Effect of TV advertising on energy intake of Georgian children: Results of an experimental study

2.1 Study design

The experimental factor investigated was the exposure to TV and TV advertising, organized in five levels: “no exposure to TV” (NoFilmNoSpot), “exposure to TV without advertisement” (FilmNoSpot), “exposure to TV and one advertisement” (FilmLowSpot), “exposure to TV and two advertisements” (FilmMediumSpot), “exposure to TV and three advertisements” (FilmHighSpot). The Film was a cartoon lasting about 22 minutes, chosen to be a non-spoken cartoon.

The main character was the Disney cartoon character Pluto ©, and the movie was made up by a set of episodes, according with a previous study [12]. The advertisements lasted 30 seconds and were in the country-specific language [13]. They referred to the snack given to the children during the experimental session. The advertisements were delivered at 0′ for the FilmLowSpot level, at 0′ and 6′15″ in the FilmMediumSpot level, and at 0′, 6′15″, and 10′35″ for the FilmHighSpot level. The children had never seen the advertising before.

2.2 Randomization and ethical conduction of the study

Each child was randomized to one of the 5 cells of the full-factorial design. Randomization was performed using a software integrated with the data collection and study conduction system. Randomization was stratified by age (two groups of children 3–6 and 7–11 years old) and by gender (male and female) to ensure complete balance for the potential confounding factors. Exclusion criteria were: suffering of cognitive disorders or metabolic diseases or allergies to the snack presented during the experiment.

Parents were asked to sign an informed consent before child’s participation in the study. Participant treatment followed the guidelines and ethics issued by American Psychologist Association. Appropriate permission was obtained.

2.3 Study setting

A sample of 60 children was enrolled in a school of Tbilisi in Georgia. The experimental session was conducted during the afternoon break, in a quiet room within the school structure. All children had a standard meal (provided by the school) and practiced regular physical activity.

Parents were asked to sit on a lateral or back position to fill out the parents’ questionnaire, without being seen by the children, who were sitting at a table, facing a monitor (if applicable according to the group to which they were randomized). Children were video-recorded through two hidden HD cameras, positioned to record frontal and side images, while the researcher assumed a back-screen position to easily interact with the child.

2.4 Parents’ questionnaire

A standardized questionnaire, consisting of different parts, was administered to parents. The first section assessed family’s socio-demographic characteristics, habits and anthropometrics (number of siblings, with whom the child lives, parents’ educational level, working status, family members’ weight and height, and physical activity frequency). The second part of the questionnaire aimed at assessing child’s food habits, (duration of breastfeeding, actual meal frequency and food consumption). The last investigated child’s physical activity (frequency and weekly hours of physical activity).

2.5 Anthropometrics

Weight and height of children, wearing light clothes and without shoes, were taken using an electronic stand-up balance scale and a rigid metric belt. BMI was computed as weight divided by height squared and BMI z-scores were computed for each child according with CDC Pediatric Growth Charts [14].

2.6 Brand awareness assessment

Children brand awareness was assessed using a validated tool, the IBAI questionnaire [15]. The questionnaire consisted of 12 food brand logos (including both international and country-specific brands).

Children were asked to recognize the brand presented in the food image, to match it with a second image chosen among 4 different options, and finally to report the specific name of the food product. Brand Awareness Scores (IBAI-score) could range from a minimum of 0 to a maximum of 36. Four different categories of brand awareness were identified: 0–12 very low brand awareness; 13–18 medium-low brand awareness; 19–24 medium-high brand awareness; 25–36 very high brand awareness [16].

2.7 Study conduction

Children were measured (weight and height) to calculate their BMI. Afterwards, the IBAI questionnaire was administered to children by the interviewer. Once the assessment of brand awareness was completed, the interviewer explained the progression of the study to the child.

