The Association of Maternal Perceived Stress With Changes in Their Children’s Healthy Eating Index (HEI-2010) Scores Over Time

Abstract

Background

Few studies have examined the role of maternal stress in relation to their children’s dietary quality and its trajectory over time.

Purpose

The objective of this longitudinal study was to examine the effect of baseline maternal stress on the change in their 8- to 12-year-old children’s dietary quality over 1 year.

Methods

Mother–child dyads (N = 189) from the greater Los Angeles area participating in the Mothers’ and their Children’s Health (MATCH) study in 2014–2016 completed assessments at baseline, 6 months, and 12 months. At baseline, mothers (mean age = 41.0 years, standard deviation [SD] = 6.1) completed the 10-item Cohen’s Perceived Stress Scale (PSS). At each time point, children (51% female, mean age = 9.6 years, SD = 0.9) completed up to two 24-hr dietary recalls. Dietary data were used to calculate each child’s Healthy Eating Index 2010 (HEI-2010) score at each time point. Multilevel models examined the effect of time on the patterns of change in children’s HEI-2010 scores over 1 year and the cross-level interaction between baseline maternal PSS score and time on the change in children’s HEI-2010 scores.

Results

On average, there was no significant linear change in child HEI-2010 across 1 year (b = −0.410, p = .586). Controlling for covariates, the rate of change in HEI-2010 differed depending on mother’s baseline PSS (i.e., significant cross-level interaction effect) (b = −0.235, p = .035).

Conclusions

Our results showed that higher-than-average maternal stress at baseline was associated with greater decline in children’s dietary quality over 1 year. Family-based dietary interventions that incorporate maternal stress reduction could have positive effects on children’s dietary quality.

Children ages 8-12 whose mothers had higher than average perceived stress at baseline had greater declines in their 24-hour recall assessed dietary quality over one year.

Introduction

Although the Dietary Guidelines for Americans (DGA) recommend sufficient intake of health-producing food groups and nutrients and limited intake of nutrient-poor and energy-dense foods, children in the USA fall short of meeting the dietary recommendations in nearly every category. Studies show that approximately 40% of children’s daily energy intake comes from “empty calories,” or foods and beverages high in solid fat and/or added sugars [1], and just one in three children consume the minimum recommended amount of fruits and vegetables (F&V) [2]. Children’s failure to meet dietary guidelines is problematic, because a poor diet has been linked to elevated risk of chronic diseases, including diabetes, cardiovascular disease [3], and obesity [4], while sufficient intake of F&V may be beneficial for disease prevention, longevity, and greater overall health [5, 6]. Thus, children’s dietary intake quality is an important health behavior that may play a role in numerous health outcomes.

Numerous factors may play a role in shaping children’s diet quality. Evidence from cross-sectional studies shows that children’s dietary quality tends to decline with increasing age, particularly during the transition from childhood to adolescence [7–9]. A recent analysis of NHANES data showed poor overall dietary quality across all age groups of children. Within the sample, diet quality was higher among children aged 4–8 and comparatively lower across children aged 9–13 and 14–18 years [7], a finding that was echoed in a similar study [8]. An examination of the proportion of youth meeting dietary guidelines indicated that the proportion of children meeting the daily F&V recommendations decreased as children get older, ranging from 49.8% (fruits) and 21.7% (vegetables) in 2- to 5-year-old youth, followed by 25.9% (fruits) and 16.2% (vegetables) in 6- to 11-year-old youth, to 19.5% (fruits) and 10.5% (vegetables) in 12- to 18-year old youth [9]. Characteristics that may protect against the decline in children’s dietary quality include being female [10], higher socioeconomic status [11], and lower stress [12], suggesting that some influences are stable while others are potentially modifiable.

Beyond the characteristics of the child, several maternal factors play an important role in the quality of their children’s diets. Factors such as higher maternal education, mothers’ own dietary quality and maternal modeling of healthy eating have all been linked to the dietary intake and dietary quality of her children [13–16]. Another important factor that may influence children’s health is the parents’ experience of perceived stress, which occurs when demands upon an individual exceed the resources available to meet those demands [17]. Short-term elevations in perceived stress may be beneficial, allowing for an individual to meet immediate demands; however, chronically elevated stress, which is higher among women than men [18] can become maladaptive, negatively impacting immune and inflammatory processes [19]. Chronic stress might also have a negative impact on other family members, including children. For example, previous research has shown that elevated maternal stress is associated with elevated body mass index (BMI) in their children [20–22]. As such, an emerging body of research has begun to examine whether elevated perceived stress in mothers may also affect their children’s dietary intake [21, 23–25].

