Abstract
One in three college students have overweight or obesity and are in need of brief, simple weight loss interventions. Implementation intentions, a strategy that connects a goal-aligned behavior to a cue, facilitate goal attainment of health behaviors but have not been applied as a standalone treatment for weight loss. The purpose of this study was to examine the efficacy of an implementation intention weight loss intervention in college students. In this three-arm, proof-of-concept, randomized controlled trial, college students with overweight/obesity (N = 95) were randomized to one of three conditions: an implementation intention group (IMP), an enhanced implementation intention group (IMP+) that included text message reminders and fluency training (i.e., training for speed and accuracy), and a control goal intention group (GOL) for 4 weeks. Participants completed anthropometric and self-report assessments pretreatment and posttreatment and experience-sampling assessments during the study to assess how implementation intentions contribute to behavior change. Across the sample, IMP and IMP+ groups reported significantly more goal-congruent behaviors than the GOL group. However, no between-condition differences emerged for weight and diet outcomes. Across conditions, students lost a statistically significant amount of weight, improved diet quality, and reduced caloric intake (ps < .05). Setting implementation intentions was associated with increased behaviors consistent with weight loss goals. Moreover, participants in all groups lost a statistically significant amount of weight. Incorporating implementation intentions into weight loss interventions, and testing the efficacy of this approach on weight loss over a longer duration, may be beneficial for college students with overweight/obesity.
Keywords: Obesity treatment, College students, Implementation intentions
Implications.
Practice: Implementation intentions can improve behavior change for weight loss and low-intensity interventions may be effective for weight gain prevention in college students with overweight and obesity.
Policy: Universities should consider offering easily disseminable weight loss interventions for students to buffer against weight gain that occurs throughout college.
Research: Future research is needed to identify the more effective treatment components that can be used in short-term interventions to spur weight loss in college students with overweight and obesity.
INTRODUCTION
Obesity is associated with a variety of medical and psychological comorbidities [1,2]. College students are an at-risk group for the development and maintenance of obesity as the largest increases in the prevalence of obesity are in individuals aged 18–29 years [3] and 35% of college students already qualify as having overweight or obesity [4]. Poor dietary habits, such as poor food choices and large portion sizes, help to explain obesity onset and maintenance in students [5]. Few interventions have been tested for college students who have overweight or obesity [6,7], and high attrition due to time burden limits impact [7,8]. Brief interventions for weight loss developed for student lifestyles are needed to improve weight status and prevent comorbidities in later adulthood. Moreover, if effective, brief interventions that require little interventionist effort also have the potential for easier dissemination and implementation.
Using implementation intentions is an effective behavior change strategy that facilitates goal attainment of health behaviors [9]. Often taking the form of “If/when situation x arises, then I will do y!,” implementation intentions aid in preplanning and habituation of behaviors as they identify the when, where, and how of behaviors leading to goal attainment [10]. Implementation intentions help to decrease consumption of dietary fat [11] and unhealthy snacks [12,13], increase consumption of fruits and vegetables [14,15], and can contribute to weight loss [16–19].
Though implementation intentions are efficacious and widely used, significant variation in effect sizes has been documented [9], particularly in applied studies with clinical samples [20]. This may be explained by increased interference by habitual behaviors over time in naturalistic settings that reduce the strength of the paired association between cue and behaviors, as well as time-related deterioration of effects possibly due to memory decay [20–22]. The mental effort required to identify cues when more than one implementation intention is held could exacerbate memory decay problems [23]. Therefore, to be translated effectively as a primary weight loss strategy, which requires multiple behavior changes over a sustained period of time, implementation intentions must address these threats.
Two strategies may be particularly useful for this purpose. First, fluency training is a learning strategy that requires an individual to perform a skill or demonstrate knowledge repeatedly for both accuracy and response rate; the goal is to enhance automaticity of the response and promote endurance and retention of skills and knowledge over time and in the face of distractors [24,25]. Implementation intentions for weight loss could be trained to fluency through repeated practice of pairing the cue and goal-oriented behavior, which should strengthen this relationship, increase the likelihood of engaging the behavioral response, and negate the negative effects of competing habits and multiple implementation intentions. Second, external reminders, such as text messages, could prevent memory decay and aid in detecting and attending to implementation intention cues. Text messages have wide-ranging support for their positive effects on health behavior change [26]. Text messages have also been effective in some [27,28], although not all [29], implementation intention interventions for exercise. However, text messaging has not been used with an implementation intention intervention for diet change or weight loss. Thus, in addition to evaluating the use of implementation intentions for weight loss, testing the added benefit of using these strategies in an implementation intention intervention is indicated.
As such, the current study tested the efficacy of an implementation intention intervention alone (IMP), as well as an implementation intention with the addition of fluency training and text message reminders (IMP+), against a goal intention control (GOL) on weight loss and dietary behaviors of college students with overweight or obesity. An active comparator (i.e., GOL) was used to control for the effect of receiving an intervention. Given the difficulty of weight loss intervention and poor outcomes traditionally seen in young adults [8], this study aimed to serve as a proof-of-concept trial to assess the efficacy of implementation intentions over a short period of time. Moreover, in order to gain insight into the mechanisms through which weight and diet may change, in line with the National Institutes of Health Science of Behavior Change Program [30], participants in the IMP and IMP+ groups reported on the frequency of adhering to their implementation intentions (e.g., how many times they encountered their prespecified cue and how often they engaged in their prespecified behavior) and, in all groups, participants were asked how often they were engaging in goal-aligned behaviors that were not prespecified.
