Predicting congenital hypothyroidism in newborns with complex risk profiles by using thyroid-stimulating hormone variations across serial dried blood spots

Objectives Newborn screening for congenital hypothyroidism (CH) relies on thyroid-stimulating hormone (TSH) levels from dried blood spots (DBS). We investigated whether incorporating TSH variation across serial DBS could improve prediction of CH in term and preterm infants with complex risk profiles. Methods Among 207,895 newborns screened, 272 (0.13 %) were diagnosed with CH. TSH variations across 3 serial DBS were analyzed in 6,146 healthy infants (2.96 %). Predictive algorithms were developed using linear mixed-effects models in 1,968 term and 1,387 preterm infants with >= 2 and >= 3 DBS, respectively. Average DBS collection times were 58, 260, and 478 h for term and 63, 350, and 664 h for preterm infants. TSH was measured by GSP neonatal hTSH assay (Revvity). Results Daily TSH variation was influenced by the initial DBS1 value. For DBS1 >5.2 and <1.7 mUI/L, increase and decrease, respectively, in TSH level on DBS2 is detectable, and this particularly occurs in preterm infants. In preterms, CH could be excluded when TSH remained <5 mUI/L on DBS1, <6 mUI/L on DBS2, <4 mUI/L on DBS3, and with daily variation <12 % from DBS1 to DBS3 (sensitivity 100 %; specificity 77.85 %). Term infants with TSH <11 mUI/L on DBS1, <4.5 mUI/L on DBS2, and daily variation <13 % from DBS1 to DBS2 may be ruled out for CH (sensitivity 96.5 %; specificity 66.6 %). Conclusions Distinct predictive algorithms for term and preterm newborns, incorporating TSH variations as daily percentage changes, may improve CH rule out in children with complex risk profiles.