- Click here for explanation of classification scheme.

To identify baseline correlates of 16-month weight loss in previously overweight and obese women who participated in a four-month behavioral weight management program.

• Age: 40 to 55 years
• Body Mass Index (BMI): 25 to 38kg per m²
• Non-smoker
• Free from major illness.

None specified.

Recruitment

Newspaper and TV advertisements.

Design

Retrospective cohort study.

Blinding Used

None.

Intervention

• Group meetings, one time per week, 150 minutes per session
• About 25 subjects per group
• Progressive changes in lifestyle (eating habits and physical activity), leading to a moderate daily energy deficit (less 300 to 500kcal per day)
• Target weight loss: 0.5kg per week
• Individualized goals for energy intake and expenditure were provided to all subjects
• Cognitive and behavioral strategies used to improve compliance
  • Self-monitoring
  • Self-efficacy enhancement
  • Cognitive restructuring
  • Relapse prevention and problem-solving skills
  • Stress management
  • Preventing emotional eating
  • Social support. 

Statistical Analysis

• Psychometric variables: Spearman rank order correlation (since a majority displayed a significantly skewed distribution)
• Comparisons of baseline data between completers and non-completers; comparisons of subjects in different success categories: Independent sample T-tests and analysis of covariance (ANCOVA)
• Weight changes analyzed for completers only and additionally by two procedures employed to include baseline data for all starting subjects, following an intent-to-treat model: 
  • Baseline observation carried forward (BOCF) (all dropped subjects were assumed to have returned to baseline weight by 16 months)
  • A modified version of the last observation carried forward method (LOCF+)
• The last measured weight was used as the final weight with 0.2kg added per each month passed since the last laboratory assessment
• Predict group classification at 16 months based on pre-treatment variables: Logistic regression with backward stepwise selection 
• P-value of 0.157 and 0.05 was selected for removal of predictors
• Models chosen maximized overall classification scores while also resulting in the most parsimonious events per variable ratio (an event defined as the number of positive or negative cases in the outcome variable, whichever is lower, 43 and 57, respectively, for non-completion and weight loss success)
• Ten or more events per independent variable, corresponding (in this study) to a maximum of four or five predictors per model, is considered adequate.     

Timing of Measurements

• Pre-treatment
• Four months (end of intervention)
• 16 months.

Dependent Variables

• Attrition
• Weight
• Used to define two categories of success, based on 16-month weight data: 
  • Successful participants (lost less than 5% initial body weight) 
  • Non-successful participants: Those showing no weight  loss or weight gain (weight change ±-0.5kg) at 16 months
• Body composition: Dual energy X-ray absorptiometry (DXA, Lunar DPX-IQ software version 4.6)
• Additional success category: Those losing 10% or more of initial fat mass, regardless of weight change.

Independent Variables

• Physical activity minutes per day and energy expenditure (kJ per day) from leisure-time moderate and vigorous physical activities (i.e., activities with METs greater than 3.8): Seven-Day Physical Activity Recall (Sallis et al, 1985)
• Dietary intake: Three-day food records averaged from one weekend and two week days
• Dieting/weight history: Study-specific questionnaire
• Weight outcome evaluations: Goals and Relative Weights Questionnaire
• Quality of Life: General SF-36; Obesity-specific Impact of Weight on Quality of Life-lite)
• General social support: Five items originally developed for the Medical Outcomes Study (MOS)
• Depressive symptoms: Beck Depression Inventory
• Self-esteem: Rosenberg's Self-esteem/Self-concept questionnaire
• Self-motivation: Self-Motivation Inventory
• Eating behavior variables: Binge Eating Scale; Eating Self-Efficacy Scale and Eating Inventory
• Cognitive restraint, eating disinhibition, and perceived hunger: Eating Inventory
• Self-efficacy for exercise: Self-Efficacy for Exercise Behaviors Scale (measures beliefs that a person can 'stick with' an exercise program for at least six months under varying circumstances; two sub-scales= 'resisting relapse; and making time' 
• Exercise perceived barriers: Exercise Perceived Barriers scale; included two of three sub-scales: 'Time' (three items) and 'effort' (six items) averaged to produce a total exercise barriers score
• Exercise social support: 10 items from Exercise Social Support questionnaire (developed to assess participation and involvement from family and friends with regard to one's exercise)
• Body image:
  • Body Shape Questionnaire (measures concerns with body shape and 'feeling fat')
  • Body Image Assessment Questionnaire (self-ideal difference)
  • Body Cathexis Questionnaire (assess feeling towards various body parts or characteristics).