A recent systematic literature review of n = 6 studies found little evidence for an association between various types of maternal stress and children’s healthy and unhealthy dietary intake [26]. While one cross-sectional study found that children of mothers with greater perceived stress were significantly more likely to consume fast food at least twice per week [21], another study found no association of parental perceived stress with children’s added sugar intake 1 year later [23]. These existing studies are subject to several limitations, which may partially explain the mixed findings. First, studies often used a cross-sectional design, preventing causal inference. Second, studies commonly used maternal report of children’s diet, which is problematic because mothers may be unaware of intake that occurs outside of the home (e.g., at school). Finally, studies have typically isolated single aspects of child diet quality (e.g., servings of sugar sweetened beverages per week), failing to capture children’s overall dietary pattern, which may be more closely linked to health and disease than consumption of isolated nutrients or foods [27, 28]. Furthermore, no existing studies have examined maternal stress in relation to the trajectory of children’s dietary quality over time. Given the previously described declines in child diet quality in the transition to adolescence [7], it becomes even more important to understand the prospective impact of maternal stress as it relates to their children’s dietary trajectory over time.

To address the limitations of previous studies, the current study used a longitudinal design with multiple assessments of child-reported dietary intake and scores of overall dietary quality. The primary goals of our study were (a) to describe the overall pattern of change in child-reported diet quality over time, and (b) to examine the association of maternal stress measured at baseline and changes in children’s dietary quality over the following year. Based upon findings from previous studies [7–9], we hypothesized that children’s dietary quality would decline as they get older. Additionally, we hypothesized that children whose mothers reported greater perceived stress would experience greater decline in their dietary quality over time, and that this effect would persist above and beyond the effects of other demographic factors (e.g., child age and sex), as well as children’s own perceived stress and mother’s modeling of healthy eating, which have been shown in previous research to affect children’s diet quality [16, 29].

Methods

Participants

Participants included 189 mother–child dyads enrolled in the Mothers’ and Their Children’s Health (MATCH) Study. MATCH is a longitudinal, within-subjects study, consisting of multiple assessment “waves,” each separated by approximately 6-month intervals [30]. Mothers were the focus of this study, as they are most often the primary caregiver [31] and have been found to experience higher levels of stress than their male counterparts [32]. A sample of ethnically diverse dyads from the greater Los Angeles metropolitan area were recruited from local schools and community centers. Dyads were expected to reflect the typical stress and dietary behaviors of mothers and children in the area. The current study includes data from the first three waves (baseline, 6 months, and 12 months). Eligibility criteria for the MATCH Study included: (a) in 3rd to 6th grade (child), (b) living together at least 50% of time (mother and child), and (c) able to speak and read in English or Spanish (mother and child). Dyads were ineligible if they: (a) used medication for thyroid or psychological condition (mother) or oral or inhalant corticosteroids for asthma (mother or child), (b) had a health condition limiting ability to be physically active (mother or child), (c) were enrolled in a special education program (child), (e) worked more than two evenings (between 5 and 9 pm) during the week or more than one 8-hr weekend shift (mother), (d) were pregnant (mother), or (e) were underweight (BMI < 5th% for age and sex) (child).

Procedures

At each wave, mothers and children attended an in-person data collection session, followed by 1 week of at-home assessments. At the in-person visits, mothers and children were measured for anthropometrics, and completed various questionnaires. In the week following each in-person visit, children completed up to two 24-hr dietary recalls via telephone. Mothers provided informed consent for themselves and their child, and children provided assent to participate. The Institutional Review Board at the University of Southern California approved all aspects of this study.

Measures

Maternal perceived stress

Mothers’ perceived stress (predictor) was measured at the baseline in-person visit using the 10-item Cohen’s Perceived Stress Scale (PSS) [17]. The PSS is a widely used validated and reliable scale for assessing the extent to which an individual appraises their life events as stressful. Mothers were asked about their experiences over the past month (e.g., “In the last month, how often have you found that you could not cope with all the things that you had to do?”). Response options ranged from 0 (never) to 4 (very often). Responses for the PSS were internally consistent (α = 0.83), and responses for all items were summed (some were reverse coded) to create the overall PSS score, which could range from 0 to 40.

