Data source
The data on the analysis of trends of CH incidence were obtained from Chinese Newborn Screening Information System (CNBSIS) established in 2011, which requires all newborn screening centers (NBSCs) to report data. Three categories of data are mainly collected, including the essential information of NBSCs, aggregated number of screened neonates and confirmed CH cases, and individual information of CH cases (including infant's birth date, sex, gestational week of birth, initial TSH concentration, etc.). The process of data collection in CNBSIS have been well described in detail elsewhere12, 13, 18, and Informed consent for newborn screening are collected from infants’ parents. The annual coverage of newborn screening and the number of NBSCs from 2012 to 2019 have been listed in the Supplementary Table S1.
Since CNBSIS did not collect the data about clinical subtypes of the diagnosed CH cases, this study used data from literature to analyze the trend of the proportion of CH subtypes. The search strategies, the inclusion and exclusion criteria and the flow diagrams of literature selection process are shown in Method S1and Figure S1 in the supplementary.
CH screening and diagnosis
The methods and process of CH screening and diagnosis in each NBSC should comply with the Technological Guideline on National Newborn Screening 19. The assay method for detecting primary CH utilizes TSH determination test based on the dried blood spots of whole blood, heel-prick samples from newborns between 3 and 7 days after birth, and three methods are recommended for the laboratories to test TSH, which are time-resolved fluoro immuno assay (Tr-FIA), fluorimetric enzyme-linked immunoassay (FEIA) and enzyme linked immunosorbent sssay (ELISA).The cutoff thresholds used in TSH screening is center-specific, ranging from 6 to 20 mIU/L. Newborns with elevated TSH level need to be called for re-examination using a serum sample, and those with both TSH elevation and a reduction in FT4 can be diagnosed as CH19. Among the infants with CH, hypoplasia, absence, or ectopic position of the thyroid gland are usually termed as thyroid dysgenesis. While when the thyroid gland is in its normal location but cannot secrete a normal amount of thyroid hormone, it is known as thyroid dyshormonogenesis20. Thyroid dysgenesis and dyshormonogenesis are the most common etiology of permanent CH, i.e., a persistent deficiency of thyroid hormone that requires life-long treatment, and transient CH, which is the opposite, refers to a temporary deficiency of thyroid hormone21.
Data statistics
The CNBSIS provides data on the number of screened newborns and CH cases by year and sex for all NBSCs in 31 provinces. All provinces are divided into the Eastern, Central and Western regions (Supplementary Table S2). Due to limited coverage of newborn screening and insufficient population representation in Tibet, the data from Tibet are not included in the final analysis.
CH incidence is estimated by using the number of screened CH cases divided by the total number of screened newborns. We explored the trend of CH incidence using a mixed effects Poisson model using the following formula:
$$\text{ln}\left({\text{C}\text{H}}_{\text{i}\text{j}\text{t}}\right)=\text{ln}\left({\text{N}}_{\text{i}\text{j}\text{t}}\right)+\left({\alpha }+{{\gamma }}_{\text{j}}\right)+{({\beta }}_{1}+{{\delta }}_{1\text{j}})\ast {\text{y}\text{e}\text{a}\text{r}}_{\text{i}\text{j}\text{t}}+{({\beta }}_{2}+{{\delta }}_{2\text{j}})\ast {{\text{y}\text{e}\text{a}\text{r}}_{\text{i}\text{j}\text{t}}}^{2}+{{\beta }}_{3}\ast {\text{s}\text{e}\text{x}}_{\text{i}\text{j}\text{t}}+{{\beta }}_{4}\ast {\text{c}\text{o}\text{f}}_{\text{i}\text{j}\text{t}}+{\epsilon }$$
where \({\text{C}\text{H}}_{\text{i}\text{j}\text{t}}\) is the sex-specific number of CH cases in NBSCs i, province j, year t; N is the sex-specific number of screened newborns; sex is a dummy variable (0-boy, 1-girl); cof refers to the cutoff value of TSH for each NBSC; the coefficient parameter\({\alpha }\) and\({ {\gamma }}_{\text{j}}\) is fixed intercept and provincial level random intercept, respectively; \({{\delta }}_{1\text{j}}\) and \({{\delta }}_{2\text{j}}\) is the random slope for year and year2, respectively; \({{\beta }}_{1}\) to \({{\beta }}_{4}\) are fixed coefficients of the variables\(.\)
In order provide imputations of risk factors at any level of aggregation(over sex groups, year, provinces)22, we applied a Bayesian approach via the Markov chain Monte Carlo (MCMC) method to estimate the parameters of the model and stimulate a sample of 5000 draws from the posterior distribution. Annual point estimate and 95% uncertainty intervals (UIs) of CH incidence for each province was calculated from the 50th and 2.5th to 97.5th percentiles of the draws, respectively. The regional-specific and national CH incidence was estimated by aggregating the provincial CH incidence weighted by the number of live births sourced from the China Statistical Yearbook 2013–2020. Additionally, we divided the initial TSH concentration of CH cases into four intervals of less than 10 mIU/L, 10– 20 mIU/L, 20–40 mIU/L, and greater than or equal to 40 mIU/L on the basis of clinical guidelines23–25. Each TSH interval-specific incidences were separately estimated using the analogous model mentioned above. Based on the model estimates, the annual growth rate (AGR) was used to assess trends in the period. The formula used was as follows:AGR = \(\left(\sqrt[n]{\frac{EV}{BV}}-1\right)\) x 100%, where EV is the ending value; BV is the beginning value; n is the numbers of years.
A random effects meta-analysis model was used to estimate the pooled proportion of CH subtypes and a meta-regression test was used to assess the effect of year, TSH cutoff value, and region at screening on heterogeneity. The coefficient of year yielded from the model was used to measure the trend of the proportion of permanent CH and thyroid dysgenesis.
To explore the potential influential factors relative to the trend of CH incidence, (1) this study analyzed the change of the proportion of newborns screened at different TSH cutoff values to discuss its impact on the trend; (2) we assessed the proportion of preterm birth complicated with CH among the total diagnosed cases of CH each year, along with a sensitivity analysis to elucidate the relationship between increasing preterm birth rate and an increase in CH incidence (Method S2); (3) assuming that the quality of CH screening and diagnosis has gradually improved since the Technological Guideline issued in 2010, this study conducted robust linear regression of incidence AGR in 2012–2019 on the provincial CH incidence in 2012 to assess the impact of increased medical quality on the trend; (4) given that the screening coverage increased from 77.88–96.99% in China, the robust linear regression of incidence AGR on the growth rate of screening coverage was also conducted to investigate the influence of the change of screening coverage on the trend.
Statistical analysis was carried out in STATA 16.1 (StataCorp, College Station, Texas, USA) and R version 4.0.2 (R Foundation for Statistical Computing, http://www.r-project.org).