Association Between In-Utero Exposure to Antibiotics and Offspring’s Hearing Loss: A Systematic Review and Meta-Analysis

TL;DR

Exposure to antibiotics during pregnancy may increase the risk of hearing loss in children, especially with aminoglycosides.

Contribution

This study is the first to systematically review and meta-analyze the link between in-utero antibiotic exposure and childhood hearing loss.

Findings

In-utero antibiotic exposure was associated with a 20% increased risk of childhood hearing loss.

Aminoglycoside use during pregnancy showed a higher risk of hearing loss in offspring.

The overall pooled prevalence of childhood hearing loss was 0.9% among exposed children.

Abstract

Objectives: Antibiotic exposure during pregnancy is common, accounting for over 80% of all medications prescribed. Antibiotics in pregnancy are linked to increased childhood disease risk, through direct toxicity or potentially microbiome dysbiosis. This systematic review investigated the relationship between in-utero exposure to antibiotics and childhood hearing loss. Methods: We searched Ovid Medline, Embase, and PubMed for studies examining antibiotic exposure during pregnancy and its associations with hearing loss in offspring. Studies with children whose mothers had data on antibiotic exposure during pregnancy were selected. The meta-analysis calculated (1) pooled prevalence of childhood hearing loss and (2) pooled odds ratios (ORs) for associations between in-utero exposure to antibiotics and childhood hearing loss. Results: Of 1244 studies identified, 18 met the inclusion criteria. Among 161,053 children exposed in-utero to antibiotics, 4368 developed hearing loss. The pooled prevalence of childhood hearing loss was 0.9% (95% CI 0.0–2.8%, I2 = 99.6%). In-utero exposure to antibiotics was associated with an increased risk of childhood hearing loss (pooled OR 1.2, 95% CI 1.1 to 1.3, I2 = 15.4%). Aminoglycoside exposure during pregnancy was associated with a higher risk of hearing loss (pooled OR 1.2, 95% CI 1.1 to 1.3, I2 = 38.4%), while exposure to other antibiotic classes showed no association. Conclusions: The prevalence of childhood hearing loss among those exposed to antibiotics during pregnancy is high. Although the overall risk appears modest, aminoglycosides are linked to a significantly higher risk, suggesting maternal aminoglycoside exposure may indicate a risk for child hearing loss. Further research is needed to clarify causal pathways and long-term effects of in-utero exposure to antibiotics.

Figures

Full text

1. Introduction

Hearing loss is one of the most prevalent childhood diseases worldwide, with disabling hearing loss—defined as hearing level exceeding 35 decibels—affecting 34 million children in 2021 [1]. The economic burden of hearing loss is significant. According to the World Health Organization, moderate and severe hearing loss in children aged 0 to 14 incurs costs ranging from 24 to 45 billion international dollars annually [2]. The consequences of childhood hearing loss extend far beyond financial burdens, as it can lead to significant challenges such as learning difficulties, impaired communication skills, and adverse neurodevelopmental outcomes [3,4]. Notably, 60% of childhood hearing loss is preventable [5]. Therefore, identifying the causal and modifiable risk factors of hearing loss is essential.

One potential modifiable cause of hearing loss is in-utero exposure to antibiotics. In-utero exposure to antibiotics accounts for about 80% of drugs administered to pregnant women [6]. Studies suggest that antibiotic exposure during pregnancy may elevate the risk of childhood hearing loss from a direct toxic effect on the cells of the ear [7,8], which disrupts the balance of microbial communities essential for health and disruption of the “gut–brain axis” [9], a critical pathway for neurodevelopment and sensory processing. Additionally, certain antibiotics have direct teratogenic effects that may adversely impact fetal ear development, potentially leading to structural anomalies. A study by Conway et al. [10] reported increased risks of congenital ear malformations following in-utero exposure to streptomycin. Previous population-based studies have shown that children exposed to antibiotics in-utero have up to a 20% higher risk of pediatric infections [11], including middle ear infections [12], which are linked to childhood hearing loss. These findings highlighted the needs for a careful evaluation of the risks and benefits of antibiotic use during pregnancy to minimize potential long-term auditory consequences for the offspring.

However, the association between antibiotic exposure during pregnancy and childhood hearing loss risk remains unclear. The association could be influenced by various factors including antibiotic class, dosage, the timing of administration during pregnancy, different hearing loss phenotypes (e.g., severity, laterality, and types) and genetic factors [13]. If in-utero exposure to antibiotics is found to increase the risk of hearing loss in offspring, it could promote better prescription management and safer antibiotic use among pregnant women, ultimately helping to reduce childhood hearing loss.

