Child diet and health outcomes of the simple suppers program: a 10-week, 2-group quasi-experimental family meals trial

Childhood obesity is a persisting public health threat with 18.5% of US youth (2 to 19 years old) currently classified as obese [1]. Of particular concern, fluctuating rates have been identified in certain age groups of children. For instance, from 2005 to 2006 to 2013–2014, there was a positive trend among preschoolers (2 to 5 years old) and young children (6 to 11 years old), i.e., a sustained plateau, however a recent report indicates a rebound [2]. In addition, racial minority children, particularly those from low-income households, are at relatively high risk for obesity [3]. Due to the negative short- and long-term effects of obesity on children’s health, well-being [4], and academic success [5], this represents a significant public health concern and points to the urgent need for effective childhood obesity prevention interventions, especially in these populations.

The American Academy of Pediatrics recommends participation in family meals as a childhood obesity prevention strategy due to evidence demonstrating a protective effect of participation in healthy mealtime routines on child diet and weight status [6]. Family meals have been associated with increased intake of fruit and vegetables and reduced child intake of sugar-sweetened beverages as well as reduced risk of childhood obesity [7‐15]. However, the current evidence linking family meals with improved child dietary intake and weight status has limitations. The majority of the family meals literature – specifically in the area of childhood obesity prevention – are observational studies, demonstrating mixed evidence of an association between family meals and child diet and weight status [16‐21]. The existing family meals intervention research is strong with regards to study design (i.e., randomized controlled trial), however limited in number and target population (i.e., focuses primarily on White children from well-educated families) [22‐25]. In addition, the majority of the current family meals research fails to examine health outcomes beyond body mass index (BMI). Inclusion of blood pressure (BP) [26] would be particularly critical as hypertension in children is associated with obesity and onset of premature diabetes [10, 27‐29], even in very young children (i.e., < 5 years old) [30, 31]. There is need for additional research on family meals, specifically experimental studies with expanded health outcomes that focus on the at-risk populations in the greatest need of intervention (i.e., those with the highest rates of obesity) [32]. This body of research would also benefit from an expansion of the target age range to include younger children (4–7 year-old), who are laying the foundation of their eating patterns [33], capable of participating in family meal preparations [34], and may reap benefits of family meals [35].

The primary objective of this study was to assess the impact of a 10-week multi-component family meals intervention study, Simple Suppers, tailored to racially diverse low-income children (4–10 year-old) and aimed at eliciting positive changes in child diet and health outcomes [36]. The hypothesis was that: 1) Diet (daily servings of fruit, vegetables, and sugar-sweetened beverages and Healthy Eating Index (HEI) score); z-scores for BMI, waist circumference, BP; weight status (healthy weight (5th to <85th percentile), overweight (≤85th percentile), obese (≤95th percentile); food preparation skills; and family meal outcomes (weekly frequency of dinner, breakfast, TV viewing during meals, and meals in a dining area) would improve more from baseline (T0) to post-intervention (T1) among children participating in the intervention compared to waitlist control participants. In addition, it was hypothesized that these outcomes would be maintained during the follow-up period (T1:T2) among intervention participants. Also, an exploratory analysis was conducted examining a potential dose-response effect factoring in the role of program attendance. Feasibility and acceptability were also assessed.

One-hundred and forty children enrolled (109 families) and 126 children completed T1 (90% participant retention; Fig. 1). Seventy-one of 87 intervention participants who completed the T1 assessment completed T2 (81.6% retention). Descriptive summaries of participant baseline characteristics are presented in Table 1. Characteristics did not differ by group assignment (intervention and waitlist control) with the exception of race which was controlled for in subsequent analyses.

Intervention versus waitlist control children demonstrated significantly higher food preparation skills (p < 0.001; CI: 2.88, 6.58) and lower TV viewing during family meals (p = 0.04; CI: − 2.06, − 0.02) at post-intervention (T1) (Table 2). No intervention effects were found on dietary intake or BP outcomes. Regarding weight status outcomes, there were no intervention effects on BMI or waist circumference z-scores, however intervention versus waitlist controls experienced a greater change toward healthy weight (p = 0.04). The increase in food preparation skills among intervention children (T0:T1) was maintained at the 10-week follow-up (T2) (Table 3). An examination of individual food preparation skills among intervention participants during the intervention period (T0:T1) according to age group (4–5 yr old; 6–8 yr old; 9–10 yr old) demonstrated that the increases were significant only for children in the youngest age categories (4–5 yr old (set dinner table, peel soft foods, measure dry ingredients, measure liquid ingredients, cut soft foods, grease/spray pan); 6–8 yr old (set dinner table, pour beverages, measure dry ingredients, measure liquid ingredients, crack eggs, grate cheese, cut soft foods, wash hands); however, when examining change throughout the entire study period (i.e., baseline (T0) to follow-up (T2)), significant increases among the older age group (9–10 yr old) emerged (core/slice apple, grate cheese, peel potato/other vegetable, and use food thermometer) (Table 4).

