Family meals are important for the health and well-being of adolescents. Past research examining family meal frequency has found that more frequent family meals are associated with positive outcomes in youth, including better dietary intake( Reference Woodruff and Hanning 1 – Reference MacFarlane, Crawford and Ball 14 ), lower prevalence of overweight/obesity( Reference Gillman, Rifas-Shiman and Frazier 6 , Reference MacFarlane, Crawford and Ball 14 – Reference Sen 17 ), fewer disordered eating behaviours( Reference Neumark-Sztainer 8 , Reference Eaton, Kann and Kinchen 18 – Reference Neumark-Sztainer, Wall and Story 30 ) and higher levels of psychological well-being( Reference Eisenberg, Olson and Neumark-Sztainer 31 , 32 ). This body of research has stimulated interest in family meals among health professionals and the public and has led to questions on what family meals look like, particularly with regard to the types of foods being served. Learning more about what is served at family meals, and parent and family factors that are correlated with healthier meals, can help in guiding interventions to improve the home food environment.
Knowledge about how the types of foods served at family meals differ by sociodemographic characteristics can help in guiding decisions regarding who is in greatest need of intervention. Learning more about how psychosocial factors, such as parent work–life stress, parent depressive symptoms and family functioning (e.g. communication, problem solving, connectedness), are correlated with foods served at family meals may help to elucidate the types of factors that need to be taken into account in developing interventions( Reference Sen 17 , Reference Hammons and Fiese 33 ). Finally, knowledge about how meal-specific variables, such as attitudes towards family meals, food purchasing barriers, and meal planning and preparation practices, are associated with the types of foods served at family meals can help guide specific intervention messages( Reference Woodruff and Hanning 1 , Reference Sen 17 ).
The present study builds upon the extant body of literature on family meals, which has primarily focused on frequency of family meals, by addressing the following research questions:
-
1. What types of foods are served at family dinners?
-
2. How are parent and family sociodemographic characteristics, psychosocial well-being and meal-specific variables associated with the type of food served at family dinners?
We focused on family dinners, rather than on other meals, because breakfast is likely to include different types of foods( Reference Cho, Dietrich and Brown 34 ) and lunch seems less likely to be eaten on a regular basis as a family meal given school schedules. Factors hypothesized to be associated with healthier family dinners include higher levels of income and education, not working full time, higher levels of psychosocial well-being, more positive attitudes regarding family meals, fewer food purchasing barriers, enjoyment of cooking, and greater involvement in meal planning and food preparation( Reference Woodruff and Hanning 1 , Reference Sen 17 , Reference Hammons and Fiese 33 ).
Methods
Study population and design
Data were drawn from Project F-EAT (Families and Eating and Activity among Teens), a cross-sectional, population-based study of parents of adolescents. Project F-EAT surveys were completed by 3709 parents of adolescents enrolled in EAT 2010 (Eating and Activity in Teens).
The EAT 2010 study population includes adolescents from twenty public middle schools and high schools in Minneapolis/St. Paul, MN, USA, which serve socio-economically and racially/ethnically diverse students. Adolescent participants in EAT 2010 reflect the diversity of enrolled schools; 71 % qualified for free or reduced-price lunch and 81 % were from ethnic/racial minority groups. Participants had a mean age of 14·4 (sd 2·0) years; 53·2 % were girls; 46·1 % were in middle school (6th–8th grades) and 53·9 % were in high school (9th–12th grades). Adolescents from selected required health, physical education and science classes at each school completed surveys during the 2009–2010 school year. Among adolescents who were at school on the days of survey administration, 96·3 % had parental consent and chose to participate.
Each adolescent was asked to provide contact information for up to two parents (or other caregivers, as relevant) for participation in Project F-EAT. Overall, the response rate among parents in Project F-EAT was high: at least one parent responded for 85·3 % (n 2382) of the adolescents and for 67·9 % (n 1327) of the adolescents who provided information on two parents, both parents responded. Parental response rates did not differ by adolescent gender, age, socio-economic status or language spoken at home but did differ by race/ethnicity, with the highest response rates among the parents of white adolescents.
