5-2-1-0 Lifestyle risk factors predict obesity in Millennials

Abstract

The MENU GenY study is a web-based intervention trial aimed at improving food choices in those aged 21–30 years. We report baseline levels of the 5–2-1–0 healthy-lifestyle patterns to predict a body mass index ≥ 30 vs. < 30 kg ^. m^-2. Overall 1674 young adults (69% female) from 2 large health systems enrolled and completed an online survey asking questions about lifestyle habits. A multivariable binary logistic regression model was utilized to predict a body mass index ≥ 30 while controlling for known predictors of obesity. Consuming > 3 daily servings of fruits/vegetables (odds ratio = 0.90, 95% CI 0.81, 0.99), and reporting > 2.5 hours/week of vigorous physical activity (odds ratio = 0.93, 95% CI = 0.89–0.96, P < 0.001) was associated with a body mass index < 30. Conversely, time sitting (odds ratio = 1.07, 95% 1.04, 1.11) and consuming sugar sweetened beverages (odds ratio = 1.08, 95% CI = 1.00, 1.15) were related to a body mass index ≥ 30. In this cohort of 20–30 year olds, we observed a consistent relationship between obesity and the 5–2-1–0 healthy-lifestyle patterns previously reported among children and adolescents.

Trial Registration:

ClinicalTrials.gov: Encouraging Young Adults to Make Effective Nutrition Choices (MENU GenY), NCT01979809, https://clinicaltrials.gov/ct2/show/NCT01979809

INTRODUCTION

Obesity trends have risen in the United States since the 1970s for both adults and adolescents, with current prevalence rates of 37.7% and 17%, respectively (1, 2). Parallel to the rise in obesity are associated health conditions such as diabetes, hypertension, and heart disease (3–5). More alarming is the earlier onset of some of these diseases, leading some to predict that the life expectancy for current young adults will actually be less than their parents’ generation (6).

Young adulthood (i.e., age 20–30 years) is a critical time during which lifestyle patterns (e.g., nutrition, physical activity [PA], and sitting) begin to emerge, supporting a trajectory towards either a healthy adulthood or one hindered by chronic health problems (7). Unfortunately, studies in this age group (often referred to as “Millennials”) reveal an average 30 pound weight increase between the ages 18–35, which may be at least partially due to the above lifestyles (8–10). This period presents an opportunity to both learn and intervene, as young adults transition away from parental influence and become more autonomous with these crucial healthy lifestyle decisions (11).

The Making Effective Nutritional Choices Generation Y (MENU GenY) study was a web-based intervention trial that evaluated targeted behavior change information aimed at improving food choices, specifically increasing the amount of daily fruits and vegetables and reducing sugary drinks within a range of lifestyle and psychosocial measures. Here we report data measured at baseline relative to various health related behaviors in young adults. Central to this analysis, we focus on the 4 common correlates of obesity used in the nationwide initiative “Let’s go 5–2-1–0” from the Maine Youth Health Collaborative for children and adolescents. The 5–2-1–0 mnemonic stands for known behavioral correlates for reducing childhood obesity; ≥ 5 daily servings of fruits and vegetables, < 2 hours per day of screen time, ≥ 1 hour per day of exercise, and 0 sugar sweetened beverages (12, 13).

The purpose of this paper is to report the relationship between the 5–2-1–0 lifestyle habits and obesity (i.e., body mass index [BMI] ≥30 kg ^. m⁻²) in a cohort of young adults enrolled in the MENU GenY trial. Furthermore, we also examined the relationship between reported exercise intensity (i.e., vigorous vs. moderate) and the prevalence of obesity in this population. We hypothesize that these same correlates found in children and adolescents will still apply to millennials in their 20s.

MATERIALS AND METHODS

Subjects

The MENU-Gen-Y trial (NCT01979809) was a randomized internet-based, tailored intervention study to promote the intake of fruits and vegetables and improve dietary choices, specifically targeted for young adults aged 21–30 years old. Participants were recruited from 2 integrated health care delivery systems, Henry Ford Health System in Detroit, Michigan, and Geisinger Health System in north-central Pennsylvania. Website development and enrollment was performed in collaboration with the University of Michigan’s Center for Health Communications Research. The institutional review boards from all 3 participating institutions approved this study prior to data collection.