Therefore, this systematic review aimed to assess the current literature for evidence regarding the association between antibiotic exposure during pregnancy and childhood hearing loss.

2. Materials and Methods

This systematic review and meta-analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement [14].

2.1. Protocol and Registration

Methods for this systematic review and meta-analysis are detailed in the protocol registered with PROSPERO, the International Prospective Register of Systematic Reviews, on 8 June 2024, reference number CRD42024550079.

2.2. Search Strategy

We systematically searched for studies published between 1990 and 15 August 2024, in the databases of Ovid Medline, Embase, and PubMed. Medical Subject Headings and free text words were used to interrogate each database, including search terms for pregnant women and their offsprings (e.g., pregnancy, maternal, prenatal, newborns, babies, toddlers, children, adolescents, and teenagers), antibiotics (including broad classes and specific drugs, see Supplementary Table S1 for details), and hearing loss (e.g., hearing loss, hearing impairment, and deafness). The full search strategy for each database is provided in Supplementary Tables S2–S4.

2.3. Eligibility Criteria

Studies were considered eligible if they fulfilled all of the following criteria: (1) had pregnant women taking antibiotics during pregnancy, (2) recorded hearing loss outcome in offspring at birth or during childhood (0 to 21 years old); (3) study designs include case reports, case–control studies, cross-sectional studies, cohort studies, or randomized controlled trials; (4) reported the association between exposure and outcome or descriptive statistics of the number of children with or without hearing loss and the number of mothers who took or did not take antibiotics during pregnancy; and (5) had full text available in English.

We excluded studies if they met any of the following criteria (1) commentaries, letters, study protocols, treatment guidelines, conference reports, or reviews; (2) non-human studies (e.g., animals and tissue/cell lines); (3) antibiotic exposure not during pregnancy; or (4) reported hearing loss caused by risk factors other than antibiotic exposure (e.g., injury, infection, etc.).

2.4. Study Selection

Studies were uploaded to an online platform, Covidence, which automatically removed duplicates. Two independent reviewers (N.Z. and S.L.) screened the title and abstract, then the full text, of each study according to the eligibility criteria. We calculated an inter-rater reliability score from the screening results using Cohen’s Kappa for two independent reviewers in STATA SE 18. A Cohen’s Kappa value larger than 0.80 (or above 80%) was considered high agreement [15]. Any conflicts were discussed and resolved in consultation with a third reviewer (Y.J.H.).

2.5. Data Extraction

A data extraction form was used to extract study characteristics (authors, publication year, country, study design, sample size, and age of diagnosis of hearing loss in children), antibiotics exposure during pregnancy (e.g., name, dosage, and duration), hearing loss (e.g., testing frequencies, cut-off point, reported ear involved, types of hearing loss, and self-reported hearing loss), the number of reported hearing loss cases, reported associations between in-utero exposure to antibiotics and child hearing loss and confounding adjustment (if any). Data was extracted by N.Z. and independently verified by J.W and Y.J.H.

2.6. Risk of Bias Assessment Tools

Two reviewers (N.Z. and S.L) assessed each included study for methodological quality using three tools. Any discrepancy was resolved after discussion with Y.J.H. These included the Joanna Briggs Institute Critical Appraisal Checklist for Case Reports 2017 (JBICAC) [16], the Newcastle–Ottawa Scale for case–control and cohort studies [17], and the Cochrane Risk of Bias (RoB) tool for randomized controlled trials (RCTs) [18]. For cross-sectional analysis, we used a revised version of the Newcastle–Ottawa Scale. Details of each study’s risk of bias assessment are included in Supplementary Table S6.

2.7. Dealing with Missing Data

When information regarding the variables for analysis was missing from publications, the corresponding authors were emailed to request the data. If the authors did not reply after a second contact attempt, these publications were excluded from the relevant analyses.

2.8. Addressing Differences in Antibiotic Exposure Definitions

As studies varied in antibiotic exposure definitions, we divided them into subgroups to allow comparison and synthesis of results. Studies were mainly grouped according to antibiotic classes provided by the Therapeutic Goods Administration (TGA) of Australia [19] including aminoglycosides, beta-lactams, and macrolides. Antibiotics not classified by TGA were regarded as “Unspecified”, while studies examining multiple antibiotics collectively were assigned to the “Multiple Antibiotics” category. For the analysis, we combined the unspecified and multiple antibiotics groups under the category “Other antibiotics”.

