Managing Complexity in Evidence Analysis: A Worked Example in Pediatric Weight Management

J Acad Nutr Diet. 2018 Aug;118(8):1526-1542.e3. doi: 10.1016/j.jand.2018.01.016. Epub 2018 May 2.


Nutrition interventions are often complex and multicomponent. Typical approaches to meta-analyses that focus on individual causal relationships to provide guideline recommendations are not sufficient to capture this complexity. The objective of this study is to describe the method of meta-analysis used for the Pediatric Weight Management (PWM) Guidelines update and provide a worked example that can be applied in other areas of dietetics practice. The effects of PWM interventions were examined for body mass index (BMI), body mass index z-score (BMIZ), and waist circumference at four different time periods. For intervention-level effects, intervention types were identified empirically using multiple correspondence analysis paired with cluster analysis. Pooled effects of identified types were examined using random effects meta-analysis models. Differences in effects among types were examined using meta-regression. Context-level effects are examined using qualitative comparative analysis. Three distinct types (or families) of PWM interventions were identified: medical nutrition, behavioral, and missing components. Medical nutrition and behavioral types showed statistically significant improvements in BMIZ across all time points. Results were less consistent for BMI and waist circumference, although four distinct patterns of weight status change were identified. These varied by intervention type as well as outcome measure. Meta-regression indicated statistically significant differences between the medical nutrition and behavioral types vs the missing component type for both BMIZ and BMI, although the pattern varied by time period and intervention type. Qualitative comparative analysis identified distinct configurations of context characteristics at each time point that were consistent with positive outcomes among the intervention types. Although analysis of individual causal relationships is invaluable, this approach is inadequate to capture the complexity of dietetics practice. An alternative approach that integrates intervention-level with context-level meta-analyses may provide deeper understanding in the development of practice guidelines.

MeSH terms

  • Child
  • Dietetics / methods
  • Dietetics / statistics & numerical data*
  • Female
  • Humans
  • Male
  • Meta-Analysis as Topic*
  • Outcome Assessment, Health Care / methods*
  • Pediatric Obesity / diet therapy*
  • Regression Analysis
  • Weight Reduction Programs / methods
  • Weight Reduction Programs / statistics & numerical data*