*Article* **Adverse Effects of Infant Formula Made with Corn-Syrup Solids on the Development of Eating Behaviors in Hispanic Children**

**Hailey E. Hampson 1,2, Roshonda B. Jones 1, Paige K. Berger 1, Jasmine F. Plows 1, Kelsey A. Schmidt 1, Tanya L. Alderete <sup>3</sup> and Michael I. Goran 1,\***


**Abstract:** Few studies have investigated the influence of infant formulas made with added corn-syrup solids on the development of child eating behaviors. We examined associations of breastmilk (BM), traditional formula (TF), and formula containing corn-syrup solids (CSSF) with changes in eating behaviors over a period of 2 years. Feeding type was assessed at 6 months in 115 mother–infant pairs. Eating behaviors were assessed at 12, 18 and 24 months. Repeated Measures ANCOVA was used to determine changes in eating behaviors over time as a function of feeding type. Food fussiness and enjoyment of food differed between the feeding groups (*p* < 0.05) and changed over time for CSSF and TF (*p* < 0.01). Food fussiness increased from 12 to 18 and 12 to 24 months for CSSF and from 12 to 24 months for TF (*p* < 0.01), while it remained stable for BM. Enjoyment of food decreased from 12 to 24 months for CSSF (*p* < 0.01), while it remained stable for TF and BM. There was an interaction between feeding type and time for food fussiness and enjoyment of food (*p* < 0.01). Our findings suggest that Hispanic infants consuming CSSF may develop greater food fussiness and reduced enjoyment of food in the first 2 years of life compared to BM-fed infants.

**Keywords:** breastmilk; child eating behavior questionnaire; corn-syrup solids; eating behavior; enjoyment of eating; food fussiness; formula; Hispanic; infant; obesity

#### **1. Introduction**

Added sugar is widespread in our food system and is characteristic of the Western Style diet. This is a concern as added sugar is associated with obesity development and a wide range of co-morbid conditions [1,2]. Evidence suggests that chronic consumption of added sugar may affect obesity risk as it is consumed in excess and contributes empty calories to the diet [3,4]. However, it is also postulated that early exposure to sugars may indirectly influence obesity risk by shaping taste preferences, self-regulation of energy intake, and the reinforcing value of food [5–9]. These eating behaviors may ultimately lead to an overindulgence of sweet foods and rejection of bitter-tasting alternatives. This is relevant as food manufacturers incorporate sugars in the form of added corn-syrup solids into infant formulas and introductory solids, despite the most recent dietary guidelines recommending zero added sugars in the first 2 years of life [4].

Human studies have shown that the introduction of added sugar in infancy influences eating behavior in childhood. For example, infants fed added sugar before 4 months were more likely to reject bitter and sour-tasting foods in early childhood. These infants were found to be at an increased risk for childhood obesity [5]. Furthermore, infants that

**Citation:** Hampson, H.E.; Jones, R.B.; Berger, P.K.; Plows, J.F.; Schmidt, K.A.; Alderete, T.L.; Goran, M.I. Adverse Effects of Infant Formula Made with Corn-Syrup Solids on the Development of Eating Behaviors in Hispanic Children. *Nutrients* **2022**, *14*, 1115. https://doi.org/10.3390/ nu14051115

Academic Editors: Odysseas Androutsos and Evangelia Charmandari

Received: 30 January 2022 Accepted: 4 March 2022 Published: 7 March 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

are fed formulas with sweet flavors have a lower acceptance of bitter tasting foods, such as broccoli, compared to infants fed formulas with bitter flavors [10]. While the exact biological mechanism has yet to be fully elucidated, animal studies suggest that the gut microbiome may be a key link between early sugar exposure and infant food preferences, leading to later obesity risk [11].

Although several studies have examined the effects of breastfeeding and formula feeding on infant eating behaviors, no study has examined whether infant formula made with corn-syrup solids instead of lactose as the major carbohydrate source may have additional, potentially adverse effects. Various infant formulae have been developed without lactose due to potential concerns of lactose intolerance, some of which use corn syrup solids. Therefore, the aim of this study was to determine associations of breastfeeding, traditional formula, and formula containing corn-syrup solids at 6 months with changes in eating behaviors at and between 12, 18, and 24 months.

#### **2. Materials and Methods**

#### *2.1. Study Participants*

We obtained data from the Southern California Mother's Milk Study, which is an on-going longitudinal cohort study of Hispanic mother-infant dyads, as reported previously [12]. Briefly, participants were eligible if they (a) self-identify as Hispanic (mother and father), (b) had a singleton birth, (c) intended to breastfeed for at least 3 months, (d) enrolled within 1 month of the infant's birth, and (e) were willing to/had the ability to understand the procedures of the study and be able to read English or Spanish at the fifth-grade level. Participants were not eligible if they (a) had a physician diagnosis of a major medical illness or eating disorder, (b) had a physical, mental or cognitive issue that prevented participation, (c) reported chronic use of medication that may affect body weight or composition, insulin resistance or lipid profiles, (d) were a current smoker or were a user of other recreational drugs, (e) had a pre-term/low birth weight infant or diagnosis of any fetal abnormalities, or (f) were less than 18 years old.

At each study visit, mothers reported health history and information on breastfeeding, which were included as covariates in the analysis [12]. Institutional review boards at the University of Southern California and Children's Hospital Los Angeles approved the study. Written informed consent was obtained from all mothers prior to data collection. Mother–infant pairs were included in the final analysis if they completed all 6, 12, 18 and 24 month visits (Figure 1).