Between February 2014 and March 2015, recruitment letters, stratified by gender and age, were sent out to randomly selected, age eligible pool of young adults. Each letter contained study-related information, the study URL, incentive structure, a unique login ID, encouragement to read more about the study online, and a phone number of the site contact person. After 10 days, the same letter stamped with “2nd Notice” was mailed as a follow-up to those not accessing the study website. Interested individuals were instructed to go to the study website and complete enrollment, which included consent, contact information and a 139-question baseline questionnaire, for which they would receive $20 (Figure 1).

Figure 1.

CONSORT flow diagram. Abbreviations: HFHS, Henry Ford Health System; GEIS, Geisinger Health System; PA, physical activity.

Measurements

PA was assessed using the International Physical Activity Questionnaire, a validated PA instrument containing questions regarding the amounts of VPA, moderate intensity physical activity (MPA), and walking habits (2). The International Physical Activity Questionnaire also measures sedentary time with a validated question about hours of daily sitting. This question served as a surrogate for screen time (e.g., television and computer).

BMI was calculated from self-reported height and weight data collected in the baseline survey. Although known bias exists with regards to over reporting height and under reporting weight, evidence shows that self-report methods of BMI can differentiate between BMI categories (14, 15).

Sweetened beverages were measured using the Dietary Screener Questionnaire, developed for the National Health and Nutrition Examination Survey (16). The total amount of sweetened beverages was a composite of 3 questions that assessed various beverage items that contain sugar consumed over the past month (i.e., pop, flavored coffee, energy drinks, and fruit drinks).

Daily fruit and vegetable intake was measured using a 2-item screener (17). To assess fruit servings, participants were asked “On a typical day, how many servings of fruit do you eat? (A serving size equals 1 piece of fruit or melon wedge, 3/4 cup of 100% juice, 1/2 cup canned fruit, or 1/4 cup dried fruit).” Similarly vegetable consumption was assessed asking subjects “On a typical day, how many servings of vegetables do you eat? (A serving size equals 1/2 cup chopped raw or cooked vegetables, 1 cup leafy raw vegetables, or 3/4 cup of 100% vegetable juice).” Individuals who reported consuming 5 or more total servings of fruits and vegetables in the eligibility screening survey were not eligible to enroll in the MENU GenY study.

Statistical Analysis

For continuous variables, such as the number of fruit or vegetable servings, equal variance and normality assumptions were tested and independent samples t-tests were used. For categorical variables, such as race or marital status, minimum expected count assumptions were tested and chi-square tests were used. Individual logistic modeling was used to predict a BMI level ≥ 30 kg ^. m⁻² for each 5–2-1–0 variable of interest while controlling for recruitment site and known correlates of obesity (i.e., average sleep per day, race/ethnicity, marital status, educational level, and day or evening work schedule). Then a full multivariable logistic model was built to determine which factors independently predicted a BMI ≥ 30 kg ^. m^-2. To examine the relationship between the variables of interest, Spearman correlation coefficients were used. All analyses were performed using R3.2.2 (R Foundation, Vienna, Austria) with the level of significance set at P ≤ 0.05.

RESULTS

In total, 1674 participants enrolled in the MENU GenY study, of which 1542 answered the International Physical Activity Questionnaire and were subsequently included in these analysis (Figure 1). Of this cohort (78% were non-Hispanic white, 69% were women, and 55% had a bachelor’s degree or higher), 29% had a BMI ≥ 30 kg ^. m^-2. Sociodemographic characteristics are presented in Table 1.

Table 1.

Descriptive baseline self-report characteristics of enrollees

Variable	Overall N (%)
Total, N	1674
Women	1156 (69)
Race  Non-Hispanic White and Other  African American	1433 (86) 241 (14)
Highest level of education  ≤High school  Some college  College graduate  Post-college graduate	1674 136 (8) 608 (36) 647 (39) 281 (17)
Age in years Mean (SD)  20–23 N (%)  24–26  27–30	1674 25.6 (2.8) 510 (30) 493 (29) 671 (40)
Marital status  Single  Married or Committed/engaged	1670 650 (39) 1020 (61)
Work schedule  Nights  Other	1365 44 (3) 1321 (97)
BMI category  <25  25–29.9  ≥30	740 (44%) 428 (26%) 505 (30%)
MET hours per week  < 7.5  ≥ 7.5	614 (38) 992 (62)

BMI, body mass index (kg m⁻²); MET, metabolic equivalent of task; SD, standard deviation.

