Mardones 2008.

Methods	RCT
Participants	1173 women Inclusion criteria: women age 18 years and over, parity 0‐5, up to 20 weeks’ gestation (confirmed by US), non‐consumers of drugs and alcohol, and underweight (BMI ≤ 21.2 at 10 weeks' gestation, as defined by Chilean charts for pregnant women) Exclusion criteria: multiple pregnancies; suffering from chronic diseases that could affect fetal growth; smokers; and disease diagnosed during pregnancy Setting: 19 urban health clinics belonging to the Servico de Salud Metropolitano Sur‐Oriente (Southeast Metropolitan Public Health Services), Santiago, Chile (study dates not reported). Mainly low‐income, ethnically diverse families (Ameri‐Indian and Hispanic).
Interventions	SUPPLEMENTATION + OTHER AGENT: omega‐3 + omega‐6 + multiple micronutrients + milk versus milk only Group 1: milk product fortified with omega‐3 LCPUFA (0.6 g/day) omega‐6 LCPUFA (3 g/day), multiple micronutrients: total number randomised: n = 589 Group 2: regular powdered milk; total number randomised: n = 552 Timing of supplementation: < 20 weeks to birth (presumed) DHA + EPA dose/day: mid: 600 mg DHA; EPA not stated
Outcomes	Women/birth: preterm birth (< 37 weeks); preterm birth (< 34 weeks); miscarriage; PE; length of gestation; caesarean section; GWG; stillbirth; neonatal death; perinatal death Babies/infants/children: birthweight; infant growth (crown‐heel length, head circumference)
Notes	Funding: Parmalat SpA, Italy, the company that provided the Maman (intervention) product. Declarations of interest: none declared
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote: "Midwives in charge assigned the women in their initial pregnancy visit using the order of arrival: odd numbers to the experimental group and even numbers to the control group".
Allocation concealment (selection bias)	High risk	Quote: "Midwives in charge assigned the women in their initial pregnancy visit using the order of arrival: odd numbers to the experimental group and even numbers to the control group".
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Authors reported that: "the study could not be blinded because Chilean regulations do not allow delivery of food without information on its composition". However, lack of blinding of participants and personnel is unlikely to have introduced bias due to the objective nature of the outcomes measured and reported.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Quote: “For those co‐authors who performed the calculations, groups were simply labelled 1 or 2, and the coding was not known by them either. These measurements should suffice to minimise the possible effects of non‐blinding”. No information was provided on blinding of the individuals who performed the assessments.
Incomplete outcome data (attrition bias) All outcomes	High risk	There were 32/1173 post randomisation exclusions. Data were available for 333/552 (60%) and 365/589 (62%) of women assigned to the control and intervention group respectively.
Selective reporting (reporting bias)	Unclear risk	insufficient information to permit confident assessment.
Other bias	Unclear risk	Among women who discontinued the trial, GA at recruitment was lower in the intervention group.