Summary of findings 1. Comparison of early childhood education and care‐based healthy eating interventions for improving the diet of children aged six months to six years.

Comparison of early childhood education and care‐based healthy eating interventions for improving the diet of children aged six months to six years
Patient or population: children aged six months to six years Setting: ECEC services (including centre‐based and family day care) Intervention: healthy eating interventions Comparison: no intervention (including usual care, minimal support or delayed intervention) or alternative, non‐dietary intervention
Outcome	Anticipated absolute effects (95% CI)*		Relative effect (95% CI)	No of participants (studies)	Certainty of the evidence (GRADE)	What happens
	Risk with control	Risk with healthy eating interventions
Diet quality	Mean score (SD) of 57.80 (10.82)a on the Healthy Eating Index (scale 0–100 points; higher scores = better diet quality)	SMD of 0.34 is equivalent to a score 3.68 points better in diet quality	SMD 0.34 SD higher (0.04 higher to 0.65 higher)	1973 children (6 RCTs)	⨁◯◯◯ Very lowb	ECEC‐based healthy eating interventions may increase diet quality slightly but the evidence is very uncertain.
Fruit consumption	Mean (SD) servings of fruit 2.06 (1.15)c	SMD of 0.11 is equivalent to an increase of 0.13 servings of fruit	SMD 0.11 SD higher (0.04 higher to 0.18 higher)	2901 children (11 RCTs)	⨁⨁⨁◯ Moderated	ECEC‐based healthy eating interventions likely increase fruit consumption slightly. This increase corresponds to approximately 20% of the daily in‐care fruit requirements for this age group.   We were unable to pool 7 studies in the meta‐analysis; 4 studies reported positive effects of the intervention, while 3 studies reported negative effects.
Vegetable consumption	Mean (SD) servings of vegetables 1.50 (1.18)c	SMD of 0.12 is equivalent to an increase of 0.14 servings of vegetables	SMD 0.12 SD higher (0.01 lower to 0.25 higher)	3335 children (13 RCTs)	⨁◯◯◯ Very lowe	The evidence is very uncertain about the effect of ECEC‐based healthy eating interventions on child consumption of vegetable servings.   We were unable to pool 8 studies in the meta‐analysis 7 studies reported positive effects of the intervention, while 1 study reported negative effects.
Non‐core foods (i.e. less healthy or discretionary) consumption	Times non‐core foods were consumed: mean 0.6; SD 0.85f	SMD of −0.05 is equivalent to 0.04 less times non‐core foods were consumed	SMD 0.05 SD lower (0.17 lower to 0.08 higher)	1369 children (7 RCTs)	⨁⨁⨁◯ Moderateg	ECEC‐based healthy eating interventions likely result in little to no difference in non‐core foods consumption.   We were unable to pool 3 studies in the meta‐analysis; 2 studies reported favourable effects of the intervention, while 1 study reported unfavourable effects.
Sugar‐sweetened beverage consumption	Portions or servings: mean 1.85; SD 1.60c	SMD of −0.10 is equivalent to 0.16 fewer portions or servings of sugar‐sweetened beverages	SMD 0.10 SD lower (0.34 lower to 0.14 higher)	522 children (3 RCTs)	⨁⨁⨁◯ Moderateh	ECEC‐based healthy eating interventions likely result in little to no difference in sugar‐sweetened beverage consumption.   We were unable to pool 3 studies in the meta‐analysis; 2 studies reported favourable effects and 1 study reported unfavourable effects of the intervention.
Cost‐effectiveness	6 studies reported on various intervention costs (i.e. total cost of intervention, projected lifetime savings, cost for participants, average cost‐effectiveness ratio, and relative value index) as measured via surveys, records or logs. 1 study reported that healthy eating interventions were cost‐effective, 2 studies reported cost per child, and 3 studies reported intervention delivery costs, all of which varied across studies.		‐	101 services (6 RCTs)	⨁◯◯◯ Very lowi	ECEC‐based healthy eating interventions may be cost‐effective but the evidence is very uncertain.
Adverse consequences	3 studies reported no additional adverse consequences (i.e. incidents, negative feedback or parent complaints) as measured via staff self‐report.		‐	62 services (3 RCTs)	⨁◯◯◯ Very lowj	ECEC‐based healthy eating interventions may have little to no effect on adverse consequences but the evidence is very uncertain.
* The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% confidence interval). The basis of assumed risk is provided in footnotes. CI: confidence interval; ECEC: early childhood education and care; RCT: randomised controlled trial; SD: standard deviation; SMD: standardised mean difference
GRADE Working Group grades of evidence High quality: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate quality: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low quality: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low quality: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

^aBaseline estimates and standard deviation used from Vaughn 2021.
^bDowngraded one level for risk of bias: three RCTs were assessed as high risk of bias across three domains and one RCT was assessed a high risk of bias across one domain. Downgraded two levels for heterogeneity: I² = 91%; > 75% considered considerable heterogeneity. Downgraded one level as publication bias was strongly detected via visual inspection of funnel plots, which indicated two outliers.
^cBaseline estimates and standard deviation used from Kornilaki 2021.
^dDowngraded one level for risk of bias: seven RCTs were assessed as high risk of bias across three or more domains, two RCTs were assessed high risk of bias across two domains.
^eDowngraded one level for risk of bias: eight RCTs were assessed as high risk of bias across three or more domains, three RCTs were assessed high risk of bias across two domains. Downgraded one level for heterogeneity: I² = 70%; 50%‐75% considered substantial heterogeneity. Downgraded one level as publication bias was strongly suspected from visual inspection of funnel plots, which indicated an outlier.
^fBaseline estimates and standard deviation used from Yoong 2020a.
^gDowngraded one level for risk of bias: five RCTs were assessed as high risk of bias across three or more domains, two RCTs were assessed high risk of bias across two domains.
^hDowngraded one level for risk of bias: two RCTs were assessed as high risk of bias across three or more domains.
ⁱDowngraded one level for heterogeneity as there was some degree of heterogeneity of the measures used to assess this outcome. Downgraded two levels for indirectness as RCTs did not directly answer this research question, with different outcomes assessed. Downgraded two levels for imprecision due to small number of services analysed: 101 services were included in cost analysis. Downgraded one level for publication bias as RCTs were relatively small, all including fewer than 50 services; publication bias may be present.
^jDowngraded one level for indirectness as RCTs did not directly answer this research question, although similar outcomes were assessed. Downgraded two levels for imprecision due to small number of services analysed: 62 services were included in the analysis of this outcome. Downgraded one level for publication bias: all four RCTs were relatively small, all including fewer than 40 services for this outcome analysis; publication bias may be present.