**Figure 1.** Participation flow chart and distribution of participants to infant feeding type groups.

#### *2.2. Infant Feeding Modality Determination*

Two 24 h dietary recalls were conducted with mothers over the phone by trained bilingual research personnel. Dietary intake collection and analysis was performed using the Nutritional Data System for Research software (NDSR) [13]. The NDSR data were used to calculate the infants' breastmilk and formula-type groups. Breastmilk-fed infants were categorized based on predominant consumption type (receiving more than 80% of feedings from breastmilk) at 6 months. Feeding type groups included infants who received primarily human milk directly from the breast or pumped human milk from a bottle (BM, n = 43), infants who received a traditional formula, which was lactose and cow-milk based (TF, n = 41), and infants who received a lactose-reduced formula made with corn syrup solids (CSSF, n = 31), in this case Enfamil Gentlease, Similac Isomil Advance, produced by Mead Johnson & Company, LLC [14,15]. Infants in the TF group received one of the following formulas: Similac Advance 20, Similac Advance EarlyShield, Gerber Good Start Gentle, Enfamil A.R., and Enfamil Infant, made by a variety of formula companies.

#### *2.3. Eating Behavior Assessment*

To assess child eating behaviors at 12, 18 and 24 months, we used the Children's Eating Behavior Questionnaire (CEBQ), a maternal report survey tailored to children 12 months of age and older [16–18]. The questionnaire was available in English and Spanish. The CEBQ consists of 35 questions organized into 8 domains, i.e., satiety responsiveness, slowness in eating, emotional under-eating, food fussiness, enjoyment of food, food responsiveness, desire to drink, and emotional over-eating. All questions were ranked on a Likert scale of Never (0) to Always (4), and scores were averaged within domains [16–18].

#### *2.4. Participant Characteristics*

We collected anthropometrics and sociodemographic information, including maternal pre-pregnancy BMI (kg/m2) (determined from height and weight recall prior to pregnancy), infant sex, infant age (days), infant birth weight (kg), mode of delivery (vaginal vs. Caesarean Section), and socioeconomic status index (SES). We determined socioeconomic status using The Hollingshead Index. Students, stay-at-home parents, and unemployed persons do not have assigned employment categories in the Hollingshead Index, therefore, a score of zero was given to these individuals under the assumption they these participants likely to have very little or no income. This was done in order to retain these participants in the analyses.

#### *2.5. Statistical Analysis*

Basic descriptive statistics were calculated for infant feeding modality, eating behavior domains, and all covariates included in the analysis. One-Way Analysis of Covariance (AN-COVA) was performed to determine initial differences in covariates between feeding-type groups. Model diagnostics were performed and any variables not meeting assumptions of ANCOVA were transformed using the bestNormalize package in R, which selects the best method to normalize the variable [19]. One-Way ANCOVA was performed at each time point to determine differences in eating behaviors by infant feeding modality. Repeated-Measures ANCOVA was conducted including an interaction term between time (infant age) and infant feeding modality at 6 months, to determine differences in eating behaviors over time as a function of infant feeding modality. For significant between-group differences and interactions, post hoc pairwise comparisons using Tukey's Test were performed. Based on the literature, we adjusted for the following common covariates: maternal pre-pregnancy BMI (kg/m2), infant breast milk feedings and formula feedings per day (in a ratio variable of breastmilk feedings to formula feedings normalized across all children in the sample), infant sex, infant age (days), infant birth weight (kg), mode of delivery, and socioeconomic status index (SES). We performed reliability testing on CEBQ domain summary scores at all timepoints, 12, 18 and 24 months, using the Psych package in R and an inclusion alpha level of 0.7 [20]. Domains not meeting the significance level 0.7 were removed from

analysis. RStudio was used for all analyses using R version 4.0.5 [21]. We performed a sensitivity analysis to determine if results changed when we distinguished between two types of breastmilk feeding. We divided breastmilk-fed infants into two groups, those directly breastfed (BF), and those fed breastmilk pumped in a bottle (PB). We ran all statistical analyses with all four groups, TF, CSSF, BF and PB, and results did not differ from the original analysis. Therefore, we collapsed the BF and PB groups into one breastmilk group (BM).

#### **3. Results**

One-hundred and fifteen mother–infant pairs were included in this study. Fifty-six percent of infants were female, and the mean infant birth weight was 3.4 ± 0.4 kg. The mean breastmilk feedings per day at 6 months was 2.7 ± 3.4 and the mean formula feedings per day was 2.9 ± 3.0 while the number of feedings per day was not significantly different between breastfed and formula-fed infants. Initial maternal and infant characteristics were similar across groups (Table 1). Twenty-seven percent of infants were in the CSSF group, 36% in the TF group, and 37% in the BM group. Cronbach's alpha analysis for internal consistency of CEBQ domains suggested that the domains satiety responsiveness (a = 0.61–0.63) and slowness in eating (a = 0.46–0.49) did not have a high enough reliability above a = 0.70. An analysis was not performed on these domains. All other CEBQ domains had an acceptable alpha level above 0.7 (0.71–0.83).

**Table 1.** Descriptive statistics based on feeding type group at 6 months (n = 115). No significant differences were found in maternal or infant characteristics between groups based on One-Way ANOVA.


\* CSSF = Formula with Reduced-Lactose, Added-Corn Syrup Solids, BM = Breastmilk, TF = Traditional Formula.