In the exploratory dose analysis, attendance level was inversely related to post-intervention (T1) BMI z-score (p = 0.02; CI: − 0.07, − 0.01), systolic BP z-score (p = 0.03; CI: − 0.14, − 0.01), and TV viewing during meals (p = 0.02; CI: − 0.26, − 0.02), and positively related to food preparation skills (p < 0.001; CI: 0.23, 0.71) (Table 5).

Program dose, defined as mean (SD) attendance rate, was 6.98 (2.3) sessions or 70% of the sessions. The majority of children (68%, n = 55) attended 7–10 lessons, 28% (n = 24) attended 4–6 lessons, and 9% (n = 8) attended 1–3 lessons. Approximately 95% of lessons were delivered as intended and child participants were engaged in the program activities 96% of the time. At program completion (T1), 100% of participating caregivers reported their child (ren) enjoyed participating in the program.

The present study assessed the impact of a 10-week, multi-component family meals intervention study aimed at eliciting positive changes in child diet, weight status, and health among 4–10-year-old children from racially diverse families from an underserved community. It was hypothesized that diet, z-scores for BMI, waist circumference, and BP, weight status, food preparation skills, and family meal outcomes would improve more from baseline to post-intervention among children in the intervention versus waitlist control group. Results demonstrated an intervention impact on food preparation skills and certain family mealtime routines (i.e., TV viewing), but not diet or z-scores for BMI, waist circumference, or BP. However there was an impact on weight status such that intervention versus waitlist controls experienced a greater change toward healthy weight. This finding parallels results from the Korea National Health and Nutrition Examination Survey or KNHANES in which the odds of obesity was inversely associated with family meals among 2904 elementary-aged children [20], and indicate a true preventive effect of healthy family mealtime routines among children at greatest risk for obesity (i.e., racial minority children, particularly those from low-income households).

With regards to the intervention impact on food preparation skills, results parallel findings from the HOME randomized controlled trial, a childhood obesity prevention intervention aimed at increasing family meals among 8–10 year-old children, in which intervention versus control children demonstrated higher food preparation skills at post-intervention [22]. In addition, though the setting was different from the current study (school vs faith-based center), results parallel findings from a narrative review of culinary interventions for children (5 to 12 years old) in which consistent positive impacts on cooking skills were observed [51]. Given emerging evidence associating food preparation skills with better long-term health [52], this warrants further investigation. However, there is need for caution given that food preparation skills may not always guarantee a healthy diet [53‐55].

Among the family meal outcomes (weekly frequency of family meals (dinner and breakfast), TV viewing during meals, and having a family meal in a dining area), decreased TV viewing was the only consistent finding. Specifically, there was an intervention impact on TV viewing during meals (intervention participants), as well as a dose effect on TV viewing (intervention and waitlist control participants). The majority of evidence demonstrates an inverse association between TV viewing during meals and child diet quality [56]. On the other hand, the data demonstrating an association between TV during and child weight status are mixed demonstrating negative [57] or null [58, 59] effects. The mechanism is unclear, though may be due to decreased distractions and resulting improvements in family dynamics and sense of connectedness [59]. The absence of an intervention effect on family meal frequency goes against the study hypothesis, but may not be important as the number of times a family reports having a meal together doesn’t necessarily reflect the quality of the meal. It will be critical for future work in this line of inquiry to expand assessment methods to include a tool to assess the healthfulness of meals served to children [60].

The lack of an intervention impact on dietary targets (daily servings of fruit, vegetables, and sugar-sweetened beverages, and diet quality or HEI), contradicts the majority of evidence and [16‐19, 22] the theoretical underpinnings of the Simple Suppers intervention (i.e., Social Cognitive theory which purports that skills lead to behavior change) [36, 38]. However, it is possible that child dietary intake improvements did in fact occur, but changes were not detected given the low response rate for the 24-h dietary recall by caregivers. The low response rate, particularly at post-intervention (T1) and follow-up (T2), was due primarily to difficulty contacting participants (e.g., changed cellular service and number, changed address). This is not an uncommon barrier in working with economically disadvantaged populations, however may be overcome by greater reliance on the community partner with regards to contacting and monitoring participants [61] and therefore should be explored in future work. The low response rate might also have been due to the burdensome nature of dietary recalls; this is a common problem in collecting these data and therefore important for future efforts to identify ways to overcome this issue through alternative validated approaches that are less arduous and similarly accurate in terms of estimating intake (e.g., digital assessment through smartphone technology) [62‐64].

