Abstract
BACKGROUND:
Parental feeding practices can affect children’s eating behavior and diet quality. Greek children seem to have poor diet quality and high rates of obesity. Measuring parental feeding practices could facilitate in the formation of target interventions aiming at improving children’s eating behaviours.
OBJECTIVE:
The aim of the current study was to develop and validate the Comprehensive Parental Feeding Questionnaire (CPFQ) in the Greek language.
METHODS:
The sample consisted of 399 parents (71.2% mothers) of children 2–12 years old. A translation to the Greek language of the CPFQ tool, was conducted followed by a repeatability test, explanatory and confirmatory factor analysis, measures of internal consistency test and factor correlations.
RESULTS:
Factor analysis resulted in a final questionnaire of 42 items distributed over 6 factors. Cronbach alpha values were adequate (0.64–0.89) and the correlations between factors were low (rho = –0.212 – 0.405). In addition, mothers use more the “child control” feeding practice (p = 0.002), parents with girls use more the “monitoring” feeding practice (p = 0.010) and normal weight parents use less the “restriction” feeding practice (p = 0.047), in comparison to overweight parents.
CONCLUSIONS:
Results support the reliability and validity of the CPFQ for the Greek population.
Introduction
There is good evidence that human’s dietary “identity” is already established by the age of 6 years, hence, the role of parental influence in children’s healthy or unhealthy dietary behaviours, is crucial [1, 2]. Parents serve as important role models for their children, modeling selections of certain food stuffs and certain eating behaviours [2–7].
Parental feeding practices include specific behaviours that parents usually apply to control how much, what and when their children eat. There is evidence that parents usually deliberately employ certain parental feeding practices to either reduce food intake of their child, increase food intake of their child or increase the variety of foods in the child’s diet [2, 8–10]. Children mimic their parents’ eating behavior, in general. Recent studies have shown that role modeling of fruit and vegetable intake by parents was associated with increased consumption of fruits and vegetables by their children [11–13]. In addition, children’s involvement in meal planning is also considered to have a positive impact on the child’s diet [14–16]. Many recent studies have shown that parental feeding practices can have an important impact on the quality of the child’s diet and on weight gain. Usually, feeding practices that allow children to develop self-regulation by using their internal cues of hunger and satiety and concurrently encourage a healthy and varied diet, are considered more positive [17–20].
Strict controlling feeding practices, on the other hand, such as pressure to eat, restriction of food, and use of food to control behavior, seem to have a negative influence on the intake of healthy foods and may hamper the ability of the child to self-regulate food intake [21–25]. In general, parents’ restriction of food is associated with children’s excess body weight [26, 27] and pressure to eat is usually associated with children’s slower weight status [28–31].
The rates of childhood and adolescent obesity in Greece are among the highest in Europe [32] and adherence to the Mediterranean diet in the youth is low [33, 34]. This shift towards less healthy behaviours has become more profound after the beginning of the economic austerity period [35]. Measuring parental feeding practices of the population and identifying and addressing the problematic practices, could facilitate the development of targeted and more effective interventions in improving children’s eating behaviours, diet quality and weight control [36]. Currently, there is no validated tool in the Greek language for measuring parental feeding practices, hence, the aim of the current study is to develop and validate, in Greek, the Comprehensive Parental Feeding Questionnaire (CPFQ), originally created by Musher-Eizenman & Holub (2007) [37]. The Comprehensive Parental Feeding Questionnaire [37], was chosen because, a thorough review of the current literature, has shown that it is widely used worldwide [38–40] and it is considered more complete and appropriate than other similar tools attempting to measure parental feeding practices [27, 40–45].