Food fussiness was significantly different between infant feeding groups (*p* = 0.01). For the CSSF group, food fussiness significant increased from 12 to 18 months (mean increase = 0.51, *p* < 0.01), and 12 to 24 months (mean increase = 0.77, *p* < 0.001) (Figure 2). Food fussiness did not change significantly over time for the other groups. There was a significant interaction between infant feeding modality at 6 months and time (infant age) in months for food fussiness (*p* < 0.01) (Table 2). In post hoc testing, food fussiness was significantly greater in the CSSF group compared to the BM group at 24 months (mean difference = 0.54, *p* < 0.05).

Enjoyment of food was significantly different between infant feeding groups (*p* = 0.001). For the CSSF group, enjoyment of food significantly decreased from 12 to 24 months (mean decrease = 0.60, *p* < 0.001), while it remained constant in the other groups over time (Figure 3). Furthermore, there was a significant interaction between infant feeding modality at 6 months and time (infant age) in months for enjoyment of food (*p* = 0.002) (Table 2).

#### **Repeated Measures ANCOVA**

,QWHUDFWLRQ*p*7 LP H*p* )HHG LQJ7 \SH*p*

**Figure 2.** Food Fussiness increases from 12 to 18 months and 18 to 24 months for infants who consumed formula with added corn-syrup solids (CSSF) at 6 months. Food fussiness increases from 12 to 24 months for infants who consume traditional infant formula (TF) at 6 months. The model was adjusted for maternal pre-pregnancy BMI (kg/m2), infant breast milk feedings per day and formula feedings per day, infant sex, infant birth weight (kg), mode of delivery, and socioeconomic status index (SES).

#### **Repeated Measures ANCOVA**

**Figure 3.** Enjoyment of food decreases from 12 to 24 months for infants who consumed formula with added corn-syrup solids (CSSF) at 6 months. Model adjusted for maternal pre-pregnancy BMI (kg/m2), infant breast milk feedings per day and formula feedings per day, infant sex, infant birth weight (kg), mode of delivery, and socioeconomic status index (SES).

**Table 2.** Repeated measures ANCOVA shows that there was a statistically significant interaction between feeding modality and time from 12 to 24 months for food fussiness and enjoyment of food. Model includes the covariates maternal pre-pregnancy BMI (kg/m2), infant breast milk feedings per day, infant formula feedings per day, infant sex, infant birth weight (kg), mode of delivery, and socioeconomic status index (SES).


<sup>a</sup> Interaction between Time and Feeding Type. *p* < 0.01 \*\*.

#### **4. Discussion**

In this study, we found that the development of child eating behaviors in the first 2 years of life was affected by early exposure to formula made with corn-syrup solids as compared to traditional formula made with lactose and compared to breastmilk. Specifically, early exposure to formula made with corn-syrup solids was associated with increased food fussiness and reduced enjoyment of food over time. In addition, CSSF-fed infants had worsening eating behaviors over time, compared to TF-fed and BM-fed infants. Importantly, the behaviors observed in the CSSF-fed infants have been linked to poor diet quality and variety, which in turn are associated with obesity and related co-morbidities in childhood [7,22,23].

Our findings provide additional insights into the early exposure to corn-syrup solids and infant eating behaviors examined in prior studies. For example, Shepard et al. found that infants fed "gentle" formula containing corn-syrup solids demonstrated consistent eating behaviors from 3 to 5 months, in contrast to our findings that CSSF-fed infants exhibited changes in eating behaviors over a longer period of time [24]. Shepard et al. examined CSSF intake at an earlier period than the present study and included fewer formula type groups, which may explain differences in findings. Because eating behaviors emerge over time, it is plausible that the prior study did not have a long enough duration so as to capture changes in eating behavior as we did. Consistent with our findings, Murray et al. reported that infants who were exposed to sweet, non-milk solids and beverages had decreased bitter food acceptance and increased obesity risk, a trademark of picky eating behavior [5]. Similarly, we found that CSSF-fed infants had a greater increase in food fussiness over time, relative to formula fed infants and breastfed infants. It may be that early exposure to added sugar in the form of corn-syrup solids enhances the infant's affinity for sweet tastes and exacerbates innate disliking of bitter tastes, which may contribute to picky eating [5,7,10].

This hypothesis is supported by work from Mennella and colleagues, who found that infants fed sweet-tasting formula were more likely to reject broccoli than those who had exposure to bitter-tasting formula [10]. Similarly, we found that all feeding groups had a similar degree of food fussiness at 12 months of age. However, we found that there was a greater increase in food fussiness among CSSF-fed infants compared to TF-fed infants, and food fussiness remained stable among BM-fed infants. This suggests that infants exposed to diverse flavors or diminished sweet flavors in breastmilk or traditional formula are more likely to accept wide-ranging foods compared to those exposed to the sweetness of CSSF. These experiences may also influence enjoyment of food among CSSF-fed infants, as picky eating behaviors may reduce overall enjoyment of a diverse diet that comes with the introduction of solid foods [23,25].

While several mechanisms may underlie our findings, a potential explanation that has gathered the most interest is the effects of early sugar exposure on the infant gut microbiome, which may affect appetite regulation through gut-derived hormones [11,12,26–28]. Simple carbohydrates and other dietary factors can serve as energy for a harmful subset of gut microbes, which may ultimately displace beneficial bacteria [26–28]. The increased proportions of harmful microbes therefore receive more energy from the host, which has a strong influence on appetite, metabolism and food-related behaviors [28]. For example, harmful bacteria may ferment added sugar and produce metabolites, primarily short-chain fatty acids, which can directly affect hunger and signaling [27]. Through this mechanism, CSSF-fed infants may have an increased affinity for sweet-tasting foods. As infants are introduced to more solid foods with a range of flavors, they may experience a decreased preference for and enjoyment of food with bitter tastes, contributing to food fussiness [23].

