Anxiety and Food Addiction in Men and Women: Results From the Longitudinal LIFE-Adult-Study

Abstract

Background: Anxiety is a widespread phenomenon, and it is connected to disordered eating and obesity. We want to analyze the connection between anxiety and food addiction (FA) over two points in time to better understand the directionality of the association. Since there are gender differences with regard to anxiety and eating, we are also interested in differences between men and women.

Methods: We used data from the population-based LIFE-Adult-Study (N = 1,474) at time 1 (baseline) and time 2 (first follow-up) to analyze the connections between anxiety (GAD-7) and FA (YFAS) using a multiple group latent cross-lagged panel model with female and male participants as groups. We controlled for age, marital status, socioeconomic status and social support.

Results: Anxiety (women: β = 0.50, p ≤ 0.001; men: β = 0.59, p ≤ 0.001) as well as FA (women: β = 0.37, p ≤ 0.001; men: β = 0.58, p ≤ 0.001) exhibited stability over time for both genders. We found a significant association between anxiety at time 1 and FA at time 2 for women (β = 0.25, p ≤ 0.001) but not for men (β = 0.04, p = 0.10), and significant associations between FA at time 1 and anxiety at time 2 for women (β = 0.23, p ≤ 0.001) as well as men (β = 0.21, p ≤ 0.001).

Conclusion: Food addiction longitudinally affects anxiety, independent of gender and other sociodemographic variables. In addition, anxiety affects subsequent FA as well, but only in women. Interventions that address FA could reduce anxiety in men and women, while interventions that mitigate anxiety could help prevent FA in women.

Keywords: GAD-7; YFAS; anxiety; food addiction; gender; longitudinal.

FIGURE 1

Multiple-group latent autoregressive cross-lagged panel model of the association between anxiety and food addiction. a₁, a₂: autoregressive paths, b₁, b₂^: cross-lagged paths; c₁, c₂^: cross-sectional paths. The measurement models for the latent variables with pairwise correlated errors over time are not shown. Model adjusted for age, marital status, socioeconomic status and social support at time 1.

FIGURE 2

Multiple-group latent autoregressive cross-lagged panel model with standardized beta coefficients and standard errors in parentheses. The model only displays the significant paths for men and women. The effects of control variables (age, marital status, socioeconomic status, social support) on anxiety and food addiction at time 1 were included in the estimation but not shown for ease of presentation. Sample size: 1,474 (48.2% female). *p ≤ 0.05; ^***p ≤ 0.001.