Background: Data are limited as to whether participants in diet trials truly adhere to their assigned diet and the factors that affect their adherence.

Methods: We evaluated success and adherence in a two-year dietary intervention randomized controlled trial (DIRECT) in which 322 moderately obese participants (mean age 52 yrs, mean body-mass-index (BMI) 31 kg/m(2), 86% men) were randomized to one of three groups: low-fat, Mediterranean, or low-carbohydrate diets.

Results: Overall compliance at month-24 was 85%, with 90% in low-fat, 85% in Mediterranean, and 78% in low-carbohydrate diet (p = .042 between groups). Attrition was higher in women (29% vs. 14% men, p = .001) and current smokers (25% vs. 14% among maintainers, p = 0.04). In a multivariate model, independent predictors of dropping-out were: higher baseline BMI (OR = 1.11; CI: 1.03-1.21) and less weight loss at month-6 (OR = 1.20; CI: 1.1-1.3). In a multivariate model, greater weight loss achieved at month-6 was the main predictor associated with success in weight loss (> 5%) over 2 years (OR = 1.5; CI: 1.35-1.67). Self-reported complete adherence score to diet was greater on low-carbohydrate diet (p < .05 compared to low-fat) until month-6, but dropped overall from 81% at month-1 to 57% at month-24. Holidays were a trigger to a significant decrease in adherence followed by a partial rebound. Changes in diet composition from month-1 to month-12 were more pronounced in the multi-stage low-carbohydrate diet-group (p < .05). Generally, the most irresistible restricted food items were cookies (45% of dieters) and fruits (30%). Among the physically active (n = 107), 44% reported a tendency to eat less after exercising compared to 10% who tended to eat more.

Conclusion: Initial 6-month reduction in weight is the main predictor of both long-term retention and success in weight loss. Special attention is needed for women, current smokers, and during holidays. Physical activity is associated with subsequent reduction in energy intake.