In addition, animal studies have shown that rats with early sugar exposure exhibit addiction-like responses, with an increased affinity for added sugars and symptoms of withdrawal [29]. These patterns are linked to opioid-receptor binding associated with the added sugar intake and neurochemical changes that result in increased dopamine from sugar consumption [29]. Early added corn-syrup exposure from CSSF could program an infant for a heighted dopamine response that is not met with the introduction of bitter foods, thus reducing their enjoyment of food. The potential effects on brain-based eating behaviors may also be exacerbated by displacement of lactose with corn-syrup solids in the CSSF. Decreased exposure to lactose and therefore, galactose, may also have deleterious effects on infant-brain development and subsequent healthy eating behavior in children [12,14,15,30].

This study has several limitations. We used a self-report survey to assess child eating behaviors, which could result in response bias. However, we only used CEBQ domains with high validity. It is possible that the translation of the CEBQ into Spanish may have affected the validity of the questionnaire. Further, this is an observational study, and we cannot make causal conclusions based on our findings. However, our study was longitudinal in nature, which adds a further temporal component to the findings. Another important limitation is that we do not have information on why mothers chose different infant formulas. However, our sample has no significant differences in maternal or infant baseline characteristics between groups, including socioeconomic status, which may be influential for selection of infant formula type. Additionally, one of the infant formulas categorized in the TF group due to its lactose and cow's milk content is Gerber Good Start Gentle. Similar to Enfamil Gentlease in the CSSF group, this product is marketed toward colicky and fussy infants. A systematic review by Belamarich et al. reports that mothers choose these gentle formulas due to the marketing strategies that imply crying and fussiness are indicative of digestive discomfort [31]. This sheds some light on the choices that mothers may be making with regard to formula, however, given that there are gentle formulas included in both groups, we do not suspect that this contributes to, nor fully explains the differences seen in our study. However, maternal beliefs related to formula choice is an important area for further study.

These findings may have important implications for infant feeding recommendations given that early life added corn-syrup exposure may influence taste-preference formation, food rejection with the introduction of solids, and the composition of the infant gut microbiome. Food fussiness and the reduced enjoyment of food may inhibit the acceptance of fruits and vegetables and other healthful foods in the diet, leading to poor diet quality and variety. Importantly, CSSF is only consumed by <10% of infants in the general US population [15]. However, in our cohort of Hispanic mothers, 50% of formula-fed infants consume CSSF. Therefore, the results of this study are highly relevant for this subset of the population that is already at higher risk for obesity and related chronic disease development.

#### **5. Conclusions**

These findings suggests that Hispanic infants who receive formula made with corn syrup solids in place of lactose develop poor eating behavior in the first 2 years of life, including greater food fussiness and reduced enjoyment of food, compared to traditional formula-fed and breastmilk-fed infants. Further studies are needed to elucidate the underlying mechanisms by which added corn-syrup solids influence child eating behaviors as well

as clinical health outcomes. Our findings provide evidence for future studies exploring the effects of infant formulas with reduced lactose and added corn-syrup solids on children's eating behaviors and growth.

**Author Contributions:** Conceptualization: M.I.G.; Data curation, M.I.G.; Formal analysis, H.E.H. and R.B.J.; Funding acquisition, M.I.G.; Investigation, M.I.G.; Methodology, R.B.J. and M.I.G.; Project administration, M.I.G.; Resources, M.I.G.; Software, H.E.H. and R.B.J.; Supervision, M.I.G.; Writing original draft, H.E.H.; Writing—review and editing, P.K.B., J.F.P., K.A.S., T.L.A. and M.I.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Institute Diabetes and Digestive and Kidney Diseases (R01 DK110793). This work was also funded by the Gerber Foundation (15PN-013). Paige K. Berger is funded by the Eunice Kennedy Shriver National Institute of Child Health & Human Development (K99 HD098288). Tanya L. Alderete is funded by NIEHS (R00ES027853). Jasmine F. Plows is funded through a Saban Research Institute Research Career Development Fellowship.

**Institutional Review Board Statement:** Institutional review boards at the University of Southern California and Children's Hospital Los Angeles approved the study.

**Informed Consent Statement:** Written informed consent was obtained from all mothers and financial compensation was provided.

**Data Availability Statement:** Data described in the manuscript, code book, and analytic code will be made available upon request. Data requests can be made to Michael I. Goran, (323) 217–5116, goran@usc.edu.

**Acknowledgments:** We would like to thank Carla Flores, Danielle Garcia, Elizabeth Campbell, Claudia Rios, Emily Leibovitch, Rosa Rangel and the entire Goran Lab for their assistance in obtaining these data. We would also like to thank Jennifer Fogel for her editing assistance on the manuscript.

**Conflicts of Interest:** Goran is a scientific advisor for Yumi and receives book royalties from Penguin Random House. The Gerber Foundation had no role in the design, execution, analyses, interpretation of data, writing of the report, or decision to submit the report for publication.