METHODS
Participants
College students (N = 95) aged 18–25 were recruited from colleges and universities in the St. Louis area via flyers on college campuses, web postings, and a participant recruitment service. Only students with overweight or obesity (i.e., body mass index [BMI] ≥25) were eligible. Eligible students were required to be interested in changing dietary habits and losing weight, not be enrolled in another formal weight loss program, and own a mobile phone on which they were willing to receive study text messages and download a mobile application. Students were excluded if they met current criteria for an eating disorder as indicated by the clinical cutoffs on the validated and widely used self-report Eating Disorder Diagnostic Scale [31]. Students were compensated $40 for their participation with the additional opportunity to receive one of three $100 prizes if they answered ≥ 85% of mobile phone surveys.
Design and procedure
A three-group (IMP, IMP+, and GOL; described in detail below) randomized controlled intervention paradigm was used. Figure 1 presents the CONSORT diagram for participant recruitment and retention, which ran from June 2017 through July 2018. If participants were deemed eligible following an initial phone or email screening, they came into the university-associated clinic for the baseline session. Upon arrival, participants provided informed consent and their eligibility was confirmed. Participants completed baseline assessments and were then introduced to the dietary goals and asked to read through brief psychoeducational materials regarding dietary change for weight loss. Using an obscured electronic spreadsheet containing randomization assignments made prior to study initiation by J.F.H. (first author), experimenters then gave the participant their condition assignment and engaged with the respective intervention protocol, described below. Both the experimenter and the participant were blind to the assignment prior to this point. Finally, all participants received training in the ecological momentary assessment (EMA) protocol and received a scale for at-home weight monitoring if they did not have access to one. After 4 weeks, participants returned to the clinic for final weight and diet measurements, completion of a treatment satisfaction questionnaire, and debriefing. For this short-term proof-of-concept study, a 4 week intervention was selected to be sufficient to demonstrate effects, as weight change has occurred with implementation intentions over 4 weeks in general adult populations [27], while conserving resources. Enrollment ended when a previously specified sample size was reached. This study was approved by the Washington University Institutional Review Board.
Fig 1.
Study participant diagram.
Intervention groups
All participants were assigned five dietary goals: (a) avoiding high-fat foods, (b) making low-calorie substitutions, (c) limiting portion size, (d) avoiding sugar-sweetened beverages and other caloric drinks, and (e) eating five servings of fruits and vegetables a day. Participants also were assigned the goal to weigh daily as regular weight self-monitoring promotes weight loss [32].
Implementation intention group
Participants in the IMP condition formed an implementation intention for each of the six goals listed above at the baseline session. Two examples of implementation intentions were provided for each goal, so participants had a model of how to set a high-quality implementation intention. Participants thought about how they would best be able to achieve the outlined goals in their life on a daily basis (goal-aligned behavior), as well as when, where, and how they would initiate these behaviors (retrieval cue). Participants created unique implementation intentions for each of the goals using the sentence structure “If/When I ____, then I will _____.” No repetitions or combinations of implementation intentions were allowed (e.g., each participant had six nonoverlapping implementation intentions). The experimenter reviewed participant implementation intentions; if participants did not adequately identify an external or internal stimulus cue and a feasible goal-aligned behavior, the experimenter aided the participant in identifying an appropriate implementation intention. The experimenter gave feedback because the quality of implementation intentions affects outcomes [33]. This protocol mirrors typical implementation intention procedures [34].
Enhanced implementation intention group
In addition to engaging in all IMP condition procedures outlined above, individuals in the IMP+ condition received fluency training and text message reminders.
Fluency training
Fluency training occurred weekly using an online survey tool. On fluency training days, participants received a survey link via email, which consisted of six multiple-choice questions for participants’ unique implementation intentions. Specifically, implementation intention cues were provided in the question portion of the multiple-choice question and participants had to correctly identify their matching goal-aligned behavior among three distractor behaviors as quickly as possible. Matching procedures have been used to create associative links between cues and behaviors for implementation intentions [16]. In line with fluency training principles, if participants selected incorrectly, the program immediately indicated the correct answer in order to provide corrective feedback. Surveys were timed to 1 min and were set to randomly select from the question bank until the timer was up, providing random order and repetition of the questions.
Text message reminders
Text messages containing all six implementation intentions, as well as goal reminders that were obtained by asking participants to write down their reasons for wanting to lose weight, were sent on 4 days each week of the intervention (16 days total) based on a previous implementation intentions protocol in which participants requested text messages on an average of 4 days [27]. At baseline, participants chose how text messages were bundled and when they were sent (e.g., a participant who wants to eat two servings of fruit at breakfast may choose 8 am to receive the text message) [27]. Text schedules stayed constant across the study and were sent automatically via a web-based service.
Goal intention control group
Participants in the GOL condition were assigned the five dietary goals and the daily weighing goal. No additional intervention was given [9].
MEASURES
Height and weight
Height and weight were measured using a portable stadiometer and a digital scale. Two measurements were taken and averaged together to produce a final value. If measurements were discrepant by >0.5 cm for height and >0.1 kg for weight, a third measurement was taken, and the two closest values were used to create an average. Height and weight measurements were then used to calculate BMI.
Demographics
Participants reported sex, age, race/ethnicity, education status, and lifestyle factors. Socioeconomic status was measured with the Barratt simplified measure of social status [35], for which respondents identify parental educational attainment and occupation.
