Abstract
BACKGROUND: Standard behavioral obesity treatment produces poor long-term results. Focusing on healthy eating behaviors rather than energy intake may be an alternative strategy. In addition, important behaviors might differ for short- vs long-term weight control. OBJECTIVE: Our aim was to describe and compare associations between changes in eating behaviors and weight after 6 and 48 months. DESIGN: We performed secondary analysis of data collected during a randomized weight-loss intervention trial with 48-month follow-up. PARTICIPANTS: We studied 481 overweight and obese postmenopausal women enrolled in the Women on the Move through Activity and Nutrition (WOMAN) Study. MAIN OUTCOME MEASURES: We measured changes in weight from baseline to 6 and 48 months. STATISTICAL ANALYSES PERFORMED: Linear regression models were used to examine the associations between 6- and 48-month changes in eating habits assessed by the Conner Diet Habit Survey and changes in weight. Analyses were conducted in the combined study population and stratified by randomization group. RESULTS: At 6 months in the combined population, weight loss was independently associated with decreased desserts (P<0.001), restaurant eating (P=0.042), sugar-sweetened beverages (P=0.009), and fried foods (P<0.001), and increased fish consumption (P=0.003). Results were similar in intervention participants; only reduced desserts and fried foods associated with weight loss in controls. At 48 months in the combined population, weight loss was again associated with decreased desserts (P=0.003) and sugar-sweetened beverages (P=0.011), but also decreased meats/cheeses (P=0.024) and increased fruits/vegetables (P<0.001). Decreased meats/cheeses predicted weight loss in intervention participants; desserts, sugar-sweetened beverages, and fruits/vegetables were independently associated in controls. CONCLUSIONS: Changes in eating behaviors were associated with weight change, although important behaviors differed for short- and long-term weight change and by randomization group. Future studies should determine whether interventions targeting these behaviors could improve long-term obesity treatment outcomes.