#### **References**


## *Article* **Associations of Caregiver Cooking Skills with Child Dietary Behaviors and Weight Status: Results from the A-CHILD Study**

**Yukako Tani 1, Aya Isumi 1,2, Satomi Doi 1,2 and Takeo Fujiwara 1,\***


**Abstract:** We examined whether caregiver cooking skills were associated with frequency of home cooking, child dietary behaviors, and child body weight status in Japan. We used cross-sectional data from the 2018 Adachi Child Health Impact of Living Difficulty study, targeting primary and junior high school students aged 9–14 years in Adachi City, Tokyo, Japan (*n* = 5257). Caregiver cooking skills were assessed using a scale with good validity and reliability modified for use in Japan. Child heights and weights derived from school heath checkup data were used to calculate WHO standard body mass index z-scores. After adjusting for potential confounders, caregivers with low-level cooking skills were 4.31 (95% confidence interval (CI): 2.68–6.94) times more likely to have lower frequency of home cooking than those with high level of cooking skills. Children with low-level caregiver cooking skills were 2.81 (95% CI: 2.06–3.84) times more likely to have lower frequency of vegetable intake and 1.74 (95% CI: 1.08–2.82) times more likely to be obese. A low level of caregiver cooking skills was associated with infrequent home cooking, unhealthy child dietary behaviors, and child obesity.

**Keywords:** cooking skills; home cooking; vegetable intake; obesity; children

#### **1. Introduction**

In the last 50 years, people in developed countries have shifted toward meals away from home and cooking at home less [1–3]. In the United States, the percentage of daily energy consumed from home food sources dropped by approximately 25% from the 1960s to 2000 [1]. In Japan, household expenditure on precooked food increased by 26% from 1993 to 2015, and eating out is becoming more widespread among younger generations [4]. Alongside the decrease in home cooking, the idea that homemakers, especially women, should be educated to feed their families seems to have become outmoded [5]. However, the obesity epidemic has led to growing concerns about poor diets among children and increasing calls to reaffirm the importance of basic food preparation and cooking skills to prevent poor diets and chronic diet-related diseases [5].

Lifestyle changes in response to the coronavirus disease 2019 (COVID-19) pandemic suddenly increased the need for home cooking. In the United States before and during the initial peak of the COVID-19 pandemic, cooking meals at home increased from 4.49 to 5.18 days per week [6]. Canadian families with young children described that their greatest change since COVID-19 was spending more time cooking [7]. In China, even post-lockdown, 65% of people reported that they cooked more at home compared with the previous year [8]. Meanwhile, the COVID-19 pandemic is leading children toward unfavorable obesity-promoting behaviors, such as decreased physical activity, increased screen time, and greater consumption of snack foods [7,9]. Therefore, caregivers need to acquire the ability to create a healthy eating environment to prevent their child from having a poor diet and becoming obese. However, there is limited knowledge on the relationships between caregiver's ability to prepare meals and their child's diet and weight status.

**Citation:** Tani, Y.; Isumi, A.; Doi, S.; Fujiwara, T. Associations of Caregiver Cooking Skills with Child Dietary Behaviors and Weight Status: Results from the A-CHILD Study. *Nutrients* **2021**, *13*, 4549. https://doi.org/ 10.3390/nu13124549

Academic Editors: Odysseas Androutsos and Evangelia Charmandari

Received: 22 September 2021 Accepted: 13 December 2021 Published: 18 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Recently, evidence has been accumulating on the dietary benefits of home cooking. A systematic review confirmed dietary benefits of home cooking, including greater consumption of healthier food groups, enhanced healthy eating self-efficacy, and improved adherence to several healthy dietary recommendations [10]. Beyond dietary outcomes, a population-based study in the United Kingdom showed that more frequent consumption of home-cooked meals was associated with a greater likelihood of having normal weight and body fat status among adults [11]. Studies on Japanese children and adolescents showed that infrequent home cooking was associated with obesity, higher blood pressure, and lower high-density lipoprotein-cholesterol [12,13].

Cooking skills may be critical to encourage home cooking and improve the quality of meals [10]. Several studies reported that high level of cooking skills was associated with lower consumption of ready meals, convenience food, and ultra-processed food among adults [14–16]. Meanwhile, intervention studies demonstrated that improvement in cooking skills led to increased cooking confidence and consumption of vegetables and fruits [17,18]. A recent population-based study among older Japanese adults showed that a low level of cooking skills was associated with lower frequency of home cooking, vegetable/fruit intake, higher frequency of eating out, and underweight status [19]. However, most existing studies have focused on dietary benefits among adults, and limited research has examined the associations of caregiver cooking skills with child diet and weight status.

The aim of the present study was to examine the associations of caregiver cooking skills with frequency of home cooking, child dietary behaviors, and child body weight status.

#### **2. Materials and Methods**

#### *2.1. Study Design and Subjects*

The Adachi Child Health Impact of Living Difficulty (A-CHILD) study was established in 2015 to evaluate the determinants of health among children in Adachi City, Tokyo, Japan [20]. We used cross-sectional data from 2018 that covered caregivers and their children in three grades: fourth-grade, sixth-grade, and eighth-grade. The survey was conducted in all public elementary schools for fourth-grade children, nine selected elementary schools for sixth-grade children, and seven junior high schools for eighthgrade children [20]. In 2018, self-reported questionnaires were distributed to 6605 children (5311 fourth-grade, 618 sixth-grade, and 676 eighth-grade). Children and their caregivers completed the questionnaires at home and then returned the completed questionnaires to their school. A total of 5793 pairs (4605 fourth-grade, 556 sixth-grade, and 632 eighth-grade) of children and their caregivers returned the questionnaires (response rate: 88%). Of these, 5382 pairs (4290 fourth-grade, 514 sixth-grade, and 578 eighth-grade) provided informed consent, returned all questionnaires, and could be linked with health checkup data (consent rate: 93%). The present analysis was carried out using data for 5257 pairs, after the following exclusions for missing information: child age (*n* = 23); caregiver cooking skills and frequency of home cooking (*n* = 26); child month of birth, height, and weight (*n* = 10); and child dietary behaviors (frequency of vegetable intake and breakfast consumption) (*n* = 66). The A-CHILD protocol and use of the data for this study were approved by the Ethics Committee at Tokyo Medical and Dental University (No. M2016-284).

