Selection bias after entry is avoided in Dannecker 2004, Eltorkey 1994, Harrison 1984, House 1986, Sleep 1984 and Rodriguez 2008 where all the women randomized are included in the analyses. Sleep 1984 and Dannecker 2004 include long-term follow up, with a loss to follow up of about 33% and 40 % of the participants respectively. Klein 1992 shows a loss to follow up rate of 0.71% for primary outcomes to 5% for secondary outcomes. In the Argentine 1993 trial the total number of women randomized was included in the analysis of the primary outcome with a 5% loss to follow up at delivery, 11% at postnatal discharge and 57% at seven months postpartum. Intention-to-treat analysis was performed in all of the studies.