Starting university with high eating self-regulatory skills protects students against unhealthy dietary intake and substantial weight gain over 6 months.

BACKGROUND: There is consistent evidence that suggests first year students are at risk of weight gain, but the reasons for this vulnerability are still unclear. This study aimed to explore whether the ability to regulate eating behaviours is a predictor of weight and dietary changes in first year undergraduate students.
METHODS: First year undergraduate students from universities situated in London were invited to complete a survey at the beginning of the academic year and at 6-month follow-up. Each survey included the Self-Regulation of Eating Behaviour Questionnaire, food frequency questions, socio-demographic questions and anthropometric questions. Linear and logistic regressions were performed to explore the associations between baseline eating self-regulatory skills and weight and dietary changes.
RESULTS: 481 first year undergraduate students took part in the study. Students who entered university with higher eating self-regulatory skills were more likely to maintain or achieve a higher fruit and vegetable (OR = 1.8, p = 0.007) and a lower sweet and salty snack (OR = 1.9, p = 0.001) intake over the course of the first 6 months in university. Higher baseline eating self-regulatory skills were also related to lower weight changes (β = -0.15, p = 0.018) and lower likelihood of gaining 5% initial body weight (OR = 0.52, p = 0.006) at 6-month. Additionally, self-regulatory skills moderated the relationship between baseline BMI and weight changes (β = -0.25, p ≤0.001) and between baseline BMI and 5% weight gain (OR = 0.82, p = 0.008).
CONCLUSIONS: Starting university with higher eating self-regulatory skills may help students to maintain or achieve a healthy diet and protect them against substantial weight gain, especially among students with overweight.

Introduction

The transition to university is a period characterised by changes in lifestyle, environment and responsibilities. In the late 1990's, a belief that this period leads to dramatic weight gain, identified as the ‘Freshman 15 pounds (6.8kg)’ was widely spread by newspapers and academic articles (Brown, 2008; Graham & Jones, 2002). More recent studies have indicated a lower, but still significant, weight gain among students starting university (Crombie, Ilich, Dutton, Panton, & Abood, 2009; Morrow et al., 2006). A review and meta-analysis (Vella-Zarb & Elgar, 2009) found students gain on average 1.75 kg (95%CI 1.73; 1.77) over the course of their first year.

However, the reasons for this vulnerability to weight gain and individual differences in the experience are still unclear. Reviews suggest weight gain in first year undergraduate students is associated with high baseline weight, dietary changes, decreases in physical activity, living in residential halls, level of stress, and dietary restraint (Crombie et al., 2009; Vella-Zarb & Elgar, 2009). Genetic influences may also play a role (Meisel, Beeken, van Jaarsveld, & Wardle, 2015). However, higher baseline weight is not always a predictor of weight gain. A study conducted with 120 first year students from the UK found that students with a lower baseline weight actually gained the most weight over a 12-month period (Finlayson, Cecil, Higgs, Hill, & Hetherington, 2012). Regarding the relationship between dietary changes and weight gain, a study with first year students from the United States found that weight gain in male students (N = 140) was predicted by an increase in alcohol consumption whereas in female students (N = 256) it was predicted by lower fruit and vegetable intake (Economos, Hildebrandt, & Hyatt, 2008). In contrast, some studies have found that dietary behaviours neither change nor predict weight gain in first year undergraduate students (Boyce & Kuijer, 2015; Nikolaou, Hankey, & Lean, 2015). These inconsistencies may be due to a lack of power to detect changes or due to the use of different measures to assess weight, physical activity and dietary behaviours.

However, it is important to note that weight gain over the first year at university may not always represent a concerning change. Small weight gains may represent natural daily weight fluctuation (Orsama et al., 2014) or even be a positive change for people who had a very low body mass index (BMI). There is also evidence that some students may experience weight loss during this transition (Gillen & Lefkowitz, 2011; Vadeboncoeur, Foster, & Townsend, 2016). Thus, further research into the mechanisms of weight change (as opposed to just the drivers of weight gain) during the transition to university is warranted.

It has been suggested that stress may increase both risk of weight loss and weight gain (Serlachius, Hamer, & Wardle, 2007). According to Boyce and Kuijer (2015) people who enter university with higher levels of stress and lower BMI may lose weight, while those with higher BMI may gain weight. Studies have also shown that increased social support may be a possible buffer of the negative effect of stress on weight gain over the freshman year, especially among men (Darling, Fahrenkamp, Wilson, Karazsia, & Sato, 2017). Increases in physical activity and decreases in calorie intake may also lead to weight loss during the transition to university (Hootman, Guertin, & Cassano, 2017). However, the transition to university has also been linked to an increased risk of developing eating disorders (Delinsky & Wilson, 2008; Striegel-Moore, Silberstein, & Rodin, 1986). Delinsky and Wilson (2008) found that women with higher dietary restraint and concerns about their weight during the first year at university were more likely to lose weight and show disordered eating symptoms.