2.9. Statistical Analysis

Prevalence of hearing loss in children with in-utero exposure to antibiotics:

We calculated the prevalence of hearing loss using the following formulae for each outcome reported in each study:

Prevalence of hearing loss = (Number of children with hearing loss)/(Number of children with in-utero exposure to antibiotics) × 100%

In studies reporting zero events, a continuity correction of 0.5 was added to both the numerators and denominators to facilitate statistical analysis [20,21]. The pooled prevalence of hearing loss was estimated by meta-analysis, accounting for anticipated heterogeneity across studies. To address this variability, we applied a random-effects model using the DerSimonian–Laird method to ensure robust and generalizable findings [22].

Associations between antibiotic use in pregnancy and child hearing loss: Meta-analysis and descriptive analysis were used to analyze the data. Odds ratios (ORs) were calculated from studies that reported the relevant descriptive data. A random-effects model using the Restricted Maximum Likelihood (REML) method was used to generate pooled effect sizes including 95% confidence intervals and p-values. The random-effects model [23] was used over other models as it takes variance between studies into consideration.

Heterogeneity between studies was measured using the I^2^ statistic. An I^2^ value of 75% and above is considered significant heterogeneity, 30% to 60% is regarded as modest heterogeneity, and 0% to 30% is non-significant [24]. We conducted subgroup analyses based on antibiotic classes. Publication bias was conducted by using eager’s test and trim-and-fill analysis [25]. All statistical analyses were performed using Stata SE 18 (StataCorp LLC, College Station, TX, USA).

3. Results

3.1. Search Results

Figure 1 outlines the process of study selection through study identification, title/abstract screening, and full-text screening. From 1244 records identified by searching the three databases (Medline, Embase, and PubMed), 1073 papers underwent title/abstract screening after removing 171 duplicate records. A total of 267 passed title/abstract screening and underwent full-text screening, resulting in 17 studies eligible for inclusion. We further included one study from the grey literature.

3.2. Characteristics of Included Studies

Across the 18 included studies, participants were drawn from 13 countries. The studies comprised six cohort studies, five case–control studies, four cross-sectional studies, two case reports, and one RCT. Sample sizes ranged from 17 to 130,067 children in the cohort, case–control, and cross-sectional studies, one to two cases in the case reports, and 125 participants in the RCT. Among the antibiotic classes, aminoglycosides were the most commonly (n = 36,343) investigated, with four studies involving streptomycin exposure, ten including gentamicin, one on kanamycin, and one examining both gentamicin and kanamycin. In addition, one study assessed the effects of multiple antibiotics. Hearing loss was diagnosed at varying ages, with some studies focusing on newborns and others extending into childhood and adolescence. All included studies had only a single time-point of follow up (Table 1 and Supplementary Table S7).

3.3. Prevalence of Hearing Loss in Children with In-Utero Exposure to Antibiotics

Figure 2 shows that among 161,053 children who were exposed to various classes of antibiotics during their mothers’ pregnancy, 4368 of those developed hearing loss, resulting in a pooled hearing loss prevalence of 0.9% (95% CI 0.0–2.8%, I^2^ = 99.6%). When dividing the pooled prevalence by antibiotic class (Supplementary Figure S1), the pooled prevalences were: 0.1% (95% CI, 0.0–2.7%, I^2^ = 99.7%) for aminoglycosides, 3.0% (95% CI, 0.3–8.8%, I^2^ =92.1%) for beta-lactams, 0.1% (95% CI, 0.0–0.1%, I^2^ =0.0%) for macrolides, and 4.0% (95% CI, 1.9–7.7%, I^2^ = 84.6%) for other antibiotics.

3.4. Associations Between In-Utero Exposure to Antibiotics and Child Hearing Loss

Six studies were included in meta-analysis. Overall, antibiotic exposure during pregnancy was associated with a higher risk of childhood hearing loss (Pooled OR 1.2, 95% CI, 1.1–1.3, I^2^ = 15.4%) (see Figure 3). In subgroup analyses on different classes of antibiotics, aminoglycoside exposure during pregnancy was associated with higher risk of hearing loss (pooled OR = 1.2 95% CI, 1.1–1.3, I^2^ = 38.4%). However, exposure to other antibiotics classes (beta-lactams, macrolides, multiple antibiotics, and unspecified classes) during pregnancy showed no association with hearing loss (see Figure 4).

3.5. Descriptive Analysis on Timing, Dosage and Genetic Mutations of In-Utero Exposure to Antibiotics

Only nine studies reported the timing and dosage of antibiotic exposure during pregnancy and hearing loss in offspring with inconsistent findings. For example, Jafari et al. [38] demonstrated that gentamicin exposure during the first trimester was found in one of their child patients with hearing loss. Conway et al. [10] found two children suffering from mild unilateral hearing loss; Jones et al. [41] reported a case of an infant who did not react to sounds and experienced difficulties with speech difficulties after prenatal exposure to a daily dose of 1 g of streptomycin during the first and second trimesters of pregnancy. Similarly, Robison et al. [42] documented two cases of sensorineural hearing loss with the same exposure. However, Lyell et al. [34] showed that the dosage range of 1.5–5 mg/kg of daily gentamicin during active labor or induced labor did not cause hearing loss in newborns.