An initial unadjusted analysis found that eight 5–2-1–0 lifestyle variables were related to BMI (Table 2). Interestingly, reported moderate activities and hours walking were not related to BMI. Those variables associated with a BMI ≥ 30 kg ^. m⁻² were consuming more sugar sweetened beverages and sitting more hours per day. The variables associated with a BMI < 30 kg ^. m⁻² were increased hours per week of vigorous PA; increased overall MET hours per week of PA; and increased consumption of fruits, vegetables, and fruits plus vegetables, combined.

Table 2.

5–2-1–0 Lifestyle variables overall and comparison by BMI group

Variable	Overall N Mean (SD)	BMI<30 N Mean (SD)	BMI≥30 N Mean (SD)	p value
Baseline daily fruit intake	1672 1.38 (0.82)	1168 1.43 (0.81)	504 1.27 (0.83)	< 0.001
Baseline daily vegetable intake	1673 1.62 (0.84)	1168 1.66 (0.85)	505 1.53 (0.83)	0.005
Baseline F+V daily intake	1672 3.01 (1.31)	1168 3.09 (1.29)	504 2.81 (1.34)	< 0.001
No. sugar-sweetened beverages per day	1673 1.43 (1.81)	1168 1.27 (1.59)	505 1.79 (2.20)	< 0.001
No. hours vigorous exercise per week	1541 2.46 (4.26)	1081 2.67 (4.31)	460 1.97 (4.12)	0.002
No. hours moderate exercise per week	1500 2.48 (5.37)	1059 2.43 (4.84)	441 2.58 (6.48)	0.661
MET hour per week	1605 28.17 (47.57)	1121 29.81 (46.31)	484 24.38 (50.21)	0.042
No. hours walking per week	1265 5.42 (8.39)	907 5.65 (8.51)	358 4.85 (8.06)	0.118
No. hours sitting per day	1454 7.55 (4.43)	1023 7.26 (4.33)	431 8.22 (4.60)	< 0.001
METS category, N (%)
<7.5 METs/week	614 (38.26%)	378 (33.72%)	236 (48.76%)	< 0.001
≥7.5METs/week	991 (61.74%)	743 (66.28%)	248 (51.24%)

BMI, body mass index; F+V, fruit and vegetable; MET, metabolic equivalent of task; No., number; SD, standard deviation.

After adjusting for reported sleeping hours, being in a committed relationship, having a higher education level, working during the day, and being of non-Hispanic white or other (non-black) ethnicity, only consumption of vegetables (p = 0.539) was no longer associated with BMI (Table 3).

Table 3.

Individual logistic model to determine odds ratio of BMI ≥ 30 for 5–2-1–0 obesity risk factors

Variables	OR (95% CI)	p value
Baseline daily fruit intake	0.80 (0.68, 0.93)	0.004
Baseline daily vegetable intake	0.95 (0.82, 1.11)	0.539
Baseline F+V daily intake	0.90 (0.81, 0.99)	0.026
No. sugar-sweetened beverages per day	1.08 (1.00, 1.15)	0.044
No. hours vigorous PA per week	0.93 (0.89, 0.96)	0.000
No. hours moderate PA per week	0.99 (0.97, 1.01)	0.459
MET hour per week	1.00 (0.99, 1)	0.002
No. hours walking per week	0.99 (0.96, 0.99)	0.011
No. hours sitting per day	1.07 (1.04, 1.1)	0.000
METS category	0.55 (0.43, 0.72)	0.000

BMI, body mass index; F+V, fruit and vegetable; MET, metabolic equivalent of task; No., number; OR, odds ratio; PA, physical activity;

For individual models, we controlled for site, race, marital status, work schedule, education level, and average sleep per night.

For the complete multivariable analysis, hours of vigorous exercise and daily sitting time were the only 5–2-1–0 variables to independently predict a BMI ≥ 30 kg ^. m⁻² (Table 4). Other independent predictors of a BMI ≥ 30 included reporting less sleeping time, having an African-American ancestry, not being in a committed relationship, having a lower completed education status, and working a night shift.

Table 4.

