States) [31]. Graphs were made using Microsoft Office Excel (2016) and Microsoft Office PowerPoint (2016).

The participants were divided into groups according to treatment and age group: game-children (GC), non-game-children (nGC), game-parents (GP), and non-game-parents (nGP). Baseline differences between GC vs. nGC and GP vs. nGP were tested for by conducting a Mann–Whitney U test for age, gender, food neophobia, liking, and word count, and Fisher's exact test was used to test for baseline differences in willingness to taste.

A McNemar test tests if two response variables are significantly different from each other within a study sample and was used for testing significant differences in willingness to taste (yes/no) between baseline and follow-up within each of the four groups. A linear mixed model tested for differences within each of the four groups between baseline and follow-up in the FNS score, liking, and word count. The collected words were both analyzed as total word count as well as count within the word-categories *hedonic, descriptive*, and *other*. The model was further used to test the difference in change found between the treatments in both age groups (GC vs. nGC and GP vs. nGP) for the same measures. Residuals of the linear model not following a normal distribution were transformed by a log-transformation.

A Cronbach's alpha test was run on the FNS scores in each of the two age groups to test reliability. The test was run on data from baseline and follow-up test separately.

The qualitative feedback collected at the follow-up test was analyzed by using a combination of pre-set and emerging codes (willingness to taste, food language, food engagement, game related) with individual emerging sub-codes.

#### **3. Results**

Two non-game-families dropped out of the study just before the follow-up test; one due to illness and another due to scheduling issues. Seven game-families and five non-game-families completed the follow-up test. Three children who participated in the baseline test did not participate in the follow-up test due to illness. A total of 12 families and 39 participants completed the entire study; 22 in the game-group and 17 in the non-game-group. Table 2 shows the age and gender distribution in the two groups. No differences were found in age or gender distribution when comparing the treatments in both age groups.


**Table 2.** Age and gender distribution of participants who completed both baseline and follow up test.

Children (*n* = 18) and parents (*n* = 21).

#### *3.1. Quantitative Measures*

3.1.1. Food Neophobia, Willingness to Taste, and Liking

No significant differences were found at baseline between the treatments in both age groups in FNS score. All groups showed a decrease in FNS score from baseline to follow-up test, but significant reductions in FNS score were only found in nGC, GP, nGP but not in GC, as shown in Figure 2. Of the 10 GC, seven showed a decrease, one remained unchanged and two showed an increase in food neophobia at follow-up.

**Figure 2.** Mean (±SEM) range of food neophobia scale (FNS) score at baseline and follow up. Abbreviations: Game-children (GC), non-game-children (nGC), game-parents (GP), non-game-parents (nGP). Significance level estimated by a linear mixed model. Significance levels: \*\*\* *p* < 0.001, \*\* *p* < 0.01, \* *p* < 0.05.

Cronbach's Alpha was calculated for the FNS score of the children and the parents separately. At baseline α = 0.64 for both the children and parents, and at follow-up α = 0.8 and 0.7 for children and adults, respectively. These sizes indicate consistency.

No significant difference in willingness to taste was found between treatments and age groups at baseline or between baseline and follow-up in any of the groups, and only minor and scattered changes in liking between baseline and follow-up were observed as seen in Table 3.


**Table 3.** Mean (SEM) of liking of the six fruit and vegetables (F&V) at baseline and follow-up (1 = super bad; 2 = really bad; 3 = bad; 4 = neither good nor bad; 5 = good; 6 = really good; 7 = super good).

Significance level estimated by a linear mixed model. Abbreviations: Game-children (GC), non-game-children (nGC), game-parents (GP), non-game-parents (nGP). Significance levels: \*\*\* *p* < 0.001, \*\* *p* < 0.01, \* *p* < 0.05. <sup>a</sup> p-value shows significant level of change from baseline to follow-up in the groups. <sup>b</sup> p-value shows significant level of difference in change from baseline to follow-up between CG and nCG and between GP and nGP.

