Baby’s first bites: a randomized controlled trial to assess the effects of vegetable-exposure and sensitive feeding on vegetable acceptance, eating behavior and weight gain in infants and toddlers

When parents start complementary feeding, they can choose from a variety of foods to introduce to their children, including (baby) cereals, grains, fruits or vegetables [21, 22]. Already in the 1970s it was theorized that to improve the acceptance of vegetables, these should be introduced before fruits or other sweet tastes during complementary feeding because infants’ inherent preference for sweet tastes will interfere with vegetable acceptance [23]. The effects of starting complementary feeding exclusively with vegetables on promoting vegetable acceptance has, however, not been studied often [24]. Two other methods of increasing vegetable intake and liking have been studied extensively. First, repeated exposure to the taste of vegetables has been shown effective in increasing its intake and liking in infants and preschoolers [24‐32], especially for bitter tastes [33]. Second, being exposed to a variety of vegetables increases vegetable acceptance in infants [23, 29, 34, 35]. However, whether it is indeed most effective to start with vegetables only was not tested until the trial by Barends et al. in 2013 [22]. This study showed that infants exposed to a variety of vegetables during the first three weeks of complementary feeding – including a target vegetable to which they were repeatedly exposed – nearly doubled their intake of the target vegetable, whereas children who only received fruits showed increased intake of fruits but not of vegetables [26]. Shortly after this trial, another intervention study found similar results: encouraging parents from the United Kingdom to start complementary feeding with a variety of vegetables significantly increased vegetable intake compared to a control group in which parents were allowed to start complementary feeding with whatever food they wanted [36].

Thus, there is preliminary evidence that starting complementary feeding by repeatedly exposing infants to a variety of vegetables is an effective way to increase vegetable intake and liking in the first year of a child’s life. However, the beneficial effects on vegetable acceptance do not seem to last when children grow older [27, 30, 37]. This is in line with the finding that children are open to trying a variety of different tastes in their first year of life, but tend to become more selective about their diet when they become older (especially in the ‘food neophobic phase’) [24, 38, 39]. Indeed, in the Barends et al. trial, starting complementary feeding with vegetables did not predict vegetable intake at age two, whereas how selective children were about what they wanted to eat did [27]. Continuing the active promotion of eating vegetables in the first and second year of the child’s life after exposing them to a variety of vegetables at the start of complementary feeding may counteract the negative effects of the food neophobic phase and effectively boost vegetable intake throughout childhood. However, most intervention studies have been conducted with infants in the early phases of complementary feeding or preschoolers older than 2 years; few studies focus on promoting vegetable acceptance in the difficult period between 12 and 24 months when children go through the major transition of eating the same meals as their family and enter the food neophobic phase [40, 41]. Therefore, we studied the effectiveness of a more prolonged vegetable-exposure intervention throughout the whole first year of complementary feeding, well into the more ‘difficult’ second year of the child’s life to promote vegetable intake in toddlers.

Apart from what parents should offer their children during complementary feeding, how they offer this food may also strongly influence a child’s acceptance of the offered food, as well as their ability to self-regulate their energy intake. Experimental studies show that pressuring a child to eat decreases children’s ability to self-regulate their energy intake and thereby to consume appropriate amounts of calories [42]. Similarly, pressuring a child to eat vegetables has a counterproductive effect and will make a child eat and like these vegetables less [43]. Even giving subtle prompts to eat, like moving food towards a child, may have a counterproductive effect [44]. However, if children start to decrease their vegetable intake when they enter the second year of life, parents are likely to use some sort of pressure to make their child eat. Indeed, an Australian study showed that more than half of the parents of 1–3 year-olds sometimes insist on their child eating a food, and 35% reported to pressure their child often or all the time [45]. As such, it is not surprising that many parents struggle with the question how to feed their infants effectively. Indeed, 25 to 40% report feeding problems with their infants and toddlers, including picky eating and strong food preferences [46, 47].

