Effects of liberal versus restrictive transfusion strategies on intermittent hypoxaemia in extremely low birthweight infants: secondary analyses of the ETTNO randomised controlled trial

Objectives: To compare the effect of liberal versus restrictive transfusion strategies on the proportion of time (%time) spent with intermittent hypoxaemia (IH, ie, arterial haemoglobin oxygen saturation measured by pulse oximetry (SpO$_{2}$) <80% lasting ≥60 s) in the ‘Effects of Transfusion Thresholds on Neurocognitive Outcome’ (ETTNO) population, and to investigate whether infants with above-median exposure to IH might benefit more from liberal transfusion strategies than those with lower exposure. Design, setting, patients: Secondary analysis in all 554/1013 infants of <1000 g birth weight recruited into the ETTNO trial (mean gestational age 26.2 weeks) with >80% completeness of SpO$_{2}$ recordings during postnatal days 8–49. Intervention: Randomly assigned liberal (n=268) or restrictive (n=286) transfusion strategies, defining transfusion triggers based on postnatal age and health status. Main outcome measures %time with IH, rate and mean duration of IH episodes during postnatal days 8–49. Interaction between exposure to IH and transfusion strategies with respect to ETTNO’s composite primary outcome, death or disability at 24 months corrected age. Results The median (quartile 1–quartile 3) %time with IH was similar between treatment groups (0.91% (0.13%–2.83%) with liberal vs 0.79% (0.16%–2.44%) with restrictive transfusions). There was no interaction between exposure to IH and transfusion strategies on outcome at 24 months. Conclusions In infants <1000 g birth weight, a liberal transfusion strategy did not reduce IH. Blood transfusions should not be administered ‘liberally’ to reduce IH or to improve neurocognitive outcome in infants with above-average exposure to IH. Trial registration number NCT01393496.