The current analysis included only one parent (or other caregiver) per household (n 1923), to ensure independence of data. For families in which two parents responded, we selected the parent who reported that he/she usually prepares food for the family on the F-EAT survey. If neither parent reported food preparation, we excluded that household from the current analysis (n 226). If more than one parent reported that they usually prepare food (n 495), we gave priority based on relationship to adolescent (e.g. preference to parents over step-parents) and parent gender (preference to females). Further, because the aim of the current analysis was to examine types of foods served at family meals, parents who indicated that they never eat family dinner (and did not report on the types of foods typically served) were excluded (n 93). Finally, parents who reported making meals but did not live with the participant most of the time (n 39) were excluded. The final analytic sample for the current analysis included 1923 parents (87 % were mothers, 8 % fathers and 5 % step-parents or other parental figures); sample characteristics are shown in Table 1.
Table 1 Sociodemographic characteristics of the study population: primary parents who reported eating family meals (n 1923)Footnote *, Project F-EAT (Families and Eating and Activity in Teens), St. Paul, MN, USA, 2009–2010
GED, General Education Development test.
* Due to missing values, numbers do not always total 1923.
Data collection
Parents were initially mailed an invitation describing Project F-EAT and a telephone number to call if they preferred to complete the survey over the phone. A follow-up mailing included the Project F-EAT survey, a consent form and a postage-paid return envelope. Parents were mailed a reminder postcard after two weeks and a second copy of the survey if they did not respond within one month. Up to eight phone calls were made to contact non-responders and offer a telephone option for survey completion. Most (78 %) of the parent surveys were completed by mail. Both mailed surveys and phone interviews were available in English, Spanish, Hmong and Somali, and the phone interview was additionally offered in Oromo, Amharic and Karen languages. The majority of surveys (83 %) were completed in English. Data collection ran from October 2009 to October 2010 and was conducted by the Wilder Research Foundation in St. Paul, Minnesota. The University of Minnesota Institutional Review Board approved all study procedures.
Survey development
The Project F-EAT survey was designed to gather information on adolescents’ family and home environments with relevance to dietary intake, physical activity and weight-related health. Social cognitive theory( Reference Bandura 35 ) and an ecological framework( Reference Sallis, Owen and Fisher 36 ) guided the selection of constructs to be assessed, while specific survey items were drawn from a previous Project EAT parent survey, corresponding measures from the EAT 2010 student survey and existing surveys from the scientific literature. New questions were also developed to address the study aims (e.g. foods served at family meals).
Initially, content area experts reviewed a draft of the Project F-EAT survey to ensure that key constructs of relevance to adolescent weight-related behaviours and outcomes were included. Further, survey appropriateness for the major cultural groups participating in the study (i.e. Native American, Hmong, Latino, Somali and African-American groups) was addressed by having bi-cultural staff members from the Wilder Research Foundation review the survey and provide feedback. Next, three focus groups were conducted with twenty-eight parents (including both mothers and fathers) from diverse backgrounds to pre-test a draft of the survey. An additional sample of 102 parents completed the survey twice over two weeks, to examine test–retest reliability of survey questions. Finally, scale psychometric properties were examined within the full Project F-EAT sample.
Once a final version of the survey was developed in English, the written survey was professionally translated. Bilingual staff members at the Wilder Research Foundation reviewed the translated survey, edited translations when appropriate, and returned the updated surveys to the original translators for final review and approval.
Measures
Family meals
Types of foods served at family dinner. Types of foods served at family dinner were examined with six items designed to assess the frequency of foods served at family dinners, including: ‘green salad’, ‘vegetables other than potatoes’, ‘100 % fruit juice’, ‘fruit (not including juice)’, ‘milk’ and ‘sugar-sweetened beverages (soda pop, Kool-aid, etc.)’. Four response options ranged from ‘never’ to ‘always’ (individual item test–retest r values ranged from 0·56 to 0·85). An additional question was used to assess fast food: ‘During the past week, how many times was a family meal purchased from a fast-food restaurant and eaten together either at the restaurant or at home (pizza counts)?’. Four response options ranged from ‘never’ to ‘three or more times’ (test–retest r = 0·43). These items were adapted from our previous research regarding home food availability and instrument development for family meals research( Reference Fulkerson, Lytle and Story 37 , Reference Boutelle, Fulkerson and Neumark-Sztainer 38 ).
