Study design and study population
We performed secondary data analysis using data from the BeeBOFT study, which is a population-based cluster randomized controlled trial for the primary prevention of overweight among young children (0–3 years) in the Netherlands [37]. In total, 51 YHC teams covering urban and rural areas in the Netherlands participated. Each YHC organization serves a region of the Netherlands, and each YHC team within an organization serves one or more municipalities of the region [4]. A team comprises a physician, nurse, and assistant [4]. The 51 YHC teams were randomly allocated to three study arms, the “BBOFT+” intervention (17 teams), the “E-health4Uth” intervention (17 teams), or the control group (17 teams). At each routine YHC visit (scheduled at child ages of 0.5, 1, 2, 3, 4, 6, 9, 11, 14, 18, and 36 months), parents allocated to the “BBOFT+” group received an intervention on child-rearing skills concerning healthy behavioral lifestyle habits of the child from birth onward. Parents allocated to the “E-health4Uth” group received intervention twice: at child ages of circa 18 and 24 months. Parents in the control group received usual care. After reviewing the research proposal of the BeeBOFT study, the Erasmus University Medical Center Medical Ethics Committee concluded that the Dutch Medical Research Involving Human Subjects Act did not apply to it. The Medical Ethics Committee therefore had no objection to the execution of the BeeBOFT study (proposal number MEC-2008-250).
From January 2009 through September 2010, parents were invited to participate in the BeeBOFT study when one of the 51 participating YHC teams visited them at home 2–4 weeks after the birth of the child. In total, 3003 parents provided written informed consent and filled in the baseline questionnaire. At child age 6 months, all the parents were invited to complete a questionnaire regarding their child’s health-related behaviors, including timing of introduction of complementary feeding, the frequency of consumption of complementary feeding, and the determinants of these behaviors. A total of 2331 parents returned the questionnaire at child age 6 months (age range 6–8 months). The questionnaire asked about the timing of the introduction of 22 types of food. Children for whom values were missing for more than five food types were excluded (n = 48). We also excluded preterm babies (gestational age < 37 weeks, n = 126). Finally, 2157 parent-child dyads were included in the present study.
Compared with the 672 infants excluded due to non-response for the questionnaire, the infants whose parents have responded the questionnaire (n = 2331) at child age 6 months had higher educated (20.0% low educated VS 11.4% low educated, p < 0.01).
Measurements
Infant complementary feeding
Timing of complementary feeding
At child age 6 months, parents were asked to report in the questionnaire at which age the child had received the following products (Additional file 1: Table S1): fruit juice; fruit juice concentrate; soft drinks (e.g., cola, iced tea); light soft drinks; fruit cordials or syrup; sweetened dairy drinks; milk or buttermilk; yogurt; porridge; bread; baby cookies; chocolate or candy; crackers or breadsticks; fruit from a jar; fresh fruit; vegetables from a jar; vegetables with fish or meat from a jar; pasta/rice/potato; fresh vegetable; fish/meat/meat substitutes. The response categories included: “< 1 month”, “between 1–2 months”, “between 2–3 months”, “between 3–4 months”, “between 4–5 months”, “older than 5 months”, and “never given”. Parents could choose “never given” if at the time they filled in the questionnaire they had not introduced that food item. For descriptive analysis, the response categories “< 1 month”, “between 1–2 months”, “between 2–3 months”, and “between 3–4 months” were combined into “before 4 months”. The average age of the infants when parents filled in the questionnaire was 6.3 months, SD = 0.6. The drinks fruit juice, fruit juice concentrate, soft drinks, fruit cordial or syrup, and sweetened dairy drinks were combined into one category called sweet beverages. The foods baby cookies and chocolate or candy were combined into one category called snack foods. The timing of introduction of complementary feeding was defined as the earliest time point that any of the abovementioned drinks and foods were first given to the child. Early introduction of complementary feeding was defined as introduction of complementary feeding (i.e., drinks and foods) before 4 months.
Frequent consumption of non-recommended foods
The questionnaire also assessed how frequently on average the child was given the abovementioned food products when parents filled in the questionnaire at 6 months (Additional file 1: Table S1). The response categories included: “never given”, “<once per week”, “1–3 times per week”, “4–6 times per week”, “1–2 times per day”, “3–4 times per day”, and “>5 times per day”. The non-recommended foods included sweet beverages and snack foods as defined above. Frequent consumption of non-recommended foods was defined as the consumption of sweet beverages and/or snack foods ≥1 time per day.
Independent variables
Based on previous research [26,27,28,29,30,31,32, 38], the following variables were selected as potential determinants for the early introduction of complementary feeding and consumption of non-recommended foods.
