Main

Hearing loss is common and occurs in 1–3 per thousand newborns.1,2 Universal newborn hearing screening (NHS) performed shortly after birth employs bedside physiologic hearing screening modalities, automated auditory brainstem evoked response (ABR), or otoacoustic emissions (OAE), with results, pass or refer (fail) provided to parents and medical providers.3 The rationale behind universal NHS is evidence based. Timely intervention, before 6 months of age, results in better language outcomes. Infants diagnosed in toddler months generally do not achieve communication skills on par with peers.4,5

Hearing loss-associated variants (we have employed the phrase ‘hearing loss-associated variant' to express a change in DNA sequence that when present, leads to the phenotypes of deafness or hard of hearing) in the gene GJB2, encoding the gap junction protein, Cx26, are the most common genetic cause of nonsyndromic prelingual sensorineural hearing loss worldwide.6–9 Cx26 and the genomically and functionally related gene, Cx30, are abundantly expressed in the cochlea.10 In the Midwestern United States, hearing loss-associated variants in Cx26 are identified in nearly 50% of individuals with nonsyndromic sensorineural hearing loss.11

Connexin-related hearing loss is autosomal recessive; to have hearing loss, an individual must inherit two hearing loss-associated variants, one from each parent. Carrier frequencies for hearing loss-associated variants range from 1 to 4%11–13 comparable with the carrier frequency in the white population for cystic fibrosis (3–4%). Rare autosomal dominant Cx26 alleles have also been reported and are frequently syndromic with dermatologic disorders.14 There are common founder alleles identified within various ethnic groups.14,15 35delG is the most common in the white population, 167delT is the most common in the Ashkenazi population, 235delC and V37I predominate in the Asian population, and R135W in the East African population.16 A deletion of 309 kb in Cx30 was identified and is known as del(GJB6-D13S1830). This Cx30 deletion inherited in trans with a Cx26 hearing loss-associated variant results in hearing loss.17,18

Current clinical practice advocates genetic testing for infants diagnosed with hearing loss.19 Discussions have been presented toward inclusion of genetic testing into NHS protocols.20,21 One step toward inclusion of genetic testing into NHS practice requires the characterization of the prevalence of Cx26/30 hearing loss-associated variants, ideally in diverse populations. Data regarding the frequency of connexin-related hearing loss in non-white populations in the United States is largely unknown, particularly in individuals of Mexican American origin residing in Southern California.22

The onset of connexin-related hearing loss begins prelingually with suggested congenital onset.16 This would indicate that NHS would sufficiently identify all infants with connexin-related hearing loss. Recent case reports suggest that some infants with connexin-related hearing loss were not identified by NHS.23–25 Systematic correlation of NHS results and genetic testing in a series of infants has not been carried out.

The phenotype of connexin-related hearing loss is typically described as nonsyndromic and sensorineural, ranging in severity from mild to profound.15,26 Significant genotype/phenotype correlations have been established; protein truncating variants produce severe to profound hearing loss whereas amino acid substitutions produce mild to moderate hearing loss.15 However, comparisons of hearing loss severity between infants with connexin-related hearing loss versus those without connexin-related hearing loss have not been carried out to determine a priori which infants are more likely to have connexin-related hearing loss.

Risk factors associated with infant hearing loss include neonatal intensive care unit stay, aminoglycoside exposure, infections, and hyperbilirubinemia.27 Correlation of neonatal hearing loss risk factors with Cx26/30 gene status has not been performed to determine whether infants who already have hearing loss risk factors would be more or less likely to have connexin-related hearing loss.

We report the results of a study to determine the prevalence of Cx26/30 genotypes and allele frequencies in a diverse US population of infants with hearing loss. We correlated NHS outcomes, hearing loss severity and common neonatal factors such as prematurity, birth weight, Apgar scores, hyperbilirubinemia, aminoglycoside exposure, and family history with presence or absence of hearing loss-associated variants to determine factors to help predict which children are more likely to have connexin-related hearing loss. This is the first study of this kind that makes an effort to differentiate children with connexin-related hearing loss from those who do not.

METHODS

Participant recruitment

Parents of potentially eligible infants and toddlers were made aware of our study by brochures placed in nurseries, clinics, and education centers or through referral from audiology, otolaryngology, or intervention-based programs. A second route was through the newborn nursery at UCLA Medical Center. Families (parents and infant/toddler) were enrolled though an informed consent process approved by the relevant Institutional Review Boards. A board certified genetic counselor provided genetic counseling and informed consent directly in English or Spanish.

Infants and toddlers less than 3 years of age diagnosed with hearing loss (n = 99 infants) were eligible to participate. Newborns referred after failing NHS (n = 6 newborns) were also eligible to participate before diagnostic audiometric evaluation.

Parents were offered genetic testing on their infant from a buccal brush specimen. After completion of laboratory studies, nearly all parents returned in person to receive results and genetic counseling. Prenatal, birth, postnatal, NHS, and audiologic records were collected, reviewed, and abstracted throughout the 4-year course of the study. Some or all records were received for 102 of 105 infants and toddlers in the study. Of the 105 infants, 95 were confirmed to have permanent hearing loss and are reported here.