However, with respect to dietary restraint, that is - the intention to eat less in order to stay in shape (Herman & Polivy, 1975), and its relationship with weight changes, other studies have shown conflicting results. For example, Provencher et al. (2009) found in a cohort of first year students (N = 2921) from Canada that high levels of dietary restraint were related to both weight loss and weight gain. Researchers have suggested that some restraint scales, such as the Restraint Scale (Herman & Polivy, 1975), assess a range of personality traits and eating tendencies (such as the susceptibility to overeat and weight fluctuation) rather than the intent to exercise dietary restraint, and that this may have contributed to mixed results (Hagan, Forbush, & Chen, 2017; Laessle, Tuschl, Kotthaus, & Pirke, 1989; Williamson et al., 2007). As a result, researchers have developed psychometric scales assessing just dietary restraint and no other traits, but this has not solved the issue of inconsistent results for the relationship with weight control (Johnson, Pratt, & Wardle, 2012; Williamson et al., 2007). Some authors have argued that inconsistent results may be because some restrained dieters have higher eating self-regulatory skills than others and may be more capable of maintaining or losing weight (Hays & Roberts, 2008; Johnson et al., 2012; Phelan et al., 2009).

Self-regulatory skills are often conceptualized as the individual's ability to alter their behaviour, thoughts, feelings and attention in the pursuit of their personal goals (Boekaerts, Maes, & Karoly, 2005; Carver & Scheier, 2001; De Vet et al., 2014; Moilanen, 2007), for example, the ability to inhibit a desire to have a sweet in order to stay healthy. Most theoretical models define self-regulatory skills as a continual and multi-level process involving self-monitoring; appraising progress and attempting to approach or maintain the desired goal; making adjustments to it when necessary or giving up (Bandura, 1991; Baumeister, Vohs, & Tice, 2007; Rasmussen, Wrosch, Scheier, & Carver, 2006; Schwarzer, 2008).

Given the dramatic changes in routine, environment and social life experienced by first year undergraduate students, some level of self-regulatory skills may be required to keep healthy habits and/or build new ones due to disruptions of old habitual behaviours. The new environment may also increase demands on self-regulation to inhibit impulses towards food temptations, since students can experience a high exposure to unhealthy food options at university (Grech, Hebden, Roy, & Allman-Farinelli, 2016).

A recent online study conducted with 923 adults in the UK showed that higher eating self-regulatory skills were related to higher fruit and vegetables intake and to lower unhealthy snack intake and sugary drinks intake, as well as lower BMI (Kliemann, Beeken, Wardle, & Johnson, 2016). Similar results were found in studies conducted specifically with undergraduate students (Price, Higgs, & Lee, 2017; Schroder, Ollis, & Davies, 2013; Tomasone, Meikle, & Bray, 2015). However, the majority of these studies had cross-sectional designs, which cannot indicate causality. Additionally, although the transition to university tends to promote weight gain and unhealthy dietary changes (Vella-Zarb & Elgar, 2009), no study has assessed the associations between self-regulatory skills and weight and dietary changes among first year undergraduate students.

Therefore, this study aimed to examine relationships between eating self-regulatory skills and changes in weight and dietary behaviours over 6 months in an online longitudinal cohort of undergraduate students from London, UK. This study hypothesised that high eating self-regulatory skills at baseline would prevent weight gain and be related to weight loss, as well as, help people to achieve or maintain healthier dietary behaviours over the first 6 months at university. People who worsened their dietary behaviours and those who maintained an unhealthy diet over the first 6 months at university would have lower eating self-regulatory skills at baseline.

Material and methods

Participants

Participants were first year undergraduate students from 13 universities within London, chosen based on convenience and having at least one university representing each of the seven regions of London. The Departments and/or Faculties within each university were individually contacted and invited to take part in the study. All interested students aged between 18 and 30 years able to give informed consent and willing to complete the online survey twice over a 6-month period were eligible. Participants who were 30 years old or over were excluded, as older students might not be as susceptible to weight gain as younger students (Hulanicka & Kotlarz, 1983). A criterion for height changes was established to allow for reporting errors (±1 cm); participants with a height change ≤−1 or ≥4 cm were excluded from the analyses.

Procedure

The Departments or Faculties that agreed to take part in the study invited all of their first year undergraduate students to complete the online survey at the beginning of the academic year (September/October 2015) through an email circular. Interested students who consented to participate were directed to the online survey on Survey Monkey (2015). At 6-month follow-up (March/April 2016), participants were invited to complete the online survey for the second time. As an incentive, participants had the chance to enter a draw to win a £20 high street voucher. Ethical approval was granted by the University College London Research Ethics Committee.

Measures

Predictor variable

Eating self-regulatory skills at baseline was assessed using the valid and reliable 5-item Self-Regulation of Eating Behaviour Questionnaire (SREBQ) (Kliemann et al., 2016). Response options ranged from 1 (never) to 5 (always). Total mean score was calculated. The SREBQ demonstrated good internal reliability at baseline (Cronbach's alpha = 0.73).