Diet History Questionnaire II
The Diet History Questionnaire II (DHQ II) is a food frequency questionnaire that assesses nutrient and food group intake across the previous month. It was administered electronically preintervention and postintervention to measure dietary change. Primary outcomes in the current study were kilocalorie consumption and a Healthy Eating Index 2010 (HEI) score, an indicator of diet quality that ranges from 0 (lowest) to 100 (highest) [36]. The DHQ II demonstrates good validity and is the gold standard of food frequency questionnaires [37].
Ecological momentary assessment
Experience-sampling data collection occurred via an app-based program on participants’ personal cell phones. Sampling occurred over two 7 day periods to provide measurements of implementation intention use and goal-aligned behaviors at the beginning (Week 1) and end (Week 4) of the study. Participants received five surveys each day, which were distributed over a period of 15 hr based on the participant’s sleep schedule (e.g., from 9 am to 12 midnight) and were randomly distributed within 2–2.5 hr blocks spaced an hour apart to ensure that participants did not receive surveys spaced too close together (e.g., 9–11:30 am, 12:30–2:30 pm, 3:30–5:30 pm, 6:30–8:30 pm, and 9:30 pm to 12 midnight).
For those in the IMP and IMP+ conditions, each survey assessed each implementation intention. Specifically, participants responded to questions asking (a) how often a cue has been encountered and (b) how many times the preselected behaviors had been used in response since the last survey. Implementation intention success was quantified continuously using a ratio of the number of times cues were encountered and the number of times the chosen behaviors were actually used (number of behaviors/number of cues). Cues and behaviors were each summed within surveys to include all cues and behaviors reported at each time point. A third question asked about additional times participants engaged in behaviors congruent with each goal that was not specified by an implementation intention since the last survey. The GOL condition only completed the third set of questions, namely how many times since the last survey they engaged in goal-congruent behaviors.
Treatment Satisfaction Questionnaire
Participants responded to a questionnaire on satisfaction, fit, and ease of use of the intervention [38]. Items are listed in Supplementary Table 2.
STATISTICAL ANALYSIS
A web-based random number generator was used to assign conditions. One-way analyses of variance and chi-squared tests were used to determine significant differences between conditions at baseline. In the case of small cell sizes, Monte Carlo simulations with 10,000 replications were used to estimate p-values.
Intervention success was assessed using an analysis of covariance with condition (IMP, IMP+, and GOL) predicting postintervention weight and BMI and kilocalories and HEI score, controlling for baseline values using statistics program R [39]. An intent-to-treat analysis was employed. Predictive mean matching using the mice package in R was used to estimate missing data values of weight, BMI, kilocalories, and HEI [40]. The number of imputed data sets was 20, and the maximum number of iterations was set at 25. If the omnibus test was significant, contrast tests determined the differences between conditions. A completers’ analysis was also performed.
Normality of the residuals for each outcome model was assessed visually using a QQ-plot and/or histogram, as well as, statistically, using a Shapiro–Wilkes test. In the case of nonnormal residuals (e.g., variables that did not pass the Shapiro–Wilkes test and/or appeared substantially skewed), variables were transformed using a Box–Cox transformation. Given that some values with kilocalories may be improbable reporting, we tested the sensitivity of findings by analyzing kilocalorie outcome results among only plausible reporters (defined as a reported calorie intake between 500 and 3,500 both [41] preintervention and postintervention and a pre–post change not greater than 2,000 kilocalories, n = 6 at baseline and 5 posttreatment). The pattern of results was the same, so only results of the full sample are reported.
The average frequency of goal-congruent behaviors per completed survey (in the case of the IMP and IMP+ condition, this includes behaviors from implementation intentions, as well as additional goal-congruent behaviors reported) and the ratio demonstrating implementation intention adherence were tested as mediators of the effects of condition on weight and diet outcomes. In the event of no effects of condition on outcomes, assessment of individual pathways of condition on the average frequency of goal-congruent behaviors and implementation intention adherence (the “a” pathway), as well as the effect of the average frequency of goal-congruent behaviors and implementation intention adherence on outcomes (the “b” pathway), were planned to gain additional insight into the relationships among variables.
RESULTS
Preliminary analyses
Power analyses to achieve a power of 0.8 and an error probability of 0.05, estimating a conservative 7.5% proportion of variance explained indicated that an enrollment of 86 participants would be needed to detect significant effects between groups. Three additional participants per group were added to account for attrition. All enrolled participants (N = 95) completed assessments at the baseline session, and 87.4% (n = 83) and 86.3% (n = 82) completed postintervention weight assessments and DHQ II assessments, respectively (one student did not complete the DHQ II due to time constraints at the final study session). All other non-EMA data were complete. Participants had a mean age of 20.83 ± 2.01, were majority female (72.6%) and Caucasian (57.9%), and had an average BMI of 30.51 ± 4.49. No significant differences were observed in demographics by condition (Table 1).
Table 1.