#### *2.2. Frequency of Home Cooking*

Frequency of home cooking was evaluated by caregivers using the following question: 'How many times did you or someone else in your family cook meals for your children at home? Circle the answer that best applies for the past month'. A cooked meal was defined as a simple meal, such as a fried egg [12]. The five response items were: 'almost every day', '4–5 days/week', '2–3 days/week', 'a few days/month', and 'rarely'. We defined <3 times a week as low frequency of home cooking because it was reported to be associated with child obesity and cardiovascular risk [12,13].

#### *2.3. Child Body Weight Status*

Child height and weight were measured in schools during health checkups by school teachers according to standardized protocols [21]. Height was measured to the nearest 0.1 cm using a portable stadiometer and weight to the nearest 0.1 kg on digital scales, without shoes and in light clothing. Body mass index (BMI) was calculated by dividing the weight (in kilograms) by the square of the height (in meters). BMI was expressed as a z-score representing the deviation in standard deviation units from the mean of a standard normal distribution of BMI specific to age and sex, according to the WHO Child Growth Standards. Child BMI was categorized as underweight (<−2SD), mild-underweight (−2SD to <−1SD), normal (−1SD to <+1SD), overweight (+1SD to <+2SD), and obese (≥+2SD) using standard deviation cut-off points [22].

#### *2.4. Child Dietary Behaviors*

Child frequency of vegetable intake was assessed by caregivers using the question 'How often did your child eat vegetable dishes? Circle the answer that best applies for the past month'. The three response items were: 'twice/day', 'once/day', and '<3 times/week'. Respondents who ate vegetables and fruit less than once a day were categorized as having low frequency of vegetable intake. This cutoff point was defined by prevalence to the 10th percentile of the included children (Table 1). Child frequency of breakfast intake for the past month was assessed by self-reporting with responses of 'every day', 'often', and 'rarely/never', with 'rarely/never' defined as breakfast skipping.




**Table 1.** *Cont.*

BMI: body mass index. <sup>a</sup> Differences were analyzed using Pearson's chi-square test.

#### *2.5. Caregiver Cooking Skills*

Caregiver cooking skills were assessed using a modified cooking skills scale designed with consideration of basic Japanese cooking methods and typical meals [19]. The scale consisted of five items: (1) able to peel fruits and vegetables; (2) able to make stir-fried meat and vegetables; (3) able to make miso soup; (4) able to make stewed dishes; and (5) like to cook. Items 1 to 4 reflected basic cooking methods and were adopted from the Japanese cooking skills score [19]. Item 5 was newly added for the present study as an indicator of cooking skills, because a previous study on life-course trajectories of cooking skills found that liking to cook was a characteristic of people who maintained a high level of cooking skills [23]. Participants were asked to evaluate their own cooking skills on a six-point scale ranging from 'do not agree at all' (=0) to 'agree very much' (=5). A high score meant that the caregiver had high confidence in their cooking skills. In psychometric testing, one factor with eigenvalue >1 was found and accounted for 92.3% of the variance. The Cronbach's α for the cooking skills scale in the study sample was 0.78. Factor loadings ranged from 0.3 (item 5) to 0.9 (item 2). We calculated the mean scores of the five items and divided the results into two categories: high (score > 4.0) and low (score ≤ 4.0) as described previously [19].

#### *2.6. Covariates*

Child age, cohabitation status (parents, parents and grandparent(s), single parent and grandparent(s), single parent, or other), other children in household (yes or no), household annual income (<3.00, 3.00–5.99, 6.00–9.99, or ≥10.0 million Japanese yen), respondent (mother, father, or other), parental age (<35, 35–44, or ≥45 years), mother's employment status (full-time, part-time, self-employed, side work, not employed, or other), and parental height and weight were assessed via the caregiver report. Parental BMI was calculated using self-reported height in centimeters and weight in kilograms. Standard categories of BMI were used to characterize parents as underweight (<18.5 kg/m2), normal (18.5–24.9 kg/m2), overweight (25.0–29.9 kg/m2), or obese (≥30.0 kg/m2) [24]. Participants with missing data on covariates were included in the analysis as dummy variables.

#### *2.7. Statistical Analysis*

First, participants were stratified by level of cooking skills, and differences between groups were analyzed using Pearson's chi-square test. Second, multiple comparisons for the cooking skills scale were performed using a mixed linear model procedure to examine which cooking skills participants rated as difficult. The peeling scale used as a reference and the participant identification code was included as a random effect. Third, multiple comparisons between respondents (mother, father, and other) were analyzed using Dunnett's pairwise comparison method with mother as the reference category. Fourth, we calculated adjusted odds ratios (ORs) with 95% confidence intervals (CIs) for low frequency of home cooking, child low frequency of vegetable intake, and child breakfast skipping using logistic regression. Fifth, we calculated adjusted relative risk ratios with 95% CIs for underweight, mild-underweight, overweight, and obese using multinomial logistic regression, with normal as the reference category. The models were adjusted for potential confounding factors that were associated with level of cooking skills in the first analysis. Finally, we conducted a mediation analysis to determine the proportion of the association between caregiver cooking skills and child weight status mediated by frequency of home cooking. We estimated the natural direct effects, controlled direct effects, and natural indirect effects of mediators after controlling for all covariates using the Paramed package in Stata [25]. The exposure was treated as a binary variable, with 0 representing a high level of caregiver cooking skills and 1 representing a low level of caregiver cooking skills. The mediators and outcomes were treated as continuous variables. Because the association between caregiver cooking skills and child weight status was U-shaped, the mild-underweight and underweight children were excluded from the mediation analysis

of the relationship with child obesity. All analyses were conducted using Stata Version 15 (Stata Statistical Software; StataCorp LP, College Station, TX, USA).

