‘Emerging adulthood’, typically defined as ages 18–25 years, is a unique developmental period when young people’s independence and autonomy are increasing. It is also a high-risk time for adverse health behaviours(Reference Nelson, Story, Larson, Neumark-Sztainer and Lytle1, Reference Park, Paul Mulye, Adams, Brindis and Irwin2). Previous research has highlighted the need for health promotion strategies for this age group, such as tobacco cessation and other efforts aimed at promoting healthy lifestyles(Reference Green, McCausland, Xiao, Duke, Vallone and Healton3). However, scant attention has focused on nutrition and healthy eating promotion for young adults. These emerging adult years may be a particularly important time to focus on nutrition-related issues, especially given the rapid increases in obesity prevalence at this age(Reference Gordon-Larsen, Adair, Nelson and Popkin4), as well as the impact that excess weight gain in early adulthood may have on long-term health(Reference Norman, Bild, Lewis, Liu and West5, Reference Carnethon, Loria, Hill, Sidney, Savage and Liu6). The transition to adulthood is also a key developmental age when long-term weight behaviour patterns may be established(Reference Nelson, Story, Larson, Neumark-Sztainer and Lytle1). Although some existing evidence suggests that adverse shifts in diet and weight-related factors likely occur between adolescence and adulthood, few studies to date have described these changes in detail(Reference Nelson, Story, Larson, Neumark-Sztainer and Lytle1).
Furthermore, while a limited number of studies have examined differences in dietary intakes among US college students by living situation (i.e. living either on or off campus)(Reference Brown, Dresen and Eggett7, Reference Brunt and Rhee8), to our knowledge no studies to date have examined these differences within a more generalizable young adult population, including those who represent a range of lifestyle characteristics (i.e. those who attend college, as well as young adults who are not college students) and living situations (those living on their own, with parents and/or on college campuses). Previous research among other age groups, such as children and adolescents, has suggested that living situations and various facets of the home environment may have an important impact on meal patterning and dietary intake and may be important factors to consider in designing health promotion strategies(Reference Fulkerson, Nelson, Lytle, Moe, Heitzler and Pasch9–Reference Neumark-Sztainer, Hannan, Story, Croll and Perry12). In order to build current knowledge and inform the possible development of nutrition interventions for young adults, there is a substantial need for research that explores how dietary patterns may vary across different living situations of young adults and how home environment characteristics may influence dietary intake among this age group.
Thus, the objective of the present work was to examine cross-sectional differences in dietary intake, meal patterns and home food availability among emerging adults by living situation (i.e. living in a rented apartment or house, living with parents or living on a college campus) using data drawn from a large, diverse, population-based cohort of youths. The research assessed: (i) the prevalence of diet-related factors by living situation; as well as (ii) disparities in dietary factors independent of known sociodemographic differences (e.g. according to socio-economic status (SES) and race/ethnicity) in diet.
Methods
Data for the present cross-sectional analysis were drawn from Project EAT (Eating Among Teens)-II, a population-based follow-up study(Reference Neumark-Sztainer, Wall, Guo, Story, Haines and Eisenberg13). At baseline (1998–9), 3074 young people were surveyed in Minnesota high schools (mean age 15·8 years). Five years later (2003–4), participants were mailed a follow-up survey and FFQ. The sample for the current study consisted of 750 males and 937 females who completed both mailed assessments (mean age 20·5 years, range 18–23 years). A majority of the cohort completed the 5-year follow-up survey several years after graduating from high school. Study protocols were approved by the University of Minnesota Institutional Review Board. Additional study design details are reported elsewhere(Reference Neumark-Sztainer, Wall, Guo, Story, Haines and Eisenberg13, Reference Neumark-Sztainer, Story, Hannan and Croll14).