#### 3.1.2. Food Vocabulary Used by Families to Describe F&V

Food vocabulary was measured through counting the number of single words used to describe a given F&V, and categorized them in one of three word categories: *Hedonic* (e.g., "delicious"), *descriptive* (e.g., "green") or *other* (e.g., "monkey"). Only a few significant changes in number of words used to describe the presented F&V was found. When a significant change was present, it was characterized as an increase in word count in the non-game-group and a decrease in the game-group, with no general tendency of specific word groups increasing or decreasing more than others (Table S1). The changes in words were not specifically connected to any of the three word-categories hedonic, descriptive, or other.

#### *3.2. Qualitative Measures*

#### 3.2.1. Perceived Change in Food Behavior

Food behavior was measured as willingness to taste, food language, and food engagement in the qualitative questionnaire. All 12 families reported an increase in willingness to taste in the qualitative questionnaire. Six families, four game and two non-game, expressed that their food language had changed over the course of the intervention period. Eight families, five game and three non-game, indicated that they have been having food-related conversations during the intervention period. Nine families, five game and four non-game, reported an increase in the children's food engagement on one or more parameters: increased interest in food, cooking, or meal planning. When summarizing the qualitative results, parents reported improved food behavior independently on the presence of a game element.

#### 3.2.2. Motivational Effect of the Game Element

Six of the seven game-families commented on the use of a game to increase willingness to taste. Four of them reported how the game/competitive element in the game increased their children's willingness to taste the F&V in the home assignment, as expressed by one mother: "*The game*/*competitive element in the game caused our children to not want to lose and* (they) *tasted almost everything the last couple of weeks*". Two of the game families further commented that the taste tasks were the most exciting part of the game. The mother of one game-family wrote: "*During the home assignments it was obvious that the children were looking forward to tasting the food, and that the best part was when someone landed on a taste task*". Two game families further reported how their children had requested to play the game during the intervention period.

#### **4. Discussion**

#### *4.1. Food Neophobia and Willingness to Taste*

Several studies have found food neophobia to decrease from childhood to adulthood [32–34]. Based on this knowledge, the lower level of food neophobia in the parents in comparison to the children was expected.

FNS have been used to measure the effect of sensory education on food behavior, showing lower scores after intervention [35,36] but not always significantly [35]. All four groups in this study showed a decrease in FNS after intervention either significantly (GP, nGP, and nGP, *p* < 0.05) or non-significantly (GC) corresponding with existing literature. Whether the change in FNS scores is persistent is unknown, as long-term effects were not investigated in this study, but the change indicates the existence of subjective perceptions of change among the participants, at least during the intervention period. This perception of change may result from the participants' own observations of behavioral changes, such as increased courage to try new foods or being less particular about which foods to eat during and after the completion of the home assignments. This indicates that continuous use of the home assignments may potentially change food neophobia persistently, as a result of increasing willingness

to try foods and thereby increase exposure to disliked and novel foods, potentially giving rise to the effect of mere exposure [23].

All 12 families reported that they had experienced an increase in willingness to taste, independent of treatment, supporting the decrease in FNS scores and indicating that the serious game did not provide an additional effect over the tasks performed without the game element.

The game-families generally ascribed the increase in willingness to taste to the competitive element of the game due to its motivational effect. Overcoming a personal boundary of tasting something unknown or novel might function as an intrinsic motivation for the participants, caused by feelings of satisfaction and joy of self-accomplishments, or due to enjoyment of playing the game [14]. As the game further gives rise to extrinsic motivation through the possibility of earning points, winning, and receiving feedback and praise from other players, the participants might further be more motivated to engage [14]. The combination of both intrinsic and extrinsic motivations thereby seems to have resulted in a high degree of willingness to taste during the home assignments in the game-families. The game-families only reported a positive outcome of using extrinsic motivation to get their children to taste the F&V in the game, but previous studies have indicated negative outcomes, as reviewed by DeCosta et al. [6]. Using extrinsic motivation, may have an undermining effect on intrinsic motivation [37], as e.g., parental prompting and restriction of food intake have been found to cause children to override their internal cues of hunger, satiety, and pleasure [6], which could lead to overeating or other negative consequences. These findings could indicate potential negative consequences of using a game to improve food behavior and willingness to taste although this was not indicated by the results of this study. A potential explanation could be the more positive type of extrinsic motivation found in the game compared to a normal eating situation, such as the possibility of gaining rewards (point cards, praises, and cheers) and the wish to win.

