Identification of neonatal hearing impairment: experimental protocol and database management

The purposes of this article are to describe the overall protocol for the Identification of Neonatal Hearing Impairment (INHI) project and to describe the management of the data collected as part of this project. A well-defined protocol and database management techniques were needed to ensure that data were 1) collected accurately and in the same way across sites; 2) maintained in a database that could be used to provide feedback to individual sites regarding enrollment and the extent to which the protocol was complete on individual subjects; and 3) available to answer project questions. This article describes techniques that were used to meet these needs.

This study was a prospective, randomized study that was designed to evaluate auditory brain stem responses, transient evoked otoacoustic emissions, and distortion product otoacoustic emissions as hearing-screening tools, and to relate neonatal test findings to hearing status, defined by visual reinforcement audiometry at 8 to 12 mo of age. Measures of middle-ear function also were obtained at some sites as part of the neonatal test battery. In addition, other clinical and demographic data were gathered to determine the extent to which factors, other than auditory status, influenced test behavior. Three groups were evaluated: neonatal intensive care unit (NICU) infants (those who spent 3 or more days in a NICU), well babies with risk factors for hearing loss, and well babies without risk factors. Six centers participated in the trial. The testers for the project included audiologists, technicians, audiology graduate students, and medical research staff. The same computerized neonatal test program was applied at each center. This program generated the neonatal test database automatically. Clinical and demographic data were collected by means of concise data collection forms and were entered into a database at each site. After the neonatal test, subjects from the NICU and at-risk well babies were evaluated with visual reinforcement audiometry starting at 8 to 12 mo of age. All data were electronically transmitted to the core site where they were merged into one overall database. This database was exercised to provide feedback and to identify discrepancies throughout the course of the study. In its final form, it served as the database on which all analyses were performed.

The protocol was a departure from typical hearing screening procedures in terms of 1) its regimented application of three screening measures; 2) the detailed information that was obtained regarding subject clinical and demographic factors; and 3) its application of the same procedures across six centers having diverse geographic location and subject demographics. A learning curve for successfully executing the study protocols was observed. Throughout the study, monthly reports were generated to monitor subject enrollment, check for data completeness, and to perform data integrity checks. In combination with monthly data reports and checks that occurred throughout the progression of the study, miscellaneous data audits were performed to check accuracy of neonatal testing programs and to cross-check information entered in the clinical and demographic database. The data management techniques used in this project helped to ensure the quality of the data collection process and also allowed for detailed analyses once data were collected. This was particularly important because it enabled us to evaluate not only the performance of individual measures as screening tools, but also permitted an evaluation of the influence of other variables on screening test results.