The relationship of within-individual and between-individual variation in mental health with bodyweight: An exploratory longitudinal study

Background Poor mental health is associated with obesity, but existing studies are either cross-sectional or have long time periods between measurements of mental health and weight. It is, therefore, unclear how small fluctuations in mental wellbeing within individuals predict bodyweight over short time periods, e.g. within the next month. Studying this could identify modifiable determinants of weight changes and highlight opportunities for early intervention. Methods 2,133 UK adults from a population-based cohort completed monthly mental health and weight measurements using a mobile app over a period of 6-9 months. We used random intercept regression models to examine longitudinal associations of depressive symptoms, anxiety symptoms and stress with subsequent weight. In sub-group analyses, we included interaction terms of mental health variables with baseline characteristics. Mental health variables were split into "between-individual" measurements (= the participant's median score across all timepoints) and "within-individual" measurements (at each timepoint, the difference between the participant's current score and their median). Results Within-individual variation in depressive symptoms predicted subsequent weight (0.045kg per unit of depressive symptom severity, 95% CI 0.021-0.069). We found evidence of a moderation effect of baseline BMI on the association between within-individual fluctuation in depressive symptoms and subsequent weight: The association was only apparent in those with overweight/obesity, and it was stronger in those with obesity than those with overweight (BMI<25kg/m(2): 0.011kg per unit of depressive symptom severity [95% CI -0.017 to 0.039]; BMI 25-29.9kg/m(2): 0.052kg per unit of depressive symptom severity [95%CI 0.010-0.094kg]; BMI >= 30kg/m(2): 0.071kg per unit of depressive symptom severity [95%CI 0.013-0.129kg]). We found no evidence for other interactions, associations of stress and anxiety with weight, or for a reverse direction of association. Conclusion In this exploratory study, individuals with overweight or obesity were more vulnerable to weight gain following higher-than-usual (for that individual) depressive symptoms than individuals with a BMI<25kg/m(2).