In contrast to pressuring children to eat, responsive feeding is often suggested to be the optimal way to feed infants and toddlers [48‐51]. Responsive feeding is generally defined as a style of feeding in which parents correctly perceive the hunger and satiety signals of the child, and respond promptly and appropriately [50, 52]. This feeding style is suggested to promote and reinforce young children’s ability to self-regulate their energy intake, because the parent who feeds responsively will not override a child’s satiety cues [50]. Indeed, promoting responsive feeding was shown to be associated with a reduced risk of overweight and of rapid weight gain during the first years of life [50, 53, 54]. However, although attending to hunger and satiety signals may promote child self-regulation of energy intake, it may not be sufficient to promote healthy food preferences including vegetable acceptance during the first years of the child’s life. As children from the age of 1.5 years become more and more autonomous and selective about their food preferences, parents have to manage that their child eats appropriate quantities, but also the specific (healthy) foods that are served. To promote healthy food preferences, parents will need to stimulate their child to eat vegetables in a non-pressuring way that is sensitive to the child’s autonomy-related needs and behaviors. This requires more than just responsiveness to hunger and satiety cues, but also sensitive discipline strategies to challenging child behavior (e.g. when a child throws their food on the ground) and sensitive responses to distracted behavior (e.g. when a child is more interested in what is happening around them than in its plate of food). Sensitive discipline strategies that parents may use entail positive encouragement (e.g. explicitly complimenting the child for positive behavior), appropriate pacing to allow the child sufficient time to adapt to the situation, granting the child appropriate amounts of autonomy (e.g. allowing the child to eat autonomously when the child is able to and shows it wants to) and showing understanding for the child’s point of view [55]. Using these sensitive discipline strategies has been shown to promote infant’s committed compliance, i.e. internally motivated and self-regulated adherence to parental rules [56]. In the current study we introduce the concept sensitive feeding to capture this broader set of sensitive parenting skills relevant to promoting children’s committed compliance to parental attempts to feed them healthy foods. Sensitive feeding thus includes the traditional concept of responsive feeding [50, 52], but with the addition of sensitive discipline as well as autonomy support, also in response to non-food related child behaviors during feeding. We hypothesize that parents showing sensitive feeding will be more successful in increasing their children’s vegetable acceptance.

In recent years a number of randomized controlled trials to promote responsive feeding have been performed, some of which incorporated the discipline component described above [57‐62] whereas others merely focused on teaching parents how to effectively respond to the hunger and satiety cues of their child [53, 54]. However, none of these interventions focused on promoting responsive or sensitive feeding alone. Instead, they incorporated a much broader range of topics such as dietary advice, advice on general feeding practices, guidelines for physical activity, or even more broad advice on how to manage the sleeping and crying behavior of the child. As such, it is impossible to isolate the specific effect of responsive feeding on the diet and eating behavior of the child, and whether this is in fact an element that should be targeted to promote healthy eating patterns. Moreover, all previous trials evaluated changes in parenting behavior via self-report questionnaires, whereas expert observations of parent-child interaction is considered the gold standard to measure parenting behavior [63]. An important disadvantage of self-reports of parenting behavior specifically is that it is questionable whether these data represent the actual parenting behavior parents show, or rather attitudes about what they think they are or should be doing. Indeed, the correlation between self-reported and observed parenting behavior is often low, both in the field of parental feeding [64‐66] and in other fields [67]. Therefore, we will test the effectiveness of an intervention focusing solely on the enhancement of sensitive feeding, by evaluating its outcomes using repeated observations of family meals at home in addition to self-reports.

Whether a combination of repeatedly exposing infants to vegetables and encouraging sensitive feeding may lead to a better vegetable intake and liking than each of the interventions alone, has never been tested before. However, there is evidence that presentation of beneficial food choices (succeeding at the what) in a non-responsive manner (failing at the how), and the presentation of unhealthy food choices (failing at the what) in a responsive manner (succeeding at the how) may lead to overweight and eating problems in children [43, 68]. For instance, an experimental study by Galloway and colleagues showed that pressuring a child to eat, even if this pressure is mild in nature, decreases the beneficial effects of repeated exposure to the taste of vegetables [43]. This suggests that an intervention aimed at both elements may be particularly powerful.

In summary, the Baby’s First Bites (BFB) study aims to test whether promoting the what and/or promoting the how of complementary feeding will result in increased vegetable consumption and liking and a better self-regulation of energy intake in infants and toddlers up until the age of 36 months. To this end, we will perform a superiority randomized controlled trial with parallel groups, comparing a) an intervention focusing on vegetable exposure (=what), b) an intervention focusing on sensitive feeding (=how), c) an intervention focusing on vegetable exposure and sensitive feeding (=what and how), and d) a control condition. The interventions will begin when the infant starts receiving complementary food (child age 4–6 months, as recommended by the Dutch Nutrition Center) and continue until the child is 16 months old. We hypothesize that a) all interventions are more effective in improving vegetable consumption and vegetable liking than the control condition without guidance on complementary feeding; b) the sensitive feeding and combined intervention will be more effective in supporting child self-regulation of energy intake than the vegetable exposure or control conditions; and c) the combined intervention is more effective than the other two interventions alone in promoting vegetable intake and vegetable liking. As the inclusion phase of the BFB study has already successfully been completed, the present article describes the characteristics of the sample of included participants as well as the design of this ongoing study.

