Core outcome set for early intervention trials to prevent obesity in childhood (COS-EPOCH): Agreement on "what" to measure

Childhood obesity affects an estimated 39 million children under five years globally and is associated with immediate health issues (e.g., asthma, sleep apnoea, hypertension) and long-term risks (e.g., type 2 diabetes, certain cancers, cardiovascular disease). Early intervention is needed to support healthy growth trajectories and reduce social, health, and economic burdens. Although many early childhood obesity prevention trials are underway, heterogeneity in outcome selection and reporting impedes synthesis and understanding of what works, for whom, and through which mechanisms. Core outcome sets (COS) can reduce outcome heterogeneity and research waste by defining a minimum set of outcomes for measurement and reporting. A prior COS exists for infant feeding interventions (≤1 year), but given the multifactorial aetiology of childhood obesity and the broad range of behaviours established from birth to five years, a more comprehensive COS applicable across multiple risk factors and settings is warranted. This study aimed to develop the COS-EPOCH for trials of early childhood obesity prevention interventions in children aged birth to five years, covering lifestyle-related components (diet, physical activity, sedentary behaviour, sleep, parent/caregiver practices) across relevant settings (e.g., community, home, early childhood education and care).

The authors reference existing work highlighting outcome heterogeneity in obesity prevention trials and a published COS for infant feeding interventions (≤1 year) that identified 26 outcomes across nine domains. They note the complex, multifaceted determinants of childhood obesity and the need for a broader COS applicable from birth to five years and spanning multiple behavioural risk factors and settings to enable consistent and comprehensive assessment and synthesis across studies.

The study was registered on the COMET Initiative registry (registration 1679) and followed COS-STAD development and COS-STAR reporting standards. It proceeded in three stages (June 2020–March 2022) under guidance of an international Steering Group with expertise in economics, psychology, paediatric public health, behavioural epidemiology, nutrition, physical activity/sedentary behaviour, sleep, biostatistics, and evidence synthesis. Ethics approval: Deakin University HREC (HEAG-H 231_2020). Stage 1: Systematic scoping review of RCTs of early childhood obesity prevention interventions (antenatal to age 5; lifestyle-related). Searches of clinicaltrials.gov, WHO ICTRP, and Ovid Medline were performed, supplemented by a Cochrane review update. Two reviewers screened studies and extracted data using COMET-recommended tools. Outcomes were categorised iteratively with Steering Group input. Stage 2: Three-round e-Delphi survey using DelphiManager to prioritise outcomes. A priori rules limited survey burden: excluded infant feeding outcomes covered by an existing COS; excluded non-specific or study-related outcomes; excluded outcomes appearing in only one of 161 studies; merged overlapping items (e.g., height/weight with BMI into weight-based anthropometry). Stakeholder groups: policymakers/funders, parents/caregivers, researchers, health professionals, community/organizational stakeholders. Target n≈30 per group (total n≈150). Recruitment: parents via sponsored social media (Facebook/Instagram); other stakeholders via literature, websites, professional networks. Outcomes were presented by domain with randomised ordering and plain language definitions. Rating used a 1–9 GRADE scale (1–3 not important; 4–6 important; 7–9 critical). Round 1 allowed suggestions for additional outcomes. Between rounds, participants received individual prior responses and group distribution feedback. Consensus thresholds: include if ≥75% in each stakeholder group rated 7–9 and <15% rated 1–3; exclude if ≥75% in each group rated 1–3 and <15% rated 7–9; otherwise no consensus. Agreement assessed by mean absolute deviation from the median (MADM: high <1.08; moderate 1.08–1.41; low >1.41). Attrition bias was assessed by comparing responses of those completing only Round 1 vs all rounds. Stage 3: Online consensus meeting (March 2022) using nominal group technique, chaired by an independent facilitator. Participants were those who completed Round 3; aim: 2–4 per stakeholder group. Outcomes with and without Delphi consensus were presented and discussed. Anonymous electronic voting (Mentimeter) for inclusion (≥75% yes) with two voting rounds per item; final discussion and confirmation of the COS.