24-Hr dietary recalls

During each of the week-long study waves (i.e., baseline, 6 months, 12 months), children completed up to two 24-hr dietary recall assessments (outcome). Recalls were scheduled so that one weekday (Mon–Fri) and one weekend day (Sat–Sun) were recalled at each wave whenever possible. The dietary recalls were conducted via telephone by trained interviewers, who instructed children to report on all foods and beverages consumed in the previous calendar day. Using the multiple pass method, the interviewer guided children to recall as much detail as possible regarding contents, preparation, and serving size of all foods and beverages consumed. Children were equipped with assistive guides for portion size reference, and mothers were asked to be available for assistance when possible and/or necessary. While all methods of assessing diet are subject to limitations, 24-hr dietary recalls have been found to be fairly reliable for estimating children’s energy intake as compared with doubly labeled water [33] and direct observation [34], with a bias toward underreporting via omission. Although young (under 12 year) children may have difficulty in accurately reporting their past-day diet through 24-hr dietary intake, many studies have used 24-hr recall methods to assess diet in children and enlisting the assistance of parent or other household member generally improves the accuracy of children’s report [33]. The Nutrition Data System for Research (NDSR) database [35] was used to code each recall for nutrients and total energy intake.

Up to two telephone-based 24-hr dietary recalls at each wave were averaged and used to calculate each child’s wave-specific Healthy Eating Index 2010 (HEI-2010) score, a measure of dietary adherence to DGA, reflecting an individual’s overall dietary quality [36]. While the majority (>80%) of participants completed two dietary recalls at each wave, the HEI-2010 score for children who completed only one recall at a given wave (n = 24 at baseline, n = 19 at 6 months, and n = 12 at 12 months) was calculated using that one recall. The HEI-2010 contains 12 component scores, nine of which measure nutritional adequacy (i.e., dietary components to increase; e.g., whole fruits, greens and beans), and three of which assess moderation (i.e., dietary components to decrease; e.g., refined grains, empty calories) [37]. The component scores are summed to create the HEI-2010 total score. An HEI-2010 score of 100 indicates an optimal diet, in complete adherence to the DGA.

Covariates

Anthropometrics

Trained staff assessed height and weight on mothers and children using a digital scale and stadiometer at the baseline in-person visit. Measures were taken in duplicate to the nearest 0.1 kg and 0.1 cm, and in discrepant cases the average of the two measurements was taken. BMI was calculated (kg/m²) and classified according to CDC categories (e.g., underweight/normal weight, overweight, obese) for mothers and based on corresponding BMI percentile for age and sex for children [38].

Demographics

Mothers completed paper questionnaires reporting on their age, highest level of education, marital status, number of children in the household, annual household income, and ethnicity (of self and child); children self-reported their age and sex.

Parental Role Modeling of Healthy Eating was measured at baseline using a 12-item scale from the Home Environment Survey [39]. The Parental Role Modeling of Healthy Eating scale is a reliable and valid assessment of parental modeling of undesirable eating habits (consumption outside of main meals, in front of the television, in response to negative affect), which has been found to correlate with both parental and child eating habits [39]. Mothers were asked to report on their assessment of their child’s observation of her own healthy and unhealthy eating behavior (e.g., “Does your child see you eat when you are bored?”). Response options ranged from 0 (never) to 4 (almost always to always). At least 75% completion of the scale was required to calculate a score, which was internally consistent (α = 0.70), and was calculated by taking the mean across all items (some were reverse coded). Possible scores ranged from 1 to 4, with higher scores indicating more modeling of healthy eating.

Child Perceived Stress was measured at baseline using the 21-item Stress in Children (SiC) Scale [40]. The SiC scale has been validated to assess the degree of perceived stress experienced by children and correlates with other validated measures of negative affective states, anxiety, and depression. The SiC scale was developed among a sample of 9- to 12-year-old youth and is appropriate for use with school-aged children [40]. Children reported on their feelings and thoughts over the past month (e.g., “Sometimes I can’t manage with the things I have to do.”). Response options ranged from 0 (never) to 4 (very often). At least 19 items were required to calculate the score, which was internally consistent (α = 0.82) and was calculated by taking the mean across all items. Possible scores ranged from 1 to 4, with higher scores indicating higher levels of perceived stress.