The lack of an intervention effect on BMI z-score finding is similar to results from the HOME [22] and HOME Plus randomized controlled trial [24] in which there were no significant treatment group differences in BMI z-scores at post-intervention. However, results from the HOME Plus trial did demonstrate that when accounting for pubertal status, prepubescent intervention children had significantly lower BMI z-scores vs prepubescent control group children. These findings suggest that younger children may be more responsive to a weight status change resulting from healthy family mealtime routines; however, this hypothesis does not align with findings from the Simple Suppers study which primarily included prepubescent children (4–10 year-old). Regardless, it will be important for future work in this line of inquiry to examine potential BMI z-score responders (e.g., prepubescent vs pubescent) in reference to healthy family mealtime routines.

A second hypothesis in the current study was that improvements in dietary intake and quality and z-scores for BMI, waist circumference, BP, food preparation skills, and family meals outcomes would be maintained during the follow-up period among intervention participants. A retention of food preparation skills at the 10-week follow up was demonstrated. The lasting change in food preparation skills is a powerful finding and may be explained by utilization of the intervention mapping protocol (i.e., selection of theoretical methods and practical strategies targeted to determinants of behavior) in the design of Simple Suppers [36, 37]. Specifically, this Social Cognitive Theory-based intervention was designed to build behavioral capabilities (i.e., skills) in food preparation by teaching children age-appropriate food preparation skills at each lesson (facilitation), having children learn these skills by being in groups of children their own age (vicarious learning), and practicing the skills at each weekly lesson and at home (mastery experience).

A further examination of individual food preparation skills among intervention participants according to age group (4–5 yr old; 6–8 yr old; 9–10 yr old) during the intervention period demonstrated that the increases were significant only for children in the youngest age categories (4–5 yr old; 6–8 yr old). These data support that young children are indeed capable of participating in meal preparation [65], and is significant in part because they are laying the foundation for future eating patterns [33]. It was also observed that there were significant increases among the older age group (9–10 yr old) over the entire study period, from baseline to the 10-week follow-up. It is possible that because the skills were more advanced, even age adjusted, it took more time to master them [38].

The exploratory dose analysis factoring in attendance level demonstrated that each additional lesson was associated with a lower BMI z-score. Importantly, results remained similar in terms of directionality and significance when examining intervention participants alone (data not shown). This finding bears similarity to results from some observational studies [16‐20] and experimental trials demonstrating a positive impact of family or shared meals on child weight status [10‐12, 22], thereby strengthening the evidence-base in favor of American Academy of Pediatrics’ recommendation to engage in healthy family mealtime routines as a childhood obesity prevention strategy [6]. On the other hand, it is possible that a selection bias occurred whereby those who attended more versus less were different in some unaccounted way (e.g., higher motivation, fewer logistic barriers such as transportation) [66]. It is also possible that the observed positive dose-response was due to increased contact and not the intervention itself. This possibility cannot be ruled out given that the study design did not include an attention control group [67].

One of the novel aspects of the current study was inclusion of BP. Hypertension in childhood is associated with obesity and early onset diabetes [24] Given the American Academy of Pediatrics’ recommendation to engage in family meals as an obesity prevention strategy, there is need to expand the understanding of the relationship between family meals and child weight status to critical biomarkers of health like BP. In the exploratory analysis, we demonstrated an inverse association between attendance level and systolic BP z-score. These findings are consistent with results from a cross-sectional study of 723 US adolescents residing in a Midwest state demonstrating associations of family dinner from fast-food restaurants and take-out sources with increased metabolic risk cluster z-scores (which included systolic BP) [68]. These findings also bear similarity to results from a 12-week gardening, nutrition, and cooking intervention in primarily Latino elementary-age children (n = 70 control and n = 34 intervention) whereby children in the intervention group had decreased diastolic BP [69]. Worth noting, there is strong evidence in favor of systolic versus diastolic BP being predictive of mortality [70]. Regardless, these data taken together point to the need to further explore the relationship of family mealtime routines and child BP.

Another novel feature of the current study was the delivery setting – i.e., faith-based community center – as few childhood obesity interventions have occurred in this space [25, 71]. Among the limited number of studies, results demonstrate that faith-based community centers are highly effective in engaging families in programming [71, 72], particularly for the African American community in which currently and historically the church is viewed as central to a community and daily living [73]. This may at least partially explain the strong feasibility (70% attendance), acceptability (100% satisfaction), and retention (87.5%) results. It is also possible that children’s enjoyment of the program spurred engagement. Specifically, based on the caregiver-completed child satisfaction survey, and a small sample of post-intervention exit interviews (n = 4; data not shown), children reported that what they enjoyed most about the program was preparing recipes (“oh, I liked when we made dessert – the yogurt parfaits” [Black female, 7 yr] and “um … they [the recipes] weren’t hard to make!” [Hispanic White female, 8 yr]. In the future, it will be important to understand the main contributing factors for the observed high feasibility and acceptability to assure successful scale up.

