The use of a communication tool about diet at the child health centre: A cluster randomized controlled trial

Abstract Aim To investigate the effect of a communication tool about diet used in public health nurse consultations with parents compared with standard consultations concerning the 2‐year‐old child's diet. Design A cluster randomized controlled trial. Methods Ten municipalities were selected randomly and matched in pairs. In each pair, the control or intervention group was randomly allocated. Parents were recruited to participate from January 2015 to January 2017. In intervention clusters, a communication tool about diet was used to help the parents (N = 140) to focus on a healthy diet for their child. In the control clusters, parents (N = 110) attended standard consultations. The participants completed semi‐quantitative food frequency questionnaires at baseline and end point. Results No effect of the intervention was seen on the child's daily intake of vegetables or saturated fat, or body mass index. Significantly fewer parents desired more information about food for toddlers in the intervention than in the control group.

practice (Holmberg Fagerlund, Pettersen, Terragni, & Glavin, 2016;Ilmonen, Isolauri, & Laitinen, 2012;Magnusson, Kjellgren, & Winkvist, 2012). Counselling based on a one-sided cognitive approach towards eating appears to have limited impact on healthy food choices among children. A review indicated that deriving pleasure from eating healthy foods or from contextual cues generated by parental attitudes to food and feeding might encourage children to adopt a balanced diet in the long term (Marty, Chambaron, Nicklaus, & Monnery-Patris, 2018). An association between healthy dietary habits in children and parents having higher education has been shown in several studies (Luque et al., 2018;Rasmussen et al., 2006;Vepsäläinen et al., 2018). A questionnaire study among parents (N = 234) of 1-to 5-year-old children revealed that short-duration breastfeeding or the food neophobia in the child was associated with a risk of poor dietary patterns in children later. No associations were found between the dietary patterns of children and the age when solid foods had been introduced (Bell, Jansen, Mallan, Magarey, & Daniels, 2018). A longitudinal study among children (N = 633) in five European countries by Luque et al. (2018) indicated that educational interventions should focus not only on the introduction of positively weighted foods, but also on avoidance of discretionary low-quality foods at early ages. Dietary patterns, particularly between 1 and 2 years, persisted into midchildhood or 8 years of age.
A Swedish questionnaire survey among parents (N = 478) suggested that parents could become less concerned about their child over-or undereating if they were offered skills training and practical counselling on how to respond effectively to eating behaviours, regardless of the child's weight (Ek et al., 2016).
According to a review by Holmberg Fagerlund, Helseth, Owe, and Glavin (2017), there is limited research on the effect of universal food and feeding counselling involving children under 2 years and their families. At Oslo Metropolitan University, an image-based communication tool about diet was developed in a previous project named SOMAH (The Research Council of Norway, 2013). This project aimed to facilitate communication about food and feeding practices at CHCs. The target group was immigrant populations with an increased risk of developing type 2 diabetes (Garnweidner, 2013;Holmberg Fagerlund, Helseth, & Glavin, 2019;The Research Council of Norway, 2013). This tool followed the recommendations of the National Nutrition Council in Norway (2011). A selection of the SOMAH images was adjusted and integrated into a communication tool about diet for universal use at the CHC in the present study. A motivational interviewing approach developed by Miller and Rollnick (2013) was integrated into this communication tool to optimize active collaboration about the child's diet between the PHN as counsellor and the collaborating parent.
The aim of the study was to investigate the effect of a communication tool about diet used in a PHN intervention at CHC consultations with parents compared with standard consultations concerning the child's diet at 2 years of age. The study hypothesized that the children in the intervention group would have a higher intake of vegetables, lower intake of saturated fat and lower body mass index (BMI) than those in the control group.