Demographic characteristics
The demographic characteristics obtained by the baseline questionnaire were maternal age (years), maternal educational level, maternal ethnic background (native/non-native), maternal employment status (employed/unemployed), family structure (single parent/two parents), child gender (girl/boy), parity (primipara/multipara), and gestational age (weeks). Maternal educational level was categorized as high (higher vocational training, university degree), middle (> 4 years general secondary school or intermediate vocational training), and low (no education, primary school, or 4 years or less general secondary school) [39]. The mother’s ethnic background was classified as non-native if one of her parents had been born outside the Netherlands [40].
Biological factors
Maternal pre-pregnancy weight and height were self-reported in the baseline questionnaire. Maternal pre-pregnancy BMI was calculated by weight (kg)/height2 (meters). Data on child weight at birth and at age 3 months were acquired from the YHC registration files. Child weight and height were measured by YHC professionals in accordance with standardized protocols at each routine visit (set at ages 0, 1, 2, 3, 4, 6 months) [41]. Child weight for age Z-score 7was calculated using the Dutch 1997 age- and gender- specific reference values [42]. Infant postnatal weight gain between age 0–3 months was calculated by subtracting the weight for age Z-score at birth from the weight for age Z-score at 3 months.
Behavioral factors
At child age 6 months, parents were asked to report whether they had started breastfeeding (yes, no), and, if so, how old the child was when the mother stopped breastfeeding (response categories included within 2 weeks, between 2 and 4 weeks, between 1 and 2 months, between 2 and 3 months, between 3 and 4 months, between 4 and 5 months, older than 5 months, and still breastfeeding) (Additional file 1: Table S1). The responses to these two questions led us to create a new variable indicating the duration of any breastfeeding: “no breastfeeding”, “breastfeeding for 0.5–4 months”, or “breastfeeding for 4 months or longer”.
Psychosocial factors
The psychosocial factors maternal depressive symptoms, parental beliefs, and infant temperament were assessed by parental questionnaire at child age 6 months. Maternal depressive symptoms were assessed using the 10-question Edinburgh Postnatal Depression Scale [43]. Mothers scoring 10 or higher were classified as having depressive symptoms. This variable was defined as missing if the questionnaire had been filled in by the father or another care giver (n = 107).
Parental beliefs/perceptions about infant characteristics, feeding, and infant weight were assessed. The items are based on a previous study investigating parental views on child overweight-related behaviors [44]. Example items included the following statements “My child always wants to eat when he/she sees someone eating”, “Fruit and vegetables can be given to the baby freely earlier than 4 months” and “I don’t like my child to be fat”. Parents could respond on a 5-point scale ranging from “strongly agree” to “strongly disagree”. The responses were dichotomized into “1” indicating agree/strongly agree, and “0” indicating neutral, disagree, or strongly disagree.
Infant temperament, e.g., soothability, distress to limitations, and distress to novel food, was measured using subscales from the Infant Behavior Questionnaire [45]. The subscales were chosen based on previous research on infant temperament and infant feeding [34]. An example item used to measure soothability was “When part of the child’s body was patted or stroked, how often did she/he calm down immediately?”, for distress to limitations, “When having to wait for food or liquids during the last week, how often did the child cry loudly”, and for distress to novel food, “When given a new food or liquid, how often did the child accept it immediately?”. Parents rated these specific child behaviors on a 7-point scale ranging from 1 (“Never”) to 7 (“always”).
Social factors
Day-care attendance of the infants was reported by parents in the questionnaire at child age 6 months. In addition, we included a variable entitled “intervention group” for the current study. Parents allocated to the “BBOFT+” study arm were defined as the “BBOFT+ intervention” group, while parents allocated to the control group or the “E-health” intervention group were combined to form a “no intervention” group.
Statistical analysis
All statistical analyses were performed using SAS version 9.4. Descriptive statistics for the study population were presented in relation to the timing of the introduction of complementary feeding (< 4 months vs ≥4 months). Differences between the two groups were compared by independent sample t test for continuous variables, and by the x2 test for categorical variables.
Intra-class coefficients (ICC) for our outcome variables (early introduction of complementary feeding and consumption of non-recommended foods) were calculated to decide whether the outcome variables differed for the participating YHC teams. The ICCs for both outcome variables were 0.02, suggesting a very low intra-class correlation and therefore multilevel modeling was not used. In addition, we found no significant influence of the intervention group on both outcome variables (both p > 0.25). We therefore applied normal logistic regression analyses to the data on all available participants to assess the factors associated with the early introduction of complementary feeding and with the frequent consumption of non-recommended foods. First, univariate logistic regression models were fitted for each of the independent variables with the outcome variables. Second, independent variables that were significantly (p < 0.05) associated with the outcome variables in the univariate models were included in the multivariate model, to assess the independent association between the factors and outcome variables. The univariate and multivariate models were both adjusted for the exact age of the child.