Hearing loss (HL), including sensorineural hearing loss (SNHL), conductive hearing loss (CHL), and mixed hearing loss (MHL), is a congenital or perinatal condition that can profoundly impair language acquisition, cognitive development, and social integration in neonates. Infants admitted to neonatal intensive care units (NICUs) are at substantially increased risk of hearing impairment due to exposure to multiple adverse medical conditions and intensive interventions. Epidemiological evidence indicates that the incidence of SNHL among NICU neonates ranges from 1% to 10%, which is approximately 30-fold higher than the 0.1% prevalence observed in the general newborn population (1, 2). Furthermore, previous studies have reported that 26.3% of affected NICU neonates present with SNHL, 30.4% with CHL, and 20% with MHL (1–3). SNHL in this population is commonly attributed to immature cochlear development and environmental stressors, whereas CHL is more often linked to congenital structural anomalies or acquired inflammatory conditions (4). These alarming statistics underscore the critical need to systematically identify and validate key HL risk factors in NICU neonates.

Otoacoustic emissions (OAE) and auditory brainstem response (ABR) are mainstream tools for neonatal HL screening (3, 5, 6): OAE assesses cochlear function with high sensitivity (7), while ABR evaluates central auditory pathway integrity and behavioral thresholds. Per JCIH recommendations, a two-stage screening protocol, initial OAE followed by ABR for infants who fail OAE, is widely implemented in NICUs and well-baby nurseries. OAE is preferred as a first-line screening tool for its simplicity, cost-effectiveness, and time efficiency (8). The OAE-ABR combination is regarded as the gold standard for NICU hearing screening, enabling the detection of auditory neuropathy spectrum disorders (9) and highlighting the necessity of early HL detection and intervention.

Over the past decades, the JCIH has proposed 10 major risk factors for neonatal HL, including very low birth weight (<1,500 g), exposure to ototoxic medications, and mechanical ventilation for more than five days [Joint Committee on Infant Hearing (JCIH), 1994 Position Statement] (10). However, subsequent studies have yielded inconsistent results regarding the predictive value of these factors. For example, Wang et al. (11)and Lima et al. (12) confirmed a significant association between low birth weight and SNHL, while Abdullah et al. (13) found no such correlation. Arslan et al. (14) identified ototoxic drug exposure as a strong predictor of HL, while Alaee et al. (15) and Hille et al. (16) reported contradictory findings. With respect to surgical history, ductus arteriosus ligation, a common NICU procedure, has been suggested to increase the risk of central nervous system-related hearing impairment in extremely preterm infants (5), but evidence remains limited, and large-scale validation is lacking.

Notably, several critical gaps persist in the existing literature. Most studies have analyzed NICU neonates together with community-born newborns, potentially obscuring NICU-specific risk profiles. Moreover, no meta-analysis to date has systematically evaluated the associations between core risk factors, like birth weight, ototoxic medication exposure, and surgical history, and HL specifically in NICU populations. Therefore, this systematic review and meta-analysis aim to synthesize evidence from retrospective studies using multivariate logistic regression to quantify the strength of associations between these risk factors and HL in NICU neonates. By providing robust pooled estimates, this study seeks to facilitate early identification of high-risk infants and inform targeted screening and preventive strategies.

The findings of this paper were presented under the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (17) and the Cochrane Handbook for Systematic Reviews of Interventions (18). The study protocol was registered with PROSPERO (CRD420250653476).

3,219 records were identified from PubMed (n = 923), Embase (n = 1,078), the Cochrane Library (n = 342), and Web of Science (n = 876). After removal of duplicates, 2,147 records underwent title and abstract screening, of which 1,798 were excluded (1,572 for irrelevant outcomes, 91 for unavailable full text, 83 for insufficient data for meta-analysis, and 52 for non-English publications, reviews, or animal studies). The remaining 349 articles were assessed in full text, leading to the exclusion of 328 studies (97 for inappropriate study design, 82 for incomplete patient or risk factor data, and 128 for failure to meet other eligibility criteria). Ultimately, 21 observational studies were included in the meta-analysis (Figure 1). [Smith et al. (20); Maharani et al. (21); Wang et al. (11); Khairy et al. (9); Abdullah et al. (13); Hajare et al. (22); Dommelen et al. (23); Lima et al. (12); Bhat et al. (24); Gupta et al. (25); Alaee et al. (15); Hoog et al. (26); Robertson et al. (27); Cooper et al. (28); Eras et al. (29); Hille et al. (16); Leslie et al. (30); Arslan et al. (14); Chant et al. (31); Wang et al. (11); Nair et al. (32)] were included. Characteristics of these studies are listed in Table 2.

Quality assessment results are displayed in Supplementary Table S1.