The BFB study is a collaboration between Leiden University, Wageningen University and Research, Danone Nutricia Research and Nutricia Early Life Nutrition. The study is a multicenter trial that is currently being performed at Leiden University and Wageningen University and Research, using a superiority randomized controlled design. The protocol has been approved by the Ethical Review Board of Education and Child Studies, Leiden University (protocol number ECPW-2015/116) and the Medical Ethical Review Board of Wageningen University and Research (METC-WU protocol number NL54422.081.15). The inclusion phase started in May 2016 and ended successfully in November 2017. Mothers and their 4–6 month-old infant were randomly allocated to receive either repeated exposure to a variety of vegetables (RVE), the parenting intervention Video-feedback Intervention to promote Positive Parenting-Feeding Infants (VIPP-FI), RVE and VIPP-FI combined, or an attention-control intervention (see Fig.1 and Table 1). Families receiving the RVE intervention were further randomly allocated to one of two types of vegetables the infant is repeatedly exposed to (see Interventions below): green beans or cauliflower. Two target vegetables were chosen as the current feeding schedule is based on the 19-day feeding schedule as described by Barends and colleagues [26, 27]. Green beans and cauliflower are commonly consumed in the Netherlands. Randomization into these conditions was done using the online program TenALEA, which assured that the exact same randomization procedure was used at both study sites [69]. To make the groups allocated to the different conditions as comparable as possible concerning relevant potential confounders, randomization was stratified by age of the child at the start of complementary feeding (4, 5 or 6 months), gender of the child and study location, using minimization procedures. The online randomization program TenALEA has been used previously in other clinical trials [70, 71]). Participants were allocated to a condition by one of the PhD-students or research assistants at each study location.

Intervention effects are assessed both during and after conclusion of the interventions by performing a pre-test at the first two days of complementary feeding (child age 4–6 months; t₀), two assessments during the interventions (at the end of the 19-day feeding schedule (child age 5–7 months; t₁) and when the child is 12 months old (t₁₂)), a post-test at the age of 18 months (t₁₈) and two follow-ups when the child is 24 (t₂₄) and 36 months old (t₃₆). T₀ and t₁ are not scheduled at a standard, fixed child age but rather within a certain age range because we wanted to allow parents to start complementary feeding when they thought their child was ready. The other measurements are scheduled at set child ages because the intervention sessions following the very first start of complementary feeding are scheduled at fixed time points (see Timing of intervention sessions below). The timeline for participants is depicted in Table 2. Participants are allowed to stop at any point during the study if they no longer want to participate. If participants decide to withdraw from the study, discontinue an intervention or are unable to complete a specific assessment, they will be asked once whether they would still be willing to complete (parts of) the intervention, the post-test and/or follow-up assessments to come.

A power analysis was conducted to calculate the sample size necessary to detect a moderate effect size of .50, which is based on previous studies of the effects of repeated exposure to vegetables [27] and the effects of VIPP [72]. Given a power of .80 and an alpha of .05 the analysis showed that a sample size of 51 participants per group would be sufficient. Taking attrition into account, we aimed to include a total of 240 mothers, 60 per group (see Fig. 1 and Table 1).

We decided to focus all interventions on mothers, because in Dutch households women most often fulfil the role of primary caregiver. Participants were recruited from the general population in four Dutch provinces (Zuid-Holland, Noord-Holland, Gelderland and Utrecht) that are closest to the two universities performing the trial, Leiden University and Wageningen University and Research. Participants were recruited by sending emails with information about the study and a link to the website of the study to mothers of 2–4 month-old infants. Addressees included parents who had signed up for the ‘Nutricia for parents group’ or were parents who had ordered a free gift box containing baby merchandise from ‘WIJ Special Media’. All addressees had indicated that they were interested in receiving information on additional opportunities and/or activities. Names and e-mail addresses were available to only a limited number of researchers, ensuring the privacy of the addressees. Finally, we approached potential participants through handing out brochures at youth health care centers within the vicinity of Wageningen University and Research. We cannot ascertain how many families were invited at the youth health care centers, but the total number of families invited through the two e-mail lists was 5565. A total of 409 families expressed interest in our study, 255 of which fulfilled in- and exclusion criteria (see below) and were randomly allocated to the groups (62.3%; see Fig. 2).

