Prevalence of permanent childhood hearing impairment identified by universal newborn hearing screening: A systematic review and meta-analysis

Abstract

Introduction Universal newborn hearing screening (UNHS) programmes have been implemented in many very highly-developed (UN Development Programme, 2015) countries over the last 20 years. The prevalence of permanent childhood hearing impairment (PCHI) detected by these programmes has not been systematically evaluated to assess the expected burden on audiological and other clinical services. In addition, the PCHI risk associated with Neonatal Intensive Care Unit (NICU) admission, a known PCHI risk factor, has not been estimated. We addressed these research gaps using systematic review and meta-analytical techniques to calculate the population-based prevalence of PCHI (defined as more than or equal to 26 decibels of hearing loss in both ears [≥ 26 dB HL]) identified through UNHS (defined as universal screening using otoacoustic emissions and/or auditory brainstem response testing within the first 6 months of life, followed by referral for diagnostic investigations). Our secondary objective was to compare PCHI prevalence in infants who were and were not admitted to NICU. Methods In accordance with the registered protocol (PROSPERO: CRD42016051267), six electronic databases (including Embase and Medline) were interrogated in January 2017 to identify eligible studies in very highly-developed countries to calculate prevalence of PCHI ascertained through UNHS. We identified further reports through searching citations of included papers and unpublished literature (November 2017). Eligible papers could be in any language and published on any date. Papers reporting outcomes for at-risk populations only, with no English abstract (unless unpublished) or of ineligible study type (e.g. case report or letter) were excluded. One reviewer extracted data and quality assessed all papers while two further reviewers undertook independent data extraction and quality assessment of a random sample of papers to ensure consistency; differences were resolved by consensus. Quality was assessed using criteria adapted from the Newcastle-Ottawa scale, STARD and QUADAS-2 tools. Pooled prevalence was calculated using random effects modelling, Freeman–Tukey double arcsine transformation of prevalences and Wilson (Score) method for calculating 95% confidence intervals (CIs) (Stata: Release 15; StataCorp LP). Results Literature searches identified 6195 non-duplicate references, from which 41 were eligible for inclusion (including 5 reports identified from the unpublished literature and citation searching). These reports contained data on 32 separate study populations (median size = 25,945 infants, interquartile range = 11,198, 83,691 infants). Studies took place between 1990 and 2014, with the majority in Europe (n = 23). Pooled prevalence of bilateral PCHI ≥ 26 dB HL identified through UNHS in the screened population was 1.08 (95% CI: 0.90–1.28) per 1000 infants. Infants admitted to NICU had 6.9 times (95% CI: 3.8–12.5) higher prevalence of PCHI than those not admitted, based on 3 studies reporting this information. Smaller studies were associated with significantly larger prevalences (Egger's test: P = 0.017). Key strengths of this review include use of a systematic search strategy and robust statistical methods. The findings are limited to very highly-developed countries. Conclusions In very highly-developed countries, around 1 in every 1000 infants will need referral to clinical services for investigation and management of PCHI. Infants admitted to NICU have a significantly increased prevalence of PCHI. Further analyses will evaluate screen test performance and other risk factors for PCHI. ESRC-funded PhD: ES/J500185/1.