Overall healthfulness of foods served at dinner. Overall healthfulness of foods served at dinner was calculated as a summary score by adding points for each of the healthier foods (green salad, vegetables, 100 % fruit juice, fruit and milk) and subtracting points for each of the unhealthy foods (sugar-sweetened beverages and fast food). Points were added (for healthy foods) and subtracted (for unhealthy foods) based on the frequency of serving each item: 0 points for ‘never’; 1 for ‘sometimes’ (or ‘once a week’ for fast food); 2 for ‘usually’ (or ‘twice a week’ for fast food); 3 for ‘always’ (or ‘three or more times a week’ for fast food). The possible range for the summary score was −6 to 15 (test–retest r = 0·80). Negative values represent dinners that consist of greater unhealthy than healthy foods (for example, when salads, fruit, vegetables, fruit juice and milk are never served, but sugar-sweetened beverages and fast food are often served).
Sociodemographic characteristics
Educational attainment. Educational attainment was assessed with the question: ‘What is the highest grade or year of school that you have completed?’ (test–retest r = 0·84).
Household income level. Household income level was assessed with the question: ‘What was the total income of your household before taxes in the past year?’ (test–retest r = 0·94).
Race/ethnicity. Race/ethnicity was assessed with the question: ‘Do you think of yourself as (i) white, (ii) black or African-American, (iii) Hispanic or Latino, (iv) Asian-American, (v) Hawaiian or Pacific Islander, or (vi) American Indian or Native American?’ and respondents were asked to check all that apply (test–retest percentage agreement = 92 %). Due to small numbers, Hawaiian/Pacific Islanders and people who checked more than one ethnic/racial group were combined.
Work status. Work status was assessed with the question: ‘Which of the following best describes your current work situation?’. Response options included ‘working full-time’, ‘working part-time’, ‘stay-at-home caregiver’, ‘currently unemployed but actively seeking work’ and ‘not working for pay’ (test–retest r = 0·82).
Marital status. Marital status was assessed with the question ‘What is your current marital status?’ (test–retest percentage agreement = 98 %).
Number of children at home. Number of children at home was assessed with the question: ‘How many children (under the age of 18) live in your household?’ (test–retest r = 0·99).
Age. Finally, age was calculated using self-reported birth date and survey completion date.
Psychosocial variables
Work–life stress. Work–life stress was assessed using three items from previous research( Reference Marshall and Barnett 39 , Reference Marshall and Barnett 40 ): ‘Because of the requirements of my job, I miss out on home or family activities that I would prefer to participate in’; ‘Because of the requirements of my job, my family time is less enjoyable or more pressured’; and ‘Working leaves me with too little time or energy to be the kind of parent I want to be’. Four response options ranged from ‘strongly disagree’ to ‘strongly agree’. A ‘not employed’ option was also included for each statement (Cronbach's α = 0·86; test–retest r = 0·75). Categories of low, moderate and high stress were created such that approximately half of the sample was in the moderate stress group and a quarter in the high and low stress groups. This categorization allowed for the comparisons of the more extreme levels of work–life stress.
Parental depressive symptoms. Parental depressive symptoms were assessed with the five-item CES-Depression Scale: ‘I felt depressed’, ‘My sleep was restless’, ‘I felt lonely’, ‘I had crying spells’ and ‘I could not get going’( Reference Bohannon, Maljanian and Goethe 41 ). Four response options included ‘rarely or none of the time’, ‘some or a little of the time’, ‘occasionally/moderate amount of the time’ and ‘most or all of the time’ (Cronbach's α = 0·81; test–retest r = 0·76). For analyses, individuals who scored 5·5 or above were categorized as having depressive symptoms and those who scored under 5·5 were categorized as not having depressive symptoms, as described previously( Reference Bohannon, Maljanian and Goethe 41 ).