A 3-generation pedigree was obtained with careful attention to history of childhood hearing loss in first degree (parents, siblings), second degree (grandparents, aunts/uncles), and third degree (first cousins) relatives. Demographic information and ethnic and racial identification were obtained from the parents for their infants.

Connexin variant analysis

Genomic DNA was isolated from fresh buccal brush specimens from the participating infants and toddlers using standard reagents (Qiagen, Valencia, CA).

All Cx26/30 diagnostic assays were carried out under standard clinical molecular genetics protocols in the CLIA-certified Orphan Disease Testing Center and Diagnostic Molecular Pathology Laboratory at UCLA.

Variant analysis for the Cx26 gene was carried out by bidirectional sequencing of both exons. Genomic DNA was amplified using primers to exon 1 (Forward: gtgtggggtgcggttaaaaggcgccacgg and Reverse: agggaccgcgagacccagagcggttgc) and exon 2 (Forward: tttcctgtgttgtgtgcattcgt and Reverse: gccagtttaacgcattgccc [bold denotes mismatch]). Two additional internal sequencing primers for exon 2 were designed against genomic sequence (Forward: tcaagggggagataaagagt and Reverse: gaggagatcaaaacccagaag). Amplicons were subjected to dideoxy-sequencing and analysis on either on a Beckmann CEQ 8000 (Beckman-Coulter, Fullerton, CA) or an ABI 3130 (Applied Biosystems, Foster City, CA) capillary electrophoresis system.

The 309kb deletion in the Cx30 gene, del(GJB6-D13S1830), was identified by a published assay.18

Cx26 V27I polymorphism studies

One hundred purified DNA samples from the control population “The Mexican American Population of Southern California” were purchased from the Coriell Institute (Camden, NJ) consisting of genomic DNA from healthy persons having at least three of four grandparents from Mexico. Samples were assayed for the V27I polymorphism by amplification of exon 2 of the gene encoding Cx26 using primers CX26Ex2F (gttcctgtgttgtgtgcattcgt) and CX26Ex2R (gggcaatgcgttaaactggc) and subjected to digestion with BstF51 (New England Biolabs, Ipswich, MA).

Audiometric analysis

For each subject, audiology records were scrutinized to determine degree, configuration, and type of hearing loss in each ear. Thresholds were extracted initially from an electrophysiologic measure, either ABR thresholds (tonal responses if available or clicks if no frequency specific data were available) or in some cases from Auditory Steady State Responses. Thresholds obtained by behavioral evaluation such as visually reinforced audiometry or standard audiometric techniques were extracted on older children. In most cases, an ear-specific behavioral response was eventually obtained but when necessary, interpretation of sound-field presentations for determining better ear thresholds was used. Both air and bone conduction pure tone and speech thresholds were evaluated to determine type of hearing loss, when bone conduction was available. In some cases, normal tympanometry was the only indication of the status of the conductive mechanism and if so, this information was used. The OAE evaluations were used to judge test consistency and to determine whether auditory neuropathy was a possible factor.

For most infants, it was possible to compute ear-specific pure tone averages (PTAs) from at least two of four frequencies (500, 1000, 2000, and 4000 Hz) and PTA from the better performing ear was analyzed. Audiometric data from the most recent and reliable evaluation was selected for most analyses. However, multiple audiometric time points were available on many of the infants/toddlers, allowing us to determine type, laterality, and progression of hearing loss. Permanent, sensorineural hearing loss was defined as a combination of absent OAEs, air-bone gap less than 15 dB or at least one normal tympanogram. A conductive component was defined when an air-bone gap greater than 15 dB was present or when multiple abnormal tympanograms were noted. Auditory neuropathy was defined when a subject demonstrated more than a mild hearing loss, poor or absent ABR, and present OAEs. For descriptive purposes, hearing loss severity was defined based on the PTA in decibel hearing loss as ≤20 dB for normal hearing, 21–39 dB for mild hearing loss, 40–69 dB for moderate hearing loss, 70–94 dB for severe hearing loss, >95 dB profound hearing loss.

Statistical analysis

Data were double-entered to minimize error and analyzed using SAS v.9.28 Descriptive statistics were computed for detection of anomalies, examination of variable distributions, and descriptive purposes. Due to non-normality of the distributions of PTAs, gestational age, Apgar scores and birth weight, comparison of these variables as a function of number of hearing loss-associated variants was analyzed using the nonparametric Kruskal-Wallis test. The proportion of heterozygotes in our sample was compared with the expected population proportion using a test of a single proportion. The frequency of the V27I allele in our sample was compared with the frequency of the V27I allele in a control sample using the test of two proportions. Fisher exact (FE) test was used to test for independence between number of hearing loss-associated alleles and ethnic origin, frequency of newborn complications, and family history of hearing loss and between NHS result and screening method. Statistical significance was set at 0.05. For analysis of patients with Cx26/30 hearing loss-associated variants, both truncating and nontruncating were grouped together. Variants listed in Table 2 are “hearing loss-associated variants” and not to be confused with polymorphic variants such as V27I.