Outcome variables

Weight and height were self-reported, as first year students tend to provide reliable anthropometric data (Vella-Zarb & Elgar, 2009). Changes from baseline to 6-month follow-up were calculated for absolute weight in kg and categorised into 1) ≥5% initial body weight gain (substantial weight gain) or <5% initial body weight gain and; 2) ≥5% initial body weight loss or <5% initial body weight loss. These criteria for categorising weight changes were based on the current evidence suggesting health benefits of losing 5% of initial body weight, such as improvements in blood pressure, blood cholesterol, and blood sugars (Brown, Buscemi, Milsom, Malcolm, & O'Neil, 2016; Van Gaal, Mertens, & Ballaux, 2005; Vidal, 2002). Following the same principle, gaining 5% of initial body weight could be considered a significant amount of weight since it may increase individuals' risk for these health issues, especially among individuals with overweight and obesity. Additionally, BMI was calculated and categorised into underweight (BMI < 18.5 kg/m2); normal weight (BMI 18.5 to 24.9 kg/m2) or overweight or obese (BMI 25 kg/m2 or over) (WHO, 2015).

Participants were asked to answer the question ‘How frequently do you typically eat fruit and vegetables (FV)’ in both surveys (baseline and 6 months) via a valid 7-point scale that ranged from ‘less than once a week’ to ‘3 or more a day’ (Cappuccio et al., 2003). This scale was then adapted to assess the frequency of sweets and salty snacks (SSS), and sugary drinks (SD) intake. Answers were recoded to represent daily intake, for example, ‘2-3 times a week’ was coded as 0.36. High and low intake were defined using percentile ranks of the scores at baseline. For FV, the 75th percentile was the cut-off point for high intake, while scores that fell below this percentile represented a low intake. Regarding SSS and SD, the 25th percentile was the cut-off point for low intake, and scores above this percentile were classified as high intake. Participants who presented a high FV and a low SSS and SD at 6 months, where categorised as those who managed to maintain or achieve healthier dietary behaviours over 6 months.

Socio-demographic and other variables

Data on age, gender, ethnicity (White; Black; Asian; Mixed or Other), and living arrangements (living in college/university halls, renting from the local authority or privately, living with parents or owning their home) were collected.

Sample size

A sample of at least 286 participants was aimed for to detect a medium effect (R2 = 0.15) of eating self-regulatory skills on weight or dietary behaviours, when running multiple regression tests with up to 10 predictors (Field, 2012). The sample size calculation ensured 95% power, a significance level of 0.01% and allowed for 50% attrition, based on a previous online study (Boyce & Kuijer, 2015). The calculation was performed using G*Power 3.1.5 software.

Statistical analysis

Descriptive analyses were used to characterize the sample. Baseline differences between completer and drop-out participants were checked using Chi-square tests for categorical variables, and t-test or Mann-Whitney tests for continuous variables. Completers were defined as those participants with data at baseline and follow-up, while drop-outs were those with missing data at follow-up.

Pearson's or Spearman's correlations were carried out to assess associations between eating self-regulatory skills, weight, dietary intake and socio-demographic characteristics at baseline. Ethnic origin was dichotomised into white ethnicity or other ethnicity; and living arrangements into living in college/university halls or not; living with parents or not; and renting or owning a home or not.

Change in weight between baseline and 6-month follow-up was explored using paired t-tests. Cohen's effect size was calculated. Chi-square tests were used to assess differences in dietary behaviours (percentage of high and low intake) over 6 months.

Hierarchical multiple linear regression analyses explored the association between eating self-regulatory skills and weight changes. The first step included only eating self-regulatory skills, while age, gender, ethnic origin, baseline BMI and height changes were entered in step 2 and interactions between eating self-regulatory skills and covariates were entered in step 3. Only significant interactions were included.

Binary logistic regression was performed to explore the associations between eating self-regulatory skills and risk of gaining 5% of initial body weight; likelihood of losing 5% of initial body weight and maintaining or achieving the three healthy dietary behaviours at 6-month follow-up. Separate models were run for each outcome. Following the same order as in the linear regression, binary models included eating self-regulatory skills in step 1, covariate variables in step 2 and interaction terms between self-regulatory skills and covariates in step 3.

All analyses were performed using IBM SPSS statistics version 22 (SPSS Inc., Chicago, IL, USA). Due to the number of analyses, a more stringent p-value of ≤0.01 was considered statistically significant for this study.

Results

A total of 815 students were interested in taking part in the study and provided baseline data. Of these, 334 had to be excluded for the following reasons: did not accept to be contacted a second time (N = 186); were not a first year undergraduate student (N = 85); reported a height change outside the acceptable range (N = 38); were from a university based outside London (N = 13); or were 30 years or over (N = 12). The final sample consisted of 481 students, and 262 completed the 6-month follow-up survey (54.3%).