Data Availability, Preparation, and Statistical Analyses

Data Availability

A total of 202 dyads enrolled in the MATCH study. Of those, N = 189 dyads (Level 2) were included in the analytical sample for the current study, with a total of n = 412 assessment waves (Level 1). Dyads were excluded if there was no available maternal baseline PSS (n = 3), if the child had no available dietary data at any assessment wave (n = 9), or if neither maternal baseline PSS nor child dietary data were available (n = 1). Individual days of dietary recall (n = 8) were excluded from the analytical sample if kilocalorie intake on that day was implausible (i.e., kcal <500 or >4,000) [41]. Excluded dyads did not differ from included dyads in any demographic or household characteristic (p > .05). However, children in excluded dyads had significantly lower SiC scores (1.72 ± 0.29 vs. 1.99 ± 0.36, p < .05).

Data preparation

Before analysis, the following maternal variables were dichotomized: marital status (married vs. not married), education level (college vs. no college), and work status (full time vs. not full time). The following maternal and child variables were also dichotomized: weight status (overweight/obese vs. normal/underweight), and ethnicity (Hispanic vs. non-Hispanic). Annual household income was divided into quartiles (≤$35,000; $35,001–$75,000; $75,001–$105,000; >$105,000).

Statistical analyses

Analyses were conducted in SAS (V 9.4). To account for the clustering of repeated observations within dyads, multilevel linear regression models (SAS Proc Mixed) were used, which account for the nonindependence of wave-level observations (Level 1) within dyads (Level 2). To explore the effect of mother’s baseline stress on change in children’s dietary quality, mothers’ baseline PSS scores were centered to represent each mother’s deviation from the group mean, allowing us to examine the relative effect of maternal stress that is higher or lower than the mean on children’s outcomes. To test whether the change in children’s outcome depends on baseline maternal stress level, a cross-level interaction term of baseline maternal stress × time was created, by multiplying the centered baseline PSS by assessment wave (i.e., baseline, 6 months, 12 months).

In Model 1, study wave was entered as a predictor, to assess the change in child HEI-2010 score over time. In Model 2a, the cross-level interaction term was entered to test the effect of baseline maternal stress on the change in child dietary quality over time. Model 2b extends Model 2a by controlling for a set of covariates that were either selected a priori (i.e., child sex and age, household income, child’s own baseline stress, baseline maternal modeling of healthy eating), or which were screened and found to be significantly associated with either baseline maternal stress or child HEI-2010 (i.e., mother’s marital status).

Results

Participants

At baseline, mothers had a mean age of 41.0 years (standard deviation [SD] = 6.1), and children (52% female) had a mean age of 9.6 years (SD = 0.9). About half of mothers (46.7%) and children (52.9%) identified as Hispanic. The majority of mothers were married (68%). Approximately two-thirds of mothers and one-third of children were classified as either overweight or obese. Mean maternal PSS and child stress were similar to estimates from other representative samples of women [32] and children [42]. The mean child HEI-2010 total score was 48.7 (out of 100), indicating low adherence to dietary recommendations. Full participant demographic and descriptive characteristics are provided in Table 1.

Table 1.

Baseline characteristics of n = 189 mothers and children in the Mothers’ and Their Children’s Health (MATCH) Study

Variable	Mothers
	M (SD)	N (%)
Age (years)	41.0 (6.1)
Hispanic		90 (47.6)
Married		129 (68.3)
College graduate		112 (61.2)
Work full-time		109 (58.6)
Number of children	2.29 (1.0)
Overweight/obese		118 (64.8)
Household income
≤$35,000		50 (26.6)
35,001–75,000		57 (30.3)
75,001–105,000		36 (19.2)
>$105,000		45 (23.9)
Perceived stress (PSS)a	14.7 (5.4)
Modeling of healthy eatingb	2.9 (0.4)
	Children
	M (SD)	N (%)
Age (years)	9.6 (0.9)
Female		97 (51.3)
Hispanic		100 (52.9)
Overweight/obese		66 (36.3)
Perceived stress (SiC Scale)c	2.0 (0.4)
HEI-2010 total scored	48.7 (12.3)

N = 189 mother–child dyads from greater Los Angeles enrolled in the Mothers’ and Their Children’s Health (MATCH) study in 2014–2016. Some covariate data are missing, thus the total numbers for each characteristic may not add up to 189.

