Abstract
Background School-based hearing screening can identify hearing loss that develops after birth or becomes apparent later in childhood. Pure-tone audiometry (PTA) and otoacoustic emissions (OAE) are widely used, but diagnostic performance varies across thresholds, technologies, populations and reference standards. Methods We searched PubMed, Scopus, the Cochrane Library and Google Scholar from database inception to 19 September 2026, with supplementary discovery using Elicit, OpenAlex and ASTA. A single reviewer conducted screening, extraction and risk-of-bias assessment, with structured second-pass verification. Eligible studies evaluated PTA or OAE in school-based or school-entry settings against an appropriate audiological reference standard and reported interpretable diagnostic-accuracy data. Risk of bias was assessed using QUADAS-2. PTA studies with reconstructable 2 × 2 data were synthesised using a Reitsma-type bivariate random-effects model fitted by restricted maximum likelihood; OAE evidence was synthesised narratively. Results The search identified 2,198 records; 262 duplicates were removed and 1,936 unique records were screened. Fifty-five reports were sought, 26 were not retrieved and 29 full texts were assessed. Fifteen reports representing 14 unique studies were included. Six independent PTA datasets were included in the primary meta-analysis. Pooled sensitivity was 77.2% (95% CI 58.3–89.1%) and pooled specificity was 92.5% (95% CI 74.8–98.1%). Replacing the 15-dB Robler threshold with the 20-dB threshold yielded sensitivity of 75.3% and specificity of 93.2%. Prediction ranges were wide, indicating substantial between-study heterogeneity. Eight studies were classified as having high overall QUADAS-2 concern. OAE evidence was too heterogeneous and sparse for a defensible pooled estimate. Conclusions PTA shows relatively high specificity but variable sensitivity for school-based hearing screening. Performance depends on screening threshold, implementation, population and reference standard. Current OAE evidence is insufficient for a reliable pooled estimate. Future studies should use standardised thresholds, complete verification, clearly defined reference standards and full 2 × 2 reporting.