The sample's characteristics at baseline are presented in Table S1. The majority was female (76.5%), white (59.7%), living in halls (70.7%) and had a healthy weight (73.4%). The mean age was 19 years old and mean weight was 60 kg. Students reported consuming on average <2 servings of FV per day and having SSS 4–6 times per week and SD 2–3 times a week. A total of 262 participants provided data at 6-month follow-up and they did not differ significantly from non-completers at baseline for the majority of the variables, with the exception of gender, ethnicity and sugary drink intake. The completer group had a significantly higher proportion of female (80.9% vs 71.2%, p = 0.01) and white (64.9% vs 53.4%, p = 0.012) participants and tended to drink sugary drink less frequently at baseline (0.28 vs 0.37, p = 0.020).

At baseline, higher eating self-regulatory skills was associated with consuming more servings of FV (r = 0.30, p < 0.01), fewer SSS occasions (r = −0.34, p < 0.01) and lower SD intake (r = −0.22, p < 0.01). There were no significant correlations between baseline eating self-regulatory skills and baseline weight, gender, age, ethnicity or living arrangements (Table S2).

Change in weight and dietary behaviours over 6 months

Over 6 months a mean weight change of 0.66 kg (sd = 3.83) was observed, and this was statistically significant (t(254) = 2.752, p = 0.006), representing a small-sized effect (d = 0.17). The range of weight change varied widely (−11.3 kg to +26.2 kg). No changes were reported in a small number of participants (19.6%, N = 50), while about a third lost weight (30.6%, N = 78) and about half gained weight (49.8%, N = 127). Among students whose weight increased over 6 months (N = 127), the mean weight gain was 3.30 kg (sd 3.16). Around a quarter of participants (23.5%, N = 60) gained 5% or more of their initial body weight.

The percentage of people with a high FV intake from baseline to 6-month follow-up did not significantly change (25.4 to 30.5%, p = 0.14). The percentage of people with a high frequency of SSS intake increased significantly (50.1 to 59.9%, p = 0.01) over 6 months. Conversely, there was a significant decrease (55.9 to 46%, p = 0.01) in the percentage of people with a high frequency of SD intake over 6 months. About 30% of participants managed to achieve or maintain a higher intake of FV, while about 40% and 50% of participants managed to achieve or maintain a low intake of SSS or SD, respectively, over the first 6 months at university.

Eating self-regulatory skills and weight changes at 6 months follow-up

Table 1 shows that the adjusted regression model (Model 2) accounted for 6.8% of the variance in weight changes (p = 0.009). However, only baseline BMI was a significant predictor (β = −0.21, p = 0.002). The inclusion of interaction terms between Self-Regulation of Eating Behaviour (SREB) and covariates (Model 3) significantly improved the model fit by 7% (ΔF = 9.986, p < 0.001). Here, eating self-regulatory skills significantly predicted weight changes (β = −0.15, p = 0.01), alongside baseline BMI (β = −0.30, p < 0.001). There was also an interaction between baseline BMI and eating self-regulation (β = −0.25, p < 0.001) and between ethnicity and eating self-regulatory skills (β = 0.16, p = 0.01).

Table 1

Predictors of changes in weight at 6-month follow-up.

Weight changes	model 1 unadjusted			Model 2 adjusted			Model 3 adjusted
	B (SE)	β	p	B(SE)	β	p	B(SE)	β	p
Constant	0.58 (0.22)		0.009	0.59 (0.22)		0.008	0.49 (0.22)		0.025
SREBa	−0.41 (0.32)	−0.07	0.194	−0.64 (0.32)	−0.13	0.045	−0.73 (0.30)	−0.15	0.018
Age				0.09 (0.13)	0.04	0.491	0.04 (0.13)	0.02	0.748
Genderb				−0.46 (0.56)	−0.06	0.413	−0.54 (0.55)	−0.06	0.327
Ethnicityc				−0.70 (0.46)	−0.09	0.130	−0.73 (0.45)	−0.10	0.103
Baseline BMI				−0.23 (0.07)	−0.21	0.002	−0.32 (0.07)	−0.30	<0.001
Height changes				0.47 (0.23)	0.13	0.037	0.43 (0.22)	0.12	0.049
Ethinicity ∗ SREB							1.58 (0.62)	0.15	0.011
BMI ∗ SREB							0.38 (0.09)	−0.25	<0.001
Model fit	R2 = 0.007 & R2 adj = 0.003F = 1.694, p = 0.194			R2 = 0.068 & R2 adj = 0.044F = 2.909, p = 0.009ΔR2 = 0.061, ΔF = 3.137, p = 0.009			R2 = 0.14 & R2 adj = 0.11F = 4.842, p < 0.001ΔR2 = 0.07, ΔF = 9.986, p < 0.001

P-value of ≤0.01 was considered statistically significant.

Eating self-regulatory skills at baseline.

Gender, Male = 0 and Female = 1.

Ethnicity, White = 0 and Other = 1.