^aPSS = Perceived Stress Scale. Range of possible scores is 0–40.

^bRange of possible scores for modeling of healthy eating is 1–4.

^cSiC = Stress in Children Scale. Range of possible scores is 1–4.

^dHEI-2010 = Healthy Eating Index—2010. Range of possible scores is 0–100.

Covariates

A priori covariates included child age and sex, household income, child stress, and maternal modeling of healthy eating, based on previous evidence of a potential confounding effect on the relationship of interest. Married mothers reported significantly lower baseline stress than nonmarried mothers (14.1 vs. 15.9; p = .03) and marital status was included as a covariate in the final model. The remaining screened variables (i.e., maternal age, maternal education, maternal BMI category, number of children in the household, Hispanic ethnicity (of child or mother)) were not associated with either maternal stress or with child dietary quality, and thus were not included as covariates in any model.

Multilevel Models

The results of the multilevel models are displayed in Table 2. Model 1 revealed that, on average, there was no significant change in children’s HEI-2010 across the three assessment waves (b = −0.410, p = .485). Model 2a revealed a significant cross-level interaction effect of mother’s baseline PSS with wave (b = −0.229, p = .038). Figure 1 displays the expected effect of the cross-level interaction term (i.e., baseline maternal stress × time). In this figure, the three lines represent the change in children’s HEI-2010 across the three assessment waves for children whose mothers’ baseline PSS was at the group mean (PSS = 14.7), 1 SD below the mean (PSS = 9.3), and 1 SD above the mean (PSS = 20.0). As shown in Fig. 1, this interaction indicates that children of mothers who reported greater levels of stress at baseline than other mothers in the sample experienced steeper decreases in diet quality over time. Model 2b in Table 2 shows that the cross-level interaction between baseline maternal stress and time (b = −0.235, p = .035) remained significant when controlling for covariates, including children’ own stress (Table 2). As seen in Model 2b, the only covariate with a significant effect on children’s dietary quality was maternal modeling of healthy eating, which was positively associated with child diet quality (b = 3.8007, p = .0267).

Table 2.

Results from multilevel linear regression models predicting longitudinal change in 8- to 12-year-old children’s HEI-2010 scores

Variable	Model 1			Model 2a			Model 2b
	b	SE	p Value	b	SE	p Value	b	SE	p Value
Intercept	48.707	0.895	<.0001	48.699	0.899	<.0001	41.094	10.638	.000
Wave	−0.410	0.586	.485	−0.383	0.581	.510	−0.688	0.704	.329
mPSS	–	–	–	0.078	0.168	.643	0.216	0.179	.229
Wave × mPSS	–	–	–	−0.229	0.110	.038	−0.235	0.111	.035
Child sex	–	–	–	–	–	–	0.640	1.600	.690
Child age	–	–	–	–	–	–	0.510	0.782	.515
Child stress	–	–	–	–	–	–	−3.707	2.187	.092
Income quartile	–	–	–	–	–	–	−1.236	0.705	.081
Mother married	–	–	–	–	–	–	2.640	1.783	.140
Mother modeling of healthy eating	–	–	–	–	–	–	3.801	1.704	.027

The Level 2 sample size is N=189 mother–child dyads from greater Los Angeles enrolled in the Mothers’ and Their Children’s Health (MATCH) study in 2014–2016. The Level 1 sample size is n = 412 observations. mPSS = maternal perceived stress, measured at baseline with the Cohen’s Perceived Stress Scale and centered at the group mean. Healthy Eating Index (HEI)-2010 was derived from multiple 24-hr dietary recalls conducted at baseline, 6 months, and 12 months. The Wave × mPSS interaction term was created by multiplying baseline mPSS by wave, to estimate the effect of baseline mPSS on change in HEI-2010 across time. Model 2b controls for the effects of child sex, child age, child stress, family income quartile, mother’s marital status, and mother’s modeling of healthy eating. Bold values indicate statistical significance at p < 0.05.

Fig. 1.

