Participants
The survey was conducted in 14 public primary schools in a prefecture in the Kanto area, the central part of the main island of Japan, in May 2018. Seven cities and towns were chosen from five administrative districts in the prefecture, based on survey feasibility. Two public primary schools with similar characteristics (e.g., number of enrolled children, location (urban/rural)) were then selected from each city/town by the municipal boards of education. These 14 primary schools enrolled 2650 children as 5th and 6th graders in April 2018, all of whom were recruited into the study. At the same time, we also recruited the children’s guardians, most of whom were the main preparers of their meals. No exclusion criteria were set, because all children attending public schools in Japan are required to be educated equally.
Questionnaires and variables
Children and their guardians were surveyed using two questionnaires each. The first was a lifestyle questionnaire that asked about basic characteristics and nutrition knowledge, as well as attitudes and behaviors toward diet, and the second was a brief-type self-administered diet history questionnaire (BDHQ; BDHQ15y for children and BDHQ for adults) that assessed dietary intake.
The main exposure was whether or not the mother was working, and the number of working hours per day among those who were. The mothers were asked to answer the following question in the lifestyle questionnaire: “How many hours do you work a day? If you are not working, select ‘not currently working’”. Working was divided into three categories: (1) not working, (2) less than 8 h a day (<8), and (3) 8 or more hours a day (≥8). The cut-off point was based on the fact that the legal number of working hours in Japan is 8 h [11].
The outcome was children’s food intake, as estimated using the BDHQ15y, an instrument for primary, junior high, and high school students. It was developed from the BDHQ, which was designed to quantify food and nutrient intakes in Japanese adults over the preceding month through 80 questions that calculate the intake of 58 foods and over 100 nutrients. A dedicated calculation program is used to calculate food and nutrient intake. Food intake obtained from the BDHQ has been validated using food intake from semi-weighed dietary records [12]. In addition, nutrient intake obtained from the BDHQ has been validated [13]. The children responded to the BDHQ15y at home with their guardians. In the present analysis, food intake was energy-adjusted by the density method [14] and expressed as food intake per 1000 kcal of energy intake (g) [15].
The information below was collected using the lifestyle questionnaire. The children answered questions about sex, grade, nutrition knowledge, attitude toward diet, frequency of eating out for dinner, and the frequency of communication about diet with their guardians. The mothers answered questions about age, employment status and working hours, nutrition knowledge, attitude toward diet, sleeping hours, socioeconomic status, cohabitants (family structure), and the frequency of communication about diet with their child.
The nutrition knowledge questionnaire used to measure the nutrition knowledge of the children and their mothers has been validated [16]. This questionnaire includes the following sections: (1) knowledge about foods as nutrient sources, (2) physiological functions of nutrients in the body, (3) awareness of dietary recommendations (only for adults), and (4) relationship between nutrients and health outcomes. The percentage (%) of correct answers was calculated. Attitude toward diet was defined as follows. The children were asked “Do you pay attention to whether your diet is healthy or not?”, and given one answer choice (always, often, sometimes, or rarely). The children’s attitude toward diet was regarded as “adequate” for those who chose “always” or “often”, and “inadequate” for those who chose “sometimes” or “rarely.” The mothers were asked to respond to six statements: (1) I am careful to eat a balanced diet, (2) I am careful not to overeat, (3) I am careful to eat more fruits and vegetables, (4) I am careful to eat low-fat foods, (5) I am careful to avoid salty foods, and (6) I am careful to avoid foods with a high sugar content. They selected one response (totally agree, agree, not so much, or disagree) to each statement, which were scored as 3 for “totally agree,” 2 for “agree,” 1 for “not so much,” and 0 for “disagree.” These scores were then summed, giving a possible total range for attitude of 0 to 18. Two groups of equal size were then established by dividing mothers into 0–10 and 11–18 score groups, which were considered to represent “inadequate” and “adequate” attitudes, respectively. The mothers were also asked about their subjective socioeconomic status (SES), with answers selected from the five choices of “very straitened,” “straitened,” “average,” “affluent,” or “very affluent.” These were then further categorized as “straitened” for those who chose “very straitened” or “straitened,” “average,” and “affluent” for those who chose “affluent” or “very affluent.” The children were asked about the frequency of eating out for dinner, with responses from the four choices of “twice a week or more,” “between once a week and once a month,” “less than once a month,” or “almost never.” The children and mothers were both asked about the frequency of communication about diet, as follows: “Do you discuss meals, food, nutrition, etc. with your guardian?” (or “your child” in the statements for mothers), with responses from the four choices of “often,” “sometimes,” “not often,” or “rarely.” These were then further categorized as “high” for “often” and “sometimes,” and “low” for “not often” and “rarely.” The state of communication was then classified into four groups based on the combination of “high” and “low” answers by the children and their mothers (child (C):low-mother (M):low; C:low-M:high; C:high-M:low; C:high-M:high) and treated as a categorical variable.
Measurements
Body height and weight were measured to the nearest 0.1cm and 0.1 kg, respectively, with the child wearing light clothing and no shoes. Measurement was done as part of a routine health check-up by school nurses at each school from April to June in 2018. Body mass index (BMI) was calculated as body weight divided by the square of body height (kg/m2).
Statistical analysis
Among 2650 enrolled pairs of children and guardians, pairs that did not provide consent to participate or did not receive sufficient information through the questionnaire were excluded (Fig. 1). In addition, pairs in which the responding guardian was not the mother and those in which the energy intake of at least one of the pair, estimated using the BDHQ (BDHQ15y for child), was not between 600 and 4500 kcal were also excluded. Finally, 1693 mother and child pairs were included in the analysis (participation rate: 1693/2650*100 = 63.9%).
Food intakes of the children and mothers were compared by mothers’ employment status (not working, working < 8 h, and working ≥ 8 h) using a univariate linear regression model which included the children’s or mothers’ food intake as a dependent variable and mothers’ employment status as an independent variable. Trends of association were examined by assigning scores to the level of the independent variable (not working = 0, working <8h = 1, working ≥8h = 2). Then, the association between the selected food intake among children and mothers’ employment status was examined using multivariate linear regression models which included the children’s food intake as a dependent variable and mothers’ employment status as an independent variable. For these analyses, intakes of five foods were selected as outcomes, namely white rice, soy products, and vegetables, based on the results of univariate linear regression analysis in the children; and confectioneries and sweetened beverages as indicators of unhealthy dietary habits [17, 18]. Intakes of potatoes and bread were not analyzed irrespective of the results of univariate analysis because the amount of intake was relatively small and the influence of these two items on health is unclear. Further, based on the results for the relationship between food intakes and maternal employment status, the association between children’s BMI and maternal employment status was assessed.
The lifestyle factors listed below were considered possible confounders of the relationship between children’s food intake and mothers’ employment status. The multivariate models were adjusted using covariates for the child (sex, grade, nutrition knowledge, and attitude toward diet), mother (nutrition knowledge, attitude toward diet, food intake corresponding to the child [i.e., if children’s vegetable intake was a dependent variable in the model, mothers’ vegetable intake was used as a covariate], and sleeping hours on weekdays as an index of spare time), and family environment (cohabitants, socioeconomic status, frequency of eating out [as a counter index of home cooking], and frequency of communication between the child and mother). For white rice and confectioneries, trends of association were examined in the same manner as in univariate analysis. All analyses were performed with Stata/SE 15.1 for Windows (StataCorp LLC, Texas, USA). Statistical tests were two-sided, and p values of <0.05 were considered statistically significant.