Descriptive analysis on maternal genetic mutations: One study (Xiong et al.) [30] found that genetic mutations at the GJB2 gene (235delC polymorphism) together with maternal aminoglycoside use increased the risk of congenital hearing loss in a Chinese population.

3.6. Risk of Bias Assessment Results

In total, 18 studies were evaluated for quality assessment using three different assessment tools, appropriate for the specific study designs (Supplemental Material Table S6). Eight studies were categorized as low risk, seven as moderate risk, and three as high risk. Among the eight cohort studies, three were low risk (scoring 3–8 points), and five were moderate risk (5–7 points). Of the four cross-sectional studies, three were high risk (2–4 points), and one was moderate (5 points). In the three case–control studies, two were low risk (7 points), and one was moderate (6 points). Two case reports scored 6 and 7 points (both low risk), and the single RCT was classified as low risk by the Cochrane RoB tool.

3.7. Publication Bias and Sensitivity Analysis

Egger’s test showed no strong evidence of publication bias, with a bias coefficient of 0.04 (95% CI −0.010 to 0.096, p = 0.105). Further analysis using the trim-and-fill method reduced the pooled effect size from 0.12 to 0.03, indicating that correction for publication bias lowered the estimated effect. However, neither the observed effect size (0.12, 95% CI −0.02 to 0.26) nor the adjusted effect size (0.03, 95% CI −0.15 to 0.22) reached statistical significance (Supplemental Table S5).

Sensitivity analysis was conducted by excluding three case reports on maternal aminoglycoside use and associated hearing loss in children, where the total sample size equaled the number of hearing loss events. After excluding these studies, the pooled prevalence was 0.03 (95% CI 0.01 to 0.05, I^2^ = 99.6%), consistent with the direction of the original pooled prevalence (Supplemental Figure S2).

4. Discussion

4.1. Key Findings

Our systematic review found that the pooled prevalence of hearing loss in children exposed to antibiotics during their mothers’ pregnancy was 0.9%. Our study result is nearly 10-fold higher than the prevalence of hearing loss in general population of the US (1–3 per 1000) [43] and Australia (1.4 per 1000) [44]. Overall, in-utero exposure to antibiotics was associated with an increased risk of childhood hearing loss (Pooled OR 1.2, 95% CI 1.1 to 1.3). Specifically, aminoglycoside exposure during pregnancy was associated with higher risk of hearing loss (Pooled OR 1.2, 95% CI 1.1 to 1.3), while other antibiotics classes (beta-lactams, macrolides, multiple antibiotics, and unspecified classes) showed no association with childhood hearing loss.

4.2. Interpretation in Light of Other Studies

We found that maternal aminoglycosides exposure could be a risk factor for childhood hearing loss. Aminoglycoside antibiotics are widely used in pregnancy because of their remarkable effectiveness and broad-spectrum actively against Gram-negative bacterial infections, particularly in severe conditions such as sepsis, pyelonephritis, and multidrug-resistant infections [45]. According to the Therapeutic Goods Administration (TGA) of Australia, aminoglycosides are classified as Category D antibiotics [46], indicating that there is evidence suggesting they can cause fetal harm, including ototoxicity in both the mother and the fetus. However, aminoglycoside antibiotics may be used if the potential benefits outweigh the risks in serious or life-threatening situations. Similarly, the UK Teratology Information Service (UKTIS) advises that systemic aminoglycosides should be used in pregnancy only when no safer alternatives are available. This recommendation is based on the significant risk of fetal hearing loss associated with aminoglycosides. Streptomycin has been classified as FDA Pregnancy Category D with a boxed warning, following case reports of irreversible bilateral congenital deafness linked to in-utero exposure during the first trimester [47]. This is also supported by the systematic review of Yu et al., which suggested that prenatal exposure to streptomycin may be associated with a potential risk of neonatal hearing deficits, particularly with prolonged exposure [48]. Their review reported hearing deficits in 7% (4 out of 59) of maternal streptomycin exposed infants, highlighting concerns about ototoxicity [48]. Although their focus was on streptomycin for postexposure prophylaxis of plague, the overlap with included studies strengthens the evidence for this association.

