Depression, emotional eating and long-term weight changes: a population-based prospective study

This study investigated the complex relationships between depression, emotional eating, and long-term changes in body weight and abdominal fat. Specifically, the researchers tested:

**Whether emotional eating acts as a mediator:** They wanted to see if depression leads to emotional eating, which then, in turn, leads to weight gain. In this context, "mediation" means that emotional eating is a step in the chain of events connecting depression to weight changes.

**Whether other factors modify these relationships (moderation):** They explored if the strength of these connections varied depending on a person's:

* **Gender**

* **Age**

* **Night sleep duration**

* **Physical activity levels**

The primary outcomes measured were:

**7-year change in Body Mass Index (BMI):** A common measure of body fat based on height and weight.

**7-year change in Waist Circumference (WC):** A measure of abdominal fat, which is linked to various health risks.

The key variables involved were:

**Depression:** The presence and severity of depressive symptoms.

**Emotional eating:** The tendency to eat in response to negative emotions like stress, sadness, or anxiety.

**Night sleep duration:** How many hours a person typically sleeps per night.

**Physical activity:** A person's level of engagement in physical exercise.

The study included a large group of Finnish adults aged 25 to 74 years old.

**Baseline (2007):** 5,024 participants were initially included in the study.

**Follow-up (2014):** 3,735 of these participants provided data for the 7-year follow-up, representing a retention rate of approximately 74.3%.

The participants were drawn from a population-based cohort, meaning they were intended to be representative of the general adult population in Finland.

The researchers used a combination of objective measurements and self-reported questionnaires:

**At Baseline (2007):**

* **Height, weight, and waist circumference (WC):** These were objectively measured by trained health professionals during a health examination.

* **Depression:** Assessed using the **Center for Epidemiological Studies - Depression Scale (CES-D)**. This is a widely used self-report questionnaire designed to measure depressive symptoms in the general population. Scores typically range from 0 to 60, with higher scores indicating more severe depressive symptoms.

* **Emotional eating:** Measured using the **Three-Factor Eating Questionnaire-R18 (TFEQ-R18)**. This is a self-report questionnaire that assesses three dimensions of eating behavior: cognitive restraint, uncontrolled eating, and emotional eating. The emotional eating subscale specifically captures the tendency to overeat in response to emotional states.

* **Physical activity:** Self-reported by participants, likely through a questionnaire asking about the frequency, duration, and intensity of their exercise.

* **Night sleep duration:** Self-reported by participants, likely through a questionnaire asking about their typical sleep hours per night.

**At Follow-up (2014):**

* **Height, weight, and waist circumference (WC):** These measures were based on either objectively measured information (similar to baseline) or self-reported information. The abstract does not specify the proportion of measured versus self-reported data at follow-up, which is an important detail.

The use of self-report for key psychological and behavioral variables (depression, emotional eating, sleep, physical activity) is common in large population studies but introduces potential for bias, as people might not accurately recall or report their behaviors or feelings. The mixed method for follow-up physical measurements (measured or self-reported) also introduces variability and potential for error compared to purely objective measurements.

This study employed a **population-based prospective cohort design**.

**How they ran the study:**

**Baseline Data Collection (2007):** Researchers recruited a large group of Finnish adults and collected comprehensive data on their demographics, physical measurements (height, weight, waist circumference), and self-reported information on depression, emotional eating, physical activity, and sleep duration. This initial data collection established the starting point for all variables.

**Follow-up Data Collection (2014):** After a 7-year period, the researchers re-contacted the original participants to collect follow-up data. This included updated physical measurements (height, weight, waist circumference), which were either objectively measured or self-reported.

**Statistical Analysis:** The data were analyzed using **structural equation models (SEM)** with a full information maximum likelihood estimator. This advanced statistical technique is well-suited for examining complex relationships between multiple variables simultaneously, including direct effects, indirect (mediation) effects, and moderation effects. The models were adjusted for age and gender to account for their potential influence on the observed associations.

**Why this design matters:**

**Prospective Nature:** The "prospective" aspect means that the baseline data (depression, emotional eating, sleep, physical activity) were collected *before* the changes in BMI and WC were observed. This is crucial because it allows researchers to examine how baseline characteristics predict future outcomes, providing stronger evidence for temporal sequences (e.g., depression *preceding* emotional eating, which *precedes* weight gain) compared to cross-sectional studies that measure everything at one point in time.

**Cohort Study:** Following the same group of individuals over time allows for the tracking of individual changes and the identification of risk factors that emerge over the long term.

**Population-Based:** Drawing participants from the general population aims to ensure that the findings are more generalizable to the broader adult population, rather than just specific clinical groups.

**Structural Equation Modeling (SEM):** This statistical approach is powerful for testing theoretical models that involve multiple hypothesized pathways. It allowed the researchers to simultaneously test the mediation hypothesis (emotional eating as a link) and the moderation hypotheses (how gender, age, sleep, PA influence these links). Adjusting for age and gender helps to isolate the specific effects of the variables of interest by statistically removing the influence of these common demographic factors.

**What this design can and cannot prove:**

**Can Prove:** This prospective observational design can demonstrate **associations** and **temporal sequences** between variables. It can provide strong evidence for **mediation**, suggesting that emotional eating is a plausible behavioral mechanism linking depression to weight changes. It can also identify **moderators**, showing *for whom* or *under what conditions* these associations are stronger or weaker.

**Cannot Prove:** As an **observational study**, it cannot definitively prove **causation**. While the prospective nature strengthens the argument for temporal order, there's always the possibility of unmeasured confounding variables (other factors not included in the model) that could explain the observed relationships. For example, while depression might precede emotional eating, and emotional eating might precede weight gain, there could be other underlying biological or psychological factors driving all three. **Randomized controlled trials (RCTs)** are generally required to establish direct cause-and-effect relationships.

**Major methodological weaknesses:**

**Reliance on Self-Report:** Key variables such as depression, emotional eating, physical activity, and night sleep duration were all self-reported. Self-report data can be subject to recall bias (inaccurate memory), social desirability bias (tendency to report what is perceived as socially acceptable), and subjective interpretation, which can affect the accuracy and reliability of the findings.

**Mixed Measurement at Follow-up:** At the 7-year follow-up, height, weight, and waist circumference were based on either measured or self-reported information. This inconsistency in measurement methods for the outcome variables could introduce measurement error and potentially bias the results, especially if self-reported measurements are systematically different from objective measurements or if the proportion of self-reported data varied across groups. The abstract does not specify the proportion of self-reported vs. measured data at follow-up.

**Lack of Blinding:** In an observational study, blinding is not typically applicable in the same way as in an RCT. However, the lack of objective measures for the psychological and behavioral variables means participants were aware of their own status, which could influence their responses.

**Generalizability:** While population-based, the study was conducted in Finland. Cultural, dietary, and lifestyle factors specific to this population might limit the direct generalizability of the findings to other populations.

The study identified several important relationships over the 7-year follow-up period:

**Mediation Effect of Emotional Eating:**

* Emotional eating was found to **mediate** the effects of depression on changes in both BMI and waist circumference. This means that depression was associated with an increase in emotional eating, and this increased emotional eating, in turn, was associated with higher gains in BMI and WC over the 7 years. The abstract states the indirect effect of depression on BMI via emotional eating was statistically significant (P = 0.045).

* This suggests that emotional eating is a behavioral pathway through which depressive symptoms can contribute to long-term weight gain and increased abdominal fat.

**Moderation by Night