Fig. 1 illustrates that higher eating self-regulatory skills (>3.6) predicted decreases in weight among students with overweight (BMI ≥ 25 kg/m2), while those with normal weight (BMI between 18.5 and 24.9 kg/m2) and underweight (BMI <18.5 kg/m2) showed increases in weight regardless of their baseline level of eating self-regulatory skills. Lower eating self-regulatory skills predicted increases in weight among white students, while no association was found for other ethnicities (Fig. 2).

Fig. 1

Interaction between baseline BMI and baseline eating self-regulatory skills as a predictor of changes in weight at 6-month follow-up

Note: SREB = baseline eating self-regulatory skills, where low SREB indicates a score ≤ 3.6 and high SREB indicates a score > 3.6. Weight changes from baseline to 6-month follow-up. Underweight indicates a BMI < 18.5 kg/m2; Normal weight indicates a BMI between 18.5 and 24.9 kg/m2 and Overweight indicates a BMI 25 kg/m2 or over. Mean weight changes adjusted for age, gender, ethnicity and height changes.

Fig. 2

Interaction between ethnicity and baseline eating self-regulatory skills as a predictor of changes in weight at 6-month follow-up

Note: SREB = baseline eating self-regulatory skills, where low SREB indicates a score ≤ 3.6 and high SREB indicates a score > 3.6. Weight changes from baseline to 6-month follow-up. Mean weight changes adjusted for age, gender, baseline BMI and height changes.

Eating self-regulatory skills and likelihood of gaining or losing 5% of initial body weight at 6 months follow-up

The results for the likelihood of losing 5% of initial body weight, presented in Table 2, were not statistically significant for any of the 3 models. In line with this, the results for the likelihood of gaining 5% of initial body weight were not statistically significant for Model 1 unadjusted nor Model 2 adjusted for covariates. However, the model fit improved significantly with the inclusion of an interaction between eating self-regulatory skills and baseline BMI (ΔΧ2(6) = 7.23, p = 0.007). Since the inclusion of interactions between SREB and socio-demographics did not improve the model fit, these were excluded from the final model. The final model (Model 3) explained from 7% to 11% of the variance in risk of substantial weight gain, correctly classifying 77% of cases. Lower eating self-regulatory skills and BMI at baseline were associated with an increased likelihood of gaining at least 5% of initial body weight (ORSREB = 0.52, p = 0.006 & ORBMI = 0.80, p = 0.003).

Table 2

Predictors of gaining or losing 5% of initial body weight or over at 6-month follow-up.

	Model 1 unadjusted			Model 2 adjusted			Model 3 adjusted
	B(SE)	OR (95%CI)	p	B(SE)	OR (95%CI)	p	B(SE)	OR (95%CI)	p
5% Weight gain
Constant	−1.2 (0.15)		<0.001	−1.2 (0.16)		<0.001	−1.4 (0.18)		<0.001
SREBa	−0.39 (0.21)	0.68(0.44;1.03)	0.071	−0.50 (0.22)	0.60(0.39;0.94)	0.025	−0.66 (0.24)	0.52(0.32;0.83)	0.006
Age				−0.04 (0.10)	0.96(0.78;1.17)	0.684	−0.04 (0.10)	0.96(0.78;1.17)	0.697
Genderb				0.16 (0.40)	0.85(0.38;1.88)	0.696	−0.17 (0.41)	0.84(0.37;1.9)	0.679
Ethnicityc				0.28 (0.33)	0.75(0.40;1.45)	0.402	−0.36 (0.34)	0.69(0.36;1.35)	0.288
Baseline BMI				−0.13 (0.06)	0.87(0.77;0.99)	0.032	−0.21 (0.07)	0.80(0.70;0.93)	0.003
Height changes				0.14 (0.15)	1.15(0.85;1.5)	0.365	0.13(0.16)	1.14(0.84;1.5)	0.392
BMI ∗ SREB							−0.20 (0.07)	0.82(0.70;0.95)	0.008
Model fit	R2 = 0.013 to 0.020Χ2(1) = 3.290, p = 0.070			R2 = 0.043 to 0.064Χ2(6) = 10.799, p = 0.095ΔΧ2(5) = 7.509, p = 0.185			R2 = 0.070 to 0.11Χ2(7) = 18.036, p = 0.012ΔΧ2(1) = 7.237, p = 0.007
5% weight loss
Constant	−2.02 (0.19)		<0.001	−2.09 (0.29)		<0.001	−2.08 (0.21)		<0.001
SREBa	0.123 (0.28)	1.13(0.65;1.97)	0.664	0.24 (0.29)	1.27(0.70;2.28)	0.420	0.166 (0.30)	1.18(0.65;2.15)	0.587
Age				0.05 (0.11)	1.05(0.85;1.31)	0.637	0.073 (0.11)	1.07(0.86;1.34)	0.516
Genderb				−0.08 (0.50)	0.93(0.34;2.47)	0.873	−0.17 (0.51)	0.98(0.36;2.67)	0.973
Ethnicityc				0.07 (0.42)	1.07(0.47;2.47)	0.861	0.05 (0.43)	1.05(0.45;2.44)	0.911
Baseline BMI				−0.11 (0.06)	1.11(0.99;1.25)	0.060	0.16 (0.63)	1.17(1.03;1.32)	0.012
Height changes				−0.27 (0.24)	0.760(0.47;1.22)	0.255	−0.26(0.24)	0.77(0.48;1.23)	0.274
BMI ∗ SREB							0.22 (0.10)	1.24(1.00;1.54)	0.042
Model fit	R2 = 0.001 to 0.001Χ2(1) = 0.189, p = 0.664			R2 = 0.024 to 0.046Χ2(6) = 5.874, p = 0.437ΔΧ2(5) = 5.87, p = 0.338			R2 = 0.042 to 0.081Χ2(7) = 10.52, p = 0.161ΔΧ2(1) = 4.64, p = 0.031

