Treatment compliance (assessed according to the criteria described above) is shown in Table 11. We assessed compliance by looking at medication pack returns, patient diaries and asking patients what they had been taking. Prior to unblinding, we defined adequate compliance as women in whom the proportion of actual doses of study medication were 80% of those of expected doses.

TABLE 11

Treatment compliance in the ITT population

Compliance was calculated from the expected number of doses taken and the assumed number of doses taken, based on the number of doses issued (usually 84) and the number returned or reportedly lost. If the number returned or lost was not recorded, this were taken as zero. In some cases, this yields implausibly large values for compliance.

Six women had a derived compliance value of > 120%:

1. compliance = 2100% – expected four doses; number of doses returned or lost not recorded; doses taken calculated as 84
2. compliance = 158% – expected 53 doses; number of doses returned or lost both zero; doses taken calculated as 84
3. compliance = 156% – expected 53 doses; number of doses returned = 1, lost = 0; doses taken calculated as 83
4. compliance = 138% – expected 26 doses; number of doses returned = 48, lost = 0; doses taken calculated as 36
5. compliance = 135% – expected 17 doses; number of doses returned = 61, lost = 0; doses taken calculated as 23
6. compliance = 133% – expected nine doses; number of doses returned = 0, lost = 72; doses taken calculated as 12.

The compliance value for subject 1, listed above, is clearly erroneous, as the participant could not have taken all 84 doses within 4 days. This subject withdrew from study treatment very soon after randomisation, delivered shortly afterwards – at approximately 25 weeks’ gestation – and withdrew from the study. The child died within 2 weeks of birth. However, there is no information that indicates that the participant was not compliant with treatment during the time that she was supposedly taking the medication.

Compliance (excluding data from subjects who had missing compliance data) is shown in Table 12.

TABLE 12

Treatment compliance in the ITT population (missing data removed)

Of the individuals indicated below, the following remain (only the 2100% is removed):

1. compliance = 158% – expected 53 doses; number of doses returned or lost both zero; doses taken calculated as 84
2. compliance = 156% – expected 53 doses; number of doses returned = 1, lost = 0; doses taken calculated as 83
3. compliance = 138% – expected 26 doses; number of doses returned = 48, lost = 0; doses taken calculated as 36
4. compliance = 135% – expected 17 doses; number of doses returned = 61, lost = 0; doses taken calculated as 23
5. compliance = 133% – expected nine doses; number of doses returned = 0, lost = 72; doses taken calculated as 12.

Premature treatment withdrawal is shown in Table 13.

TABLE 13

Premature treatment withdrawal in the ITT population

Serious adverse events (SAEs) known to occur in the safety population in the reporting window (maximum of end of treatment date + 28 days and date of delivery + 30 days) or where it is unclear whether or not they are in the reporting window are listed in Table 14.

TABLE 14

Patients with at least one SAE by System Organ Class and preferred term

Serious adverse events known to occur outside the reporting window and those in which the timing was uncertain are also reported separately in Appendix 3.

Other prespecified safety outcomes are shown in Tables 15–17.

TABLE 15

Other preplanned safety outcomes: maternal complications

TABLE 16

Other preplanned safety outcomes: fetal and neonatal complications

TABLE 17

Further preplanned safety outcomes