Family functioning. Family functioning was assessed with the following five items: ‘Family members are accepted for who they are’; ‘Making decisions is a problem for the family’; ‘We don't get along well together’; ‘We can express feelings to each other’; and ‘Planning family activities is difficult because we misunderstand each other’( Reference Epstein, Baldwin and Bishop 42 ). Response options ranged from ‘strongly disagree’ to ‘strongly agree’ (Cronbach's α = 0·72; test–retest r = 0·67). Higher scores were indicative of higher family functioning. Categories of low, moderate and high family functioning were created such that approximately half of the sample was in the moderate group and a quarter in the low and high functioning groups.
Meal-specific variables
Importance of family meals. Importance of family meals was assessed with a four-item scale: ‘It is important that our family eat at least one meal a day together’; ‘Different schedules make it hard to eat meals together on a regular basis’; ‘In our family, it is often difficult to find a time when family members can sit down to a meal together’; and ‘In our family, children are expected to be home for dinner’. Five response options ranged from ‘strongly disagree’ to ‘strongly agree’. These items were adapted from the Family Eating Attitude and Behavior Scale( 43 ) and used in Project EAT( Reference Neumark-Sztainer, Wall and Story 30 ). Categories of low, moderate and high importance were created such that approximately half the sample was in the middle group (Cronbach's α = 0·52; test–retest r = 0·72).
Food purchasing barriers. Food purchasing barriers were assessed with a four-item scale addressing cost, variety and quality of produce( Reference Campbell, Crawford and Salmon 44 ), with response options ranging from ‘strongly disagree’ to ‘strongly agree’ (Cronbach's α = 0·79; test–retest r = 0·59).
Enjoyment of cooking and meal planning. Enjoyment of cooking and meal planning were assessed with two separate items previously included in the Nepean Kids Growing Up Today Parent Questionnaire( Reference Hardy, Baur and Garnett 45 ): ‘I like trying new recipes’ and ‘I usually know or plan in the morning what we will eat for dinner that night’. Response options for both ranged from ‘strongly disagree’ to ‘strongly agree’ (test–retest r = 0·69 and 0·62, respectively). Responses were dichotomized into high and low for analyses.
Hours of food preparation. Hours of food preparation was asked with the questions: ‘How many hours per week do you normally spend preparing food for your family?’ and ‘How many hours per week does your spouse, partner or other adult in the household spend preparing food for your family?’ (test–retest r = 0·68)( Reference Demo and Acock 46 ). Responses to the two questions were summed.
Statistical analysis
Descriptive statistics were used to examine sample sociodemographic characteristics. A series of logistic regression models adjusted for age, race, education, income, gender, work status, marital status and number of children were conducted to assess associations between parent/family characteristics and types of foods served regularly (i.e. usually or always) at family dinner. Analyses were run with and without adjustment for sociodemographic variables, which allowed for the attainment of significance tests for the remaining variables independent of sociodemographic differences. The crude estimates are useful in understanding the prevalence of a variable of interest (i.e. food served at family meals), whereas adjusted analyses can help in understanding the reasons for any differences (e.g. if differences across ethnicity/race are likely due to economic differences). Furthermore, comparisons between crude and adjusted estimates allow for a check of statistical validity; a large discrepancy between the two numbers can often be a sign of high correlation or sparse data (as well as possible confounding). In the current study, analyses generally revealed similar findings, thus we only present adjusted analyses. Using the fitted model, the probability of the dependent variable (e.g. serving green salad) was calculated for each person in the data set at a specified exposure level (e.g. less than high school education), given the individual's observed value of each adjustment variable. The average of these predicted probabilities is the adjusted percentage (e.g. for serving green salads among those with less than high school education)( Reference Localio, Margolis and Berlin 47 ). This procedure was repeated for each outcome at each level of the exposure variable. Associations between personal/demographic factors and overall healthfulness of food served at family meals were assessed using likelihood ratio tests from the logistic models. Individual pairwise tests were performed between each ethnic/racial group to allow more detailed examination of this important categorical variable. Statistical significance was defined as P < 0·05. Analyses were conducted using the Stata version 10 statistical software package.
