Data for this study were retrieved from the National Nutrition Survey on Preschool Children (NNSPC), which was conducted in September 2015 by the Ministry of Health, Labor, and Welfare of Japan (MHLW) [10].
Study population and procedure
Figure 1 shows the study population and procedure. Children aged ≤6 years as of May 31, 2015 were randomly selected from households among 1,106 districts for the Comprehensive Survey of Living Conditions, conducted by the MHLW. Three districts affected by heavy rain in September 2015 were excluded from the survey sampling. First, the MHLW explained the survey method to the prefectures. Subsequently, the prefectural public health center employed investigators to visit the households selected for this survey. The investigators asked the children’s mothers (or the caregiver involved in providing food to the child) to complete a questionnaire, which was collected at a later date. In total, 2,992 households with 3,936 children aged ≤6 years participated in the survey. The response rate of the survey was 56.8%. The questionnaires associated with 65 children were excluded because information on age was not available. Finally, 3,871 questionnaires were collected for analysis [10]. A database was prepared by the Maternal and Child Health Division, Department of Equal Employment and Children’s Family, MHLW.
The National Nutrition Survey on Preschool Children has two types of questionnaires, one of which is restricted to infants aged <2 years and the other of which encompasses children aged 2–6 years. Data obtained from the latter questionnaire were used in this study. In total, 2143 persons responded to all items consistent with the purpose of this study.
Measurement
Children’s dietary patterns in relation to the eight food groups (grains, fish, meat, eggs, soybeans/soy products, vegetables, fruit, and milk) were evaluated as objective variables, as well as their intake of processed foods, whereby four items (sweetened beverages, confectioneries, instant noodles, and fast food) were investigated. The survey inquired how often the children consumed foods in each group (≥2 times per day, once a day, 4–6 days per week, 1–3 days per week, or less than once a week or rarely) [10,16]. The FAO’s food diversity score (FDS) was applied to assess the nutritional quality of the whole diet [5].
Thirteen items assessed parents’ care behavior in relation to children’s diets. We posed the question “Are you (parent) careful about your child’s diet?” with regard to the following 13 items: 1) food (seven items encompassing nutritional balance, flavoring and seasoning, size or softness, assorted arrangements and colors, amount) and snack (contents, amount); 2) mealtime practice (three items comprising regular mealtimes, chewing well, table manners); and 3) “parent–child communication (three items encompassing enjoyment of eating, eating together, cooking together). Each item was scored based on “yes” or “no” responses.
Explanatory variables related to parents included their relationships with their children, age of mother, current employment status of mother, household structure (i.e., presence of other children, grandparents, and others), subjective economic status, leisure time (i.e., affluent, somewhat, neither, not so much, cannot afford at all), the place where the child spends time during the day (i.e., nursery school, kindergarten, centers for early childhood education and care, grandparents and relatives, or none of the above) and lifestyle regarding eating breakfast with parents. Moreover, data concerning children’s age; height; weight; nutritional status (degree of obesity); food allergies; tooth decay; and time spent on TV, video, and games were obtained.
Nutritional status of children
The nutritional status of children was determined based on body weight and height. The degree of obesity (%) was calculated using the following formula: self-reported body weight (g) − standard body weight (g) for height/standard body weight (g) for height × 100. The judgment criteria for the degree of obesity were “obese” (≥30%), “overweight” (20%–30%), “tendency to be overweight” (15%–20%), “standard” (−15% to +15%), “tendency to be underweight” (less than 15% to less than 20%), and “underweight” (less than 20%). The standard body weight was calculated using the formula of standard body weight for height in Japanese children [17,18]. The formula does not consider age because the standard body weight for height curves were almost identical for children aged 1–6 years [19].
Statistical analysis
The FDS of children comprised the total number of eight food groups (grains, fish, meat, eggs, soybeans/soy products, vegetables, fruits, and milk) being consumed at least once a day [4,10,19]. The FDS was one point if consumption occurred once or more per day or zero points if less than that. There were eight types of foods; thus, the maximum score was eight points. Once the FDS distribution was identified, the FDS was divided into two groups according to medians: 1) three or fewer points and 2) four or more points [4]. The processed food score was calculated according to the total number of four food items (sweetened beverages, confectioneries, instant noodles, and fast food) being consumed at least once a day [4]. As in the case of FDS, the processed food score was calculated as a score of one point if the food type was consumed at least once a day or zero points if less than that. There were four types of foods; thus, the maximum score was four points.
The sex of the parent who answered the questionnaire, age, and socioeconomic status as well as children’s sex; nutritional status; food allergies; tooth decay; and time spent on TV, video, and games were compared between the two FDS groups. Initially, parents’ socioeconomic status was compared between children’s FDS groups, and then, children’s health and lifestyle were analyzed by the two FDS groups. Furthermore, the total value calculated from the 13 items of the parent’s care behavior in children’s diets was compared between the two FDS groups.
Multiple regression analysis analyzed relationships between FDS scores and several variables, including the total value of the parent’s care behaviors in children’s diets; subjective socioeconomic status; food allergies; tooth decay; and time spent on TV, video, and games, as well as the child’s age and mother’s age. The continuous variables included the total value of the parent’s care behaviors in children’s diets, the child’s age, and the mother’s age. The nominal variables such as subjective socioeconomic status; food allergies; tooth decay; and time spent on TV, video, and games were converted to an ordinal scale.
Next, we used logistic regression to specifically analyze the relationship between the variables and the higher FDS group. Multivariate analysis was performed for each of the 13 items measuring parental care in relation to children’s diets using a logistic regression model that adjusted for the parent’s (i.e., mother or father) relationship with the child, child’s sex (i.e., male or female), employment status of the mother (i.e., yes or no), and household structure (i.e., presence of other children, grandparents, and others) (Model 1).
Additional multivariate analysis was performed for each of the 13 items measuring parental care in relation to children’s diets using a logistic regression model that adjusted for the parent’s (mother or father) relationship with the child, child’s sex, employment status of the mother (i.e., yes or no), household structure (i.e., presence of other children, grandparents, and others), subjective economic status (i.e., affluent, somewhat, neither, not so much, or not able to afford at all), leisure time ( i.e., affluent, somewhat, neither, not so much, or not able to afford at all), and place where the child spends time during the day (i.e., nursery school, kindergarten, center for early childhood education and care, with grandparents, with relatives, staying at home) (Model 2).
All statistical analyses were performed using SAS software, version 9.2 (SAS Institute, Inc., Cary, NC, USA). A probability (p) value of <0.05 was considered statistically significant.