The snack employed for the study was a chocolate based product, mimicking Easter egg (nutritional facts per 100 g (1 snack = 20 g): 545 Kcal, 8.3 g protein, 55 g carbohydrate, 51 g sugar, 32.4 g fat, 14 g saturated fat, 0.083 g sodium, 1.8 g alimentary fiber). It was prepared ad hoc for the experimental study in Georgia, and was chosen because the particular selling combination of the snack allowed a precise weight of the content, both before and after the study, giving, in this way, an accurate estimate of the amount of calories eaten by every child.

The part containing the chocolate was previously weighted and then offered to children. At the time of the first offering of the snack to the child, if the child was assigned to a TV exposed group, the movie was started and not interrupted for any reason thereafter. After the first snack, the investigator could present a further snack one by one (always waiting for the child’s request). Each child could eat ad libitum up to a maximum of 12 snacks. Snacking offering was performed using a pre-developed protocol, in a way to control, as much as possible, investigator influence on the child behavior.

Once the experimental session was concluded, the investigator weighed the chocolate’s residues per each snack, recording all weights in the study software.

All the sessions were digitally recorded for subsequent examination and data quality assurance.

2.8 Statistical analysis

Descriptive statistics were reported using median (I-III quartile) for continuous variables and percentages (absolute numbers) for categorical variables, whenever appropriate. Wilcoxon-Kruskal-Wallis test was performed for continuous variables and Pearson chi square test for categorical ones.

The analysis was based on a linear model where blocking factors (age, gender, and FilmSpot) were considered in the model. This is the base model used in the analysis. Specific investigations on single factor-level effects were conducted using appropriate linear contrasts.

To further check for additional confounding factors, six models were developed. In each model, variables were added to the base model:

Base: Age+Gender+FilmSpot

M1: BMI+Breast Feed+Hours/Week TV+Physical activity (hours/week)

M2: BMI Father+BMI Mother+number Brothers

BA: IBAI-Score

M3: Number of rooms in the house+number of TV in the house+Educational level mother

M4: breakfast in the morning+fruit portions/day+vegetables portions/day

Each model was estimated and the AIC criterion in the backward fashion was used for variables selection. For each model, significance of the experimental factor was assessed.

Children were identified as “high consumers” if their caloric intake during the experimental session exceeded the III quartile of the observed distribution.

Analyses were performed using the R System [17].

3.1 Sample characteristics

Fifty-seven children took part to the experimental session. Three children did not take part to the experimental session due to late refusal of consent. Sample characteristics are summarized in Table 1. Children’s parents BMI showed a median of 22 for the mothers and 26 for the fathers. The median weekly hours of TV watching was 15. The analysis of eating habits demonstrated that most of children ate only one serving of fruit and vegetables per day. Referring to brand awareness, the IBAI score showed a normal distribution (Fig. 1, on the left side) with a median value of 13 (which corresponds to low brand awareness).

OPEN IN VIEWER

Fig.1

On the left side, IBAI box plot. On the right side, energy intake box plot.

No significant differences were found according with the experimental factor (TV viewing and TV advertising).

The energy intake reported a median value of 220.72 kcal (Fig. 1, on the right side), roughly corresponding to a median of 2 snacks for each child (median of 40.50 g).

Data on energy intake (both absolute values and per BMI) in specific study factors’ subgroups are presented in Table 2. The higher absolute caloric intake was found among children in the “exposure to TV without advertising” group (median of 307.30 Kcal), which is shown also in Fig. 2. However, the assessment of the relationship between energy intake and different levels of TV viewing and TV advertising exposure revealed no significant association, even after the adjustment for all the described potential confounding factors (Table 3).

The post-hoc power referred to the anova test, for the energy intake and the different levels of TV and TV advertising exposure, was equal to 0.22.

OPEN IN VIEWER

Fig.2

Energy intake related to TV and advertising exposure.

The third quartile of energy intake corresponded to an amount of 319.4 kcal. This value was chosen as the cut-off to identity the subgroup of children with high energy intake during the experimental session.

Only 14 children were found to be high consumers according with this cut-off. Table 4 shows the main characteristics of such “high consumer” children, who were not significant different for any of the characteristics analyzed compared to “low consumers”, except for gender (male children were more likely to be “high consumer” compared to female ones, p-value 0.011).