Results
Types of foods served at family dinner
Green salad was served at family dinner on a regular basis (usually or always) in 28·2 % of the homes (Table 2). In contrast, the majority of parents (70·1 %) reported serving vegetables (other than potatoes) on a regular basis. Fruit and 100 % fruit juice were served on a regular basis in about a third of the homes. Milk was served on a regular basis in about half of the homes. About one-fifth (22·0 %) of the parents reported usually or always serving sugar-sweetened beverages at family dinners. Similarly, 21·0 % of the parents reported having fast food at family meals two or more times per week.
Table 2 Frequency of foods served at family dinners, as reported by primary parents (n 1923)Footnote *, Project F-EAT (Families and Eating and Activity in Teens), St. Paul, MN, USA, 2009–2010
* Only parents who reported eating family dinners were included in the analytic sample, see Methods section for details.
Parent/family characteristics: associations with foods served at family dinner
Sociodemographic characteristics
Parent educational attainment showed strong inverse associations with overall healthfulness of foods served at family meals in analyses mutually adjusted for all sociodemographic characteristics (Table 3). Parents with lower levels of education were significantly less likely to serve vegetables (P = 0·002) and more likely to serve sugar-sweetened beverages (P = 0·013) and fast food (P = 0·029) on a regular basis at family meals. Ethnic/racial differences were found in the overall healthfulness of family meals and for specific foods. In general, whites and Asian Americans had the highest overall scores, although differences between the groups were not all statistically significant. Given the large ethnic/racial differences in serving milk at family meals, and both cultural and biological reasons for these practices, analyses for overall healthfulness scores were also run without including milk; patterns were found to be similar and are not presented here. Consistent associations were not found between work status and different types of foods served at family meals. The association between work status and the summary score for overall healthfulness of foods served at family meals bordered on statistical significance (P = 0·058), with the lowest scores reported by parents working full time.
Table 3 AdjustedFootnote * percentages of families serving specific foods at family dinner on a regular basis (i.e. usually or always)Footnote †, and overall mean score for healthfulness of foods served, by parent/family sociodemographics, psychosocial factors and meal-specific variables, for parents involved in food preparation who reported eating family dinner (n 1923), Project F-EAT (Families and Eating and Activity in Teens), St. Paul, MN, USA, 2009–2010
a,b,c,d Values within a column with unlike superscript letters were significantly different in pairwise tests (P < 0·05).
* All models are adjusted for age, race, education, income, gender, work status, marital status and number of children. The P value listed is from a likelihood ratio test examining an overall difference in proportions between the levels of a covariate.
† Or at least two times per week for fast food.
‡ Not including potatoes.
§ Summary score for all foods shown. Salad, vegetables, fruit juice, fruit and milk are coded as healthy and sugared drinks (i.e. sugar-sweetened beverages) and fast food as unhealthy (possible range: −6 to 15).
Psychosocial factors
Higher levels of work–life stress, the presence of depressive symptoms and low levels of family functioning were all strongly associated with lower healthfulness of family meals in adjusted analyses (P values ranged from P = 0·003 to P = 0·001; Table 3). Parents who reported higher levels of work–life stress were less likely to serve vegetables on a regular basis (P < 0·001) and more likely to serve fast food (P = 0·020). Parents reporting depressive symptoms were more likely to serve sugar-sweetened beverages and fast food (both P < 0·001). Finally, family functioning was positively associated with serving green salad (P = 0·022), vegetables, fruit (both P < 0·001) and milk (P = 0·012) on a regular basis and inversely associated with sugared drinks (P = 0·013).
Meal-specific variables
Perceiving family meals as important, enjoyment of cooking, meal planning and hours in food preparation were all positively associated with overall healthfulness of family meals (all P < 0·001), while reporting more food purchasing barriers was inversely associated with healthfulness scores (P < 0·001). Additionally, many associations were also seen between these variables and specific types of foods served at family meals, as shown in Table 3.