R2 = ‘Cox & Snell R2’ to ‘Nagelkerke R2’. Mean self-regulatory skills among students who gained 5% of their initial body weight or over was 3.30 (sd = 0.71). Mean eating self-regulatory skills among students who did not gain 5% the mean was 3.50 (sd = 0.70). P-value of ≤0.01 was considered statistically significant.

Eating self-regulatory skills at baseline.

Gender, Male = 0 and Female = 1.

Ethnicity, White = 0 and other = 1.

These results also suggest that self-regulatory skills moderated the relationship between baseline BMI and 5% weight gain (OR = 0.82, p = 0.008). As shown in Fig. 3, students with overweight (BMI ≥ 25 kg/m2) and normal weight (BMI between 18.5 and 24.9 kg/m2) had higher baseline eating self-regulatory skills (>3.6), and lower risk of gaining at least 5% of their initial body weight over the first 6 months at university than those who had lower baseline eating self-regulatory skills.

Fig. 3

Interaction between baseline BMI and baseline eating self-regulatory skills as a predictor of gaining 5% of initial body weight or over at 6-month follow-up

Note: SREB = baseline eating self-regulatory skills, where low SREB indicates a score ≤ 3.6 and high SREB indicates a score > 3.6. Underweight indicates a BMI < 18.5 kg/m2; Normal weight indicates a BMI between 18.5 and 24.9 kg/m2 and Overweight indicates a BMI 25 kg/m2 or over. Predicted probability of gaining 5% of initial body weight adjusted for age, gender, ethnicity and height changes.

Eating self-regulatory skills and dietary behaviours at 6 months follow-up

Table 3 shows the results for the logistic regressions. The interactions were not significant for any model, and therefore, only the results for the two-step models are presented. In the unadjusted model, eating self-regulatory skills at baseline significantly predicted higher FV intake (p = 0.008). The inclusion of socio-demographic variables improved the model fit significantly (ΔΧ2(4) = 18.907, p = 0.001), and this final model explained from 9% to 14% of the variance in FV intake and classified 66% of the cases correctly. Greater baseline eating self-regulatory skills (OR = 1.8, p = 0.007) and being female (OR = 4.3, p = 0.002) were associated with an increased likelihood of maintaining or achieving a higher consumption of FV at 6 months follow-up.

Table 3

Predictors of maintaining or achieving a healthier dietary intake at 6-month follow-up.

	Maintained or achieved healthier dietary behaviours
	Model 1 unadjusted			Model 2 adjusted
	B(SE)	OR (95%CI)	p	B(SE)	OR (95%CI)	p
High F&V intakea
Constant	−0.79 (0.14)		<0.001	−0.987 (0.16)		<0.001
SREBd	0.54 (0.20)	1.71 (1.1; 2.5)	0.008	0.59 (0.22)	1.8 (1.1; 2.7)	0.007
Age				−0.19 (0.10)	0.82 (0.66; 1.0)	0.060
Gendere				1.4 (0.47)	4.3 (1.7; 10.9)	0.002
Ethnicityf				−0.57 (0.31)	0.56 (0.30; 1.0)	0.066
BMI baseline				0.03 (0.05)	1.0 (0.93; 1.13)	0.511
Model fit	R2 = 0.029 to 0.041Χ2(1) = 7.402, p = 0.007			R2 = 0.09 to 0.14Χ2(5) = 26.308, p < 00.001ΔΧ2(4) = 18.907, p = 0.001
Low SSS intakeb
Constant	−0.43 (0.13)		0.001	−0.43 (0.13)		0.001
SREBd	0.64 (0.19)	1.9 (1.2; 2.7)	0.001	0.64 (0.20)	1.9 (1.3; 2.8)	0.001
Age				−0.05 (0.08)	0.95 (0.80; 1.1)	0.551
Gendere				−0.24 (0.34)	0.78 (0.40; 1.5)	0.479
Ethnicityf				−0.09 (0.28)	0.91 (0.52; 1.6)	0.737
BMI baseline				0.01 (0.04)	1.0 (0.93; 1.1)	0.789
Model fit	R2 = 0.044 to 0.059Χ2(1) = 11.307, p = 0.001			R2 = 0.048 to 0.065Χ2(5) = 12.343, p = 0.030ΔΧ2(4) = 1.035, p = 0.904
Low SD intakec
Constant	0.19 (0.13)		0.140	1.44 (0.13)		0.275
SREBd	0.37 (0.18)	1.45 (1.0; 2.1)	0.041	0.36 (0.18)	1.4 (0.99; 2.01)	0.053
Age				0.03 (0.08)	1.0 (0.88; 1.2)	0.688
Gendere				0.80 (0.34)	2.2 (1.1; 4.3)	0.017
Ethnicityf				−0.15 (0.27)	0.86 (0.50; 1.5)	0.581
BMI baseline				−0.02 (0.04)	0.98 (0.90; 1.0)	0.685
Model fit	R2 = 0.017 to 0.023Χ2(1) = 4.291, p = 0.038			R2 = 0.044 to 0.058Χ2(5) = 11.226, p = 0.047ΔΧ2(4) = 6.935, p = 0.139