In the non-game-families, several of the parents reported that they were impressed by how many of the novel F&V their children were willing to taste during the home assignment. The intrinsic motivation of self-accomplishment might likewise have occurred in the non-game-families during the home assignment. The surprise from the parents' side, that their children were willing to taste the large variety of F&V presented, could also be an example of the discrepancy in expected pickiness between children and parents found by the Danish Agriculture and Food Council [38], as parents perceive their children to be pickier than the children themselves are. The children could have been willing to taste such F&V before the intervention period but may not have been served it due to the parents' expectations of their refusal to taste it. In a review by Scaglioni et al. [21], parental food habits were shown to be one of the most important determinants of children's food choice and behavior. Together with the gap between parental and child beliefs about picky eating, this could be part of the reason why the F&V consumption of European children does not meet the recommendations [2,3]. The children are to a large extent limited in their F&V selection to what is available in the kitchen at home—and what is available and served at home might be limited to what the parents believe to be what their children like [7]. The F&V used in this study were selected to be a mix of well-known and novel stimuli and thereby expose the families to F&V other than what they usually eat. Simply tasting and experiencing novel F&V could be a way to enlighten parents of their children's higher willingness to taste and eat new F&V than what they believe and thereby be a motivation to incorporate such new foods in the kitchen, which will likely lead to increased F&V intake.

The taste samples used in the game-group were kept a secret until the point of tasting (hidden in a container with lid), whereas the samples were visible for the non-game-families from the beginning of their home assignment (placed in a container without lid). This additional element of secrecy present in the game-families' assignment may have increased the level of arousal before the reveal of the F&V [22]. Even if the F&V hidden were well-known or at least known to some degree, the participants would not know before opening the container in which they were hidden. The anticipation of what was hidden could possibly increase the arousal, in contrast to the non-game-group where the participants were able to see the F&V before engaging in tasting.

No changes in willingness to taste were found, which does not align with the reduction in FNS scores and the qualitative perception of increased willingness to taste. This lack of difference in willingness to taste could be caused by the high willingness present at the baseline test. Using willingness to taste might thereby not be the best measure for investigating a change in courage to taste different foods, when the stimuli is mostly well-known F&V and when the participants are not neophobic. Other studies have found varying effects of using willingness to taste, ranging from positive effects [36] and temporary effects [35] to no change [39]. This line of thought was recently shared by Olsen [40], who suggests that the focus in this area of research is too narrow and could benefit from using a broader specter of outcome measures, including qualitative ones. If this is the case, the varying effects of sensory education on willingness to taste [35,36,39] may be explained by the inadequacy of the measurement approach rather than the sensory education itself. Other approaches have been made in an attempt to develop a behavioral food neophobia measure for children, such as using wiliness to taste where the children were to taste an unknown food based on their own previous indication of willingness [35], and correlating it with the FNS, but the correlations found between the two tests were generally weak [35,41]. The poor correlation between the FNS and the behavioral food neophobia tests indicates that the two tests may measure two different things. More research into how to effectively measure these, which are considered closely related concepts, is required in order to perform this kind of studies.

The difficulties of using willingness to taste as a behavioral measure of food neophobia may indicate that willingness to taste is a far more complex concept than simply a yes/no question. It may be assumed that different levels of novelty and resistance towards certain foods exist, which may mean that the action of tasting a novel food is rejected but does not necessarily mean that other forms of interactions with the novel food are rejected. Such other interactions could potentially result in willingness to taste at a later time point because of increased familiarity [42]. This speculation is supported by the findings of Dazeley and Houston-Price [43] and Coulthard and Sealy [10], who both found that non-taste sensory interaction increased children's tasting afterwards.

