Meta-analysis or Systematic Review

M - Click here for explanation of classification scheme.

POSITIVE: See Quality Criteria Checklist below.

To evaluate the effectiveness of behavioral weight management programs and examine how program characteristics affect mean weight loss.  

• Randomized and quasi-randomized controlled trials
• Terms for overweight and obesity, diet and physical activity, and weight loss interventions
• Adults (18 years or older)
• BMI 25kg/m² or higher
• Interventions that involved multiple contacts with a provider or multiple web sessions; if an Internet intervention, must have been a clearly defined multi-component weight management programs with a detailed description of each component
• Required to include a measure of weight change at 12 months or greater from baseline
• Had to have a non-BWMP control arm or provide direct comparison.  

• Studies in pregnant women
• People with eating disorders
• Those where weight loss was used as a treatment for a medical condition (e.g., diabetes)
• Programs that involved surgery or medication or incorporated other lifestyle changes (e.g., smoking cessation). 

Recruitment

Databases searched included:

• BIOS
• Cochrane Database of Systematic Reviews
• CENTRAL
• Conference Proceedings Citation Index
• Database of Abstracts of Reviews and Effects
• EMBASE
• Health Technology Assessment Database
• MEDLINE
• PsycINFO
• Science Citation Index.

Design

• Systematic review
• Meta-analysis.

Intervention

A variety of interventions focused on multi-component behavior change and weight management.

Statistical Analysis

• Random effects meta-analyses:
  • Effectiveness of behavioral weight management program (BWMP) compared to non-BWMP
  • BWMP with and without a particular component.
• Meta-regression:
  • Across trials comparing BWMP with control
  • Comparing effect sizes in BWMP that included a pre-specified characteristic of interest to effect sizes in those that did not. 
• Ran multi-variable models adding the variable with the strongest association in univariate analysis first and adding all others in turn, regardless of significance in the univariate model. These were retained in the model if they were statistically significant (P<0.05), building the model in steps until no further variables were significant.

Dependent Variables

• Delivery of data components: Participant was asked to monitor E-intake; was the dietary program delivered by an RDN?
• Delivery of physical activity components: Advice was supported by supervised activity sessions; advice required specialized equipment or setting to enact
• Intervention format:
  • Delivery method
  • Length of intervention up to 12 months
  • Contact type
  • Number of sessions offered in first 12 months of the program
  • Frequency of contact.
• Participant was given a weight loss goal
• Use of FU prompts
• Behavioral change techniques used.

Independent Variables

Behaviors grouped in behavioral domains:

• Goal and planning
• Reward and threat
• Regulation
• Antecedents
• Identify
• Self-belief
• Covert learning
• Feedback and monitoring
• Social support
• Shaping knowledge
• Natural consequences.

• Initial N: N=2,210; N=206 screened, with 53 meeting inclusion criteria (37 studies total). Thirty studies included non-BWMP control, 29 of which had sufficient outcome data to be included in meta-regression; 10 studies directly compared BWMP on a variable of interest.
• Attrition (final N): N=37 studies, with 16,000 participants; 13,453 were included in primary meta-analysis and meta-regression, with a mean 378 subjects per study
• Age: Mean age ranged from 32 years to 70 years; a majority were female (68%)
• Location: A total of 53% were conducted in the United States.

 

 

Key Findings

• Direct comparison BWMP vs. controls:
  • Weight loss in BWMP vs. control: Pooled mean was -2.8kg (95% CI: -3.6kg to -2.1kg)
  • Intervention format: No significant difference in weight lost at 12 months detected for more vs. less contact time.
• Indirect comparisons (meta-regression):
  • Diet, PA and intervention format:
    • Longer programs (-0.3kg, 0.5kg to -0.1kg, P=0.009) and especially kcal-counting programs (-3.3kg, -4.6kg to -2.0kg, P=0.0027) associated with greater weight loss; more sessions associated with lower weight loss.
    • Involving dietitian in delivery was a strong predictor (-1.5kg, -2.9kg to -0.2kg, P<0.001).

Most but not all behavioral weight loss programs are effective, with more effective programs leading to 8kg of weight loss in 12 months. Counting kcal and contact time with a dietitian were associated with modest but worthwhile improvements of 1.5kg or greater weight loss.  

Government:	National institute for health and care excellence (NICE); UK Medical research Council
University/Hospital:	University of Oxford