Perceived stress as a predictor of eating behavior during the 3-year PREVIEW lifestyle intervention

Abstract

Background: To better support participants to achieve long-lasting results within interventions aiming for weight loss and maintenance, more information is needed about the maintenance of behavioral changes. Therefore, we examined whether perceived stress predicts the maintenance of changes in eating behavior (flexible and rigid restraint of eating, disinhibition, and hunger).

Methods: The present study was a secondary analysis of the PREVIEW intervention including participants with overweight (BMI ≥ 25 kg/m²) at baseline and high risk of type 2 diabetes (n = 1311). Intervention included a 2-month low-energy diet phase and a 34-month subsequent weight maintenance phase. The first 6 months were considered an active behavior change stage and the remaining 2.5 years were considered a behavior maintenance stage. Eating behavior was measured using the Three Factor Eating Questionnaire and stress using the Perceived Stress Scale. The associations between stress and eating behavior were analyzed using linear mixed effects models for repeated measurements.

Results: Perceived stress measured after the active behavior change stage (at 6 months) did not predict changes in eating behavior during the behavior maintenance stage. However, frequent high stress during this period was associated with greater lapse of improved flexible restraint (p = 0.026). The mean (SD) change in flexible restraint from 6 to 36 months was -1.1 (2.1) in participants with frequent stress and -0.7 (1.8) in participants without frequent stress (Cohen's d_s (95% CI) = 0.24 (0.04-0.43)). Higher perceived stress at 6 months was associated with less flexible restraint and more disinhibition and hunger throughout the behavior maintenance stage (all p < 0.001).

Conclusions: Perceived stress was associated with features of eating behavior that may impair successful weight loss maintenance. Future interventions should investigate, whether incorporating stress reduction techniques results in more effective treatment, particularly for participants experiencing a high stress level.

Fig. 1. Estimated marginal means (EMM) and 95% confidence intervals (CI) of eating behaviors throughout the behavior maintenance stage of the PREVIEW intervention by stress groups.

Individuals with frequent high stress levels (n = 132) reported high Perceived Stress Scale scores (≥20) 2 or more time-points during the behavior maintenance stage (6–36 months). Determined using linear mixed effects models with maximum likelihood estimation adjusted for age, sex, diet (high vs medium protein), and baseline (month 0) eating behavior and BMI as fixed effects and participant ID and intervention center as random effects, including group*time interaction term. Pairwise comparisons were conducted at each time point, and significantly different (p < 0.05, Bonferroni adjusted) mean values are indicated with different letters.

Fig. 2. Estimated marginal means (EMM) and 95% confidence intervals (CI) of eating behaviors throughout the PREVIEW intervention by 3-year weight reduction success groups (successful weight reduction = weight loss from baseline to end of study >8% of baseline weight, n = 293; partially successful weight reduction = weight loss 1–8%, n = 405; unsuccessful weight reduction = weight loss <1%, n = 264).

Determined using linear mixed effects models with maximum likelihood estimation adjusted for age, sex, diet (high vs medium protein), and baseline (month 0) BMI as fixed effects and participant ID and intervention center as random effects, including group*time interaction term. Pairwise comparisons were conducted at each time point, and significantly different (p < 0.05, Bonferroni adjusted) mean values are indicated with different letters.