Table 2

 Methodological quality of the studies included in the review

StudySample size estimateMethod of randomisationAllocationConcealmentBlinding of interventionProportion (%) included in analysis*Blinding of outcomeBaseline variables comparedAssessment of confounding
*This is the percentage included in the analysis after excluding losses to follow-up (for which there were very few for the short-term outcomes) and those with missing values for some outcomes.
Davies9NoRandomised but method not statedNot statedNot stated100%Not statedYesStratification by gestational age
Raiha et al10NoBlocked randomisationCoding of bottlesNot stated95%Not statedYesStratification by gestational age
Schultz et al11NoRandomised but method not statedNot statedNot stated100%Not statedYesNo
Gross12NoRandom number tableNot statedNot stated100%Not statedYesNo
Lucas et al13–22YesBlocked randomisationOpaque, sealed, numbered envelopesNo100% for short term outcomes69% at 9 months 86% at 18 months 82% at 8 yearsFor some outcomesYesFor some outcomes
Cooper et al23NoNot random (alternating assignment)Not statedNot stated100%Not statedYesNo
Putet et al24NoNot stated (may not be randomised)Not statedNot stated92%Not statedYesNo