Participant demographic information at baseline (mean [SD] or N [%])
| Variable | IMP+ (n = 31) | IMP (n = 34) | GOL (n = 30) | Comparison statistics |
|---|---|---|---|---|
| Age | 21.19 (2.01) | 20.65 (1.94) | 20.67 (2.12) | F(2, 92) = 0.74, p = .48 |
| Gender | χ 2 (4, N = 95) = 6.58, p = .16a | |||
| Male | 6 (19.4%) | 8 (23.5%) | 10 (33.3%) | |
| Female | 25 (80.6%) | 26 (76.5%) | 18 (60.0%) | |
| Other | 0 | 0 | 2 (6.7%) | |
| Race | χ 2 (6, N = 95) = 12.08, p = .06a | |||
| Asian/Asian American | 2 (6.5%) | 3 (8.8%) | 9 (30%) | |
| Black/African American | 5 (16.1%) | 8 (23.5%) | 14 (46.7%) | |
| White/Caucasian | 19 (61.3%) | 22 (64.7%) | 14 (46.7%) | |
| Mixed | 5 (16.1%) | 1 (2.9%) | 2 (6.7%) | |
| Ethnicity | χ 2 (2, N = 95) = 1.06, p = .69a | |||
| Hispanic | 3 (9.7%) | 3 (8.8%) | 1 (3.3%) | |
| Non-Hispanic | 28 (90.3%) | 31 (91.2%) | 29 (96.7%) | |
| Socioeconomic status (possible range: 8–66) | 46.63 ± 14.15 | 52.82 ± 14.05 | 53.18 ± 12.64 | F(2, 92) = 2.27, p = .11 |
| BMI | 31.49 (4.75) | 30.23 (4.71) | 29.82 (3.88) | F(2, 92) = 1.16, p = .32 |
| Class | χ 2 (10, N = 95) = 11.72, p = .48a | |||
| Freshman | 4 (12.9%) | 7 (20.6%) | 6 (20.0%) | |
| Sophomore | 5 (16.1%) | 5 (14.7%) | 6 (20.0%) | |
| Junior | 6 (19.4%) | 10 (29.4%) | 5 (16.7%) | |
| Senior | 6 (19.4%) | 9 (26.5%) | 8 (26.7%) | |
| Graduate/professional/CE | 9 (32.3%) | 3 (8.8%) | 5 (16.7%) | |
| Housing | χ 2 (4, N = 95) = 9.36, p = .12a | |||
| Residence hall/Greek house | 10 (32.3%) | 15 (44.1%) | 17 (56.7%) | |
| Off campus | 12 (38.7%) | 15 (44.1%) | 11 (36.7%) | |
| At home with family | 9 (29.0%) | 4 (11.8%) | 2 (6.7%) | |
| Meals | χ 2 (8, N = 95) = 7.41, p = .51a | |||
| Dining hall | 6 (19.4%) | 7 (20.6%) | 11 (36.7%) | |
| Residence | 18 (58.1%) | 22 (64.7%) | 12 (40.0%) | |
| On-campus eateries | 3 (9.7%) | 2 (5.9%) | 2 (6.7%) | |
| Off-campus eateries | 2 (6.5%) | 3 (8.8%) | 4 (13.3%) | |
| On the go | 2 (6.5%) | 0 | 1 (3.3%) |
BMI body mass index; CE continuing education; IMP+ enhanced implementation intention group; GOL goal intention group.
a Monte Carlo simulations (10,000 replications) were used to estimate chi-square tests given low cell values.
Supplementary Table 1 illustrates correlations of continuous demographic variables and outcome variables at baseline. We observed a positive correlation between age and HEI score and a negative correlation between BMI and HEI score, suggesting that older students and students with lower BMIs were eating a healthier diet. Socioeconomic status was negatively associated with BMI, indicating that students of higher socioeconomic status weighed less.
For the fluency training in the IMP+ group, the average number of correct responses for multiple-choice matching questions in 1 min was 11.00 ± 3.31. A fixed-effects linear mixed model with time (four possible weekly surveys) predicting number correct also demonstrated that the number of questions correct improved over time, b = 0.94, standard error = 0.23, t(99) = 4.11, p < .05. No adverse events were observed throughout the trial.
Intervention effects on weight and diet
The results of the effects of condition on postintervention weight and diet outcomes are presented in Tables 2 and 3. For missing data analysis, no baseline variables were related to missing posttreatment variables, with two exceptions. A larger ratio of Hispanic students compared to non-Hispanic students had missing follow-up data, and a larger ratio of students eating primarily at on- and off-campus eateries than other locations dropped out. Visual diagnostic tests using density plots, box plots, and correlation plots showed imputation data sets accurately reflected the collected data distributions. Models generally fit normality assumptions; however, models of HEI scores required transformation.
Table 2.
ANCOVAs on outcome measures using data imputation
| Outcome | Factor | Estimate (SE) | t | p |
|---|---|---|---|---|
| Kilograms (4 weeks) | Kilograms (bl) | 0.99 (0.02) | 65.32 | <.001 |
| IMP+ | −0.07 (0.26) | −0.26 | .79 | |
| IMP | 0.01 (0.26) | 0.03 | .97 | |
| BMI (4 weeks) | BMI (bl) | 0.96 (0.03) | 32.76 | <.001 |
| IMP+ | 0.01 (0.18) | 0.06 | .95 | |
| IMP | −0.02 (0.18) | −0.12 | .90 | |
| Kilocalories (4 weeks) | Kilocalories (bl) | 0.45 (0.06) | 7.66 | <.001 |
| IMP+ | −7.31 (61.53) | −0.12 | .91 | |
| IMP | −46.46 (60.90) | −0.76 | .45 | |
| HEI (4 weeks) | HEI (bl) | 0.39 (0.10) | 3.89 | <.001 |
| IMP+ | 1.07 (1.65) | 0.65 | .52 | |
| IMP | 0.07 (1.57) | 0.05 | .96 |
ANCOVA analysis of covariance; bl baseline; BMI body mass index; HEI Healthy Eating Index; IMP+ enhanced implementation intention group; SE standard error.
Table 3.