#### *4.2. Food Vocabulary Used by Families to Describe F&V*

The limited changes in word count when describing the F&V are in accordance with previous studies, where 11- to 13-year old children showed a decrease in number of words used to describe bread in both an intervention and control group after sensory education [44]. Mustonen et al. [44] expected the lower number of words to be partly due to restlessness in the children during the follow-up test, which was also observed in this study in several children. Likewise, several parents showed signs of restlessness and appeared to use less time on this task at the follow-up test. As the baseline and follow-up test questionnaires were identical, an explanation could be that the writing tasks at follow-up test perhaps were perceived as tedious and not as exciting and fun as tasting the unknown F&V. This speculation is backed up by the qualitative data, where the tasting part was described by two families to be the best part of the game. The tendency to an increase in word count in the control groups and decrease in the intervention groups could indicate that the game element was interfering with the descriptive tasks. The game players might have been eager to move on to other parts of the game that they found more fun, as indicated by some of the game-families, as described in Section 4.4. Focus of the non-game-families, on the other hand, may have been more on the task itself, as there were no game elements. Due to the simpler nature and the limited number of tasks (each participant only having to answer four questions per session), more effort may have been put into solving the non-game. If the outcome is to achieve a more nuanced food language through increasing vocabulary and ability to describe F&V, better results might be achieved by completing the tasks without a game element.

#### *4.3. Qualitative Measures*

The qualitative feedback received form the parents showed a positive improvement in food behavior in both groups, indicating that the specific tasks (describing, tasting, and being creative with F&V) present in both the game and non-game assignments possibly are sufficient on their own to improve food behavior. It is not possible to tell which element of the home assignment caused the improvements seen in both groups or if it was a collaborative effect.

Both the game tasks and the non-game tasks caused the families to designate time specifically to explore F&V by sensory and mental interactions together which may be a contributing factor to why a positive effect was found in both groups. The positive effect of designating time to these types of tasks has also been found in other studies [10,43,45]. The social situation can also affect willingness to taste through listening to others' reflections and expectations and observing their behavior towards the F&V [46,47].

#### *4.4. Can Serious Games Improve Food Behaviour in Families?*

The results of this pilot study do not show any additional effect of using a serious game to improve food behavior in families compared to performing similar non-game tasks, despite the fact that other studies [10,48] have found an effect of physical games on vegetable consumption in children, indicating a potential effect of games. As this pilot study failed to show an effect, further research into this specific segment and topic should be done in order to fully understand the possible outcomes. Conducting a similar study on a larger scale with increased intervention time is recommended in order to investigate if long-term use of the intervention materials would show additional differences between the groups. The additional motivators [14,15] of the game may give rise to continuous use of the game as indicated in the qualitative feedback where the GC requested to play the game during the intervention period. Such continuous use would result in continuous exposure to F&V and here potentially facilitate long-term effects through mere exposure [23]. On the contrary, the non-game tasks may become more tedious in the long run due to fewer motivators. This speculation was substantiated in the qualitative feedback by several of the parents in the game group, mentioning the game as an important motivator. Two game-families reported how their children had requested to play the game again, substantiating the motivational effects of a game.

On the other hand, it is still worth considering the possibility of a long-term effect of the non-game tasks as well, as they are less confined to a specific situation (a game situation), and therefore may be more easily incorporated into a busy lifestyle. Although playing *The Kingdom of Taste* has the potential to be more motivating over time, the non-game tasks may become integrated into the family's food habits more effortlessly and thereby constitute easy and accessible tools to introduce novel foods. If elements of the home assignments are adopted as new habits in everyday meal situations, rather than requiring the family to set aside time specifically to do the tasks in the format used in this study, the positive effect may occur more automatically and effortlessly [49]. The results of the simple tasks performed by the non-game-families in this study are an example of how little effort it takes to improve food behavior. It seems that it is a matter of making a habit of tasting and discussing ingredients, flavors, etc., of F&V together in an explorative manner—leading to a continuous introduction to and integration of novel F&V, which could cause a shift in food choice and behavior. Ultimately, this could result in overcoming the limiting factors faced by parents of introducing novel vegetables [7]. These speculations on turning the elements of the home assignments into everyday habits are not substantiated by the collected data, as long-term effects were not investigated, causing a need for further research.

