1. Methods planned in the protocol but not used in this review.

Topic	Method
Unit of analysis issues	Cluster‐randomized trials We will then combine the estimates and their corrected SEs from the cluster‐randomized trial with those from trials of parallel‐group design using the generic inverse variance method in Review Manager 2014
Dealing with missing data	For interventions in which there is substantial attrition (≥ 15% for at least 1 of the groups) of trial participants (caregivers, children, or caregiver‐child units), we will report the attrition rate and perform sensitivity analyses (see "Sensitivity analysis")
Assessment of heterogeneity	If we identify unexplained heterogeneity, we will not pool results into an overall effect estimate but instead will present the individual effect sizes per study for the specific outcome, in a table
Assessment of reporting biases	If we have 10 or more studies included for an outcome, we will use funnel plots to assess the possibility of small‐study effects. In the case of asymmetry, we will consider various explanations such as publication bias, poor study design, and the effect of study size
Subgroup analysis and investigation of heterogeneity	When data allow, we will perform the subgroup analyses listed below, to explore substantial and considerable heterogeneity across studies Age (e.g. 2 to 5 years of age vs 6 to 12 years of age vs 13 to 18 years of age) High‐income countries or settings vs low‐ and middle‐income countries or settings (according to the World Bank country and lending group classifications [data.worldbank.org/about/country‐and‐lending‐groups] per the year of publication). If there is a multicenter study with sites in countries classified in different income categories, we will consider the study in a subgroup of its own in the meta‐analysis Active caregiver interventions vs inactive caregiver interventions Duration or intensity of intervention (e.g. short term vs long term, one‐off vs multiple sessions) Individual context vs group context (i.e. children receive the intervention individually and with a caregiver vs children receive the intervention in a group and with caregivers) Diet only vs physical activity only vs both behaviors BCT cluster vs no BCT cluster (e.g. techniques from "reward and threat" cluster vs no techniques from "reward and threat" cluster)
Sensitivity analyses	When data allow, we will perform sensitivity analyses to assess the following, and will report results in tables Influence of studies’ risk of bias (first pool all relevant studies per outcome, then pool only studies where the random allocation sequence was appropriately concealed) Influence of attrition (first pool all relevant studies per outcome, then pool only studies where there was less than 15% total attrition or less than 10% differential attrition) Study design (first pool all relevant studies per outcome, then pool only individually randomized trials and cluster‐RCTs where the primary trial authors appropriately adjusted for clustering in their analyses, i.e. cluster‐RCTs where we did not have to calculate effective sample size)

BCT: behavior change technique.

RCT: randomized controlled trial.

SE: standard error.