Paired-samples t-tests of outcome measures using completers
| Outcome | Baseline mean (SD) | 4 week mean (SD) | df | t | p | Cohen’s D |
|---|---|---|---|---|---|---|
| Kilograms | 86.56 (12.80) | 86.09 (12.22) | 82 | 3.16 | .002 | 0.35 |
| BMI | 30.51 (4.49) | 30.25 (4.58) | 82 | 2.97 | .003 | 0.33 |
| Kilocalories | 1,824.94 (751.34) | 1,132.08 (540.02) | 81 | 10.31 | <.001 | 1.14 |
| HEI | 63.32 (11.25) | 66.66 (10.70) | 81 | −2.95 | .004 | 0.33 |
BMI body mass index; df degrees of freedom; HEI Healthy Eating Index; SD standard deviation.
Significant effects of condition were not observed on postintervention weight, BMI, kilocalories, or HEI score in the intent-to-treat analysis or in the completers analysis. Two exploratory follow-up analyses were conducted. First, because everyone received an active intervention and no between-group differences were observed, we explored changes in outcomes across conditions over time using paired-samples t-tests. All outcomes improved significantly over the 4 week intervention (Table 3). Specifically, weight decreased by approximately 0.5 kg or 1 lb, BMI decreased by approximately 0.25 points, kilocalories decreased by approximately 700 kcal, and HEI scores increased by approximately 3 points. Second, exploratory paired-samples t-tests within groups showed that weight and BMI trended (all ps between .074 and .112; Cohen’s D between 0.31 and 0.35) toward decreasing for each group, kilocalories significantly decreased in each group (all ps < .001), and HEI decreased significantly in the IMP, t(1, 26) = −2.12, p = .043, and IMP+ groups, t(1, 26)=−2.21, p = .036, but showed no change in the GOL group, t(1, 27) = −0.70, p = .489.
Pathway analysis
Given the absence of condition effect on weight and diet outcomes (“c” pathway), a mediation analysis was not warranted. However, a planned analysis of “a” and “b” pathways was performed. On average, participants across conditions completed 3.14 ± 1.71 goal-aligned behaviors per survey. Two individuals who completed the study did not complete any EMA surveys due to technical difficulties. For the “a” pathway, a linear regression using condition to predict average goal-aligned behaviors completed at each survey showed significant results, F(2, 90) = 3.35, p = .04. Post hoc analyses with the GOL condition as a comparison group showed individuals in the IMP, F(1, 91) = 3.95, p = .049, and IMP+, F(1, 91) = 5.93, p = .016, groups were completing significantly more goal-aligned behaviors than those in the GOL condition, although no differences were observed between the IMP and IMP+ groups, F(1, 91) = 0.22, p = .604. For the analysis of the “b” pathway, controlling for condition, there were no significant relationships when using average goal-aligned behaviors to predict any outcome, with the exception of average behaviors F(1, 91) = 4.13, p = .046 on posttreatment kilocalories showing that individuals reporting more goal-aligned behaviors were consuming more calories (all other ps ≥ .77).
Looking at only the IMP and IMP+ groups, participants had an implementation intention adherence ratio of 0.64 ± 0.17 (i.e., participants responded with planned behaviors when cues were encountered 64% of the time). For the “a” pathway, a linear regression using condition to predict average implementation intention adherence ratio showed trend-level results, F(1, 61) = 3.32, p = .07, suggesting that individuals in the IMP+ group had a trend toward having a higher implementation intention adherence ratio. For the analysis of the “b” pathway, no significant relationships were observed when using implementation intention adherence ratio to predict outcomes, with the exception of HEI score, F(1, 52) = 5.03, p = .030, showing that individuals reporting more goal-aligned behaviors had a better quality diet (all other ps ≥ .27).
Treatment satisfaction
Participants had similar responses to treatment satisfaction questions across conditions, were generally satisfied with their intervention, and found it easy to use (Supplementary Table 2).
DISCUSSION
Implementation intentions have moderately sized effects on weight-related health behaviors [9,15] but have rarely been translated to target weight. This proof-of-concept study assessed the use of a low-intensity weight loss intervention using implementation intentions, both alone (IMP) and enhanced (IMP+), on weight loss and dietary outcomes in college students, a population that struggles to engage with and derive benefit from traditional obesity treatments [8]. Results demonstrated that neither of the implementation intention interventions improved outcomes compared to an active goal intention control condition. However, combined across conditions, participants lost a statistically significant amount of weight over the intervention, suggesting that psychoeducation and goal setting around dietary behaviors with college students with overweight/obesity may be beneficial in promoting short-term weight loss. Moreover, both intervention groups engaged in more weight loss-consistent behaviors than the goal intention control condition, indicating that implementation intentions were useful in changing dietary behaviors, although they did not translate into effects on weight or summary diet measures.
The primary results showing no group differences between conditions are in contrast to a number of studies showing that implementation intentions promote change in individual dietary behaviors [11–15] and can aid in weight loss [16,17,19], although studies using implementation intentions for weight loss have varied in magnitude and duration of effects. Differences in study outcomes may relate to differences in intervention delivery, such as using implementation intentions as a supplement to a more intensive behavioral weight loss program [16,17,19]. Implementation intentions are possibly more effective for weight loss if participants are well practiced or are receiving coaching in weight loss behaviors. Furthermore, in creating implementation intentions that could be employed at least daily, they may have lacked cue or behavior specificity, which is important for successful implementation intentions [33].