Discussion
Findings from the current study suggest that, in many homes, healthy foods such as vegetables and milk are commonly served at family meals. But findings also indicate that in one-fifth of households, sugar-sweetened beverages and fast food are served on a regular basis. Some differences in the overall healthfulness of foods served at family meals were found across sociodemographic characteristics, in particular across levels of parent education. However, psychosocial factors and meal-specific variables were more consistently associated with the types of foods served at meals. Psychosocial factors including high levels of work–life stress, reported depressive symptoms and low levels of family functioning were strongly associated with lower healthfulness of family meals. Finally, all of the meal-specific variables assessed were strongly correlated with healthfulness of foods served at family meals. Findings provide insight into the types of factors to be addressed within interventions aimed at improving the food environment of family meals.
Associations between various sociodemographic characteristics and types of foods served at family meals were examined in order to identify population groups that might benefit most from interventions. Some differences in the overall healthfulness of foods served at family dinners were found across sociodemographic characteristics; in particular, parent educational attainment was inversely associated with the healthfulness of foods served at family meals. Given that the association was strong even after mutual adjustment for other sociodemographic characteristics, this finding suggests both the importance of parent education for family health and a need for ensuring that interventions around family meals are suitable for, and reach out to, parents with lower levels of educational attainment. Some differences across ethnicity/race were also observed; for example, sugared drinks were served most often in the homes of African Americans and American Indians, while sugared drinks and fast food were served least often in the homes of Asian Americans. Moreover, previous research has shown that weekly purchases of fast food for family dinner is significantly and positively associated with mean percentage body fat, metabolic risk scores and insulin levels among adolescents( Reference Fulkerson, Farbakhsh and Lytle 48 ). These findings suggest a need for the tailoring of intervention messages for different ethnic/racial subgroups. Interestingly, associations between work status and the types of foods served at family meals were much weaker and less consistent than associations with work–life stress. In a previous examination we similarly found that the subjective experience of work–life stress was more strongly correlated with a number of measures of parent dietary intake and the home food environment than the actual work status of parents( Reference Bauer, Hearst and Escoto 49 ).
The strong associations found between psychosocial factors including work–life stress, depressive symptoms and family functioning, and overall healthfulness of food served at family meals, highlight the importance of addressing factors beyond those directly related to food (e.g. cooking skills) that may be serving as barriers to the preparation of healthy foods at family meals. Findings from the current study are in line with previous research findings indicating that stress from competing demands on parents’ time may influence family meals( Reference Fulkerson, Story and Neumark-Sztainer 50 – Reference Devine, Farrell and Blake 53 ). Past research has shown a relationship between parent depressive symptoms and decreased positive parenting behaviours (e.g. related to communication, emotional closeness and setting limits) and increased negative parenting behaviours (e.g. conflict, disengagement, inability to attend to daily routines)( Reference Lovejoy, Graczyk and O'Hare 54 , Reference Wilson and Durbin 55 ); thus it may be the case that parents struggle with maintaining daily routines such as serving healthy foods at family meals because they are challenged with managing depressive symptoms. Additionally, all of the meal-specific variables examined in the current study were strongly associated with the healthfulness of foods served at family meals. Meal planning was associated with all of the specific types of foods served at family meals; more planning was always correlated with healthier options (e.g. higher levels of vegetables and lower levels of fast food). Similarly, McIntosh and colleagues found that meal-specific variables, such as mothers’ belief in the importance of family meals and meal planning practices, are correlated with the frequency of family dinners( Reference McIntosh, Kubena and Tolle 56 ). These findings indicate a need for the development of interventions that are broad in scope, and either directly address or take into account psychosocial stressors facing families. Findings also suggest a need for interventions that provide specific skills related to meal planning and preparation.