R2 = ‘Cox & Snell R2’ to ‘Nagelkerke R2’. P-value of ≤0.01 was considered statistically significant.

Maintaining or achieving a consumption at least 2.25 servings of fruit and vegetable per day.

Maintaining or achieving a consumption of a maximum of 0.36 occasions of sweet and salty snacks per week.

Maintaining or achieving a consumption of a maximum of 0.1 occasions of sugary drinks per week.

Eating self-regulatory skills at baseline.

Gender – Male = 0 and Female = 1.

Ethnicity – White = 0 and Other = 1.

With respect to the logistic regression model for maintaining or achieving a low consumption of SSS, the unadjusted model showed that eating self-regulatory skills was a significant predictor (OR = 1.9, p = 0.001). Although the inclusion of socio-demographic variables did not significantly improve the model fit (ΔΧ2(4) = 1.035, p = 0.904), the likelihood ratio test increased. Model 2 explained from 4.8% to 6.5% of the variance in SSS intake and correctly classified 62% of the cases. The results indicated that higher baseline levels for eating self-regulatory skills was related to a greater likelihood of maintaining or achieving a lower consumption of SSS over 6 months. None of the covariates were found to be related to the outcome.

Finally, the results for the unadjusted model for a low SD intake at 6-month follow-up indicated that greater eating self-regulation was related to an increased chance of maintaining or achieving a low SD intake (OR = 1.45, p = 0.041), however this did not reach the stringent cut-off for significance established for this study (p ≤ 0.01). The inclusion of covariates (Model 2) did not improve the model fit (ΔΧ2(4) = 6.935, p = 0.139). The model explained from 4.4% to 5.8% of the variance in SD intake and classified 59% of cases correctly.

Discussion

This is the first study to assess eating self-regulatory abilities using a valid scale and to examine the impact of self-regulation on weight gain and healthy dietary behaviours among first year undergraduate students. As hypothesised, students who entered university with higher eating self-regulatory skills were more likely to maintain or achieve a healthier diet over the course of the first 6 months in university. Additionally, higher eating self-regulatory skills were related to decreases in weight and lower likelihood of gaining a substantial amount of weight among students with overweight.

Although weight gain (0.6 kg) was modest, around a quarter of the students gained a substantial amount of weight. This is in line with a recent study in which 301 first year students in London were weighed and measured over 7 months and found a weight gain of 0.54 kg, and that one in five gained at least 5% of their initial body weight (Meisel et al., 2015). However, this still conflicts with results from other studies (Vella-Zarb & Elgar, 2009) and there is also little consistency around whether weight gain is related to a lower or higher baseline BMI in first year students (Finlayson et al., 2012; Mihalopoulos, Auinger, & Klein, 2008; Vella-Zarb & Elgar, 2009). According to a recent study, a potential explanation for these inconsistencies is the fact that baseline BMI appears to interact with other factors in order to promote weight gain (Boyce & Kuijer, 2015). This is in line with findings from the present study, which showed that higher eating self-regulatory skills protected against substantial weight gain among students with overweight and normal weight. On the other hand, students with underweight gained weight regardless of their level of eating self-regulatory skills.

However, it is important to note that weight gain in the underweight and normal weight group could represent a positive outcome. On the other hand, weight gain could represent a negative outcome for those with a BMI on the borderline of normal weight/overweight or for those with overweight and obesity. Therefore, the prevention of weight gain in this group is particularly relevant, since people with higher BMIs may be more genetically predisposed to gain weight in an obesogenic environment (Kautiainen, Rimpela, Vikat, & Virtanen, 2002; Wardle & Boniface, 2008). Self-regulation is therefore a potential target for interventions seeking to prevent substantial weight gain among people predisposed to obesity.