Plot of longitudinal change in 8- to 12-year old children’s HEI-2010 scores as a function of the interaction of baseline maternal stress and wave. Note: Figure of the change in children’s HEI-2010 scores as a function of the interaction of baseline maternal stress and wave, based on data from N = 189 mother–child dyads from greater Los Angeles enrolled in the Mothers’ and Their Children’s Health (MATCH) study in 2014–2016.

Discussion

The present analyses examined the pattern of change in 8- to 12-year-old children’s dietary quality as well as the association of baseline maternal stress with individual trajectories of change in children’s dietary quality over 1 year. Overall, the negative linear trend in children’s dietary quality over 1 year was not significant. However, our findings showed that children of mothers who reported greater baseline perceived stress experienced steeper declines in diet quality over time, relative to children of mothers with average level of baseline stress, regardless of the child’s age at baseline.

Maintaining a healthy dietary pattern, full of fiber and micronutrients is important for lifelong health, while consumption of added sugars, saturated fat, and refined foods has been linked to elevated risk of diet-related chronic diseases [27, 28]. The HEI-2010 reflects guidelines for a healthy diet and is a useful tool for determining an individual’s overall dietary quality. While a minimum HEI-2010 score of 80 is recommended for health benefits and disease prevention, the mean HEI-2010 score among our sample was far below this minimum threshold, with a mean of 48.7, and ranging from 48.7 (SD: 12.2, range: 23–77) at baseline, to 48.0 (SD: 12.6, range: 18–78) at 6 months, to 47.9 (SD: 12.7, range: 23–80) at 12 months. In our sample, none of the 189 children consumed on average a health-promoting diet as defined by a mean HEI-2010 score of at least 80; instead, children had dietary quality that would be classified as “poor” (HEI < 50) or “needs improvement” (HEI = 50–80) [43], with one child’s HEI-2010 score reaching the minimum recommended value of 80 at the 12-month assessment wave.

Based on recent USDA and NHANES data, the average HEI-2010 score for children is approximately 50 [7, 44], indicating that the dietary quality of our sample was representative of children in the USA, and highlighting the insufficiency of healthy diet among the children in our sample as well as U.S. children overall [7, 44]. Additionally, although children’s diet tends to decline in the transition to adolescence, the lack of significant linear change observed here is similar to the lack of linear change observed in a longitudinal study of older adolescents’ HEI across 4 years, from 10th grade of high school to 1 year after high school graduation [45]. As the current study assessed children across 1 year, it is possible that the relatively short time frame missed any systematic declines in child diet associated with increasing age. Nevertheless, in the current sample of children aged 8–12 at baseline, children’s age was not a significant covariate in our final model, indicating that it was not significantly associated with change in dietary quality across the 1-year study period.

Our study revealed that elevated baseline maternal stress was associated with prospective decreases in the quality of their child’s diet over 1 year. Although at the group-level children’s diet did not change over time, children of mothers with baseline stress greater than the group mean experienced significant decreases in dietary quality across this period. To illustrate the meaning of this interaction effect, children whose mothers experienced low stress at baseline had HEI scores that declined about 4.5 points over the year. However, children of mothers experiencing high stress at baseline had HEI scores that declined twice as much (about 9 points over the next year) (Fig. 1). This difference between the high and low stress groups represents almost 10% of the entire HEI scale range (and is equivalent to a little less than 1 SD in terms of the HEI score distribution of our sample). Ours is the first known study to use multilevel modeling to examine the slope of change in children’s diet quality over time in relation to maternal stress. Previous studies have examined the prospective effects of maternal stress on child diet using other approaches; for example, one study found that maternal report of >2 social risk factors (a measure of stress) over a period of 2 years was associated greater odds of >3 daily servings of soda and juice in children 2 years later [24]; however, this study did not take into account baseline intake or change in intake over time.

By controlling for several factors that are known to impact children’s dietary quality including age [7–9], sex [10], SES [11], child stress [46], and maternal modeling of healthy eating [16], we are able to isolate the independent effect of maternal perceived stress on the change in children’s dietary quality over 1 year. The longitudinal nature of our dataset also allowed us to examine whether reverse causality may be playing a role in our findings, in which poorer quality diet in children may lead to elevated stress in mothers. However, ancillary results from multilevel models found that child diet quality at baseline did not predict changes in maternal stress across the study period. Interestingly, in our final model using maternal baseline stress to predict children’s dietary quality, maternal modeling emerged as a significant covariate (b = 3.8007, p = .0267), suggesting that maternal modeling of healthy eating may be an important factor to consider when disentangling the complex relationship between maternal stress and child diet. Taken together, these findings suggest that maternal factors may play a significant role in shaping children’s dietary patterns.