Multivariable model predicting BMI ≥ 30

Variables	OR (95% CI)	p value
Baseline F+V daily intake	0.91 (0.80, 1.04)	0.168
No. sugar-sweetened beverages per day	1.09 (0.98, 1.2)	0.099
No. hours vigorous exercise per week	0.95 (0.91, 0.99)	0.028
No. hours walking per week	0.98 (0.96, 1.00)	0.124
No. hours sitting per day	1.05 (1.02, 1.09)	0.006
Average sleep per night	0.77 (0.66, 0.89)	0.001
Race: African-American (vs. others)	1.73 (1.04, 2.87)	0.034
Marital status: committed relationship (vs. others)	0.64 (0.46, 0.90)	0.010
Education level: college grad or greater (vs. others)	0.43 (0.31, 0.6)	0.000
Work schedule: nights (vs. others)	4.295 (1.7, 11.01)	0.002

BMI, body mass index; F+V, fruit and vegetable; OR, odds ratio; PA, Physical activity.

All the individual models are controlled for site, race, marital status, work schedule, education level, and average sleep per night.

Weekly walking and VPA were related to each other as were the amount of daily fruits and vegetables consumed (Figure 2). Conversely, daily servings of sugar sweetened beverages was inversely related both to VPA and the amount of daily fruits and vegetables (Figure 2). Hours per day spent sitting was also negatively associated with VPA, walking, and daily fruits and vegetables consumption.

Figure 2.

Correlation matrix for 5–2-1–0 lifestyle variables. Presented in each cell is the Spearman correlation coefficients. Cells without an ‘X’ through them are significant at p ≤ 0.05. Abbreviations: FV, fruits and vegetables (number per day); SSB, sugar sweetened beverages (number per day); VPA, vigorous physical activity (hours wk-1);

DISCUSSION

In this large cohort of young adults, we observed similar relationships between the risk of obesity and the 5–2-1–0 recommended health-lifestyle patterns previously reported (12, 13). Specifically, we found that individuals in their 20s who consume additional servings of fruits, drink less sugary beverages, participate in vigorous exercise, and spend less time sitting have a reduced risk of obesity. We also observed a pattern showing lifestyle variables related to BMI are associated with each other.

Fruits vs. Vegetables

Whereas we did find a relationship between fruit consumption and obesity in the adjusted analysis, we did not observe a similar association with vegetable consumption. One possibility, and also a limitation with this study, is that individuals consuming large amounts of fruits or vegetables (i.e., meeting the American Heart Association dietary recommendations of > 5 servings per day) were excluded from the study, thus biasing the sample to those who consume lower amounts. The average of fruit and vegetable intake for the entire cohort was 1.4 and 1.6 servings, respectively, well below the current dietary guidelines (18), which may also explain why neither fruit or vegetable consumption predicted BMI in the complete multivariate analysis. However, with less than 10% of the US population achieving the national recommendations for fruit and vegetable servings, and only a small number of ineligible subjects excluded for consuming too many fruits or vegetables, these results are likely representative of the general young adult population (14).

A possibility for why an additional serving of fruit contributed to a 22% less likelihood of obesity in the adjusted analysis could be that individuals who reported eating more fruit are substituting the fruit for other higher caloric options. In addition to reducing the caloric content during a snack, substituting fruit has been shown in other studies to reduce the caloric intake of a subsequent meal, possibly due to increased satiety from higher fiber content (19).

Moderate vs. Vigorous Activities

Another unexpected finding was that engaging in vigorous activities were associated with a 7% less likelihood of obesity, while moderate activities were not. With the increased popularity of higher-intensity interval training, more studies have recently looked at it as a possible strategy for weight loss. While a meta-analysis by Wewege et al. (20) reported no difference in magnitude of weight loss when comparing higher-intensity interval training vs. moderate-intensity exercise, they did observe that higher-intensity interval training participants required 40% less time commitment for the same amount of weight loss. Given the busy time constraints of modern life, incorporating VPA as a time effective strategy (i.e., more calorie expenditure compared to MPA for a given amount of time) may be more appealing and practical for many millennials to reduce the risk of obesity.

In addition to being a time-saving strategy, a twin-study by Wang et al. (21) also showed that vigorous exercise mitigated the genetic influence found with elevated BMIs. Other studies have also supported the notion that vigorous exercise may lessen the genetic influence on BMI, possibly through activation of certain mediators (e.g., genes and hormones) that help maintain body weight (22).

While MPA was not statistically related to a BMI < 30, we do acknowledge that moderate-intensity activities are important in the prevention of obesity in this population. Either form of PA (i.e., MPA or VPA) can be used to achieve a caloric balance or deficient, as evidenced in this study by the strong relationship between achieving a weekly MET/hour total > 7.5, which is a product of both MPA and VPA. One possibility for the non-significant effect of moderate exercise to BMI risk was that individuals may have over-reported what constituted as activities that took MPA, especially in this younger cohort.