#### **3. Results**

The characteristics of the children and caregivers are summarized in Table 1. Half of the children were girls, about 80% were fourth-grade and lived with their parents, 81% had siblings, and 11% had families with annual incomes below 3.00 million yen. A total of 10.8% ate vegetable dishes less than three times a week, 2.5% were breakfast skippers, 2.3% were underweight, and 5.5% were obese. The majority of the caregiver respondents were mothers (91%); 8% were fathers. The most common mother's employment status was part-time (48%), followed by full-time (21%). A total of 2.6% of households cooked less than three times a week. Approximately 5% of caregivers (*n* = 247) were classified as having a low level of cooking skills. Children who had caregivers with a low level of cooking skills tended to live with grandparent(s), have no other children in the household, have father respondents, and have full-time working mothers (Table 1).

The mean score for caregiver cooking skills was 5.5 points among all participants (Table 2). The mean caregiver cooking skills score was lower for fathers (5.0 points) than for mothers (5.5 points). For each item in the cooking skills scale, mother scored higher than father, except for 'like to cook' (item 5). Compared with mothers, other respondents gave higher scores for the item of 'like to cook'. Among the four cooking methods, fathers and other respondents rated stewing as more difficult than peeling, while mothers rated all methods as being of similar difficulty, although they had significant differences among the four methods (Table 2). 'Like to cook' was correlated with other cooking methods (*r* = 0.20–0.26, *p* < 0.0001), and especially highly correlated among fathers (*r* = 0.47–0.51, *p* < 0.0001) (Supplementary Table S1).

A low level of caregiver cooking skills was associated with low frequency of home cooking and child low frequency of vegetable intake (Table 3). After adjusting for potential confounders, caregivers with low-level cooking skills were 4.31 (95% CI: 2.68–6.94) times more likely to have lower frequency of home cooking than those with high level of cooking skills. Children with low level of caregiver cooking skills were 2.81 (95% CI: 2.06–3.84) times more likely to have low frequency of vegetable intake. Low level of caregiver cooking skills was not significantly associated with child breakfast skipping (AOR = 1.61, 95% CI: 0.97–3.53).

A U-shaped association was found between caregiver cooking skills and child weight status (Table 4). Children with low level of caregiver cooking skills were 1.74 (95% CI: 1.08–2.82) times more likely to be obese and 1.84 (95% CI: 0.88–3.83) times more likely to be underweight, although the association with underweight status was not statistically significant. The mediation analysis showed that 91% of the association between low level of caregiver cooking skills and child obesity was mediated by frequency of home cooking.


**Table 2.** Cooking skills scale scores of Japanese caregivers (*<sup>n</sup>* = 5257).

Multiple comparisons between items on the cooking skills scale were analyzed using a mixed linear model procedure. Participant identification code was included as a random\*\*\*\*\*

 effect.

> Multiple comparisons between respondents were analyzed using Dunnett's pairwise comparison method. *p* < 0.01, *p* < 0.001, versus mother.


**Table 3.** Adjusted odds ratios of low frequency of home cooking, child low frequency of vegetable intake, and child breakfast skipping according to levels of caregiver cooking skills (*n* = 5257).

AOR, adjusted odds ratio; CI, confidence interval. The models were adjusted for cohabitation status, siblings, respondent, mother's age, and mother's employment status.

**Table 4.** Adjusted relative risk ratios of child obese, overweight, mild-underweight, and underweight status according to levels of caregiver cooking skills (*n* = 5257).


ARRR, adjusted relative risk ratio; CI, confidence interval. The models were adjusted for cohabitation status, siblings, respondent, mother's age, and mother's employment status.

#### **4. Discussion**

To the best of our knowledge, this is the first study to investigate the associations between caregiver cooking skills and weight status of school children. Using a modified version of the existing cooking skills scale for use in the Japanese population, we found that a low level of caregiver cooking skills was associated with low frequency of home cooking and low frequency of vegetable intake in the child. Regarding child weight status, a U-shaped relationship was observed and a significant association was found between a low level of caregiver cooking skills and child obesity.

A low level of caregiver cooking skills was positively associated with child obesity and most of this association was explained by the frequency of home cooking. These findings are consistent with a previous study showing that infrequent home cooking was associated with child obese status [12]. A systematic review confirmed dietary benefits of home cooking, including greater consumption of healthier food groups, although most of the included studies involved adults [10]. In a study on child diets, home cooking was associated with higher vegetable consumption among children in the United Kingdom [26]. Consistent with that study, we also found that low level of caregiver cooking skills was associated with child low frequency of vegetable consumption.

A U-shaped relationship was observed, which indicates that low level of caregiver cooking skills also tended to be associated with child underweight status. This finding is consistent with a study on older Japanese adults showing that low level of cooking skills can lead to under-nutrition [19]. Children have difficulty preparing their own meals, and therefore their meals depend on their caregivers. Thus, low cooking frequency arising from low level of caregiver cooking skills may mean that children skip meals or eat low-energy diets, leading to them becoming underweight. To test this hypothesis, future studies are warranted to conduct more detailed dietary surveys on the frequency of children eating out.

