Background Newborn screening programs for congenital hypothyroidism (CH) using thyroid stimulating hormone (TSH) as a primary screening test has led to a drastic decrease in untreated CH. There is no optimum screening cut-off level of TSH to trigger investigations for diagnosis and treatment. In different health systems, the cut-off value could be a standard (fixed) TSH level (such as 10.0 mIU/L) or based on the 97.5th, 98.5thor even 99thcentile of a normative TSH distribution. An informal survey of Singapore neonatologists showed that a standard 25.0 mIU/L is used as the screening cut-off. A pilot study conducted at Singapore General Hospital (SGH) in 2013 showed that umbilical cord TSH values formed a normal distribution curve, with the 97.5th centile falling at 16.8 mIU/L. Therefore, cord TSH exceeding 16.8 mIU/L formed the basis for screening cut-off TSH values at SGH since 2013.

Objectives The objective of our study was to determine the prevalence of CH amongst newborn infants born at SGH who had umbilical cord TSH levels in the borderline range 16.8 - 25.0 mIU/L.

Methods In this retrospective cohort study, livebirths from 1 January 2015 to 31 October 2020 who had borderline cord TSH levels of 16.8 – 25.0 mIU/Lwere included. Neonates with raised cord TSH ≥16.8 mIU/L had thyroid function repeated on day 6–8 and 2-weekly thereafter. Demographic and birth data were collected, including cord free thyroxine (fT4) and TSH levels, aetiology, age at initiation of treatment, initial dose and length of outpatient follow-up.

Results Our screening program based on cord TSH cut-off ≥16.8 mIU/L resulted in a positive screening rate of 3.63%. A total of 276 of 10,172 neonates (2.71%) born during the study period was found to have cord TSH levels in the range 16.8–25.0 mIU/L. The mean (±S.D.) gestational age was 38.2 (±1.9) weeks and mean (±S.D.) birth weight was 3040 (±486) grams. There were three cases of congenital hypothyroidism amongst 276 livebirths, giving a prevalence rate of 1 in 92. All 3 neonates were born at term with mean birth weight 3,287 grams at a mean gestational age 38 weeks. Of the 3, one was diagnosed postnatally with trisomy 21 complicated by patent ductus and ventricular septal defect, while another was born to a mother with Graves’ disease. All had thyroid dysgenesis and were given daily L-thyroxine replacement.

Conclusions Our current newborn screening program for congenital hypothyroidism based on an umbilical cord TSH cut-off ≥16.8 mIU/L resulted in a positive screening rate of 3.63%. Amongst 276 neonates with TSH levels of 16.8 - 25.0 mIU/L, three term infants were diagnosed with congenital hypothyroidism, giving a prevalence rate of 1 in 92. These 3 cases of CH would have been missed if the screening cut-off level were increased to 25 mIU/L. A better screening TSH cut-off at the 97.5th centile of the distribution curve appeared to be 18 mIU/L instead of 16.8 mIU/L.