Sedentary Behaviors

Beyond the known health benefits associated with low levels of PA, more recent attention has examined the health risk of sedentary behaviors (23, 24). Sedentary behavior has received greater attention, not simply due to the obvious connection with total energy expenditure, but also, because, some studies have suggested that PA behaviors themselves may have a genetic or epigenetic link (25).

With respect to the 5–2-1–0 literature, sedentary behavior has been captured by the hours of screen time, with the number of daily hours positively associated with a greater risk of weight gain in children and adolescents (26–28). Sedentary time risk persists, demonstrated by a study that reported that 50% of college students surveyed watched 2 or more hours of television per day (8). In our study, in place of screen time we used the surrogate of hours of sitting, which has been shown as an independent predictor of obesity and health status regardless of time spent performing leisure time PA (10). Our data confirmed that the BMI < 30 group averaged approximately 1 hour less per day spent sitting.

Interrelated Lifestyle Variables

Many behavioral risk factors (e.g., PA, eating habits and smoking) cluster together along the continuum of the health spectrum (10, 24). Using the 5–2-1–0 model as a guide for evaluating risk behaviors, our work supports this focus, as we showed that individuals who consumed more fruits and vegetables were also more likely to report more minutes per week participating in vigorous exercise or walking. Conversely, unhealthy behaviors, such as consuming more sugary drinks or spending more time sitting, were related to fewer minutes per week of VPA and lower fruit and vegetable consumption.

Knowledge of these relationships may be helpful when developing behavior change interventions. To date, there are few prospective studies that have specifically targeted 1 or more behavioral risk factors to determine the effect on another. Spring et al. (29) did such an intervention that sought to increase fruit and vegetable intake while decreasing sedentary behavior. Those randomized to that synergistic intervention parenthetically reported a concurrent decreased consumption of saturated fats, which was not the focus of the intervention.

Limitations

Due to the nature of this cross-sectional secondary data analysis, causality cannot be determined. Also the study sample was largely non-Hispanic white, which may limit generalizability. Self-report of the various intensities of PA may have been influenced by social desirability or reporting bias, which could influence interpretation of findings.

While the majority of participants were women, this study included a large sample with nearly 30% men and 2 geographic regions with diversity in socioeconomic status. Further, this sample of generally healthy young adults enrolled in a health promotion intervention provides a valuable picture of lifestyle behaviors and their relation to body weight.

CONCLUSION

In a large cohort of Millennials living in the diverse areas of metro Detroit or rural north-central Pennsylvania, we showed that after controlling for education, race, sleep, relationship status, and work schedule that predictors associated with obesity to be less reported time performing vigorous exercise, eating fewer daily servings of fruit, consuming more sugar sweetened beverages, and increased time sitting. Further, these lifestyle risk factors were found to be associated with one another, indicating that healthy (or unhealthy) habits were clustered together in this population. This finding provides more evidence to the lifestyle correlates associated with a higher BMI in 21–30 year olds. Future research in this population should consider a longitudinal analysis of these clustered 5–2-1–0 predictors and assess interventions to improve them with a goal of improving weight management. Additionally, because of the strong correlation between genetics and obesity, it would be interesting to examine the role genetics might have on the lifestyle phenotypes themselves.

Our study is one of the first to report the findings that established risk factors related to obesity in youth are meaningful risk factors in young adulthood, highlighting the influence of sedentary behavior on obesity. Future research should follow young adults over time to assess changes, if any, in physical lifestyle activities as subjects pursue efforts to improve food choices.

What is already known about this subject?

• We know that the correlates of obesity are multifactorial and complex, consisting of lifestyle, socioeconomic, and genetic factors.

• Specifically, in children and adolescents common lifestyle factors contributing to obesity are consumption of sugary beverages, increased sedentary behaviors, lack of physical activity, and a low consumption of fruits and vegetables.

What this study adds?

• Here we apply findings from pediatric studies to Millennials between the ages 20–30.

• Specifically, we are reporting on the relationship between obesity and lifestyle factors (i.e. physical activity, sedentary behaviors, consumption of fruits and vegetables, and consumption of sugary-beverages) while accounting for other know predictors (e.g. education, race, and average sleep).