KICKING BAD HABITS MAY BE EASIER THAN DEVELOPING GOOD ONES: A SYSTEMATIC REVIEW OF RANDOMIZED TRIALS OF BEHAVIORAL INTERVENTIONS TO PREVENT CHILDHOOD OBESITY
Celia C. Kamath, PhD, Kristin S. Vickers, PhD, Angela Ehrlich, PhD, Lauren McGovern, MD, Jonathan Johnson, MD, Vibha Singhal, MD, Remberto Paulo, MD, Allison Hettinger, MD, Patricia Erwin, and Victor M. Montori, MD, MSc. Mayo Clinic College of Medicine, Rochester, MN
Purpose: Pediatric obesity interventions vary widely, consisting of single or multiple cognitive, behavioral, environmental, or social support components, with different treatment durations, behavioral targets and patient populations. The purpose of our study was to summarize evidence on efficacy of interventions aimed at changing lifestyle behaviors (increased physical activity (PA), decreased sedentary activity (SA), increased healthy dietary habits (HD) and decreased unhealthy dietary habits (UD)) to prevent pediatric obesity. Methods: We searched all randomized trials identified through librarian designed searches of MEDLINE, ERIC, EMBASE,CINHAHL, PSYCInfo, DISSERTATION abstracts, Science Citation Index, Social Science Citation and the Cochrane Central Database till February 2006 supplemented with bibliography searches of included articles, reviews and content expert files. Two reviewers abstracted data from eligible studies independently and in duplicate on methodological quality, study characteristics, and treatment effects. We determined intervention components according to a theory-derived classification developed for this review. For each outcome, we conducted random-effects meta-analyses, quantified inconsistency using I2, and explored a priori subgroup analyses to explain observed inconsistencies. Results: We found 31 eligible trials. Trials were of brief duration, enrolled small samples and had poor allocation concealment and blinding. The pooled effect size for increased physical activity (n=9891) was 0.12 (95% confidence interval: 0.04, 0.20; IČ=63.2%); decreased sedentary activity (n=3003) was -0.29 (-0.35, -0.22; IČ=0.2%), increased healthy dietary habits (n=5468) was 0.00 (-0.20, 0.20; IČ=83.3%) and decreased unhealthy dietary habits (n=9578) was -0.20 (-0.31, -0.09; IČ=33.6%). Subgroup analyses indicated greater effect sizes for interventions with cognitive components relative to those without, for PA (0.15 vs. 0.00; p<0.06) and SA (-0.31 vs.-0.09; p<0.06) and for interventions with reinforcement components relative to those without, for PA (0.24 vs. -0.07; p<0.07) and HD (0.41 vs. -0.03; p<0.05). Conclusion: Pediatric interventions were more effective at reducing unhealthy behavior (SA and UD) than at increasing healthy behavior (PA and HD), particularly among children rather than adolescents. Interventions with cognitive components work better for targeting physical activity (PA and SA), while those with reinforcement are effective in increasing healthy behavior (PA and HD). The heterogeneity across studies in terms of interventions, delivery methods, outcome measures and participants make inferences to specific populations difficult. Studies should detail intervention methods and findings with greater diligence to enable better comparisons across studies.