This study used a cross-sectional survey to collect the tabulated data on NHS and diagnosis, which were implemented in 2012 and 2018, respectively, by the medical institution. The data in the 2012 survey covered the period from 2008 to 2010, and the data in the 2018 survey covered the year 2016.
The two surveys were legal and mandatory data collections with the permission of the NHCC, and all the original data were tabulated data that did not involve individual information. The publication of the study has received permission from the NHCC. It meets the conditions of ethical exoneration according to the Council for International Organizations of Medical Sciences [14].
Subjects
The two surveys included all licensed medical institutions providing NHS or a diagnosis of congenital hearing impairment in China. According to the Regulation of Neonatal Disease Screening promulgated by NHCC in 2009 [15], all the medical institutions providing NHS or hearing impairment diagnosis should be licensed by the provincial health administrative department. The licensed NHS institutions are responsible for the primary screening and secondary screening of newborns who fail the primary screening, and the referral of newborns who fail the secondary screening. However, licensed hearing impairment diagnosis institutions are responsible for those who fail the re-screening. The process of hearing screening or diagnosis of hearing impairment should conform to the national technical guidelines for neonatal disease screening issued by the NHCC in 2010 [16]. The operation process specified by the technical guidelines can be seen in Appendix 2.
The 2012 survey included 7,001 institutions that provided NHS and 116 that diagnosed infant hearing impairment, which were located in 30 provinces comprising 1,657 districts/counties. The 2018 survey included 11,875 institutions that provided NHS and 214 that diagnosed impairment, located in 31 provinces comprising 2,664 districts/counties (Appendix 1).
Data collection
This study collected the tabulated data from the two types of institutions using two unified questionnaires. The data on the number of live births, initial screening rates (total and failed), and secondary screening rates (total and failed) were collected from the licensed NHS institutions; data on rates of hearing impairment diagnosis among infants who failed secondary screening were collected from the licensed hearing impairment diagnosis institutions.
The questionnaires were released to provincial, municipal and county health administrative departments step by step, and ultimately to the licensed medical institutions. The staff in charge of the data management (usually the nurse) at each institution were responsible for completing the questionnaire. Then, the health administrative department at each level authorized the MCH hospital/station at the corresponding level to collect and audit the questionnaire (Figure 1).
Quality control
According to the requirements of the Regulation of Neonatal Disease Screening [15], all licensed institutions are responsible for routine data collection, although there is no need to report regularly. To ensure the accuracy of the data, hierarchical data auditing was implemented by a multilevel MCH hospital/station. This means that the county-level authorized MCH hospital/station should audit the data from the institutions in the jurisdiction, and municipal and provincial institutions and the NCBDMC audit the data step by step (Figure 1). NCBDMC interviewed staff members cmpleting the questionnaires at 45 medical institutions in 3 provinces (Guangdong, Chongqing, and Shaanxi provinces) to further confirm the accuracy of the data.
Statistical analysis
UNHS was defined as well-born babies who received NHS before discharge and babies in the neonatal intensive care unit who received hearing screening based on the results of an automatic auditory brainstem response test before discharge [17]. Diagnosis of hearing impairment was defined as well-born babies who failed secondary NHS and were diagnosed with hearing impairment within 3 months after birth, and babies in the neonatal intensive care unit who failed auditory brainstem response screening and were immediately diagnosed with hearing impairment.
To calculate UNHS coverage, the number of newborns who received NHS within four weeks after birth was divided by the number of live births. The detection rate of hearing impairment was calculated by combining the rates of primary and secondary screening failure with the rate of newborns who failed secondary screening and were later diagnosed with hearing impairment (Appendix 3). This aggregate approach to calculating the detection rate was necessary because individual-level data on detection were unavailable. This aggregate detection rate does not take into account neonates admitted to the intensive care unit after birth.
Data were analysed for all 31 provinces in China, which were stratified into three geographic areas based on social and economic development [18]: eastern provinces included Beijing, Tianjin, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, and Guangdong; central provinces included Hebei, Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan, and Hainan; and western provinces included Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang, and Tibet. All data were entered into Epidata 3.0 (The Epidata Association, Odense, Denmark) and analysed using R version 3.6.1 (R Foundation for Statistical Computing, http://www.r-project.org).
Estimated detection rates of hearing impairment in our study were calculated based on data from the 2018 survey. Rates were calculated for all provinces except Chongqing and Tibet, because data were unavailable on the number of Chongqing newborns diagnosed with hearing impairment among newborns who failed secondary screening, and only 1979 of 50,896 live births received NHS in Tibet in 2016. National detection rates were estimated together with 95% confidence intervals (CIs), which were calculated by performing multiple sampling 1000 times based on provincial detection rates.