Context. Transient hypothyroxinaemia is common in infants <30 weeks gestation and is associated with neurodevelopmental deficits; it has no consensus definition. We previously suggested that appropriate ranges for postnatal serum T4 values are at least cord levels corrected to an equivalent gestational age had the fetuses remained in utero.

Objective. To investigate the contribution of prenatal and intrapartum factors (n = 27) to the variations in cord levels of iodothyronines, TBG and TSH; to provide an appropriate definition of transient hypothyroxinaemia.

Design. Cohort study (n = 620) in eleven Scottish neonatal intensive care units

Patients Infants delivered at 23-42 weeks gestation recruited between January 1998 and September 2001.

Results. Using -2SD of adjusted T4 cord levels applied to postnatal day 7 values of equivalent gestational age, 14% of the 23-27 weeks group, 1% of the 28-30 weeks group and 3% of the 31-34 week group are hypothyroxinaemic; using -1SD the respective figures are 41%, 23% and 12%.

Conclusions In the absence of neurodevelopmental follow-up studies to quantify transient hypothyroxinaemia, a pragmatic criterion is necessary for selection of extreme preterm infants into clinical trials of thyroxine supplementation. We suggest the use of serum T4 levels on postnatal day 7 which are below -1SD of adjusted cord T4 levels of equivalent gestational age. This criterion avoids over-recruitment of the more mature infants where universal thyroxine supplementation is detrimental to neurodevelopmental outcome but still allows selection of the least mature entrants where universal thyroxine supplementation is beneficial.