It will also be important to examine ways to further minimize programmatic costs in future scale-up studies. In the current study, $25 per family was budgeted for weekly programmatic expenses. While this amount ($25) is the same or less than as what it would cost for a family to eat out at a fast food or fast casual restaurant [74] and this expense decreased throughout the study (data not shown) due to improvements in the program staff’s ability to ‘grocery shop on a budget’, additional cost savings strategies such as assessing a nominal fee to participating families should be explored.

The present study had multiple strengths. First, the Simple Suppers intervention was designed utilizing the intervention mapping protocol, peer-reviewed by a team of experts in the field, and pilot tested for feasibility [36]. In addition, outcomes were assessed using reliable questionnaires and established protocols by trained researchers. In addition, this study was the first family meals intervention to expand health outcomes beyond weight status (i.e., BP). The setting of a faith-based community center was also novel, and facilitated engagement in programming. Additionally, the study population was broader than the populations in previous family meals research. The current study included greater racial and socioeconomic diversity, allowing for increased generalizability to minority and at-risk groups. The inclusion of all children in the family who met age criteria was also unique to the current study.

A major limitation of the current study was lack of randomization and thus potential introduction of selection bias. That said, between-group differences were assessed at baseline and differences (child race) were accounted for in all statistical models to help minimize any bias that may have been introduced. Recruiting at various (multiple) times throughout the year helped to reduce the chance of bias because much of the programming at the center was seasonal and therefore there was a regular influx of new families to the center (e.g., basketball in winter, soccer in spring). Regardless, the difficulty with randomization, inherent to community-based research, needs to be better understood and factored into future research [75]. In addition, the study was designed for low-income family households; identification of this specific target population was assured by recruiting from a community center whose mission is to serve families in economic and social need and is situated in a part of the City surrounded by this target population. However, data from the current study demonstrated recruitment of a mixed income population. That said, the measure of income status has imprecisions (i.e., reliance on federal food assistance programs). It will be important for future work to include a more sensitive or comprehensive assessment of income status [76]. Use of a self-report dietary assessment method (i.e., 24-h diet recall) represents another study limitation due to the well-documented issues of subject recording bias, subject selection bias, and motivation [77]. Additional issues arise in assessing diet in children, particularly those between the ages 4 and 10 years old, due to issues of literacy and writing skills, limited skills in recognizing foods, memory constraints, difficulty conceptualizing time, limited attention span, and inability of parents to give a detailed assessment of what their children eat when away from home or in the care of others [77]. There is consensus that the caregiver-assisted 24-h multiple pass recall conducted over at least a 3-day period (weekdays and weekend days) is the best method for overcoming these issues, and accurately estimating dietary intake for children in this specific age group (4 to 10 or 11 years old) [40]. Importantly, the current study relied on this methodology. However, in many cases only a single dietary recall was collected due to difficulty contacting participants, ultimately reducing the ability to capture a ‘picture’ reflective of usual or typical intake. A separate but related issue was the relatively high proportion of participants with no dietary recalls at post-test (T1) or follow-up (T2), also due to difficulty contacting participants. It is possible though not likely that this introduced a nonresponse bias as precautions were taken to protect against high respondent burden, judgmental comments by interviewers, or language/cultural barriers (i.e., extensive piloting and staff training) [78]. Regardless, it will be critical for future research in this line of inquiry (i.e, family meals interventions in underserved racially diverse children and caregivers occurring in the faith-based community setting) to develop strategies for overcoming barriers to achieving complete dietary data collection. A final limitation was the use of only one closed-ended question to assess program acceptability. A forced-choice response may have made participants feel obligated to answer “yes” due to social desirability bias or to please researchers. Future interventions would benefit from expanding the assessment of program acceptability via in-depth interviews to better understand what participants did or did not enjoy and allow them to expand upon their answers.

High feasibility of the Simple Suppers study was demonstrated, as well as improvements in children’s weight status, food preparation skills, and TV viewing during meals among intervention versus control participants, although no overall impacts on diet or z-scores for BMI, waist circumference, or BP. Increased food preparation skills and decreased TV viewing during meals were retained among intervention participants at study follow-up, and change in food preparation skills throughout the study period varied by age group. In addition, a dose response was observed on z-scores for BMI and systolic BP, as well as TV viewing during meals, indicating that the program did have an impact among participants who attended more versus less. To our knowledge, this is the first study of its kind in that it included racial and income diverse children and families. Collectively, findings from the current study indicate the importance of targeting populations at increased risk for childhood obesity.