Use of T4 to Predict Outcome in Very Low Birth Weight Infants

Abstract 1487 Poster Session III, Monday, 5/3 (poster 29)

Low T4 has previously been associated with IVH and mortality in very low birth weight infants (VLBW). It remains unclear whether transient hypothyroxinemia is causal in these conditions. Low total T4 in VLBW infants may also be indicative of systemic illness as it is affected by gestational age, brain maturity, nutrition, and other factors. Therefore, we postulated that T4 could be used to predict outcomes. In this investigation we compared the accuracy of T4 and the Score for Neonatal Acute Physiology (SNAP) in predicting outcomes in VLBW infants. Methods: retrospective analysis of infants in a single level III NICU from 7/96-7/98. Infants were included in the analysis if their birth weight was <1500g, received a minimum of 1 cranial sonogram, and 1 newborn screen for thyroid function (n=234). None of the infants were diagnosed with congenital hypothyroidism or received supplemental thyroid hormone. SNAP scoring was calculated on data from the 1st 24 hours of life, and total T4 was collected routinely on the 5th day of life. Statistical analysis included Student's t-test, Pearson correlation, and logistic regression. A p <.05 was considered significant. Results: gestational age of the cohort was 28.5 ± 2.7 wks and birth weight 1066 ± 268 g., mean 24 hr SNAP score was 10.9 ± 5.7, and T4 was 6.8 ± 3.4 µg/dl. SNAP score at 24 hrs correlated with T4 on initial newborn screen (R= -.44, p<.01). Gest. age correlated with T4 (R=.6, p<.01) and SNAP (R = -.46, p<.01). Infants with IVH had a lower T4 (6.0 ± 3.4 vs 7.3 ± 3.2 µg/dl, p=.01) and higher SNAP (13.3 ± 6.3 vs 9.7 ± 4.3, p<.01) compared to infants without IVH. Similarly, infants who died had a lower T4 (4.4 ± 2.9 vs 7.0 ± 3.2 µg/dl, p<.01) and a higher SNAP (17.3 ± 6.3 vs 10.4 ± 4.7, p<.01) compared to surviving infants. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of a T4< 4 µg/dl for predicting IVH (n=67) was 27%, 85%, 42%, and 74% and for a SNAP>15 was 40%, 83%, 50%, 78% respectively. The sensitivity, specificity, PPV, NPV of T4<4 µg/dl for predicting mortality (n=16) was 63%, 84%, 24%, and 97%, and for SNAP>15 were 69%, 81%, 23%, and 97% respectively. After controlling for gestational age, T4 < 4µg/dl was associated with increased odds of IVH (odds ratio 2.1, 95% CI 1.3-3.4, p<.01) and mortality (odds ratio 6.4, 95% CI 3.1-13.2,p<.01). SNAP >15 was also associated with an increased odds of IVH (odds ratio 2.9, 95% CI 1.5-5.5, p<.01) and mortality (odds ratio 5.6, 95% CI 1.9-16.6, p<.01). Conclusions: in our population of VLBW infants, T4 from initial newborn screen can be used to predict IVH and mortality with similar accuracy of SNAP. Further study is needed to validate the ability of T4 to predict outcome.

Author information

Additional information

(Spon by: John L. Stefano)

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Paul, D., Leef, K., Voss, B. et al. Use of T4 to Predict Outcome in Very Low Birth Weight Infants. Pediatr Res 45, 253 (1999).

Download citation