To investigate the perinatal risk factors for conductive hearing loss (CHL) in infancy and develop an initial prediction model to facilitate accurate diagnosis and early detection of CHL.
This retrospective study utilized data from the Newborn Cohort Study of Hearing Loss (ChiCTR2100049765). Infants who underwent diagnostic audiological assessments at our hospital between January 2003 and June 2024 were included. Data analysis was conducted using R (version 4.4.1) to construct an initial prediction model for CHL in infancy, applying the LASSO regression technique.
A total of 661 infants (1322 ears) were included, with 1253 ears in the normal hearing group and 69 ears in the CHL group. Statistically significant differences were observed between the groups in the following factors: parent-reported infant response to sound, craniofacial deformities, neonatal hemolysis, jaundice treatment, and neonatal hypoglycemia. A multivariate prediction model and nomogram for CHL in infancy were developed and validated, achieving an accuracy of 92.5% and a specificity of 91.3%.
This study identified key risk factors for CHL in infancy and developed a preliminary predictive model, improving the diagnostic accuracy for CHL. Improved diagnostic precision can decrease misdiagnoses, reduce delays in treatment, and limit unnecessary antimicrobial prescriptions for infants.