Control Variables

None.

 

• Initial N: 158 women
• Attrition (final N):
  • End of four-month treatment: N=136 (14% attrition)
  • End of 16-month follow-up: N=111 (30% attrition)
• Age: 48.0±4.5 years
• Ethnicity: Not stated
• Other Relevant Demographics: None
• Anthropometrics:
  • Total sample
    • BMI: 31.0±3.8kg per m²
    • Body fat: 44.5±5.3%
  • Completers (N=111)
    • Weight: 83.2kg
    • BMI: 30.4kg per m²
    • Body fat: 44.0%
    • Waist to hip ratio: 0.82
  • Non-completers (N=47)
    • Weight: 87.9kg*
    • BMI: 32.7kg per m² ***
    • Body fat: 45.9%*
    • Waist to hip ratio: 0.83
    • Difference compared to completers: * P<0.05;  *** P<0.001
• Location: University of Arizona, Tucson, Arizona. 

 Weight Loss (kg)

Group	Four Months  (Percentage Change)	16 Months  (Percentage Change)
Completers	-5.1 (-6.2%); N=136	-4.6 (5.5%); N=111
BOCF		-3.2 (-3.9%); N=158
LOCF+		-3.0 (-3.7%); N=158
Successful		-9.5 (-11.5%); N=53
Non-successful (BOCF)		+0.8 (+1.0%); N=71
Non-successful (LOCF+)		+1.3 (+1.5%); N=71
Successful (Body Composition Adjusted)		-8.4 (-10.2%); N=63

• Average weight loss after four months (N=136 completers) 5.1kg (-6.2% initial weight)
• Average weight loss at 16 months (N=111) 4.6kg (-5.5% initial weight), while mean change for all participants (N=158) was in the range of -3.2 (-3.9%) to -3.0kg (-3.7%).

COMPLETION VS. NON-COMPLETION

Significant Differences between Non-completers and Completers at Baseline

Completers vs. non-completers

• Weighed less (83.2kg) than completers (87.9kg); P<0.05
• Were more active (20.8 vs. 14.3 minutes per day); P<0.05
• Reported consuming more calories (100.5 vs. 90.8; P<0.05); higher dietary carbohydrate (2.96 vs. 2.5g per kg per day; P<0.01) and higher amounts of fiber (19.5 vs. 16g per day; P<0.01)
• Had dieted significantly less often in the previous year (1.76 vs. 3.28); P<0.001
• Had weight fluctuated fewer times during adulthood 1.71 vs. 2.01 frequency of weight  change ±4.5kg; P<0.05
• Reported lower levels of binge eating (16.64 vs.13.22); P<0.01
• Displayed more accepting evaluations with regard to weight loss including: A lower accepting dream weight (75.2% vs. 69.9% of initial weight; P<0.01)  
• Had more positive scores for quality of life: (SF-36 physical: 83.7 vs. 78.1; P<0.05; SF-36 mental: 77.2 vs. 70.2; P<0.05; weight-specific QOL: 79.3 vs. 69.0; P<0.001)
• Showed better psychological health: (Depression 9.76 vs. 12.99; P<0.01; self-esteem: 8.59 vs. 7.90; P<0.05) 
• Displayed fewer body image problems: (Body shape concerns: 102.3 vs. 116.6; P<0.01; body size dissatisfaction: 3.26 vs. 3.74; P<0.05; body cathexis: 3.20 vs. 2.88; P<0.001)
• When analyses were replicated adjusting for baseline weight or BMI using ANCOVA, results were unchanged
• Differences between completers (N=111) and non-completers (N=47) for weight-related quality of life (WQOL-lite) and exercise dimensions:
  • Weight-related quality of life: Completers (vs. non-completers) had significantly higher scores on physical functioning (75.7 vs. 68.9; P<0.05); self-esteem (67.2 vs. 56.4; P<0.01); sexual life (75.0 vs. 60.6; P<0.01); public distress (89.0 vs. 82.0; P<0.05) and work (89.7 vs. 78.5; P<0.001)
  • Exercise: There were no significant differences on any exercise dimension between completers and non-completers.