Our meta-analysis found no association between childhood hearing loss and other antibiotic classes, such as beta-lactams and macrolides. This is further supported by a systematic review by Alsowaida et al. [49], which found no significant association between macrolide exposure and hearing loss, as well as findings from Arısoy et al. [50], indicating that ototoxicity side effects of beta-lactam antibiotics are uncommon. These antibiotics fall under Categories A or B according to the TGA [46], indicating they are generally considered safer to use during pregnancy compared to Categories C or D. Category A and B antibiotics are preferred when treatment is necessary. Our findings offer reassurance regarding their safety during pregnancy from a hearing loss perspective when used appropriately.

4.3. Implications and Future Research Directions

This review reinforces the known ototoxicity of aminoglycosides on the developing fetal auditory system [51]. The observed associations suggest that maternal aminoglycosides exposure could be a risk marker of child hearing loss. These findings carry significant clinical implications, particularly regarding the cautious use of antibiotics during pregnancy. While the overall risk of hearing loss is low, the association with aminoglycosides emphasizes the need for judicious prescribing practices. Clinical guidelines should prioritize safer alternatives when available, balancing the benefits of treating maternal infections against the potential risks to fetal auditory development.

The findings of this review also underscore several critical research gaps. First, studies capturing precise antibiotic dosages are essential to elucidate any dose–response relationships for aminoglycoside-induced ototoxicity. Further investigation is needed into the interplay of clinical factors and co-administered treatments that may exacerbate ototoxicity. For example, conditions such as fever, hypoxia, impaired renal function, and nutritional or antioxidant deficiencies could amplify the effects of aminoglycosides. Additionally, the concomitant use of other drugs, including loop diuretics or neuromuscular blockers, may substantially increase aminoglycoside-induced auditory damage [52]. Third, the dependence on single follow-up points in the included studies limits our understanding of long-term effects. Longitudinal studies are needed to comprehensively evaluate the extended impacts of prenatal antibiotic exposure on child auditory outcomes. Finally, recent genetic studies [13] highlight the role of mitochondrial mutations, such as the m.1555A>G variant, in increasing susceptibility to aminoglycoside-induced hearing loss [53]. These findings suggest a potential role for genetic screening in pregnancy to identify at-risk individuals and enable early interventions, such as avoiding aminoglycosides or implementing enhanced fetal monitoring. However, the prevalence of aminoglycoside-related MT-RNR1 variants varies significantly across populations [54]. Therefore, understanding the distribution of genetic polymorphisms in diverse populations, particularly in multicultural settings like Australia, is essential to inform targeted clinical strategies.

4.4. Strengths and Limitations

Our systematic review employed a comprehensive search strategy and included a range of study designs, including randomized controlled trials, cohort studies, case–control studies, and case reports. With a large overall sample size, our study offers (1) a robust meta-analysis assessing the prevalence of hearing loss among children exposed to antibiotics in-utero; (2) a random-effects model to estimate the pooled effect sizes, presenting the associations between in-utero exposure to antibiotics and offspring hearing loss; (3) a subgroup analysis evaluating the impact of different antibiotic classes on childhood hearing loss prevalence and their effect sizes; and (4) a descriptive analysis of the timing, dosage, and genetic mutations associated with in-utero antibiotic exposure. Among the eighteen included studies, eight were classified as having a low risk of bias, and seven as having a moderate risk, highlighting the high quality of evidence and minimal selection bias.

Our review has several limitations. First, we might have underestimated the associations between in-utero exposure to antibiotics and childhood hearing loss, particularly for non-aminoglycoside antibiotics, due to the limited number of studies reporting detailed timing and dosage of exposure and the inconsistent findings among these studies. Second, the absence of information on the severity of maternal infections complicates efforts to disentangle the potential interplay between infection-related factors and ototoxicity [45,55]. Third, significant heterogeneity across included studies limited the generalizability of our findings. Our findings likely reflect variations in study design, types and timing of antibiotic exposure, definitions of hearing loss, and sample sizes. Furthermore, of the eighteen included studies, only eleven specified the audiological tests used to diagnose hearing loss, and just seven provided clear definitions of hearing loss, which varied widely. This inconsistency hindered sub-analyses based on severity or phenotype and limited cross-study comparability.

5. Conclusions

This study provides evidence of an increased risk of childhood hearing loss following maternal exposure to aminoglycosides, while no significant association was found for other antibiotic classes. Despite high heterogeneity and a notable prevalence of hearing loss among exposed children, these findings underscore the importance of caution when prescribing aminoglycosides during pregnancy. High-quality, long-term studies are needed to clarify if there is a dose–response relationship and inform the development of refined clinical guidelines to optimize antibiotic use in pregnant populations.