IDENTIFYING PREDICTIVE VARIABLES FOR LONG-TERM WEIGHT CHANGE AFTER PARTICIPATION IN A WEIGHT-LOSS PROGRAM

Objective and design To determine if there was an association between weight change and 31 independent variables among obese persons 2 years after a weight loss program. Data were obtained from subjects' records and from questionnaires administered at enrollment and after a 2-year follow-up. Setting The 8-week weight control program was taught by registered dietitians and developed by the staff at the Sid Richardson Institute for Preventive Medicine, Houston, Tex. Subjects/samples Of the 1,460 subjects who attended at least one of eight classes, 509 subjects (123 men and 386 women) responded to the mailed follow-up questionnaire. Main outcome measures Associations between weight change and the 31 independent variables were assessed. Heights and weights were measured by the dietitians during treatment. Two-year follow-up weights were self-reported. Statistical analyses performed Analysis of variance was used for 16 of the independent variables. For the remaining variables we performed a test of the null hypothesis that the correlation coefficient was 0 based on the test of the regression coefficient between the independent and dependent variable. A stepwise regression process was used to determine the best combination of variables predictive of weight change. Results Of the 31 independent variables, 16 were significantly predictive of weight change. The adjusted R2 for the entire group of 16 variables was .379. Thus, 37.9% of the variance was explained by the joint efforts of the 16 variables. Eight variables with an adjusted R2 of .371 (accounting for 37.1% of the variance) were most important: feeling in control of eating habits, percentage over ideal body weight at enrollment, percentage of weight lost during the 8-week treatment, frequency of weight measurement, increase in physical activity, frequency of eating in response to emotions, number of pounds gained before subject resumed diet, and occupation. Application/conclusion The predictive variables for weight change may be useful to professionals who treat obese clients and may improve success rates of long-term weight loss.