Methods
The comprehensive parental feeding questionnaire (CPFQ)
After obtaining written approval by the author of the original questionnaire CFPQ, the validation process began. The Comprehensive Parental Feeding Questionnaire [37] is a self-report instrument that examines a wide range of feeding behaviours in parents of children 2–8 years old. It consists of twelve factors deriving from forty-nine items, in total. The possible answers are based on a five-point scale, from 1 to 5 (“never” to “always” for the first 13 items and “disagree” to “agree” for the rest) and in 3 items the answers must be reversed. The higher score means that parents apply this feeding practice more intensely. The twelve factors (child parental feeding practices categories) derived from this instrument are: “Encourage Balance and Variety” indicates how much a parent promotes well-balanced food intake, including the consumption of varied foods and healthy food choices, “Environment” shows if a parent makes healthy foods available at home, “Involvement” shows how much a parent encourage child’s involvement in meal planning and preparation, “Modeling” indicates how much a parent models healthy eating for the child, “Monitoring” shows the degree to which the parent keeps track of their child’s consumption of unhealthy foods, “Teaching about Nutrition” indicates how much a parent uses explicit didactic techniques to encourage the consumption of healthy foods, “Emotion Regulation” shows how much a parent uses food to regulate child’s emotion, “Food as Reward” shows how much a parent uses food as reward for desired behaviour in their child, “Pressure” which investigates how much a parent pressure a child to eat, “Child Control” determines how much a parent allows the child to make decisions around what and when they eat, “Restriction for Health” measures how much a parent restricts child’s food intake, focusing on healthy eating, and “Restriction for Weight Control” how much a parent restricts child’s food intake in order to limit excess weight gain [44, 45].
Subsequently to its publication, the instrument has been validate and widely used in different countries such as the United States [27, 37], France [37], Iran [46], New Zealand [42], Brazil [45], Norway [43], Jordan [41] and Malaysia [44].
Procedures
The study consisted of two separate phases. Firstly, a translation to the Greek language of the CPFQ tool, was conducted followed by a thorough statistical analysis of the results which included a repeatability test, explanatory and confirmatory factor analyses, measures of internal consistency test and factor correlations.
Translation and validation of the CPFQ-Gr
The English version of the CPFQ was firstly translated to Greek and then a back translation of the Greek version to English was performed, by two independent nutrition scientists fluent in both languages. Finally, a native English speaker made the last translation from English to Greek. The two Greek translations were evaluated by an academic expert in the field of nutrition, for possible discrepancies. After minor appropriate corrections, the final Greek version of the CPFQ-Gr was created.
Participants
The sample of the study consisted of 399 parents (71.2% mothers) with children between the ages of 2 and 12 years old. All participants were recruited from the general population, by word of mouth, from the urban regions of Attica and Korinthos, in Greece, on a voluntary basis. The period of the questionnaires’ collection was from January to December 2017. A total of 450 CPFQ-Gr questionnaires were distributed, either in a printed (33%) or in a web-based form (67%). Seventy six out of the 399 parents who consented in participating to the study, filled out the questionnaire, on two separate occasions, 15 days apart, in order for the research team to examine test-retest reliability of the instrument. Participants were also asked to report their age, weight, height, education and occupation, together with their child’s weight and height. Obesity categories for children were calculated according to the Extended International (IOTF) Body Mass Index Cut-Offs for Thinness, Overweight and Obesity in Children [47]. The response rate of the participants was 88,65%. The inclusion criteria were being parents with children 2–12 years old. Parents with more than one child in the above age range were asked to complete separate questionnaires for each child.
All participants were provided written information on the purpose and the procedures of the study. A cover letter which informed the participants for the purpose of the study, was also given. The collected data was confidential and all procedures were approved by the Institutional Ethics Review Board of the Harokopio University (no 53/ 11-11-2016).
Statistical analysis
Data are presented as mean values (SD) for the quantitative variables and frequencies (%) for the qualitative variables. The repeatability of the CPFQ-Gr questionnaire was examined by comparing baseline and after two-week values in 76 individuals, using paired Student’s t-tests. Paired Student’s t-tests was applied in each of the 49 questions and in each of the 12 subscales of the questionnaire.