Families that showed interest in our study received a phone-call from one of our trained researchers/students, explaining the study in detail. Families still expressing interest in the study at the end of the call received a detailed information brochure as well as consent forms. Both mothers and fathers were asked to sign and return the consent forms. After receiving the signed consent forms, mothers were asked to fill out an online screening questionnaire which assessed inclusion criteria. Families had to fulfil the following inclusion criteria: a) first-time mothers; b) healthy term infants (37–42 weeks of gestation); c) planning to start complementary feeding at child age of 4–6 months (families that already started complementary feeding were excluded) and d) sufficient knowledge of the Dutch language to receive advice on complementary feeding in Dutch and to be able to fill out Dutch questionnaires. Mothers with major psychiatric diagnoses (e.g., depression, schizophrenia or borderline personality disorder) were excluded, as these may affect parenting [73]. Following the study protocol of Barends and colleagues [26], families were also excluded when the first-borns were twins or in the case of medical problems in the infants that influence the ability to eat, such as food allergies, swallowing or digestion problems. Finally, for standardization purposes, mothers who were not willing to commit to the outcome of the randomization procedure were excluded, e.g. the child was assigned to a VIPP-FI group, but the mother was objecting to being video-taped. A flowchart of the inclusion phase can be found in Fig. 2.

In total, 255 first-time mothers and their babies were randomly allocated to the various conditions. Directly after randomization, prior to starting the intervention-phase, 12 mother-infant dyads dropped out (for reasons, see Fig. 2). A total of 243 families successfully started the intervention-phase. Mean age of the mothers was 30.4 years (sd = 4.7, range 18–44). Concerning educational level, 41.6% of mothers had a lower education (finished high school or vocational school), 38.7% finished higher education (higher vocational school) and 19.8% finished university. The trial was thus successful in including a large group with lower education, which is generally considered a risk factor for having less healthy eating habits [74] and less beneficial parental feeding styles [75]. About 18% of mothers worked fulltime, and 63 worked part-time, and 19% did not have paid work Gender of the child was roughly equally distributed (47.3% boys); mean age of the children at the start of the intervention-phase was 4.68 months (sd = .42, range 3.98–6.38 months); median age was 4.57 months.

The specific content and timing of the RVE and VIPP-FI interventions are specified in Table 1. To control for possible placebo-effects due to receiving attention from researchers/interveners, the number of contacts with researchers/interveners and time in between contacts are the same for all conditions. The interventions in all groups as well as the attention control condition is performed by trained researchers or Master’s students in the fields of nutrition or child and family studies. Participants in all conditions are allowed to seek any type of concomitant advice on infant feeding during the trial; to control for potential co-intervention bias we ask participants after the interventions are completed whether they sought advice concerning feeding elsewhere, and if so, where and how often.

Prior to the start of each intervention, all mothers are instructed to give their infant rice-flour porridge with a spoon for at least five days, to accustom the infant to eating food from a spoon [26]. Each intervention starts with providing infants their first bites of complementary foods according to a specific 19-day feeding schedule (see Table 3). The infants in the repeated exposure and combined conditions receive a variety of commercially available jars of vegetable purees, whereas the infants in the VIPP-FI and attention-control condition receive similar jars containing both fruits and a sweet vegetable puree (carrots). During the first two days and the last two days of the feeding schedule, the target and control vegetables (cauliflower and green beans) are provided to infants in all conditions. During these days, families are visited at home by the research team and the feed is videotaped; researchers measure at home how much the child has eaten (see Measures). During the other days of the feeding schedule, the mother feeds her child at home without the presence of the researchers. To facilitate compliance to the feeding schedule, mothers receive a printed overview of the feeding schedule indicating which puree to feed their child on each of the 19 days. In addition, each jar of food is labelled with a sticker indicating the day of the feeding schedule.

After this feeding schedule has been completed, all families are provided with a total of 100 jars of age-appropriate fruits and/or meals with vegetables, depending on the condition they are in, up until the child is approximately 12 months of age (distributed on five different occasions; 20 jars per occasion). Parents are free to decide whether they want to feed their baby using homemade foods or the jars provided to them. The provision of these foods serves as a means to facilitate prolonged exposure to vegetables in the repeated vegetable exposure conditions by making sure age-appropriate meals containing vegetables are available to the families. Whether or not families use these jars and how much the child eats of these jars is reported by the mother.

The five sessions of each intervention and the phone calls in the control condition are timed to take place when the infant goes through major transitions in eating (see Table 2). It was decided to give advice specifically during these major transitions to optimize the potential effectiveness of the interventions. The first two sessions are scheduled when the infant has just started eating complementary foods (approximately one and two weeks after the start). The third session is scheduled when the child reaches the age of 8 months and parents should start introducing their child to more lumpy foods to facilitate their infants’ acceptance of different food textures [76]. The fourth session is scheduled when the child is approximately 13 months and is allowed to eat the same foods as the rest of the family. Finally, the fifth session is scheduled when the child is 16 months of age to prepare parents for the potentially difficult ‘food neophobic phase’ that infants tend to reach in their second year [39, 77].