Study strengths include the high response rate, the large and diverse nature of the study population, the assessment of different types of foods served at family dinner, and the inclusion of various measures of parent characteristics (e.g. work–life stress, depressive symptoms). However, the study also had limitations that should be taken into account in interpreting the findings. Many of the variables were assessed with brief tools, in order to reduce participant burden and enhance parent response. For example, a short scale was used to assess depressive symptoms, some variables were measured with only one item (e.g. enjoyment of cooking), and the measure of types of foods served at family meals only assessed seven fairly broad categories of foods. Thus, conclusions from the current study should be made cautiously; for example, some of the fast food options served at family meals may have been healthy choices. Further replication studies that utilize more comprehensive measures are needed. Furthermore, all items were assessed via self-report, and assuming some degree of social desirability, it may be that the healthier food options are served less frequently and the poorer food options served more frequently than described here. Finally, the study included cross-sectional data and thus conclusions about the temporality of associations cannot be drawn.
The current study expands upon the existing research on family meals, which has primarily focused on the frequency of family meals and associations with child and adolescent outcomes. Little research has focused on the types of foods served at family meals and most existing studies have included very limited measures of the types of foods being served( Reference Vejrup, Lien and Klepp 11 , Reference MacFarlane, Crawford and Ball 14 , Reference Boutelle, Fulkerson and Neumark-Sztainer 38 , Reference Jacobs and Fiese 57 ). In an informative study on family meals among adolescents in Australia( Reference Gallegos, Dziurawiec and Fozdar 58 ), adolescents were asked to describe what they ate the previous day at dinner; this strategy offers a simple solution to learning more about the types of foods consumed at meals, although quantifying results can be difficult from such an open-ended question. Further research is needed to better understand the types of foods being served at family meals using more comprehensive measures and objective assessments such as unobtrusive observations of family meals. Research should also explore associations between the types of foods served at family meals and overall dietary intake and health outcomes (e.g. weight status), since different foods served at family meals may be consumed by different family members, and given that family members eat in various settings throughout the day (e.g. school, work). It is also important to learn more about those families who never eat family dinners together. In the current study, parents who reported they never ate family dinners did not differ from other parents in terms of age, employment status, marital status or number of children, but did have lower levels of educational attainment and income, and were more likely to be white and male. More information is needed to guide the development of interventions for these families. Finally, research is also needed to explore how to improve the dietary quality of family meals, taking into account the realities faced by modern-day families, particularly those with fewer resources (e.g. lower educational background) and greater stresses (e.g. due to challenges faced in balancing work–life stresses).
Findings from the current study suggest a need for interventions aimed at improving the healthfulness of foods served at family meals. Dietitians and other health-care providers working with parents should take the time to explore the types of foods being served at family meals, as well as barriers to eating together as a family and serving healthful meals. Given the high percentage of fast foods being eaten at family meals, it is important to discuss healthier options for food choices at fast-food restaurants (e.g. pizza with vegetables instead of sausage, smaller portion sizes) and healthier alternatives to fast food that can be made quickly at home (e.g. foods that can be made quickly such as bagged and prewashed salads, sandwiches made at home with wholegrain bread). Community interventions and public policies are also important in order to ensure that affordable, healthful food options are readily available and help families work healthful family meals into their routines( Reference Story and Neumark-Sztainer 59 , Reference Fiese and Schwartz 60 ). Interventions should be appropriate for parents with low levels of education, address different eating patterns across ethnic/racial groups, and take into account the busy lifestyles and other stressors facing parents. Finally, findings suggest the importance of enhancing parental attitudes toward family meals and providing specific skills to make it easier to plan and prepare healthful family meals.
Acknowledgements
Sources of funding: The project described was supported by Award Number R01HL093247 from the National Heart, Lung, and Blood Institute. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Heart, Lung, and Blood Institute or the National Institutes of Health. Conflicts of interest: The authors certify that they have no conflicts of interest. Authors’ contributions: D.N.S. is the Principal Investigator on the study and wrote the manuscript. R.M. conducted data analysis. K.L., J.A.F., M.E.E. and J.B. were involved in study implementation and contributed to the interpretation of results and manuscript revisions.
