Contradictory mortality results in early 2-dose measles vaccine trials: interactions with oral polio vaccine may explain differences

•A 2-dose vs 1-dose measles vaccine (MV) schedule was beneficial in the first RCT.•However, in two later RCTs mortality tended to be higher in the 2-dose group.•We examined whether oral polio vaccine campaigns (C-OPV) explained the pattern.•With no C-OPV-before-enrolment, the hazard ratio (HR) was 0.79 (0.62-1.00).•When C-OPV was received before enrolment the HR was 1.39 (0.97-1.99). Between 2003 and 2019, three trials (randomised controlled trials [RCTs]) in Guinea-Bissau randomised infants to an early 2-dose measles vaccine (MV) schedule at 4 and 9 months vs standard MV at 9 months. The RCTs produced contradictory mortality results; the effect being beneficial in the 2-dose group in the first but tending to have higher mortality in the last two RCTs. We hypothesised that increased frequency of campaigns with oral polio vaccine (C-OPV) explained the pattern. We performed per-protocol analysis of individual-level survival data from the three RCTs in Cox proportional hazards models yielding hazard ratios (HR) for the 2-dose vs the 1-dose MV group. We examined whether timing of C-OPVs and early administration of OPV0 (birth to day 14) affected the HRs for 2-dose/1-dose MV. The combined HR(2-dose/1-dose) was 0.79 (95% confidence interval: 0.62-1.00) for children receiving no C-OPV-before-enrolment, but 1.39 (0.97-1.99) for those receiving C-OPV-before-enrolment (homogeneity, P = 0.01). C-OPV-before-enrolment had a beneficial effect in the 1-dose group but tended to have a negative effect in the 2-dose group, especially in females. These effects were amplified further by early administration of OPV0. In the absence of C-OPVs, an early 2-dose MV strategy had beneficial effects on mortality, but frequent C-OPVs may have benefitted the 1-dose group more than the 2-dose MV group, leading to varying results depending on the intensity of C-OPVs.