Although the IMP and IMP+ interventions did not differ from the GOL condition on weight and diet, individuals in the IMP and IMP+ groups reported engaging in more goal-congruent behaviors than in the GOL condition. Consistent with the goal of identifying mechanisms of treatment [30], this finding reinforces that implementation intentions can impact behaviors that should improve health outcomes. However, when looking at the relationships between the number of goal-aligned behaviors and behavior change, most were null. Goal-aligned behaviors were measured broadly in the study and it is possible that participants were not completing enough goal-aligned behaviors to have an impact on weight or diet quality more generally. Of note, the number of goal-aligned behaviors was positively related to calorie consumption. It is possible that these individuals had better goal achievement in things that may have increased calories, such as eating more fruits and vegetables (while not cutting back on other food items), which is consistent with findings of improved diet quality, or that they were also doing a number of behaviors that were not goal aligned, which was not measured. Other research suggests that small changes in diet and regular self-monitoring of weight can produce significant weight loss [32,42] and, despite previous studies suggesting 4 weeks as a feasible intervention length [18], the relatively short duration of the current study may not have allowed these changes to take effect, particularly in a young adult population in which weight loss intervention is difficult [8]. As such, implementing goal-aligned behaviors and measuring outcomes over a longer duration may have an impact on weight or diet quality and is a consideration for future research.
The addition of supplementary treatment components, such as text messaging to bolster the underlying mechanisms of implementation intentions, was meant to address challenges of utilizing implementation intentions in a weight loss intervention, although the results showed they were not effective in producing better weight and diet outcomes compared to either a control or an implementation intention condition alone. Compared to other implementation intention studies that used text messaging successfully [27], the eating behaviors targeted in the current study may be more difficult constructs to change than other health behaviors, such as exercise. Indeed, to lose weight, unhealthy eating practices must be reduced, while healthier eating behaviors must be increased. However, improving physical activity primarily focuses on increasing or adding new exercise behaviors. Previous research has found that implementation intentions are more effective for increasing or initiating behaviors than reducing unhealthy behaviors, which may explain more limited impact when used to change eating habits for weight loss [15].
Participants in the IMP+ condition also completed a fluency training exercise weekly to strengthen the association between the implementation intention cue and goal-aligned behavior, promoting habit formation, and to promote endurance and retention of their plans. However, this, in combination with the text messages, did not produce any benefit for the implementation intention intervention on weight and diet outcomes. No other studies have utilized fluency training with implementation intentions, but the current study would suggest that the additional rehearsal of implementation intentions by a matching procedure is not helpful to enhance effects.
Although groups did not differ on weight or diet postintervention, participants, when combined across conditions, improved on all outcome measures over the course of the 4 week intervention. On average, students lost approximately 1 pound. This is less weight loss than in a more intensive behavioral weight loss program [43] but similar in magnitude to other low-intensity interventions without interventionist contact for college students [6], as well as a 4 week technology-based implementation intention intervention in adults [18]. Moreover, college is a period when many students experience weight gain [4], with estimates of approximately 3 pounds over the course of 5 months [44]. Therefore, a small loss or prevention of further weight gain may be beneficial, particularly among students with overweight or obesity. Additionally, participants decreased their daily energy consumption and improved HEI scores, which should lead to weight loss and lower risk of chronic disease over time [45]. These positive weight and diet effects for all groups are likely due to the active control group. All participants received education about eating for weight loss, adopted goals to improve eating habits and monitor weight, and completed daily surveys about how often they were engaging in goal-congruent behaviors. As such, these strategies alone may be effective for students to manage their weight short term, although this cannot be documented without an inactive control group.
Finally, all interventions were designed to be simple, low effort, and appropriate for a busy college student lifestyle. Across conditions, participants found the interventions easy to use, believed that they fit into their lifestyles, and were generally satisfied with their experience. Treatment satisfaction also is reflected in this study’s low attrition rate relative to previous studies with students [7], and the generally good acceptability, feasibility, and appropriateness ratings suggest that a low-intensity implementation intention intervention was well-received.
Strengths and limitations
This randomized controlled trial has a number of strengths. It is one of relatively few weight loss intervention trials for college students, and the interventions were designed to be simple, easy and low cost, making them optimal for implementation in a college population. The current study also utilized objective measurement tools for BMI and diet, which are less commonly used in implementation intention studies [14].
However, study limitations should be noted. The inclusion of common intervention factors across all groups does not provide insight into how the implementation intention interventions function in relation to a less active control group. Additionally, although allowing participants to create and tailor their own implementation intentions for prespecified goals more closely mimics clinical care and promotes their effectiveness [12], this invites variability into standardized procedures. Some participants with greater knowledge of diet change for weight loss than that provided may have produced higher quality or more effective implementation intentions, although experimenter review and guidance was meant to limit this occurrence. Finally, as was intended by proof-of-concept study design, the duration of intervention was relatively short and did not provide follow-up time points to assess whether changes reported in goal-oriented behaviors via EMA had an effect over time. Notably, the assessment measure used for the EMA protocol has also not been validated and may be prone to participants inaccurately reporting perceived actions as goal-aligned behaviors when they were not. Moreover, behaviors that were not aligned with goals or that may sabotage weight loss efforts were not measured.
CONCLUSION
The current study suggests that a weight loss intervention for college students utilizing implementation intentions can aid in increasing goal-aligned dietary behaviors; however, results did not demonstrate significant effects on weight and summary diet outcomes when compared to an active goal intention control condition. Nevertheless, all students when looked at combined lost approximately one pound over the 4 week study, suggesting that low-intensity interventions that include psychoeducation, goal setting, self-monitoring, and, in some cases, implementation intentions, may be helpful for slow weight loss over time or as a method to prevent weight gain in college students. Continued exploration into intervention protocols for short and simple interventions and dissemination among college student populations may provide opportunities for students with overweight and obesity to buffer weight gain that occurs in college.
