Results
Between 26/05/2015 and 14/08/2017, 1,182 women were screened for trial eligibility at our antenatal assessment unit. A total of 304 women were randomised, nine participants withdrew prior to the infusion, two did not receive the infusion and one entered labour before the infusion. 292 participants received an infusion and 278 were included in the analysis. The trial flow is shown in Figure 1.
The demographic and clinical characteristics of study participants at baseline are presented in Table 1. Participants were enrolled into the trial on average at 32 weeks gestational age. Approximately half of all participants were anaemic at study entry. The first infusion was received on average 11-12 days after screening.
The proportion of participants assigned to the two different trial doses of IVI were not equivalent within a 5% margin at any time point (Table 2). More participants assigned to 500mg IVI required repeat infusions at 4 weeks post infusion compared to those assigned to 1000mg IVI (26/73 (36%) vs 5/67 (8%); estimated difference in proportions, 0.28, 90% CI (0.18, 0.39)). Between 4 weeks post infusion and 6 months postpartum, the difference between the groups reduced but the 90% confidence interval did not lie within (-0.05, 0.05), therefore equivalence was not achieved within a 5% margin (15/88 (17%) vs 9/67 (13%) at 6 months postpartum; estimated difference in proportions 0.037, 90% CI (- 0.059, 0.133)) (Figure 2). The proportion of participants requiring a repeat infusion increased over time in the 1000mg IVI group from 5/67 (8%) at 4 weeks post infusion to 14/70 (20%) at 12 months postpartum. Equivalence was not achieved in the sensitivity analyses using linear mixed models, GEEs with different correlation structures, participants who received repeat infusions or were treated per protocol (Table S1) and the secondary analysis of all randomised participants (Table S2).The proportion of participants who required a repeat infusion at any time during follow-up was also not equivalent (83/152 (55%) vs 34/126 (27%), estimated difference in proportions 0.288, 90% CI (0.191, 0.384)) ( Table S1).
Compared to participants assigned to 1000mg IVI, participants in the 500mg IVI received more than twice the repeat infusion rate prior to delivery, after delivery and overall (0.81 (0.84) vs 0.40 (SD= 0.69), Rate Ratio 2.05, 95% CI (1.45, 2.91), p < 0.001) (Table 3).
Participants who received 1000mg IVI had significantly higher ferritin levels up to 6- months postpartum, compared to those receiving 500mg IVI (Table 3). Similarly, serum iron and transferrin saturation were higher in participants who received 1000mg IVI than participants who received 500mg IVI at 4 weeks post infusion, while transferrin was lower (Table 3). Between group differences had disappeared for all markers of iron status by 12 months postpartum.
No serious adverse events were observed. Minor adverse events were observed in 3% (n=8/276, 2 with missing data) of all participants during the first infusion, including 5/126 participants (4%) who received 1000mg iron (dizziness n=2, hypotension n=1, nausea n=1, chest tightness n=1) and 3/150 participants (2%) who received 500mg iron (nausea n=2, hypotension n=1).  There was no difference between the groups (Fisher’s exact test, p=0.48). No adverse events were observed on subsequent infusions.
PPROM occurred in 6/121 participants in the 1000mg arm only. The likelihood of other maternal complications did not differ between participants who received 500mg and 1000mg IVI (Table 3). Similarly, no difference was found between the length of gestation, neonatal outcomes, or child neurodevelopment between the two doses (Table 3).
A post-hoc analysis of haemoglobin levels indicated these were significantly higher in participants in the 1000mg IVI group compared to the 500mg IVI at 4 weeks post infusion ( Table S3). At each time point, the mean Hb level was in a range indicating sufficiency (>115) in both the 500mg and 1000mg IVI iron groups.