Measures
Living arrangement
Living arrangement was self-reported on the Project EAT-II survey. Living arrangement was defined as where the participant lived for a majority of the time in the past year, including: rented apartment or house, parent’s home or on campus. Campus options included residence hall and fraternity/sorority; these were collapsed due to the relatively low prevalence of students living in fraternity/sorority housing. Participants indicating other types of living situations, such as ‘own a house’ (n 48) or ‘other’ (n 49), were not included in these analyses due to small sample sizes.
Dietary intake
The Youth and Adolescent FFQ was used to assess usual past year intake of a wide range of foods and nutrients(Reference Rockett, Breitenbach, Frazier, Witschi, Wolf, Field and Colditz15–Reference Perks, Roemmich, Sandow-Pajewski, Clark, Thomas, Weltman, Patrie and Roqol17). Dietary intakes (servings/d) of fruits, all vegetables (excluding French fries), dark green and orange vegetables only, Ca-rich foods, whole grains (excluding chips) and soft drinks were measured. Specific foods and beverages included in each category have been described elsewhere(Reference Neumark-Sztainer, Story, Hannan and Croll14, Reference Larson, Neumark-Sztainer, Hannan and Story18). Ca intake, total energy and the percentages of total energy intake from total fat, saturated fat and trans fat were also assessed. Research describing the reliability and validity testing of this questionnaire has been detailed extensively in previous publications(Reference Rockett, Breitenbach, Frazier, Witschi, Wolf, Field and Colditz15–Reference Perks, Roemmich, Sandow-Pajewski, Clark, Thomas, Weltman, Patrie and Roqol17).
Eating patterns
Meal frequencies were assessed using the EAT-II survey. Participants were asked separate items for: how often they ate breakfast, lunch and dinner during the past week (response categories included never, 1–2 d, 3–4 d, 5–6 d and every day); how often they ate at a fast-food restaurant (like McDonald’s, Burger King, Hardee’s, etc.) in the past week (response categories included never, 1–2 times, 3–4 times, 5–6 times, 7 times and >7 times); and how often they snacked (ate between meals) yesterday (response categories included none, 1 time, 2–3 times, 4–5 times and >5 times). Test–retest reliability for these items was evaluated at baseline (1998–9) of Project EAT (breakfast, r = 0·77; lunch, r = 0·71; dinner, r = 0·72; fast food, r = 0·46; snacking, r = 0·47).
Home food availability
Participants were also asked to report the availability of healthy and unhealthy foods and drinks where they live. Response categories ranged from 1 (never available) to 4 (always available). A score for home food availability of healthy foods (range 5–20) was created by summing responses to five items that assessed the availability of: (i) fruits and vegetables at home; (ii) vegetables at dinner; (iii) fruit juice at home; (iv) milk at dinner; and (v) dark bread at home. A score for unhealthy food (range 4–16) was created by summing the home availability of four items: (i) ‘junk food’; (ii) potato chips or other salty snack foods; (iii) chocolate or other candy; and (iv) soda pop. For both scales, higher numbers indicated higher availability of the food items assessed. These scales have been used in previous work and found to be predictive of young adults’ food choices(Reference French, Story, Neumark-Sztainer, Fulkerson and Hannan19–Reference Larson, Neumark-Sztainer, Harnack, Wall, Story and Eisenberg21).
Sociodemographic characteristics
Gender, age, parental status, race/ethnicity, SES and student status were self-reported on the Project EAT-II survey. Parental status was defined as having at least one child (yes/no), including stepchildren and adopted children. SES was based primarily on parental educational level, defined by the higher level of either parent at baseline(Reference Neumark-Sztainer, Story, Hannan and Croll14). Student status was defined as: not a student, student at a community or technical college (including full-time and part-time 2-year college students) or student at a 4-year college.
