Oken 2013.

Methods	3‐arm trial; NCT01126762: 'Food for thought' (pilot study)
Participants	61 women randomised Inclusion criteria: women 12‐22 weeks GA, consuming ≤ 2 fish servings/month, ≥18 years of age, singleton pregnancy, planning to remain in Boston till the birth, women with no contraindications to fish consumption such as allergy, or self‐restrictions such as vegetarian diet Exclusion criteria: as indicated above Setting: Harvard Medical School or participant’s homes, greater Boston, USA (recruited April to October 2010)
Interventions	DIETARY ADVICE: Advice to consume fish versus advice + vouchers versus generic advice Group 1: advice to consume low‐mercury/high‐DHA fish; 8‐page booklet and resources on safe fish consumption and health benefits; weekly emails; total number randomised: n = 20 (17) Group 2: advice and grocery store gift cards to purchase fish; 8‐page booklet on safe fish consumption and health benefits; weekly emails; USD 120 value in gift cards (USD 40 at baseline, first month and second month = USD 10 a week); total number randomised: n = 20 (18) Group 3: generic advice (control): 8‐page generic advice on pregnancy nutrition; weekly emails; total number randomised: n = 21 (20) All women: USD 25 gift card at baseline and completion DHA + EPA dose/day: unclear
Outcomes	Women/birth: fish intake (using 1 month fish intake FFQ) and fruit, vegetables, dairy, nuts and meat; use of DHA supplements; women’s opinions and attitudes about fish consumption; DHA (plasma); mercury (blood and hair); preterm birth; maternal mortality; stillbirth; gestational diabetes; PE; gestational hypertension; induction of labour; caesarean birth; postpartum depressive symptoms
Notes	Funding: NIH; HSPH‐NIEHS Center for Environmental Health, Harvard Clinical Nutrition Research Center; Harvard Pilgrim Health Care Institute Declarations of interest: none declared
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number table
Allocation concealment (selection bias)	Low risk	Sealed opaque envelopes, sequentially opened
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: “All study staff were blinded to group assignment before baseline measures were collected. To minimize bias introduced by non‐blinding of the single research assistant who both delivered the intervention and collected follow‐up data, self‐reported data were collected by self‐administered questionnaire rather than by interview. Laboratory staff, statistical analysts, and study investigators remained blinded to group assignment throughout data collection and analysis.”
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Laboratory staff, statistical analysts, and study investigators remained blinded to group assignment throughout data collection and analysis.”
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	6/61 (10%) women were lost to follow‐up (2 in the advice group discontinued intervention (1 died)); and 1 and 3 in the control and advice + gift card groups respectively Post birth outcomes were not available for 13/61 (21%) women – 9/40 in the 2 intervention arms and 4/21 in the control arm.
Selective reporting (reporting bias)	Unclear risk	Specific figures for most birth outcomes not reported (only direction of effect).
Other bias	Unclear risk	Baseline characteristics similar, except for a higher proportion of women working full time in the advice + gift card group.