To assess the relationship between self-reported snacking and weight gain and obesity in a Spanish middle-aged free-living cohort of university graduates.

• Used data from Seguimiento Universidad de Navarra-University of Navarra (SUN) project follow-up.
• Participants who had completed a baseline assessment by September 2005 were included in this study.

• Did not answer any follow-up questionnaires
• Reported extremely low values (less than 500kcal per day for women, less than 800kcal per day for men) or extremely high values (greater than 3,500kcal per day for women, greater than 4,000kcal per day for men) for total energy intake
• Followed a special diet
• Were pregnant at baseline or during follow-up
• Were missing information about snacking or covariates of interest in questionnaires.

Recruitment 

• Recruitment began in December 1999 for SUN project
• Participants who had completed a baseline assessment by September 2005 were included in this study.

 Design

• Prospective longitudinal cohort study
• Baseline data collection included FFQ, medical history, health habits, lifestyle and sociodemographic variables (including smoking, physical activity)
• Self-reported weight was recorded at baseline and every two years during follow-up. 

Blinding used 

Not used

Statistical Analysis

• Least-squares regression models used to evaluate association between snacking and weight change per year
• Analyses of covariance used to adjust the average weight change per year for potential confounders
• Non-conditional logistic regression models assessed relationship between snacking and increment of weight of at least 3kg per year and 5kg per year
• Odds ratio and 95% CI calculated using non-snacker participants and BMI categories
• Multivariate Cox's regression analyses used for the association between snacking and risk of incident obesity (BMI ≥30); excluded obese participants at baseline; converted hazard ratios to rate advancement period
• Multi-collinearity tested using tolerance and its reciprocal the variance inflation factor
• Adjusted for age, sex, physical activity, smoking, TV viewing, sitting hours, baseline BMI, fast-food consumption, alcohol, fiber, and total energy intake; in additional multivariate models controlled for physical activity, smoking habits, consumption of meats and fruits/vegetables.
• Sensitivity analyses conducted excluding participants with chronic disease at baseline or during follow-up
• Analyses performed using SPSS version 15.0; P values significant at P<0.05.

Timing of Measurements

Data were collected at baseline and every two years after baseline

Dependent Variables

• Weight (self-report), weight change (calculated), and BMI (calculated)
• Risk of incident obesity (using multivariate Cox's analyses after excluding obese participants at baseline).

Independent Variables

• Self-reported usual snacking habits and dietary habits assessed through questionnaire and included the following:
  • Nutrient intake (total energy, carbohydrate, protein, fat, saturated fat, monounsaturated fat, polyunsaturated fat, glycemic load, alcohol intake, fiber intake)
  • Food intake (fruits and vegetables, nuts, biscuits, non-handmade pastries and bakery, chocolates and cakes, sugar-sweetened beverages, fruit juices, alcoholic beverages, fast-food, processed meat, chip potatoes)
• Usual snackers was defined as those participants who answered yes to the following question: "Do you usually eat between meals (snacking)?"

Control Variables

• Age
• Sex
• Physical activity
• Smoking
• TV viewing
• Sitting hours
• Baseline BMI
• Fast-food consumption
• Alcohol, fiber, total energy intake, consumption of meats and fruits/vegetables.

• Initial N: 20,095 in SUN cohort
• Attrition (final N): 10,162 university graduates included in the analysis
• Age: Mean age 39 years (SD: 12)
• Ethnicity: Not reported exactly but predominantly Spanish
• Other Relevant Demographics and Anthropometrics:
  • Baseline for overall sample:
    • 54% female
    • Average age 39 years
    • Average weight 69kg
    • Average BMI 23.8
    • Smokers 23%
    • Ex-smokers 29%
  • Baseline for non-snackers vs. usual snackers:

• Location: Spain.

Key Findings

• Usual snackers had a higher total energy intake, total fat intake, higher glycemic load, and were more likely to consume nuts, non-handmade bakery foods, chocolates, sugar-sweetened beverages, fast-food and processed meat
• Usual snackers had a statistically significant higher mean weight gain per year during follow-up in multivariate analyses and were at higher risk of gaining substantial weight during follow-up
• After adjusting for potential confounders, self-reported between-meal snacking was significantly associated with a higher risk of substantial weight gain (more than 3kg per year; P<0.001; more than 5kg per year, P<0.001, >10% of baseline weight, P<0.001)
• The habit of self-reported between-meal snacking was significantly associated with a higher risk of becoming obese during follow-up
• Among participants with a baseline BMI lower than 30kg/m², there were 258 new cases of obesity
• Usual snackers presented an adjusted 69% higher risk of becoming obese during follow-up (hazard ratio=1.69, 95% confidence interval: 1.30-2.20)
• Controlling for chronic disease did not change results significantly.

Baseline nutrients and food composition by snacking category

Alcohol intake, fiber intake, fruits and vegetables, biscuits, fruit juices, alcoholic beverages, and chip potatoes were not statistically significant.

^*Fast-food was the sum of hamburgers, sausages, and pizza consumption.

^a Adjusted for age (years), sex, physical activity (METs-hours per week, continuous), smoking (non-smokers, smokers, ex-smokers, missing), TV viewing (hours per day, continuous), sitting (hours per day, continuous), total energy intake (kcal per day, continuous), baseline body mass index (tertiles), fast-food consumption (tertiles), sugar-sweetened soft drinks consumption (tertiles), alcohol intake (tertiles), total fiber intake (tertiles).

^bAdjusted for the variables of multivariate model 1 plus changes in physical activity (no change, increase, decrease, not known), in smoking habits (non-smokers, ex-smokers, smokers, quitters, starters, missing), and in the consumption of meat and fruits/vegetables (no change, increase, decrease, not known) during follow-up.

In conclusion, between-meal snacking has been associated with a subsequent higher body weight and with an increase in the risk of becoming obese in a free-living cohort of healthy middle-aged Spanish university graduates. Thus, this dietary trend of unstructured eating episodes with food of low nutritional value is a potential risk factor to be discouraged in the promotion of healthy lifestyles to tackle the obesity epidemic. At least, the promotion of alternative nutritious snacks (e.g. nuts, fresh fruit and vegetables) should be considered from a Public Health perspective.

Authors note the following limitations:

• Cultural differences between countries with regards to the definition of what constitutes a usual snack compared with a main meal can contribute to the results
• Snacking based on participant self-report
• Potential selection bias with difference in eligible population (n=15,742) and participants included in the analysis (n=10,162), however, the cohort is large and homogeneous in terms of demographic characteristics and baseline exposure.