F I G U R E 1
The timeline according to the child's age in the clusters consisting of intervention and control municipalities

| Design
The study design was a two-armed parallel cluster randomized controlled trial (cRCT). Clusters of municipalities were randomly assigned to two groups, intervention municipalities and control municipalities ( Figure 1). Parents within the clusters answered a semiquantitative food frequency questionnaire (SFFQ) on behalf of their child at baseline (T0) and end point (T1), on average 8-11 months after the end of the intervention. This clinical trial is registered at ClinicalTrials.gov, Identifier: NCT02266953.

| Sample
In total, five matched pairs of municipalities (  If one cluster in a pair declined to participate, this led to a new draw to obtain a systematic match to the remaining cluster. Thus, obtaining the sample of 10 clusters required contact with 72 municipalities ( Figure 2). Twenty-five of the originally selected municipalities declined to participate in the research project. In addition, 20 of the drawn municipalities could not participate because they had not implemented the healthcare programme at the CHC in a way consistent with the authorities' regulations. Seventeen were excluded because their CHCs did not practise individual 10-month consultations.
Municipalities with fewer than 100 births in 2012 and municipalities in the three northernmost counties of Norway were excluded from the draw. Three municipalities were excluded because they had been involved in the development of the intervention. In total, 139 municipalities were available for the sampling.
The municipalities were contacted through their head of the CHC. Oral information about the research project was provided, and if accepted, written information about the project was sent. The study's participants consisted of parents with young children who had consented to participate in the trial before their child reached the age of 10 months. Parents were recruited during their visits at the CHC. The parents received oral and written information about the study from their PHN. The only exclusion criterion was parents with insufficient Norwegian skills to understand the written information about the study. Participants were recruited continuously from 5 January 2015 to 31 January 2017.

| Control municipalities
The participants and their children in the control municipalities The image presenting: 1. "Infants learning to feed themselves" A 1-year-old infant sitting in a high chair and using a bib, customized dishes and cutlery, finger food and appropriate food on a plate 2. "The Plate Model" (Camelon et al., 1998) Proportions of the three food groups: carbohydrates (e.g. potatoes, pasta and bulgur), vegetables and proteins (e.g. meat, fish, beans, peas and eggs) and a spoon of plantbased oil 3. "Five a Day" Depicting vegetables and fruit to encourage consumption of at least five portions of them each day 4. Whole grain bread Depicting healthy alternatives and one example of an unhealthy choice 5. Healthy bread spreads Depicting healthy alternatives and one example of an unhealthy choice 6. Healthy flavourings of natural yoghurt Inspiring alternatives of berries and fruit to use as flavouring in yoghurt 7. Natural yoghurt as an alternative to sugary yoghurt Comparing natural yoghurt without added sugar in relation to the amounts of added sugar in sweet yoghurt types Themes at 12 months: 1. "Infants learning to feed themselves" The same image as at 10 months

| Intervention municipalities
In addition to the content mentioned above, the participants in the intervention municipalities were exposed to the intervention, the PHN's use of the communication tool about diet. In this intervention, the PHN presented to the parents six or seven printed images per consultation about different nutritional themes related to the child's age and developmental stage ( Table 1). The images were A4 size and presented on a flip-chart stand at the PHN's desk. To ensure the selection of appropriate image material to be integrated into the communication tool about diet, a feasibility and acceptability test as outlined by Richards (2015) was performed in June-September 2014. Based on these images, parents were invited to discuss relevant themes regarding food and feeding practices, adapted to their family and child.
An aim of the communication tool about diet was to help the family choose an optimal diet for the child, raising the caregiver's awareness of food habits as central to the child's recent and long-term health.
A further aim was to help the family adjust meals to give their child the opportunity to develop skills in eating and get used to different tastes. The authorities' labelling schemes designating healthy foods and dietary factors according to the authorities' recommendations were emphasized in the images and during the use of the communication tool about diet.

| Preparations for implementation
Before implementation of the trial, the first author had visited all cooperating CHCs to prepare the PHNs by explaining about the trial and about their tasks. This preparation lasted 2-3 hr on average.
In intervention municipalities, a standardized 1-hr introduction on the use of the communication tool about diet and the corresponding user's manual was included. To minimize likely performance bias and expectation bias related to awareness, information about the study outcomes was withheld during the preparations. This was to enhance objectivity among the cooperating PHNs and participants as described by Polit and Beck (2017).
To accelerate recruitment effort, the PHNs were kept motivated and reminded about this project through monthly contact with their managers by email or telephone. For the same reason, the first author visited all CHCs during the recruitment process.