Statistical analyses
Descriptive statistics were calculated to examine demographic and other characteristics of young adults by living situation. Given previous evidence documenting differences in diet-related factors between males and females at this age(Reference Larson, Neumark-Sztainer, Story, Wall, Harnack and Eisenberg22–Reference Davy, Benes and Driskell24), gender-stratified linear regression models were used to examine differences in meal frequency, dietary intake and home food availability outcomes according to living situation. Regression models were first examined unadjusted. Then a second model was examined adjusting for demographic variables (race/ethnicity, SES, age, parental status). Given our a priori hypothesis that sociodemographic differences exist across living situations (e.g. with lower SES groups and racial/ethnic minorities being more likely to live with their parents after high school), the purpose of these adjusted analyses was to assess whether or not there are disparities in dietary intake by living situation independent of these hypothesized differences in sociodemographic factors. These findings provide a valuable contrast to those of the unadjusted analyses, which serve primarily to identify which young adult subgroups may be most at risk and thus may be the most effective targets for intervention strategies. A 95 % confidence level was used to interpret the statistical significance of probability tests. Whenever the outcome variable exhibited positive skewness, testing was carried out under the square root transformation.
Participants were excluded from analyses if they did not complete the Project EAT-II survey (n 10) or the Youth and Adolescent FFQ at follow-up (n 5) or if they reported a biologically implausible level of energy intake (n 8); a plausible range of 1·7–29·3 MJ (400–7000 kcal) was selected a priori. Analyses were weighted to adjust for differential response rates to Project EAT-II using the response propensity method(Reference Little25), described in detail in a previous publication(Reference Neumark-Sztainer, Wall, Guo, Story, Haines and Eisenberg13). Analyses were conducted using the SAS statistical software package version 8·2 (SAS Institute, Cary, NC, USA).
Results
Significant differences (P < 0·01) by living situation were observed for gender, SES, race/ethnicity, age and student status (Table 1). Males were more likely to be living with their parents, whereas a greater proportion of females were living in rented homes/apartments and/or living on campus. Lower SES and racial/ethnic minority groups were also more likely to be living with parents. Young adults living with their parents were more likely to be non-students or students attending a 2-year community or technical college.
Table 1 Demographic and other characteristics of young adults by living situation: Project EAT (Eating Among Teens)-II, 2003–4
Mean estimates are unadjusted. Although the final sample size is 1687, sample sizes of individual analyses vary slightly due to a small degree of missing data.
Unadjusted analyses among young adult males and females indicated that across all types of living situations, dietary intake was generally not in line with the Dietary Guidelines for Americans(26) (see Tables 2 and 3). For both males and females, meal patterns differed by living situation, with those living on their own or with parents reporting eating meals (particularly breakfast and dinner) less frequently compared with those living on campus. Fast-food intake also differed by living situation, with the lowest frequencies being reported by those living on campus (P = 0·02 for males, P < 0·001 for females).
Table 2 Mean meal frequencies, dietary intake and home food availability of young adult males by living situation: Project EAT (Eating Among Teens)-II, 2003–4
Mean estimates are unadjusted. Although the final sample size is 705, sample sizes of individual analyses vary slightly due to a small degree of missing data.
a,b,cMean values with unlike superscript letters were significantly different in the unadjusted model (P < 0·05). Only presented for unadjusted models in which living situation was statistically significant.
*Adjusted for race/ethnicity, socio-economic status, age and parental status (i.e. if participant has children).
Table 3 Mean meal frequencies, dietary intake and home food availability of young adult females by living situation: Project EAT (Eating Among Teens)-II, 2003–4
Mean estimates are unadjusted. Although the final sample size is 878, sample sizes of individual analyses vary slightly due to a small degree of missing data.
a,b,cMean values with unlike superscript letters were significantly different in the unadjusted model (P < 0·05). Only presented for unadjusted models in which living situation was statistically significant.
*Adjusted for race/ethnicity, socio-economic status, age and parental status (i.e. if participant has children).