#### *4.5. Strengths and Limitations*

The use of mixed methods in this study constituted a strength, as the qualitative data provided insights that would otherwise not have been discovered through the quantitative data, markedly changing the discussion and conclusion of the study. The major limitation to the study was the small sample size, which over the course of the intervention period was reduced from 49 to 39 individuals, and as mentioned previously, it would be valuable to conduct a similar study with increased sample size and time span. Another limitation was that the families were recruited from a small geographical area in or around Copenhagen, potentially limiting the diversity of family lifestyles and social and

environmental surroundings. Copenhagen has a large percentage of people with high educational levels compared to other parts of Denmark, which is expectedly reflected in this study sample and can have caused a bias as educational levels have been found to correlate to diets and health. Furthermore, as participating families were recruited through social media and newsletters by Taste for Life, they can be expected to have a higher interest in food than the general population, which implies that results may not extrapolate to all families. Different family compositions (varying from two parents and three children to two parents and one child) and children's age will likely have an impact on the effect of the game concerning level of help provided and adaptation of tasks. Another limitation arises from the younger children being able to receive help from their parents or the experimenter to fill in the test questionnaire. Although parents were requested not to help their children with anything other than writing, it is uncertain whether patents fully complied with the instructions, and they may also have suggested responses to their children. This could potentially cause a difference between children able to write by themselves and children not able to write by themselves. It is not possible to know if the children, who requested to play the game again, did it to gain attention from their parents or if it was because they wanted to play the game. Due to the limited period of time available to complete the study, it was not possible to measure if the effect of the intervention was persistent over time. It is recommended that future studies contain a control group with no tasks or tasting of F&V to be able to measure any possible differences in effect between using the game and not doing any tasks.

#### **5. Conclusions**

In conclusion, most families reported improved food behavior towards F&V in the children—regardless of the presence of a serious game. This indicates that designating time as a family to taste and discuss attributes and handling of F&V is enough to improve food behavior. However, the quantitative results were not as clear, as most measures showed no or limited change. A decrease in food neophobia score was seen in all four groups; however, it was only significant for the parental groups and the non-game-children, indicating no difference between the treatment groups. The lack of complete alignment between the quantitative and qualitative results raises the question of whether current quantitative measures are capable of truly reflecting concepts as complex as willingness to taste and food behavior. Based on these findings, conducting a similar study of larger scale to investigate if these results are persistent is recommended. Results of such a study could be used to consider if future research in this area should initially focus on developing new and better ways of measuring the complex concepts within this field of study by adopting a broader approach of both quantitative and qualitative measures.

#### **6. Future Perspectives**

Based on the discussion of appropriate measures to investigate a change in food behavior, a better approach may be to evaluate the journey towards willingness to taste, instead of the end point (tasting). The authors of this article therefore suggest that *food exploration* could serve as a new concept, which through both quantitative and qualitative measures allows the assessment of many different ways of interacting with novel food. Exploring foods can take place in several ways, both as a sensory interaction (tactile, olfactory, visual, auditory, gustatory) or as a mental interaction (e.g., using one's imagination to compose a meal, associating one food with another food, memory, etc.). Examples of exploring a novel or disliked food could for instance be a sensory-based description of a food based on flavor and appearance or a combination of sensory and mental interaction, as seen in the study by Coulthard and Sealy [10], where pre-school children created pictures using F&V. The concept of food exploration acknowledges the existence of different levels of novelty and resistance towards certain foods. If a person is not comfortable with tasting a novel food, he or she might be comfortable with interacting with the food in other ways. Engaging in non-taste sensory or mental interaction with foods might give rise to willingness to taste at a later time point.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/12/5/1415/s1. Table S1: Mean (SEM) number of words used to describe the six F&V in the baseline and follow-up test and significance in difference for the treatment and age groups.

**Author Contributions:** Conceptualization and methodology, S.S., A.S. and A.O.; data collection, and data analyses, S.S. and A.S., interpretation of data, S.S., A.S. and A.O., writing of the manuscript, S.S., review and editing, S.S. and A.O., supervision and funding acquisition, A.O. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was part of the "Taste for Life" project, which is financed by Nordea-fonden. The foundation had no involvement in the work.

**Acknowledgments:** We would like to thank Cathrine Terkelsen for her help and guidance with working with children and qualitative data in this project, Anna Skouw Nielsen for illustrating the game board and the booklet, and Peter Willer Hansen for producing the physical elements of the game.

**Conflicts of Interest:** All authors declare that there are no conflicts of interest regarding the publication of this paper.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Nutrients* Editorial Office E-mail: nutrients@mdpi.com www.mdpi.com/journal/nutrients

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18