| The semi-quantitative food frequency questionnaires
The SFFQ at baseline, designed to investigate feeding practices At baseline, the parents who had consented to participate received an SFFQ when the child was approximately 8.5 months old.
They were asked to complete and return this SFFQ just before the child's 10-month consultation at the CHC. At end point, the SFFQ was sent to the parents just after the 2-year consultation at the CHC.

| Age-and gender-related body mass index
Overweight was estimated at the cut-off point of BMI 25. BMI 25 is equivalent to BMI 18 adapted to the child's age and gender among 2-year-old children (Cole, Bellizzi, Flegal, & Dietz, 2000).

| Primary outcome
The primary outcome was daily intake of vegetables measured as grams of consumed vegetables per child, based on the completed SFFQs.

| Secondary outcomes
The study evaluated several secondary outcomes: the percentage of energy intake (E%) of saturated fatty acids of the total daily energy intake of the child, based on the completed SFFQs; the child's BMI, based on the completed SFFQs; and lastly the number (proportion) of parents who reported a wish to obtain more information about their toddler's diet, based on the completed SFFQs.

| Sample size consideration
The study was powered to reveal a predefined change in vegetable intake between the intervention and control groups. According to the literature, children consume on average 50 g vegetables daily at the age of 2 years ). Thus, we expected this daily intake of vegetables to increase by 15 g in the intervention group as compared to the control group. To keep the level of statistical significance at 5% and statistical power of 80% (beta = 20%), we would need 176 children in each group to determine whether the change described above was statistically significant. Attrition was expected, and 300 children were enrolled in each group to make sure our study was sufficiently powered.

| Ethics
All participating parents gave their written informed consent.
Participation was voluntary, and the participants could withdraw without giving a reason. All data were treated as confidential.
Participant anonymity was guaranteed. Regional Committees for

| RE SULTS
Analysis of completers versus non-completers was conducted at baseline. This revealed comparable groups among the completers and non-completers of the SFFQ at baseline concerning the background variables, except for fewer married and single mothers and more cohabitant mothers among the completers. This analysis also showed a somewhat higher educational level among mothers and TA B L E 2 Differences between completers and non-completers at baseline fathers who completed the SFFQ (Table 2). According to process evaluation, the median time for performing the intervention was 10 min in the intervention municipalities.

| Participant flow
Two hundred and thirty-two participants responded to the SFFQ in the intervention municipalities (65% response rate) and 208 in the control municipalities (63% response rate) at baseline (Figure 2). At the end of follow-up, 140 participants in intervention municipalities (39% response rate) and 110 participants in control municipalities (33% response rate) had completed the SFFQ (Figure 2). Table 3 shows that the distribution of the selected background variables was similar in the intervention and the control municipalities.

| The main outcome
The median age of the included children was 2.2 years in both groups at the time of completion of the SFFQ at end point (Table 4). Our data revealed only small differences in the background characteristics between the groups at 2 years, except for a significantly higher proportion of the mothers in the intervention municipalities who were either married or cohabitants compared to the control group, 98.5% versus 93.6%.
We did not find any statistically significant differences between the intervention group and the control group concerning the main outcome, the mean daily intake of vegetables, 64.5 g versus 68.7 g (Table 5).

| Secondary outcomes
Our study did not reveal any statistically significant differences between the groups regarding consumption of saturated fat or BMI among children aged 2 years. However, a statistically significant difference was revealed concerning the desire for more information. Fewer of the parents in the intervention municipalities than in the control municipalities reported that they desired more information about food for toddlers, 24.4% versus 41.1% (Table 5).