Although we found a significant longitudinal effect of baseline maternal stress on change in children’s dietary quality across time, there was no effect of maternal perceived stress on children’s initial baseline dietary quality (b = 0.05, p = .77) as examined in ancillary analyses. This finding is in accordance with much of the existing literature to date, which has primarily relied on cross-sectional data [26]. In the present study, the robust longitudinal finding lends credence to the possibility that elevated maternal stress may play a long-term role in shaping children’s dietary patterns. Ancillary analyses revealed a lack of association between children’s dietary quality and children’s BMI at baseline, as well as a lack of prospective association between baseline maternal stress and children’s BMI trajectory; this suggests that the effects of maternal stress on child behavioral patterns may operate over longer periods of time, and that children’s diet may not directly associate with their weight status at a given time. Similar to the observed relationship between exposure to stressors in early childhood and elevated stress reactivity [47] and risk of obesity later in life [48], exposure to elevated maternal stress may lead to alterations in biological processes [49] or behavioral patterns that are cumulative in nature, manifesting in declining dietary quality with passing time.

Strengths of our study include the use of longitudinal design to measure diet at three time-points over 1 year, which allowed us to study the long-term trajectory of children’s diet across 1 year. In contrast to previous studies which rely on mothers’ report of features of child diet (e.g., servings of F&V), we used multiple interviewer-assessed child-led dietary recalls paired with the HEI-2010 to estimate overall dietary quality. The HEI-2010 is appropriate for use in children [50], and provides a useful tool for summarizing overall dietary quality and energy density, making it suitable for studies of obesity [36], and allowing for cross-study comparison.

Although our study contains many strengths, there are also some limitations to note. In general, 24-hr dietary recalls underreport total intake, which may be a limitation in the present study [51]. Additionally, children may have difficulty in accurately and completely recalling dietary intake; however, by using parental assistance and guides we attempted to mitigate this limitation [33]. Furthermore, recalls with implausible energy intake (less than 2% of all recalls) were excluded from analyses. Although the HEI-2010 is a useful index for assessing children’s overall dietary quality, the present study did not examine specific component scores that comprise the HEI-2010 (e.g., whole fruits, empty calories), which may be important for designing interventions to buffer the effects of maternal stress on children’s dietary quality. However, ancillary analyses revealed no significant associations between maternal stress and the change in any of children’s HEI-2010 component scores over time. These findings suggest that the effect of maternal stress on children’s overall dietary quality may be cumulative. These results, however, should be interpreted with caution, given the non-normal distribution of several of the component scores within our sample, which may have been driven by participants with one recall per assessment wave (high proportion of zero values), and is a limitation of the current study. Future research should continue to disentangle the specific dietary components that may be altered by elevated maternal stress. Additionally, although the HEI-2010 accounts for recommended calorie intake in its scoring algorithm, the final score is a measure of energy density that is independent of energy intake and thus reflects an individual’s dietary quality, as opposed to diet quantity [36]. Total kilocalorie intake (i.e., quantity), another important dietary feature, is not captured.

Conclusions

Ours is the first study to examine the prospective association of mothers’ perceived stress on changes in children’s dietary quality over time. Findings are relevant to childhood obesity prevention and interventions programs and suggest that maternal stress reduction may help to promote children’s healthy diet.

Compliance with Ethical Standards

Authors’ Statement of Conflict of Interest and Adherence to Ethical Standards Dr G. F. Dunton reports consulting fees from the Dairy Council of California and the National Collaborative on Childhood Obesity Research, and travel funding from the National Physical Activity Plan Alliance, none of which are directly related to the submitted work.

Authors’ Contributions S.G.O. formulated the research question, analyzed the data, and wrote the manuscript. J.H. provided critical guidance on statistical analysis and interpretation of results. S.M.S. and N.V.L. contributed to the development and critical revision of the manuscript. G.F.D. conceptualized and designed the study, and contributed to the development and editing of the manuscript.

Informed consent Informed consent was obtained from all individual participants included in the study.

Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.