Population-based newborn screening (NBS) has been carried out in developed nations for the past 40 years [1]. Many factors have limited the establishment of such a service in less developed countries: the high birth rate, low number of institutional births, and poor awareness about preventive health, among others. Most published studies on NBS in India have been carried out in a single-hospital setting [2-8]. For a national NBS program to be initiated, its feasibility has to be assessed in the framework of a regional network, with one center providing laboratory as well as scientific knowhow, serving as the node for transport of samples from all the maternity centers of the region [9]. We carried out such an outreach program for NBS in a predominantly rural and inner city population in and around Lucknow, to assess feasibility and recall rates for congenital hypothyroidism (CH), galactosemia and biotinidase deficiency (BTD) screening.

Biotinidase deficiency: Initial cut offs for recall were ≥77U (Equivocal 50-77 U, presumptive positive <50U). After interim analysis on data of initial 1680 babies, these cut-offs were changed to ≤45 U (presumptive positive < 36 U, partial BTD: 17-35.9 U/l, complete BTD <17 U, equivocal 36.1-45 U). For biotinidase deficiency confirmation testing, repeat heel prick sample and direct enzyme assay in serum was performed by spectrophotometric assay (Synergy HT, Biotek instruments, USA).

Statistical analysis: All statistical analyses were performed using SPSS Version 16. Comparison of means was done by Mann Whitney test and correlation by Spearman test as the TSH data were not normally distributed.

NBS was carried out for 13426 neonates. Seventy-three percent of delivered babies could be successfully sampled. The mean reporting time of results was 8.8 days of age.

Table I shows the trend of mean TSH values; with increasing age of sampling, there was a decline in the mean values . Interim analysis showed our recall rate for CH to be 1.4% using a fixed cut off of TSH >20mIU/L. Using the prevalence of approximately 1:1100 from recent Indian reports [13,14], this recall rate meant that for every infant diagnosed to have CH, a large number (15) were recalled who were found to be normal during confirmatory test. In the interests of using our resources for those babies most likely to have permanent severe CH, after 6500 babies we used the age-related cut offs mentioned above. Using these, recall rate fell to 0.84%.

TABLE I:	Mean Screen TSH (Serum Units) at Various Sampling Ages

Neonates with male sex, low birth weight and those born by vaginal delivery had a significantly higher TSH. There was no effect of gestational age and season of sampling on TSH levels (Table II). Birth weight was the only variable which was significant for correlated with TSH (r= -0.031, P<0.001).

TABLE II Parameters Affecting  TSH Levels on Screening

A total of 139 babies were recalled for confirmatory samples. There was 85.7% response to recall. Eleven children were diagnosed to have CH. Their characteristics are shown in Web Table I. The male to female ratio was 1:2.6. Mean (SD) age at start of therapy was 17.7 (4.6) days. One child was lost to follow up. Of the remaining 10, two were diagnosed to have transient CH. They could be taken off therapy at 5 and 9 months of age, respectively, due to high T4 and suppressed TSH values on minimal doses of thyroxine (12.5µg). They have remained euthyroid on follow-up till 18 months of age.

For the initial 1791 samples, recall rate for galactosemia was 4.1%. After revising cut-offs for recall (as described in methodology), it was 0.32% (43/13426). All neonates who were tested by confirmatory tests were negative for galactosemia after repeat galactose assay and/or GALT assay.

In the initial 1680 samples analyzed, recall rates for biotinidase deficiency was 8%. Based on calculation of our own data, cut-offs for recalling patients was revised. Using these newer cut offs, over all recall rate of biotinidase deficiency was 0.16%. All these 22 children had biotinidase enzyme in the range of partial deficiency in the initial screening sample. None of the babies was found to be affected with biotinidase deficiency on confirmatory test.

In a regional network catering to predominantly rural and low socioeconomic strata population, we were able to demonstrate more than 70% success in sampling, more than 80% success in recall and 90% success rate in follow-up of neonates screened for hypothyroidism. None of the screened infants had galactosemia or biotinidase deficiency.