Variables Completers (N = 250) Non-completers (N = 190) p-value*
Age of the child (months) c , mean (SD) 9.8 (0. The child referred to in the food frequency questionnaire.
*p-values <.05 considered as statistically significant differences between groups.

| D ISCUSS I ON
Our findings showed no statistically significant differences between the two groups on the predefined outcome variables: daily vegetable intake, daily intake of saturated fat and BMI. We do not know if the communication tool about diet was used as intended, to promote a dialogue. The tool might have been used purely for one-sided information giving, corresponding exclusively to a cognitive parental approach to eating and thus with less impact on their healthy food choices (Marty et al., 2018). Consequently, no effect was attained on the outcomes related to the child's healthy diet. It is not known whether skills training and practical guidance on how to respond effectively to the child's eating behaviours (Ek et al., 2016) were highlighted during the intervention.
The choice of the study's nutritional outcomes, the child's daily intake of vegetables and saturated fat, was seen as reasonable because they are associated with the occurrence of cardiovascular disease in adulthood (Kaikkonen et al., 2013;World Health Organization, 2017). A modest increase in vegetable and fruit intake could have an impact on population health, particularly prevention of deaths from cardiac heart disease (Boeing et al., 2012;Tobias et al., 2006). The current study focused exclusively on vegetables because bitter-tasting vegetables might usually be harder to accept than fruit initially during the transition to the family's food, due to infants' innate preference for sweet flavours (Birch & Ventura, 2009). In a previous national dietary survey in 1999 among 2-year-old children, the mean daily vegetable intake was 33 g .
Eight years later in 2007, a corresponding national dietary survey among the same age group showed a mean vegetable intake of 54 g/ person/day ). Related to the findings one de- Withholding of information about the outcomes of the trial among participants to prevent performance bias and expectation bias might have been one reason why the intervention did not contribute to the parents changing their child's diet in terms of vegetable intake.
In a systematic Cochrane Collaboration systematic review on "Interventions for increasing fruit and vegetable consumption in children aged 5 years and under," nine studies were related to children younger than 2 years of age (Hodder et al., 2018). None of these in- The child referred to in the food frequency questionnaire.
*p-values <.05 considered as statistically significant differences between groups.

| 1227
HOLMBERG FAGERLUND Et AL.  & Birch, 1994;Vazir et al., 2013;Verbestel et al., 2014;Watt et al., 2009). According to this Cochrane review, the daily vegetable intake in children younger than 5 years of age increased on average by 3.50 g based on child-feeding and multicomponent interventions (Hodder et al., 2018). Such a small effect size might limit potential public health benefits from implementing these types of interventions (Hodder et al., 2018).

TA B L E 4 Differences between intervention and control municipalities at end point
Our study revealed an intake of ≥10E% of saturated fatty acids among 89%-91% of the 2-year-olds and a mean saturated fatty acid intake of on average 13E% in both groups. This is higher than the