Supplementary Material
Acknowledgments
Mariana Alisio and Sophia Rotman provided invaluable contributions to study implementation.
Funding:
This study was completed with funding from the National Institutes of Health (F31 DK113700, K01 DK116925).
Compliance with Ethical Standards
Conflicts of Interest: The authors declare that they have no conflicts of interest.
Authors’ Contributions:
J.F.H., K.N.B., A.K.G., M.J.S., W.K.B., D.E.W. contributed to the design of the experiment; J.F.H. collected the data; J.F.H. and M.J.S. analyzed and interpreted the data; J.F.H. drafted the article; K.N.B., A.K.G., M.J.S., W.K.B., and D.E.W. provided critical revision of the article; All authors read and approved the final version.
Ethical Approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed Consent: Informed consent was obtained from all individual participants included in the study.
References
- 1. Must A, Spadano J, Coakley EH, Field AE, Colditz G, Dietz WH. The disease burden associated with overweight and obesity. JAMA. 1999;282(16):1523–1529. [DOI] [PubMed] [Google Scholar]
- 2. Simon GE, Von Korff M, Saunders K, et al. Association between obesity and psychiatric disorders in the US adult population. Arch Gen Psychiatry. 2006;63(7):824–830. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Mokdad AH, Serdula MK, Dietz WH, Bowman BA, Marks JS, Koplan JP. The spread of the obesity epidemic in the United States, 1991-1998. JAMA. 1999;282(16):1519–1522. [DOI] [PubMed] [Google Scholar]
- 4. American College Health Association. The American College Health Association National College Health Assessment (ACHA-NCHA): Fall 2012 Reference Group Executive Summary.Silver Spring, MD: American College Health Association ; 2014. [Google Scholar]
- 5. Kasparek DG, Corwin SJ, Valois RF, Sargent RG, Morris RL. Selected health behaviors that influence college freshman weight change. J Am Coll Health. 2008;56(4):437–444. [DOI] [PubMed] [Google Scholar]
- 6. Napolitano MA, Hayes S, Bennett GG, Ives AK, Foster GD. Using Facebook and text messaging to deliver a weight loss program to college students. Obesity (Silver Spring). 2013;21(1):25–31. [DOI] [PubMed] [Google Scholar]
- 7. Ames GE, Perri MG, Fox LD, et al. Changing weight-loss expectations: A randomized pilot study. Eat Behav. 2005;6(3):259–269. [DOI] [PubMed] [Google Scholar]
- 8. Lanoye A, Brown KL, LaRose JG. The transition into young adulthood: A critical period for weight control. Curr Diab Rep. 2017;17(11):114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Gollwitzer PM, Sheeran P. Implementation intentions and goal achievement: A meta‐analysis of effects and processes. Adv Exp Soc Psychol. 2006;38:69–119. [Google Scholar]
- 10. Gollwitzer PM. Implementation intentions: strong effects of simple plans. Am Psychol. 1999;54(7):493–503. [Google Scholar]
- 11. Armitage CJ. Evidence that implementation intentions reduce dietary fat intake: A randomized trial. Health Psychol. 2004;23(3):319–323. [DOI] [PubMed] [Google Scholar]
- 12. Adriaanse MA, de Ridder DT, de Wit JB. Finding the critical cue: implementation intentions to change one’s diet work best when tailored to personally relevant reasons for unhealthy eating. Pers Soc Psychol Bull. 2009;35(1):60–71. [DOI] [PubMed] [Google Scholar]
- 13. Allan JL, Sniehotta FF, Johnston M. The best laid plans: Planning skill determines the effectiveness of action plans and implementation intentions. Ann Behav Med. 2013;46(1):114–120. [DOI] [PubMed] [Google Scholar]
- 14. Armitage CJ. Effects of an implementation intention-based intervention on fruit consumption. Psychol Health. 2007;22(8):917–928. [Google Scholar]
- 15. Adriaanse MA, Vinkers CD, De Ridder DT, Hox JJ, De Wit JB. Do implementation intentions help to eat a healthy diet? A systematic review and meta-analysis of the empirical evidence. Appetite. 2011;56(1):183–193. [DOI] [PubMed] [Google Scholar]
- 16. Armitage CJ, Norman P, Noor M, Alganem S, Arden MA. Evidence that a very brief psychological intervention boosts weight loss in a weight loss program. Behav Ther. 2014;45(5):700–707. [DOI] [PubMed] [Google Scholar]
- 17. Benyamini Y, Geron R, Steinberg DM, Medini N, Valinsky L, Endevelt R. A structured intentions and action-planning intervention improves weight loss outcomes in a group weight loss program. Am J Health Promot. 2013;28(2):119–127. [DOI] [PubMed] [Google Scholar]
- 18. Veling H, van Koningsbruggen GM, Aarts H, Stroebe W. Targeting impulsive processes of eating behavior via the internet. Effects on body weight. Appetite. 2014;78:102–109. [DOI] [PubMed] [Google Scholar]
- 19. Luszczynska A, Sobczyk A, Abraham C. Planning to lose weight: Randomized controlled trial of an implementation intention prompt to enhance weight reduction among overweight and obese women. Health Psychol. 2007;26(4):507–512. [DOI] [PubMed] [Google Scholar]
- 20. De Vet E, Oenema A, Sheeran P, Brug J. Should implementation intentions interventions be implemented in obesity prevention: The impact of if-then plans on daily physical activity in Dutch adults. Int J Behav Nutr Phys Act. 2009;6(1):11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. McDaniel MA, Guynn MJ, Einstein GO, Breneiser J. Cue-focused and reflexive-associative processes in prospective memory retrieval. J Exp Psychol Learn Mem Cogn. 2004;30(3):605–614. [DOI] [PubMed] [Google Scholar]