Researchers coding video data are blinded for intervention-allocation. It is impossible to blind participants for intervention-allocation, because they will be informed prior to randomization about what types of advice they can receive in the study and it will be clear after randomization what type of advice they are receiving.

As a compensation for the time and effort participants invest in our study, families receive several compensations. Apart from the pureed vegetables or fruits during the feeding schedule and the 100 jars of baby foods until the infant is 12 months of age, families receive gift tokens of 25 euros and a gift for the child of approximately 5 euros at t₁₈, t₂₄, and t₃₆. Additionally, all videos made throughout the study are shared with the families at completion of the study, and families randomly allocated to receive VIPP-FI receive the videos used for the intervention during the last session of the intervention.

To involve participants in the study we will send families biannual newsletters about the study, mentioning interesting facts (e.g., inclusion rates, presentations at symposia, pictures of researchers/students involved in the project). Also, we aim to stimulate a pleasant relationship between researchers and participating mothers by for example sending birthday cards to the family when the child will have its birthday. In a similar effort, and to diminish any additional burden for the participating families, we will strive to provide continuity in the researchers/students that are in direct contact with a family (e.g., at home visits or telephone calls). Moreover, we will make sure during every home visit to check whether participants have any questions about the measurements and/or interventions and to provide assistance in filling out questionnaires or dietary recalls whenever needed.

All data will be stored using numbers to identify participants at the secured databases of Leiden University and Wageningen University and Research. Only one document exists that links participant numbers to personal data, and this file is only available to the main researchers performing data collection at Leiden University and Wageningen University. Data that need to be entered manually (e.g., measured weight and height during home visits, codes of video material) will be entered in the latest version of the statistical software package IBM SPSS Statistics by trained researchers/students. The quality of this data entry will be checked regularly by another (independent) trained researcher/student.

As detailed in the consortium agreement-contract of the project, only researchers and students involved in the project working at any of the academic parties (Leiden University, Wageningen University and Research) will be allowed access to the data. With the exception of the video-recordings (VIPP-FI), which contain privacy-sensitive information, research data will be open access where possible (e.g. when a peer-reviewed journal requests or offers the uploading of anonymized datasets into an open access database. In these cases, all personal information will be removed from data files and replaced by participant identification numbers. The file linking these numbers to personal information will be stored digitally in a separate password protected file that will only be accessible to the researchers). Large video-files will be shared between the two universities by making copies on external pass-word protected hard-drives and personally exchanging these hard-drives.

The intention-to-treat principle will be applied to all analyses. Whether the interventions differentially affect primary and secondary outcomes over time will be analyzed using linear mixed models analyses, a technique that makes use of every data point for every participant, irrespective of their missing data. The three intervention groups will be compared to the control group, and the combined group will be compared to the repeated exposure and the VIPP-FI group. A significance level of α = .05 will be used. The analyses will be corrected for relevant covariates such as family socioeconomic status, maternal consumption of vegetables, parental body mass index (BMI), child temperament, etc.

Participants will be asked to fill out an evaluation form concerning the interventions following the last session. These forms will assess participants’ satisfaction with the intervention as well as with the person delivering the intervention. In addition, participants will be asked to note any other comments about the interventions, allowing for spontaneously reported adverse events. As the interventions are not invasive and merely provide parents support, advice and commercially available foods with a history of safe use, no adverse events are expected and no stopping guidelines are formulated. For the same reasons, a data management committee is not needed. Principle investigators at each study site (i.e. JM, SV, KG, JV and GJ) will supervise data collection and data management. We will not perform any interim analyses as we want to avoid the risk of the results of such analyses influencing the overall results of the trial. No explicit trial conduct audit is planned; however, yearly reports on the progress of the project will be sent to the major funder of the trial (The Netherlands Organization for Scientific Research). If any major changes will occur in the study protocol (e.g., changes to outcomes or assessment periods) the ethical review boards that approved the study as well as the funder of the trial will be notified of these changes.

It is planned to publish the results of our trial in peer-reviewed journals, as well as present the results at (inter) national conferences. Also, participants will receive a report of the results of our study after completion of the study. Publication in magazines for healthcare professionals and the general public are also intended. Authorship to any publications will be granted to those who fulfil the ICMJE recommendations [109]. We will not hire any professional writers.