TA B L E 5 Differences in outcome variables between intervention and control municipalities at end point
This intervention might have been too weak to have an impact on any parental food preferences related to saturated fatty acids. To achieve an effect regarding parental selection of less saturated fat in the child's diet, a more assertive intervention has shown positive results. In this longitudinal Finnish study by Kaitosaari et al. (2006), the intervention group parents received individualized dietary counselling twice a year by a physician and a dietitian from when their child was 7 months old. This intervention focused on supporting parents in adopting a healthy low-saturated-fat and low-cholesterol diet for their child. The control group received general health education at the CHC as usual before school age and no in-depth dietary counselling when the child grew older. This intervention results in children consuming less saturated fat than control children at the age of 9 years (Kaitosaari et al., 2006).
The BMI levels of 11.7% overweight children in the interven- where similar dietary food patterns were tracked between the age groups of 2 and 8 years (Luque et al., 2018). According to the World Health Organization (2018), increased BMI is a major risk factor for non-communicable diseases such as cardiovascular diseases, diabetes, musculoskeletal disorders and some cancers. Childhood obesity is associated with a higher risk of obesity, premature death and disability in adulthood. In addition, obese children experience an increased risk of breathing difficulties, fractures and hypertension, early markers of cardiovascular disease, insulin resistance and psychological effects (World Health Organization, 2018). According to a Norwegian population-based longitudinal study, being overweight or obese at the age of 8 years was associated with an increased BMI throughout infancy and childhood. Hence, interventions to prevent children becoming overweight should start at an early age (Glavin et al., 2014). For instance, increased consumption of vegetables and fruit and regular physical activity could help at the individual level to prevent overweight and obesity (World Health Organization, 2018).
Individual-level interventions targeting healthy eating and physical activity usually have no statistically significant effect on clinical measures of obesity in children (Nigg et al., 2016). The current study's results showing no impact of the individual-level intervention on the child's BMI were therefore as expected.
Dietary assessments of infants and preschool children appear complicated because their dietary habits often change rapidly (Andersen, Lande, Arsky, & Trygg, 2003). Food served during this age is often not consumed, and gaining an overview of total food intake might be challenging for the parents because of the child's day care (Andersen et al., 2003). Regarding reliability concerns as discussed by Polit and Beck (2017), the accuracy and consistency of information obtained from the completed SFFQs have been central.
The chosen age-specific SFFQs were considered suitable for assessing dietary intake in large groups (Andersen, Lande, Trygg, & Hay, 2004). The SFFQ used among 2-year-old children as the basis for the current SFFQ at end point was validated as valuable for measuring average intakes of energy, macronutrients and several food items.
Its validity was not influenced by length of the parents' education or whether the child was attending day care (Andersen et al., 2004).
Our data did not reveal any statistically significant differences at baseline between the groups, so we did not adjust for any possible confounders, and thus, no multiple models were fitted. Further, as the consumption of vegetables was close to zero at baseline, we were not able to model possible changes in this consumption using repeated-measures methodology.
Use of the communication tool about diet had positive effects on parents in their search for information about food for toddlers.
Significantly fewer of the parents desired more information about food for toddlers in the intervention group than in the control group.
This might suggest that PHNs in control municipalities spent less time counselling on food and feeding practices because they were not obliged to use any communication tool about diet. PHNs in intervention municipalities reported a median of 5-min longer con-

| Limitations
Although power analysis was carried out before the study, we did not achieve a large enough sample size based on the preceding calculations because of high attrition among participants who had consented initially. However, we can speculate that even if the calculated sample size had been achieved, it would not have been possible to conclude on any effect of the intervention because the observed differences in the main outcome between the groups were much smaller than anticipated. Based on the literature regarding vegetable intake among 2-year-olds ), we initially anticipated that there would be a difference of 15 g/ person/day between the groups. According to a recent Cochrane Collaboration systematic review, child-feeding interventions appear to increase vegetable intake in children by 3.50 g on average (Hodder et al., 2018) Polit and Beck (2017).
The fact that the recruitment process lasted for 2 years might have influenced internal validity, because during this time products with a healthy diet focus have been introduced to the market continuously. However, this would have affected both groups concurrently.
The presented sample was similar to the general population in Norway except for the level of education and ethnic background.
Only two per cent among mothers and 2%-3% among fathers had The participants having almost exclusively non-immigrant background and underrepresentation of education below upper secondary level compared to the general Norwegian population might limit the generalizability of our findings. The unknown response rate, based on how many participants were initially invited to participate in the study, also implies a limitation to the generalizability of the results.

| CON CLUS ION
Our study did not reveal any differences between the groups on the outcome variables, daily vegetable intake, daily intake of saturated fat or BMI of the child at the age of 2 years. Thus, the PHN using a communication tool about diet in three CHC consultations did not influence parents to choose more vegetables and less saturated fat in their children's diet when compared with the control parents.
There were, however, significantly fewer of the parents who desired more information about food for toddlers in the intervention group than in the control group. Our findings indicate that regular nutritional CHC counselling, despite its positive contribution to parents' information search, will not have any particular impact on the daily intake of specific components in the child's diet such as vegetables and saturated fat.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

R E S E A RCH E TH I C S CO M M IT TE E A PPROVA L
The study was approved by the Regional Committees for Medical and