Table 3: Overall and subgroup analyses for primary outcomes
Outcome; group or subgroupMeta-analysis, standard mean difference (95% CI)Statistical heterogeneity (within-group) Test for between-group differencesNo. of participantsNo. studiesGRADE quality of evidence*
p valueI2 value, %p valueI2 value, %
Change in BMI and BMI z-score (baseline to end of intervention period)
Overall–0.53(–0.69 to –0.36)< 0.00183NA390830Moderate
Behavioural–0.54(–0.73 to –0.36)< 0.001850.370334628Low
Pharmacological + behavioural–0.43(–0.60 to –0.25)0.3405622Moderate
Behavioural — diet–0.36(–0.65 to –0.06)0.27190.366.82702Moderate
Behavioural — exercise–0.43(–0.65 to –0.21)NA3221High
Behavioural — diet + exercise–1.09(–1.84 to –0.34)< 0.001946846Moderate
Behavioural — lifestyle–0.42(–0.61 to –0.23)< 0.00176207019Moderate
Behavioural ≤ 12 mo–0.54(–0.73 to –0.35)< 0.001840.970305625Low
Behavioural > 12 mo–0.53(–1.31 to 0.26)< 0.001902903Low
Behavioural — aged 2–12 yr–0.54(–0.76 to –0.32)< 0.001860.810261222Low
Behavioural — aged 13–18 yr–0.59(–0.92 to –0.25)< 0.001787346Moderate
Behavioural — individually-focused–0.90(–1.27 to –0.53)< 0.001890.00786.2134711Moderate
Behavioural — family-based–0.34(–0.52 to –0.16)< 0.00173199917Moderate
Behavioural — low risk of study bias–0.41(–0.59 to –0.22)0.9200.5104793High
Behavioural — unclear risk of study bias–0.49(–0.68 to –0.30)< 0.00181263822Low
Behavioural — high risk of study bias–1.24(–2.79 to 0.32)< 0.001962293Very Low
Weight loss maintenance: change in BMI and BMI z-score
Baseline to end of intervention period–0.51(–0.86 to –0.16)0.00181NA7164Moderate
End of intervention to 6–12 mo follow-up0.08(–0.07 to 0.23)0.4406864Low

Note: GRADE = grading of recommendations assessment, development and evaluation, NA = not applicable.
*Rating reflects confidence in the estimate of effect assessed through 5 domains of the evidence (risk of bias, indirectness, imprecision, inconsistency and reporting bias) as follows: high = no downgrades, moderate = downgraded for risk of bias, low = downgraded for risk of bias and imprecision, very low = downgraded for very serious risk of bias (double downgrade) and imprecision.