Overall, young adults living with their parents and/or living in a rented apartment or house appeared to have poorer dietary intake than those living on campus. After adjusting for sociodemographic factors, females living on campus reported the highest intakes of fruit (P = 0·003), vegetables (P = 0·007), dark vegetables (P < 0·001), Ca (P = 0·02) and whole grains (P < 0·001), as well as the lowest percentage of energy from fat (P < 0·001) and saturated fat (P < 0·001). Results among males were observed in similar directions, although these differences were not statistically significant after controlling for sociodemographic factors. Among females, findings showed that on average females living with parents consumed more fast food and a greater percentage of energy from fat, as well as fewer whole grains and vegetables, compared with females living on their own in rented apartments and houses.
The availability of both healthy and unhealthy foods at home also differed by living situation among males and females. Adjusted analyses indicated that males and females living on campus reported the greatest availability of healthy foods where they lived (P < 0·001), as well as the greatest availability of unhealthy foods (statistically significant among females only, P = 0·01; P = 0·16 for males).
Overall, despite adjusting for sociodemographic covariates in our models, where differences across living situation were shown to exist (Table 1), a majority of the differences detected in unadjusted models remained significant (sixteen of the twenty-two crude associations that were significant, across both genders). Adjusting for sociodemographic factors appeared to result in greater attenuation of estimates among males compared with females.
Discussion
The findings from the present study showed that young adults living with their parents and those living on their own (i.e. in rented apartments or houses) exhibited the poorest dietary intake. These patterns were evident even after adjusting for sociodemographic characteristics such as race/ethnicity, SES and age. In contrast, young adults living on college and university campuses reported the most favourable diet and meal patterns. Since a large proportion of students living on campus may participate in some type of structured meal plan or other food service option, it is likely that this consistent, institutional meal provision could play an important role in providing access to an abundant array of healthy foods(Reference Brown, Dresen and Eggett7). In general, post-secondary institutional food service is common on college and university campuses in the USA, but may vary substantially across other international settings. Our findings suggest that without college food service providing consistent and readily accessible meals, dietary quality among US young adults might be substantially lower.
Our findings are consistent with previous research illustrating poor dietary intake among young adults who have begun to live on their own and are responsible for their own meal provision and food acquisition for the first time in their lives(Reference Brown, Dresen and Eggett7, Reference Brunt and Rhee8, Reference Brevard and Ricketts27). Previous research from Project EAT has indicated that many young adults at this age are ill-prepared for home meal preparation and lack the skills and/or the resources to regularly prepare food at home(Reference Larson, Story, Eisenberg and Neumark-Sztainer28), and have a high propensity for ‘eating on the run’ rather than eating in more traditional meal settings. These factors have been associated with poor diet quality among young adults(Reference Larson, Story, Eisenberg and Neumark-Sztainer28) and are likely linked to characteristics of one’s living situation.
Our findings may not only have important implications for young adults moving out on their own for the first time, but also for young adults continuing to live with their parents after high school. For example, previous research in the USA(Reference Brown, Dresen and Eggett7, Reference Brunt and Rhee8, Reference Brevard and Ricketts27) and in other countries(Reference Suzuki, Murashima and Hoerr29–Reference Kremmyda, Papadaki, Hondros, Kapsokefalou and Scott31) suggests that young adults either living on post-secondary campuses or continuing to live with their parents may experience benefits in terms of both nutrition and weight status compared with those young adults who are living independently. However, a major limitation of these previous studies is their focus on traditional college students and their failure to capture a young adult population representing a range of lifestyle characteristics. In contrast, the findings presented in the current paper are drawn from a population-based sample of young adults, including both those who attend college and those who do not. Results from our young adult cohort suggest that those individuals continuing to live with their parents may not be exhibiting as favourable diet-related profiles as previous research might imply(Reference Suzuki, Murashima and Hoerr29, Reference Papadaki, Hondros, Scott and Kapsokefalou30). These findings may reflect heightened levels of away-from-home food consumption for those living in their parents’ homes, possibly due to greater scheduling demands from work and/or school, or greater discretionary income due to living at home, although the factors underlying these differences are not clear. Baseline disparities in dietary intake among families may also account for these trends of poor diet quality among young adults living with their parents, even after adjusting for SES and racial/ethnic differences. Future longitudinal research exploring these issues is needed.