A low level of caregiver cooking skills was associated with low frequency of home cooking. This finding is consistent with the previous study among older Japanese adults [19]. However, this result may be underestimated because there may have been more than one

person in charge of cooking at home, such as the mother, father, grandmother, and older siblings, or a person different from the respondent may be the main cook. Furthermore, low level of caregiver cooking skills was not significantly associated with child breakfast skipping. One possible reason is that breakfast is generally a simple meal in Japan [27], and thus does not require a high level of cooking skills. Otherwise, it may be due to the child's lack of time or appetite.

The validity of the cooking skills scale needs careful consideration. In the present study, mothers who tended to prepare food scored significantly higher on the cooking skills scale than fathers who were less likely to prepare food. This suggests that the modified version of the cooking skills scale used in the study had notable discriminant validity. The observed sex difference is consistent with previous findings on confidence regarding cooking skills, in which women were found to be more confident in their cooking skills than men [15,28,29]. We included four basic cooking methods in the cooking skills scale. Consistent with the previous study among older Japanese adults [19], fathers rated stewing as more difficult than peeling and boiling. Compared with the results for the Japanese older adults in the previous study, middle-aged women (mothers in the present study) scored almost the same as older women, while middle-aged men (fathers in the present study) tended to score higher than older men [19]. This generation difference among men may be explained by opportunities to learn cooking skills in school. In Japan, cooking education in schools for men started in 1947 and became compulsory in 1989 [30]. Therefore, older men had less opportunity to learn cooking in school.

We confirmed that 'like to cook' was correlated with other cooking methods, and especially highly correlated among fathers. This is plausible because women need to cook regardless of whether they like it because of the social norm [10], and as a result, their cooking skills will improve. We further confirmed that 'like to cook' was important for prevent low frequency of home cooking. As a result of analyzing the association between the single item 'like to cook' and the frequency of home cooking, caregivers with low level of liking to cook (score ≤ 4.0) were 2.21 (95% CI: 1.52–3.19) times more likely to have lower frequency of home cooking than those with high level of liking to cook (score > 4.0) after adjusting for potential confounders (data not shown). Given the importance of liking to cook for maintenance of a high level of cooking skills during the life course [23], it may be critical to examine subjects for liking to cook when examining the associations of cooking with diet-related outcomes.

There are some limitations to the present study. First, child frequency of vegetable dish intake was assessed using a single simple item. Future studies should use more detailed validated questions to assess which food groups and nutrients are associated with cooking skills. Second, we were only able to evaluate a limited number of child eating behaviors. Future studies are warranted to investigate the relationships between caregiver cooking skills and other aspects of child diets, such as amounts of energy and foods other than vegetables consumed, to understand the mechanisms. Third, we observed a ceiling effect for caregiver cooking skills, especially among mothers, similar to the findings in previous studies using the original cooking skills scale [15,19]. Given that mothers are often working, it may be useful to investigate not only their cooking methods (such as stewing), but also their ability to cook well in a short amount of time. In addition, more comprehensive validated measures are now available for assessing confidence in food and cooking skills in United Kingdom populations [31]. Therefore, it may be possible to use these measurement methods in the future. Fourth, the generalizability of the results may be low because our sample of school children was located in only one city in Japan. Fifth, we lacked data on some potentially confounding factors, such as caregivers' nutritional knowledge and food preference. There may be a much more dynamic association between child obesity, caregiver cooking skills, and their liking for cooking. Finally, we were unable to assess causality because this was a cross-sectional study. However, in a previous study that examined the acquisition of cooking skills, more than half of the respondents reported that they had learned most of their cooking skills when they were teenagers and that these

cooking skills were mainly taught by their mothers [32]. Randomized controlled trials in the younger generation before having children are needed in the future to clarify the effectiveness of caregiver's ability to prepare meals for preventing obesity in children.

We found that a low level of caregiver cooking skills was associated with low frequency of home cooking, low frequency of child vegetable intake, and child obese status. Most of the association between low level of caregiver cooking skills and child obesity was mediated by the frequency of home cooking. The present findings are important for preventing unhealthy eating behaviors and obesity because COVID-19 is increasing the demand for home cooking. In addition, poor caregiver cooking skills can cause not only obesity in children but also less opportunity to learn cooking skills from caregivers, which may have an impact on the next generation (i.e., grandchildren of the current parents) due to the poor cooking skills of the children when they become parents [32]. In the future, it is necessary to clarify the causal relationships and promote research on support to improve caregiver cooking skills.

**Supplementary Materials:** The following is available online at https://www.mdpi.com/article/10.3 390/nu13124549/s1, Table S1: Spearman correlation coefficients for items on the cooking skills scale.

**Author Contributions:** Conceptualization, Y.T. and T.F.; Data Curation, Y.T. and T.F.; Methodology, Y.T.; Formal Analysis, Y.T.; Investigation, Y.T., T.F., A.I. and S.D.; Writing—Original Draft Preparation, Y.T.; Writing—Review & Editing, T.F.; Supervision, T.F., A.I. and S.D.; Project Administration, Y.T.; Funding Acquisition, Y.T. and T.F. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (JSPS KAKENHI) grant number 16H03276, 16K21669, 17J05974, 19H04879, 19K20109, 19K19309, 19K14029, 19J01614, 19K14172, 20K13945 and 21H04848.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of Tokyo Medical and Dental University (No. M2016-284).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Acknowledgments:** We are especially grateful to the central office Adachi City Hall and its staff members for conducting the survey. We would also like to thank everyone who participated in the survey. Additionally, we would particularly like to thank Yayoi Kondo, Syuichiro Akiu, and Yuko Baba from Adachi City Hall, who contributed significantly to the completion of this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