- 22. Chapman J, Armitage CJ. Evidence that boosters augment the long-term impact of implementation intentions on fruit and vegetable intake. Psychol Health. 2010;25(3):365–381. [DOI] [PubMed] [Google Scholar]
- 23. Guynn MJ, McDaniel MA, Einstein GO. Prospective memory: When reminders fail. Mem Cognit. 1998;26(2):287–298. [DOI] [PubMed] [Google Scholar]
- 24. Binder C. Behavioral fluency: Evolution of a new paradigm. Behav Anal. 1996;19(2):163–197. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Lindsley OR. Ten products of fluency. J Precis Teach Celeration. 1995;13(1):2–11. [Google Scholar]
- 26. Head KJ, Noar SM, Iannarino NT, Grant Harrington N. Efficacy of text messaging-based interventions for health promotion: A meta-analysis. Soc Sci Med. 2013;97:41–48. [DOI] [PubMed] [Google Scholar]
- 27. Prestwich A, Perugini M, Hurling R. Can the effects of implementation intentions on exercise be enhanced using text messages? Psychol Health. 2009;24(6):677–687. [DOI] [PubMed] [Google Scholar]
- 28. Mistry CD, Sweet SN, Rhodes RE, Latimer-Cheung AE. Text2Plan: Exploring changes in the quantity and quality of action plans and physical activity in a text messaging intervention. Psychol Health. 2015;30(7):839–856. [DOI] [PubMed] [Google Scholar]
- 29. Schwerdtfeger AR, Schmitz C, Warken M. Using text messages to bridge the intention-behavior gap? A pilot study on the use of text message reminders to increase objectively assessed physical activity in daily life. Front Psychol. 2012;3:270. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Nielsen L, Riddle M, King JW, et al. ; NIH Science of Behavior Change Implementation Team . The NIH Science of Behavior Change Program: Transforming the science through a focus on mechanisms of change. Behav Res Ther. 2018;101:3–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Stice E, Telch CF, Rizvi SL. Development and validation of the Eating Disorder Diagnostic Scale: A brief self-report measure of anorexia, bulimia, and binge-eating disorder. Psychol Assess. 2000;12(2):123–131. [DOI] [PubMed] [Google Scholar]
- 32. Steinberg DM, Tate DF, Bennett GG, et al. The efficacy of a daily self‐weighing weight loss intervention using smart scales and e‐mail. Obesity. 2013;21(9):1789–1797. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. de Vet E, Oenema A, Brug J. More or better: Do the number and specificity of implementation intentions matter in increasing physical activity? Psychol Sport Exerc. 2011;12(4):471–477. [Google Scholar]
- 34. McDaniel MA, Howard DC, Butler KM. Implementation intentions facilitate prospective memory under high attention demands. Mem Cognit. 2008;36(4):716–724. [DOI] [PubMed] [Google Scholar]
- 35. Barrat W. The Barratt simplified measure of social status. 2006. http://socialclassoncampus.blogspot.com/2012/06/barratt-simplified-measure-of-social.html. Accessibility verified June, 2015.
- 36. Guenther PM, Reedy J, Krebs-Smith SM, Reeve BB. Evaluation of the Healthy Eating Index-2005. J Am Diet Assoc. 2008; 108(11):1854–1864. [DOI] [PubMed] [Google Scholar]
- 37. Subar AF, Thompson FE, Kipnis V, et al. Comparative validation of the Block, Willett, and National Cancer Institute food frequency questionnaires: The Eating at America’s Table Study. Am J Epidemiol. 2001;154(12):1089–1099. [DOI] [PubMed] [Google Scholar]
- 38. Larsen DL, Attkisson CC, Hargreaves WA, Nguyen TD. Assessment of client/patient satisfaction: Development of a general scale. Eval Program Plann. 1979;2(3):197–207. [DOI] [PubMed] [Google Scholar]
- 39. R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2013. [Google Scholar]
- 40. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained equations in R. J Stat Softw. 2010;45(3):1–68. [Google Scholar]
- 41. Rhee JJ, Sampson L, Cho E, Hughes MD, Hu FB, Willett WC. Comparison of methods to account for implausible reporting of energy intake in epidemiologic studies. Am J Epidemiol. 2015;181(4):225–233. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Crane MM, Lutes LD, Ward DS, Bowling JM, Tate DF. A randomized trial testing the efficacy of a novel approach to weight loss among men with overweight and obesity. Obesity (Silver Spring). 2015;23(12):2398–2405. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Moyer VA; U.S. Preventive Services Task Force . Screening for and management of obesity in adults: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2012;157(5):373–378. [DOI] [PubMed] [Google Scholar]
- 44. Vadeboncoeur C, Townsend N, Foster C. A meta-analysis of weight gain in first year university students: Is freshman 15 a myth? BMC Obes. 2015;2:22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Chiuve SE, Fung TT, Rimm EB, et al. Alternative dietary indices both strongly predict risk of chronic disease. J Nutr. 2012;142(6):1009–1018. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.