The incidence of HL among neonates admitted to the NICU (1%–10%) is substantially higher than that observed in healthy neonates (approximately 0.1%) (1, 2). As the first comprehensive meta-analysis to systematically evaluate risk factors for HL in NICU neonates, this study provides robust evidence to inform clinical risk stratification and surveillance strategies. Our findings reinforce that genetic susceptibility and environmental injury constitute the two principal dimensions underlying HL in NICU neonates. Genetic risk is exemplified by a family history of HL and pathogenic variants like GJB2 mutations, with previous reports indicating that up to 56% of nonsyndromic HL cases have a positive family history (33). Environmental factors identified in this meta-analysis include prolonged mechanical ventilation and hyperoxia exposure, which may damage cochlear hair cells through hypoxia, oxidative stress, and excessive noise exposure (12, 34). In addition, hyperbilirubinemia (total serum bilirubin >10 mg/dL) exerts neurotoxic effects on retrocochlear auditory pathways, significantly increasing HL risk (35), supporting enhanced auditory monitoring in these neonates.

Subgroup analyses based on birth weight revealed that very low birth weight (VLBW) infants (<1,500 g) have a significantly elevated risk of HL, whereas infants with birth weights ≥1,500 g do not demonstrate a statistically significant increase in hearing impairment. This disparity likely reflects differences in biological maturity and cumulative exposure to pathological and iatrogenic insults. VLBW infants are predominantly premature, with incomplete maturation of cochlear hair cells and delayed myelination of the auditory nerve, rendering them particularly vulnerable to hypoxia and oxidative stress. These infants frequently experience complications like respiratory distress syndrome and severe hyperbilirubinemia, which may disrupt cochlear microcirculation and induce neurotoxic injury to brainstem auditory nuclei. Moreover, the frequent need for prolonged mechanical ventilation and high-concentration oxygen therapy further exacerbates inner ear injury through acoustic trauma and free radical accumulation. However, newborns with a birth weight of >1,500 g, due to higher developmental maturity and less exposure to pathological and iatrogenic injuries, do not show obvious hearing risks (36, 37).

With respect to ototoxic medications, loop diuretics are widely used in the NICU for fluid management, yet their potential contribution to HL warrants careful consideration. Pooled analysis of studies by Eras, Hoog, and Robertson indicated that loop diuretic exposure may increase the risk of HL in premature NICU neonates. Experimental and clinical data suggest that intravenous administration at concentrations exceeding 50 μg/mL can alter endolymph composition and impair cochlear function (38, 39). Similarly, our meta-analysis confirmed a statistically significant association between aminoglycoside exposure and HL, reflecting their well-established ototoxic potential. Aminoglycosides induce cochlear injury primarily through disruption of cell membranes and interference with cellular metabolism (40). Ototoxicity varies by agent, with amikacin and netilmicin exhibiting relatively lower toxicity. Importantly, drug-induced HL is strongly modulated by genetic susceptibility, most notably the mitochondrial 12S rRNA m.1555A>G mutation, and by physiological vulnerability associated with prematurity and systemic inflammation, which can markedly amplify ototoxic effects (41). In adult studies, Vancomycin's HL risk shows a concentration dependence feature (42). When the serum concentration exceeds 40 mg/L, it may cause temporary SNHL, while a concentration of ≥80 mg/L may lead to permanent damage (43). Given the prolonged drug clearance and heightened vulnerability of extremely low birth weight infants, particularly those <1,000 g, subgroup analyses stratified by birth weight are likely essential for accurately evaluating vancomycin-associated HL risk in neonates. However, current evidence remains insufficient, highlighting a critical gap in the literature.

Beyond pharmacological factors, surgical ligation of PDA emerged as an important risk factor for HL. This association may reflect the combined effects of anesthetic neurotoxicity, intraoperative cerebral hemodynamic fluctuations, and underlying congenital vulnerabilities. The impact appears especially pronounced in extremely premature infants, consistent with previous reports (19).

This study has several notable strengths and innovations. First, it systematically validates the applicability of the JCIH risk indicators within the NICU population, clearly demonstrating the clinical relevance of core factors like craniofacial anomalies and family history of HL. Second, it expands the existing risk factor framework by identifying specific ototoxic agents (loop diuretics, aminoglycosides, and vancomycin) and NICU-specific interventions (e.g., PDA surgical ligation) not explicitly addressed in current JCIH guidelines, thereby filling an important gap in neonatal HL risk assessment.

Nevertheless, several limitations should be acknowledged. The limited number of eligible studies precluded subgroup analyses distinguishing sensorineural from conductive HL. Additionally, the small sample sizes in studies examining vancomycin exposure and PDA ligation possibly cause bias. Due to insufficient data, the association between small-for-gestational-age status and HL could not be evaluated, nor could the potential synergistic ototoxic effects of combined aminoglycoside and diuretic therapy be assessed. Future research should prioritize large-scale, multicenter prospective studies to elucidate the synergistic pathogenic interactions among genetic susceptibility, ototoxic medications, and hypoxic injury. Such studies are essential to define safe exposure thresholds for interventions like vancomycin administration and PDA surgery and to establish a more comprehensive, evidence-based risk assessment framework for HL in NICU neonates.