Thyroid hormone for preventing of neurodevelopmental impairment in preterm infants

Observational studies have shown an association between transiently low thyroid hormone levels in preterm infants in the first weeks of life (transient hypothyroxemia) and an abnormal neurodevelopmental outcome. Thyroid hormone therapy might prevent this morbidity.

To assess whether thyroid hormone therapy in preterm infants without congenital hypothyroidism results in clinically important changes in neonatal and long term outcomes in terms of both benefits and harms.

The standard search strategy of the Neonatal Review Group was used. This included searches of the Oxford Database of Perinatal Trials, Cochrane Controlled Trials Register, MEDLINE, previous reviews including cross references, abstracts, conferences, symposia proceedings, expert informants and journal handsearching in the English language.

All trials using random or quasi-random patient allocation, in which thyroid hormone therapy (either treatment or prophylaxis) was compared to a control in premature infants.

Primary clinical outcomes included measures of neurodevelopmental outcome and mortality. Assessment of trial quality, data extraction and synthesis of data, using relative risk (RR) and weighted mean difference (WMD), were performed using standard methods of the Cochrane Collaboration and its Neonatal Review Group.

Eight studies were identified that compared thyroid hormone treatment to control. Four randomized or quasi-randomized studies met inclusion criteria (Chowdhry 1984, Amato 1989, van Wassenaer 1997 and Vanhole 1997). All studies enrolled preterm infants < 32 weeks gestation, but used different timing, dose and duration of treatment with thyroid hormones. Three studies used thyroxine, whereas Amato 1989 used triiodothyronine. Only two studies with neurodevelopmental follow-up were of good methodology (van Wassenaer 1997 and Vanhole 1997). All studies were of small size with the largest, van Wassenaer 1997, enrolling 200 infants. A lack of comparability of data (neurodevelopmental test or timing of follow-up) prevented meta-analytic pooling of the studies for neurodevelopmental outcomes. There was no significant difference in mortality to discharge (typical relative risk 0.74, 95% CI 0.44, 1.26) in infants who received thyroid hormone treatment compared to controls. In individual studies, no significant differences were found in neurodevelopmental outcomes including risk of abnormal neurological outcome, and Bayley Mental or Psychomotor Development Indices. No data were available for the incidences of cerebral palsy or sensorineural impairment. Fraction of inspired oxygen was lower in infants receiving triiodothyronine in one small quasi-randomized study (Amato 1989), but not in infants receiving thyroxine in a randomized study (Vanhole 1997). No other differences were found to suggest a reduced severity of respiratory distress syndrome in infants receiving early thyroid hormone therapy.

This review does not support the use of thyroid hormones in preterm infants to reduce neonatal mortality, improve neurodevelopmental outcome or to reduce the severity of respiratory distress syndrome. The a posteriori subgroup analyses of data from one study (van Wassenaer 1997) which showed benefits in infants 24-25 weeks gestation should be treated with caution. The small number of infants included in trials incorporated in this review limits the power of the meta-analysis to detect clinically important differences in neonatal outcomes. Future trials should be of sufficient size to detect clinically important differences in neurodevelopmental outcomes. They should consider enrolling those infants most likely to benefit from thyroid hormone treatment such as infants born at less than 27 weeks gestation and use thyroid hormones as treatment instead of prophylaxis.