Exploratory and confirmatory factor analyses (EFA and CFA) were performed to reveal the dimensions of the CPFQ questionnaire, based on the 399 parents. At EFA principal components method was used, based on eigenvalue greater than 0. Also, the varimax method was used in order to extract a factor and for the retainment of the factor, factor loadings had to be greater than 0.3. In order to evaluate the inter-correlation of the CPFQ items the Kaiser-Myer-Oklin criterion and the Bartletts’s test of sphericity were calculated. Finally, CFA was used in order to verify the results from the EFA.
To evaluate the internal consistency of the exported factors, the Cronbach’s alpha was calculated. Spearman’s correlations were performed in order to test overlapping between factors. A value >0.85 indicates a strong overlap.
The STATA software, version 14 (MP & Associates, Sparta, Greece) was used for all statistical analyses.
Results
Table 1 shows parents’ and children anthropometric characteristics. Parents’ average age was 40.03 years (SD = 6.64). A forth of the parents had a University degree (23.4%) and they were mainly working as Public or Private Office employees and scientists (24.8% and 19.7%, respectively). Almost 4 out of 10 parents were either overweight or obese (29.3% and 10.6% respectively). In the case of children, 54% of the sample consisted of girls and the mean age was about 6.88 years (SD = 3.10). Children’s overweight and obesity rates were 27.4% and 16.2%, respectively, whereas the 17.5% of the children were thin.
Parents’ and children Anthropometric and Sociodemographic Characteristics (n = 399)
Parents’ and children Anthropometric and Sociodemographic Characteristics (n = 399)
Table 2 shows the repeatability of the 12 subscales of the CPFQ (baseline vs. 2 weeks). Mean and standard deviation of each subscale is described. No significant differences were found in the CPFQ subscales between the two assessments, however in the subscale “Restriction for Health” the above difference was marginally insignificant (p = 0.051).
Test-re-test analysis of the 12 factors of the CPFQ (n = 76)
The KMO value was 0.781 and the sphericity test was significant, indicating that the CPFQ items had good inter-correlation and that they are acceptable.
Table 3 shows the results of the EFA. The extracted factors were 6 and they are described with each items’ loading (greater than 0.3). Also the EFA demonstrated that 7 items, item No 15, 16, 17, 18, 37, 42 and 43 from the original questionnaire should be excluded. The extracted factors were:
Healthy Eating Guidance, describing how much a parent creates a healthy environment for the child, including modeling, teaching, encouraging and child’s involvement to the whole process. This factor includes the original subscales “Encourage Balance and Variety”, “Modeling”, “Environment” (minus 2 questions), “Involvement” (minus 1 question) and “Teaching about Nutrition” (minus 1 question).
Emotion Regulation/ Food as Reward, describing how much a parent uses food to control child’s behavior. This factor includes the original subscales “Emotion Regulation” and “Food as Reward”.
Monitoring, describing how much a parent keeps truck of the child’s unhealthy food consumption. This factor includes the original “Monitoring” subscale.
Child Control, describing the degree to which parent allows the child to decide about what and when to eat. This factor includes the original “Child Control” subscale.
Pressure, describing the degree to which the parent pressures the child to eat. This factor includes the original “Pressure” subscale (minus 1 question).
Restriction, describing how much the parent restricts the child’s food intake. This factor includes the original subscales “Restriction for Health” (minus 1 question) and “Restriction for Weight” (minus 1 question).
Factors, items and loadings from exploratory analyses in the full sample (n = 399)
Table 4 includes the CFA results, which was contacted for each of the above 6 factors and their significance was confirmed (p < 0.05). CFA revealed that the model with 6 factors had chi-square equal to 658.05 (p < 0.0001).
Confirmatory Factor Analysis (CFA) results for the new 6 factors
Cronbach’s a and Spearman’s correlations are presented in Table 5. Cronbach’s a values were higher than 0.64 (0.64–0.89). The correlations between the factors were low and there was no overlap (rho = – 0.212–0.405).