Although no association between self-regulation and the likelihood of losing at least 5% of initial body weight was found, the results for weight gain suggest that higher eating self-regulatory skills are related to lower likelihood of 5% weight gain in individuals with overweight and normal weight. Further studies should explore this in samples that include more participants affected by overweight and obesity. It is possible that among people with normal weight, a lower likelihood of 5% weight gain may have occurred as a consequence of factors other than their capacity to regulate eating behaviours. Studies have suggested that eating disorders may affect 8 to 49% of undergraduate students (Eisenberg, Nicklett, Roeder, & Kirz, 2011; Lipson & Sonneville, 2017; Prouty, Protinsky, & Canady, 2002). These disorders usually involve symptoms such as concern about body image, body image distortion and worrying about losing control over their eating (Eisenberg et al., 2011). This group of people tend to present rigid control over their eating, rather than flexible control. The latter is more representative of the ability to self-regulate eating behaviours (Johnson et al., 2012) and may explain why self-regulation was not found to be a predictor of weight loss among those with lower BMIs.

Previous studies have shown that ethnicity does not predict weight changes (Gillen & Lefkowitz, 2011; Roane et al., 2015), and this was also the case in the present study. However, a significant moderating effect of eating self-regulatory skills on the relationship between ethnicity and weight changes was found. White students who had lower eating self-regulatory skills experienced greater increases in their weight compared to those with higher eating self-regulatory skills, while a smaller association was found for people classified as ‘other ethnicities’. A previous study found that white female students tend to be more concerned about gaining weight during the first year of university than black students (Webb, Butler-Ajibade, Robinson, & Lee, 2013). It is possible, therefore, that white students tend to apply more self-regulatory skills to control their weight and their capability may reflect their level of success.

With respect to dietary behaviours, the level of eating self-regulatory skills at baseline was related to higher baseline FV intake and lower baseline SSS and SD intake, in line with results found in a cross-sectional study with UK adults (Kliemann et al., 2016). As anticipated, higher baseline eating self-regulatory skills also predicted higher FV and low SSS intake at 6-month follow-up. Although lower SD intake was also related to higher eating self-regulatory skills, it did not reach the significance established for this study. However, this study only assessed differences in the frequency of SD intake. A systematic review has suggested that sugary drinks tend to be consumed in large portion sizes, due to their lower satiety effect compared to solid foods of the same energy density (Malik, Schulze, & Hu, 2006). Therefore, future studies should explore the effect of eating self-regulatory skills on the amount of sugary drinks consumed.

This study had limitations. For convenience, only students from universities based in London were included. As a consequence, the sample may not be representative of UK first year students, because London tends to have a lower percentage of students with overweight and obesity compared to other regions of the UK (Public Health England, 2015). In fact, individuals with overweight and obesity were under-represented in the sample, which may explain the modest weight gain found in this study. Men were also under-represented, suggesting that the participants who decided to take part in the study may differ from the general student population regarding their interest in a healthy diet and weight control.

The use of self-report measures to assess dietary intake is also a limitation. Although the FV measure has been validated (Cappuccio et al., 2003), the SSS and SD measures have not, although they have been used in several previous studies (Croker, Lucas, & Wardle, 2012; Kliemann et al., 2016; McGowan, Croker, Wardle, & Cooke, 2012). In order to promote high retention rates, the online surveys were kept short and only four questions on food frequency were included. However, they lacked portion size information, were related to groups of foods rather than specific foods, and responses options ranged from 1 to 7. Also, as a retrospective measure, this food frequency questionnaire is also limited in that it relies on individuals' memory. However, its unannounced and self-administered features as well as the fact that it captures habitual behaviours are important strengths of this method (Walton, 2015). Additionally, previous studies using these questions have shown that they can provide valid data on habitual dietary intake (Kliemann et al., 2016; McGowan et al., 2013).

Although there are still several aspects about the susceptibility to weight gain among first year undergraduate students that need to be further investigated, this study provides some initial evidence for the role of eating self-regulatory skills in protecting students against substantial weight gain and unhealthy dietary changes. There is some evidence that interventions using goal-setting, planning, self-monitoring and feedback on performance techniques may potentially promote self-regulatory skills and weight loss among adults with overweight and obesity (Annesi, Johnson, Tennant, Porter, & McEwen, 2016; Crane, Ward, Lutes, Bowling, & Tate, 2016; Kolodziejczyk et al., 2016; Norman, Kolodziejczyk, Adams, Patrick, & Marshall, 2013). Also, a recent study showed that habit-based interventions promoting the repetition of target behaviours in a consistent context hold promise for enhancing self-regulatory skills among adults with obesity (Kliemann et al., 2017). Habit-based interventions are of particular interest because they are considered to be scalable, and are designed to promote lasting behaviour changes. Future studies should investigate whether these techniques may also enhance self-regulatory skills among undergraduate students and the effect of improving these skills on their weight and diet over the course of their studies at university. Additionally, future powered studies should further investigate the potential impact of ethnicity on the relationship between self-regulation and weight changes, exploring this relationship in different ethnic groups.

Conclusions

This study provides evidence that higher baseline eating self-regulatory skills may help students to maintain or achieve a healthy diet and protect them against substantial weight gain, especially among students with overweight. Weight gain prevention initiatives that include eating self-regulatory skills training should be tested among individuals with overweight or predisposed to overweight and obesity.