Chaffee 2013.

Methods	Cluster‐RCT (randomisation by health service unit): NCT00635453 (Porto Alegra Early Life Nutrition and Health Study), with 3‐year follow‐up of infants
Participants	715 mothers and pregnant women and their fetuses/infants from 20 health service clusters were randomised. Inclusion criteria: all pregnant women with scheduled clinic visits from April to December 2008 (and their foetuses/infants) in the selected study services (births occurred from May 2008 to February 2009) Exclusion criteria: for health centres, ≤ 100 infant patient visits in 2006; staff‐sharing among clinics or participation in a contemporaneous community‐based dietary programme; for participants: HIV+ mothers; and infants with congenital malformations Setting: Health units in Porto Alegra, Rio Grande do Sul, Brazil (women were enrolled from April to December 2008, child caries assessments occurred from August 2011 to June 2012).
Interventions	Group 1 (n = 360 pregnant women from 9 clusters randomised) Women received dietary advice from healthcare workers who were trained in infant feeding guidelines, namely the "Ten steps of Healthy Diet for Brazillian Children under Two Years of Age", plus written material relating to the dietary advice. The recommendations in these guidelines included: (1) exclusive breastfeeding to 6 months of age; (2) continued breastfeeding to 2 yrs of age, with gradual introduction of complementary foods; (3) at 6 months, start complementary feeding (grains, meat, fruits) 3 times daily while continuing breastfeeding; (4) mealtimes at regular intervals, adjusted to the child’s internal hunger cues; (5) new foods should gradually get thicker until the child is able to eat a family meal, but foods should never be liquefied; (6) provision of a variety of healthy foods every day; (7) daily intake of different fruits and vegetables; (8) avoidance of sugar, sweets, soft drinks, salty snacks, and processed and fried foods; (9) implementation of good hygiene practices in food preparation and handling; and (10) adequate, responsive feeding during illness. The guidelines contained no specific oral heath messages. Group 2 (n = 355 pregnant women from 11 clusters randomised) Women received standard care. Timing: counselling was provided when mothers attended clinics for pre and postnatal visits; no further details on timing of intervention were provided (> 6 months intervention duration). Theory or model used as a basis for intervention: clinical guidelines for early infant feeding, more specifically the “Ten steps for healthy feeding of children younger than two years”
Outcomes	Data in meta‐analysis for: primary: caries presence in primary teeth, dmfs index; secondary outcomes: none Narrative text for: none Tabulated data for: none Additional outcomes that had not been prespecified: Child: none reported. Mother: none reported
Notes	Funding: "The Brazilian Ministry of Health, the Rio Grande do Sul Research Support Foundation (FAPERGS), and NIH‐NIDCR grant F30DE022208 (to BWC) supported this research". Declarations of interest: none declared
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Of the 31 eligible health centres, 16 were initially selected via a witnessed drawing, by the principal investigator, of labelled markers from an opaque container, such that 2 health centres would be included from each of the city’s 8 geo‐administrative districts. Following a stratified randomisation scheme, health centres were block‐randomised by district, with one health centre per district allocated to the intervention and another to the control. To increase statistical power, 4 additional health centres from the original 31, regardless of district, were randomly drawn. Health centre size differed, and thus, to maintain a balanced number of births by group, these additional 4 health centres were block‐randomised at a 1:3 ratio. This yielded 9 intervention and 11 control group health centres".
Allocation concealment (selection bias)	Low risk	Opaque container used to ensure allocation concealment during randomisation
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "the health centers were invited to participate without disclosure of allocation status". It is likely that participants and study personnel were aware of their group assignment.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Quote: "Dental outcomes were available for 64.1% (458/715) of the initial sample. Losses were principally due to withdrawal from the study or inability to locate and did not differ significantly by allocation status...Children available for analysis differed statistically significantly from those lacking dental information for 3 variables: mean maternal age (26.4 yrs intervention vs 25.2 yrs control), proportion having fathers with ≤ 8 yrs of education (49.9% vs 43.3%), and proportion low social class (78.3% vs 82.4%)".
Selective reporting (reporting bias)	Unclear risk	Caries in infants/children was not prespecified as an outcome in the study protocol. Quote: "mother‐child pairs were enrolled at baseline, prior to the decision that dental outcomes would be assessed".
Other bias	Low risk	No signs of other bias