Cronbach’s a and Spearman’s Correlations for the 6 factors
Correlations between the 6 factors of the CPFQ and the sociodemographic characteristics of the sample are presented in Table 6. Mothers use the “child control” feeding practice more than fathers (p = 0.002) and parents with girls use the “monitoring” feeding practice more than parents with boys (p = 0.010). As for parental obesity status, parents of normal weight use less the “restriction” feeding practice in contrast to overweight parents (p = 0.047). Moreover, parents with a MSc/PhD degree use more the “monitoring” feeding practice than others of compulsory education (p = 0.038), parents of higher education use more the “pressure” feeding practice than others of compulsory education (p = 0.05) and parents with an MSc/PhD degree use less the “restriction” feeding practice in contrast to those with compulsory or higher education (p = 0.001). Finally, semi-skilled parents use the “Emotion Regulation/ Food as Reward” feeding practice more than skilled parents (p = 0.010).
Correlations between the 6 factors of the CPFQ and sociodemographic characteristics in the full sample (n = 399)
p < 0.05, Anova, t-test, *differences between groups.
The aim of this study was to test the validity and reliability of a Greek version of the CFPQ within a sample of Greek parents of 2-to-12-year-old children. The translation, adaptation and factor analysis produced a final questionnaire of 42 items distributed over six factors. Cronbach’s alpha values were satisfactory and correlations between factors were low. Our findings do not confirm the structure of the original CPFQ in the Greek version of the questionnaire. This was also the case of previous studies aimed the validation of the tool, in other languages and countries, such as the United States [27, 37], France [27], Iran [46], New Zealand [42], Brazil [45], Norway [43], Jordan [41] and Malaysia [44].
Following our analyses, five original subscales (“Encourage Balance and Variety”, “Modeling”, “Environment”, “Involvement” and “Teaching about Nutrition”) were put together and concluded on a single factor named “Healthy Eating Guidance”. Other factors were the combination of the original subscales “Emotion Regulation” and “Food as Reward” and the combination of the original subscales “Restriction for Health” and “Restriction for Weight”. The other three factors remained as in the original questionnaire (“Monitoring”, “Child Control” and “Pressure”) [37]. Moreover, items No 15, 16, 17, 18, 37, 42 and 43 from the original questionnaire were excluded because they did not load onto any of the factors.
As it happened in the original tool, the correlations between the factors of the CPFQ were low [37]. The highest positive correlation was observed between “Healthy Eating Guidance” and “Monitoring” (rho = 0.405, p < 0.05). These two practices measure healthier eating and, as a result, parents may be advised to use them in parallel. This positive correlation was observed also in the validation of the CPFQ in parents of preschool children in Brazil [45] and in a large sample of New Zealanders [42]. The highest negative correlation was observed between “Emotion Regulation/Food as Reward” and “Monitoring”. This was also expected due to the fact that “Monitoring” is a more positive feeding practice in contrast to the other.
The strengths of this study are the relatively satisfactory sample and the test-retest reliability in a subset of the sample.
The main limitations of the study are the following: the participants were recruited from urban areas (not rural areas) and the mothers were predominantly middle aged and of high education level.
Conclusion
Results support the reliability and validity of the CPFQ for a Greek population, allowing for comparisons across cultures. The CPFQ-Gr may be used to identify and address problematic parental feeding practices at an early stage, and hence, facilitate in the prevention of poor diet quality, unhealthy eating habits and excess body weight in children.
Source of funding
The research work was supported by the Hellenic Foundation for Research and Innovation (HFRI) and the General Secretariat for Research and Technology (GSRT), under the HFRI PhD Fellowship grant (GA. no. 949).
Conflict of interest statement
None to declare.
Contributor statement
MM, VC & MY led the study design. MM and EM conducted the field data collection. DB & MM conducted the statistical analysis of the derived data. MM & VC wrote the first draft of the manuscript. All authors have read, approved contributed to the writing of the final draft of manuscript.
Footnotes
Acknowledgments
Special thanks to all the participants for their valuable contribution to the study.