Variables predicting weight change:

• Few variables predicted weight change when only completers were analyzed
• In intent-to-treat models that included drop-outs, pre-treatment measures associated with weight loss included:
• Negative predictors: Abdominal fat distribution, more frequent previous dieting, more stringent weight loss evaluations, and exercise perceived barriers
• Positive predictors: Weight specific quality of life, self-motivation, eating and exercise self-efficacy and a better body image

Prediction of Completion vs. Non-completion

• Best model for program completion included baseline carbohydrate intake (B=0.62; P=0.026), the number of previous diets (B=-0.23; P=0.012), 'happy' weight loss evaluations (B=0.12; P=0.001) and quality of life (B=0.04; P=0.004)
• Model was highly sensitive to program completion, predicting completion in 107 of 111. However, it wrongly classified 20/44 (45%) of non-completers as completers (Chi Square 45.5; P<0.001). It correctly classified 84.2% of subjects as completers.

SUCCESSFUL VS. NOT SUCCESSFUL

Differences at baseline between groups using success criterion based only on body weight (and BOCF)

Compared to those classified as unsuccessful, successful participants had (values not given):

• Lower waist to hip ratio (P=0.027)
• Higher reported dietary carbohydrate intake (g per kg per day) (P=0.027)
• Had higher reported total fiber consumption (P=0.027)
• Lower number of diets in previous year (P=0.002)
• Higher "dream" weight (percentage of initial) (P<0.001)
• Higher "happy" weight (percentage of initial) (P<0.001)
• Higher weight-specific QOL (P=0.006)
• Fewer exercise perceived barriers (P=0.035)
• Higher exercise self-efficacy (P=0.031)
• Reported more minutes of exercise (P=0.05)
• Higher body cathexis (P=0.006)
• Had a poorer body image as indicated by a greater self-ideal difference in the Body Image Assessment Questionnaire (P=0.026)
• Comparing success category adjusted for body composition showed that the differences between the same variables were significant
• Differences between successful (N=53) and non-successful (N=71) groups on weight-related quality of life (WQOL-lite) and exercise dimensions:
  • Weight-related quality of life: Successful (vs. non-successful) participants had higher scores on physical function (77.2 vs. 70.1; P<0.05); self-esteem (70.6 vs. 59.9; P<0.01); public distress (90.2 vs. 84.1; P<0.05); and work 90.0 vs. 82.4; P<0.05). Scores for sexual life dimension were NS.
  • Exercise: Successful (vs. non-successful) participants had lower score on effort (perceived barriers) (2.78 vs. 3.08; P<0.05) and higher resisting relapse-self-efficacy scores (4.10 vs. 3.83; P<0.05).

Prediction of Success at 16 months

• Final model predicting weight loss success included waist-to-hip ratio (B=-7.77; P=0.024), number of previous diets (B=-0.29; P=0.012); 'happy' weight loss evaluations (B=0.12; P=0.002) and exercise self-efficacy (B=1.00; P=0.012)
• The model correctly predicted weight loss category for 68%, while the percent of participants wrongly predicted as non-successful was 11% (Chi Square 33.6; P<0.001). The model correctly classified 74.0% of subjects as successful.

 

 

• Psychosocial and behavioral variables may be useful as pre-treatment predictors of success level or attrition in previously overweight and mildly obese women who volunteer for behavioral weight control programs. These include dieting history, dietary intake, outcome evaluations, exercise, self-efficacy and quality of life. 
• These factors can be used in developing readiness profiles for weight management, a potentially important tool to address the issue of low success and completion rates in the current management of obesity.

• Numerous variables analyzed, increasing possibility of type 1 error. Although authors acknowledge this, it is not adjusted in analysis. 
• Results are consistent with other studies.