In our program, early discharge of mothers resulted in an unacceptably high recall rate; this could be due to a residual neonatal TSH surge at the time of sampling. To combat this problem, we adopted age-related cut-offs being used by other NBS programs, who also face the problem of early discharge of mothers [11,12]. Though some neonates with CH may have been missed by these higher cut-offs, a high false positive rate causes unnecessary anxiety to parents and also burdens the system. The additional yield with lowered TSH cut-offs in various programs has been mainly (but not solely) in the category of subclinical or mild or transient hypothy-roidism [15]. Such considerations have led many developed countries to also adopt higher cut-offs [16]. A study from India also showed a high incidence of transient hypothyroidism in children who were being treated for CH, when they were re-evaluated at age of 3 years [17]. After applying age-related cut-offs, our recall rate reduced from 1.4% to 0.84%; however, this is still high in comparison to some other countries using similar cut-offs [18]. Similar high recall rates have also been described from other Asian countries [19]. The high recall rate in our study could be explained by earlier age of sampling, iodine deficiency, and inherently higher TSH in Asian and Hispanic populations [20].

Our finding of significantly higher TSH in low birth weight infants, in male babies, and in those born by vaginal delivery is in consonance with other reports [21,22]. We did not find any significant correlation between gestational age and TSH [23]. It is possible our program had too few samples from preterm infants to bring out significant associations.

Our study had some potential limitations. The sample size was relatively small and may not be adequate for calculation of a true prevalence. We were unable to do maternal TSH levels, maternal anti-TSH receptor antibody and urinary iodine levels which would have been useful for determination of etiology in cases of transient CH. Further, we were able to do repeat sceening at two weeks for babies <35 week gestation and <1.5 kg weight as our program was not equipped to handle the numbers of second screen for all babies between 35-37 weeks gestation and 1.5-2.5 kg weight. We could not detect any patient of biotinidose deficiency or galactosemia to make any interpretation about these disorders.

To conclude, population-based newborn screening could be successfully carried out in our predominantly rural and inner-city population. Age-related cut-off for TSH is useful in dealing with the problem of a high recall rate due to early discharge. Irrespective of the age at sampling, TSH in our newborns appears higher than those in Caucasian populations in developed countries. These results need to be corroborated with larger studies from our country.

Acknowledgements: Participating investigators: Drs. Abhay Verma, Amita Joshi, Anil Kumar Verma, Chitra, Cynthia M, Manu Bhatnagar, Meenu Sagar, Nilamber, Pradeep Tandon, Rajinder Chawla, Ramakant, Rita Raman, Sandhya Pradhan, Sarita Saxena, Savita Bhat, Sr Immaculate, SK Singh, SK Tandon and Vijayalakshmi of the following hospitals: District Women’s Hospital, Barabanki; Ram Manohar Lohia Hospital; Veerangana Avanti Bai Hospital and Veerangana Jhalkari Bai Hospital, Lucknow; and Bishop Conrad Memorial Hospital, Khairabad, Uttar Pradesh. We wish to thank following field workers, nurses and data entry operators: Abhishek Verma, Akhilesh Yadav, Anil Yadav, Ankur Saxena, Arjun Singh, Kamla Singh, Manavendraji Tiwari, Pallavi Tiwari, Richa Pandey, Shiny Davis, Shivam Tiwari and Vibhuti Gupta. For technical help, we thank Pramod Kumar Dubey, Divya Singh and PK Awasthi.

Contributors: VB, SP: conceived and planned the study; VG, KJ, VB, SP, PD, MA, VD, SJ, SG, BG, AP, DK, MK: collected, analyzed the data and participated in patient care; VG, KJ, VB, SP, PD, MA: wrote the manuscript; VB: will act as guarantor of the paper.

Funding: Department of Biotechnology, Government of India (BT/PR11395/SPD/24/335/2008); Competing interests: None stated.