- Scoping review: Identified 18 outcome domains and 221 unique outcomes from 161 trials; 112 outcomes included in Round 1 after applying a priori rules and merging. - e-Delphi participation: Round 1 n=206 participants from 25 countries; Round 2 n=143 (69% of R1); Round 3 n=96 (67% of R2). Stakeholder composition in R1: researchers 85, parents/caregivers 32, health professionals 47, policymakers/funders 20, community/organizational stakeholders 22. - Round 1 consensus to include: child time spent sedentary; child physical activity; child diet quality; child dietary intake; household food security. Seventy-eight additional outcomes were suggested; seven substantively new outcomes added in Round 2. - Round 2 consensus to include: above five plus parent/caregiver role modelling of healthy eating. - Round 3 consensus to include: above six plus child screen time; sedentary behaviour or physical activity home environment; child non-core beverage intake; food environment. No outcomes met exclusion criteria across rounds. Attrition analyses suggested similar scoring patterns between completers and non-completers. - Consensus meeting: 9 participants (2 policymakers, 2 researchers, 2 parents/caregivers, 1 health professional, 2 community/organizational; 8 Australia, 1 South Pacific). Final COS of 22 outcomes across nine domains; “parent/caregiver role modelling of healthy eating” refined under “parent/caregiver nutrition parenting practices.” - Final COS outcomes: Anthropometry: child weight-based anthropometry. Dietary intake (ages 1–5): child diet quality; child dietary intake; child fruit and vegetable intake; child non-core food intake; child non-core beverage intake; child meal patterns. Sedentary behaviour: child screen time; child time spent sedentary. Physical activity: child physical activity; infant tummy time (for infant trials). Sleep: child sleep duration. Parent/caregiver outcomes: physical activity parenting practices; sleep parenting practices; nutrition parenting practices. Environmental: food environment; household food security; family meal environment; sedentary behaviour or physical activity home environment; early childhood education and care (ECEC) environment. Emotional/cognitive functioning: child wellbeing. Economic: economic evaluation.

The COS-EPOCH provides a stakeholder-derived minimum set of 22 outcomes spanning key domains relevant to early childhood obesity prevention. It complements the existing infant feeding COS by specifying dietary intake outcomes for ages 1–5 and extends to additional domains (anthropometry, sedentary behaviour, physical activity, sleep, parent/caregiver practices, environmental factors, wellbeing, and economic evaluation). By reflecting multiple levels of the socio-ecological model (child, family, environment, policy), the COS supports comprehensive and comparable assessment across diverse interventions and settings. While it recommends 22 outcomes, trialists may tailor application based on intervention focus, age group relevance (e.g., tummy time only for infants), and feasibility, with justification for any omissions. Adoption of the COS is expected to reduce heterogeneity, improve synthesis of effects, and guide future intervention development. Dissemination strategies and subsequent work to identify validated measurement instruments (per COSMIN guidance) are planned to support implementation and uptake.

COS-EPOCH was developed using rigorous, stakeholder-informed methods and identifies 22 core outcomes across nine domains for early childhood obesity prevention trials (birth to five years). Implementing this COS will enhance consistency in outcome selection, measurement, and reporting, facilitating more efficient evidence synthesis and translation. Future research will recommend specific measurement instruments for each outcome to further standardise practice.

- Participant representation skewed towards Australia, the United Kingdom, and the United States; limited participation from LMICs and non-English-speaking contexts. - e-Delphi parent/caregiver participation lower than expected; only two parents/caregivers in consensus meeting. - Virtual consensus meeting held in Australia likely restricted broader international participation due to time zone differences, contributing to an Australian-centric meeting. - The Steering Group lacked non-academic stakeholder members, though non-academic perspectives were included among Delphi and consensus participants. - COVID-19-related disruptions may have affected recruitment and participation. - Consensus meeting had a small sample (n=9), though NGT does not rely on statistical power.