Given that a growing number of young adults live with their parents after graduating from high school(Reference Matsudaira32), it is important to understand how family dynamics and meal-related influences change during the pre-high school to post-high school years. Although previous research has suggested that living with one’s parents after high school may limit some types of young adult risk behaviours (e.g. alcohol and drug use)(Reference White, McMorris, Catalano, Fleming, Haggerty and Abbott33), other risk behaviours – such as poor dietary patterns – may increase, particularly as young adults adopt busier schedules and begin to turn to convenience and fast foods as a replacement for home meal preparation. While previous research among college students has suggested that young adults still living at home with their parents after high school may be engaging in healthy eating behaviours due to home- and family-related factors (e.g. less responsibility for purchasing food for themselves and/or preparing meals), our data do not support this assertion.
To our knowledge, this is the first study of its kind to compare dietary factors across living situations drawing from such a diverse, young adult cohort. An additional strength of the present study includes the collection of data on a broad array of meal- and diet-related factors. Despite these strengths, our findings should be interpreted with several caveats in mind. For example, our cross-sectional sample was drawn from a large, longitudinal cohort study consisting of individuals who had been high school students in one Midwestern metropolitan area in 1998–9, which may limit generalizability. Furthermore, our self-reported dietary data were collected using an FFQ. While validated food frequency tools are a well-recognized method for dietary assessment in large-scale studies, it can be difficult to accurately estimate intake frequency and serving sizes with a limited number of response options. In general, the nature of the self-reported measures used here, as well as the cross-sectional design, may also limit the conclusions one can draw from these results. Finally, although our analyses adjusted for important sociodemographic factors such as SES, it is possible that these factors have not been fully captured using the covariates available in our data and that residual confounding may be resulting in some of the differences observed between living situations in adjusted models.
In conclusion, our findings indicate that those young adults living on their own and with their parents report less frequent meals and poorer dietary intakes compared with those living on campus. Despite the disparities in diet by living situation, it is important to recognize that overall few young people at this age are consuming optimal diets. Although students living on campus reported the best overall dietary intake, they were, on average, far from meeting national health recommendations (e.g. consuming only 1·6–2·0 servings of vegetables and 1·0 serving of whole grains daily)(26, 34). While college food service and meal plans may be an important environmental influence that effectively improves the quality of dietary intake among many young adults living on post-secondary campuses, these findings overall suggest that effective health promotion efforts are needed for all young adults across a variety of living situations.
Little research to date has sought to understand the lifestyle characteristics of emerging adults. Additional work is needed to explore the modifiable determinants, environmental influences and broader lifestyle characteristics influencing dietary intake among this age group, particularly among those who are still living with their parents and those who have recently begun to live independently. Such work will inform the development of intervention strategies promoting healthy dietary intake and overall wellness during this important age in a wide range of settings. Ultimately, as young adults begin to lead independent lives outside the previous constraints of their family home, the health behaviours followed during this period may set the stage for establishing long-term behaviours that have an important impact on lifetime disease risk. Given that poor dietary intake is one of the leading modifiable contributors to mortality and that behaviours during young adulthood may have long-term impacts on chronic disease risk, nutrition promotion among emerging adults is an important focal area for future research(Reference Mokdad, Marks, Stroup and Gerberding35).
Acknowledgements
Project EAT was supported by Grant R40 MC 00319 from the Maternal and Child Health Bureau (Title V, Social Security Act), Health Resources and Services Administration, Department of Health and Human Services. Additional salary support for the analysis of these data was also provided by Award Number K07CA126837 from the National Cancer Institute. The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the National Institutes of Health. There are no conflicts of interest. All authors were responsible for interpreting results and revising the manuscript. M.C.N. wrote the manuscript and developed the analysis plan. N.I.L. conducted the statistical analysis. D.N.-S. and M.S. designed the Project EAT study and were responsible for data collection.