Table 2 Genetic testing results (n = 95 families; n = 93 genetically independent infants)
Full size table

RESULTS

Over a three and one half year period, December 2002 through June 2006, infants and toddlers were enrolled into the study. Six newborns referred for outpatient NHS entered the study before diagnostic audiometric evaluation. Of these infants, five passed outpatient NHS and one infant had moderate unilateral sensorineural hearing loss. Ninety-nine infants and toddlers entered the study after parental report of a diagnosis of hearing loss. Diagnosis of hearing loss could not be confirmed in three infants because of unavailability of audiology records and two infants were determined to be hearing.

The final analysis group consisted of 95 infants with confirmed permanent hearing loss. Two sets of siblings are included in this sample for a total of 93 independent families. Table 1 provides descriptive information on this sample.

Table 1 Demographic of participants
Full size table

The race and ethnicity of the infants of this cohort was as follows: Hispanic origin (37%) with bi-parental origins primarily from Mexico (n = 21). The mean age at diagnosis for 88 infants for whom the initial diagnostic audiology records were available was 7.4 months (SD = 8.3). The mean age of enrollment was 13.2 months (SD = 10.8) (Table 1).

Genotypes consistent with autosomal recessive connexin-related hearing loss were identified in 23 families yielding a frequency of 24.7% (93 genetically independent infants). Genotypes are listed in Table 2.

In 10 infants, only one hearing loss-associated variant was identified, yielding a heterozygote frequency of 10.8% in the sample of 93 independent infants. This proportion of infants carrying only one hearing loss-associated variant is significantly greater than the reported United States carrier frequency of 3.01% (P < 0.001).11 For 64.5% of the infants in this study, no hearing loss-associated variants were identified (Table 2).

We found connexin-related hearing loss in 33.3% of infants of non-Hispanic white origin, in 46.2% of infants of Asian origin, and 9.1% in infants of Hispanic, primarily Mexican, origin. The frequency of connexin-related hearing loss in infants of Hispanic origin is significantly lower than that found in infants of non-Hispanic white and Asian origin (FE, P = 0.04), suggesting that biallelic changes in Cx26/30 are less frequently observed as the cause of hearing loss in infants of Hispanic origin (Fig. 1).

Fig 1
figure 1

Hearing loss-associated variant detection rate by ethnicity. These data exclude the second sibling with hearing loss in the two families with two participating infants. **Stars denote a statistically significantly lower frequency of biallelic hearing loss variants in persons of Hispanic origin compared to non-Hispanic white and Asian infants. Infants of African American heritage and mixed heritage were excluded from this analysis.

Full size image

In infants of Hispanic origin, 21 of whom were of Mexican origin, a high frequency of the Cx26 V27I allele not in association with any other allele was observed (supplemental Table 1, available online only). To determine whether this allele is associated with hearing loss we tested 100 control samples. We found that the frequency of V27I in our sample of infants of Mexican origin was 19% and not significantly different than the prevalence in control samples at 24% (P = 0.6). V27I is a polymorphic variant and not associated with hearing loss.29

In this study, we calculated the allele frequency for the hearing loss-associated variants identified over the group of 93 independent infants. As expected, 35delG was the most commonly encountered allele in this population with a frequency of 0.13 of total possible alleles including “wildtype,” followed by V37I (0.05), and 167delT (0.03) (supplemental Table 2, available online only). Other alleles, 235delC, del(GJB6-D13S1830), V27I;E114G (in cis)30 and M34T were each present at a frequency of 0.01. All other alleles were present at a frequency of <0.01. The distribution of alleles between ethnic groups showed differences. The 35delG was highly prevalent in infants of non-Hispanic white origin (allele frequency = 0.25) but also prevalent in patients of Hispanic origin (0.08). The allele 167delT was found in both infants of non-Hispanic white and Hispanic origin (0.08 and 0.02, respectively). The alleles 235delC, V37I, and V27I;E114G (in cis) were only found in infants of Asian origin (supplemental Table 2, available online only).

In this study, 82 infants underwent bedside NHS (Table 3). Information on screening method was available on 69 of these infants: 64 of 69 (92.8%) were screened with ABR, four (5.8%) with OAE, and one (1.5%) with a combination of ABR and OAE. When stratified by screening result, screening method data were available on 87% of those who failed NHS and on 67% of those who passed NHS. There was no statistical association between screening result and use of ABR or OAE (FE, P = 0.36). There was no statistical association between screening result and number of connexin-related hearing loss associated variants (FE, P = 0.38).

Table 3 Newborn hearing screening results in 82 infants
Full size table

Seventy of the 82 screened infants failed NHS, with a mean age at hearing loss diagnosis of 4.9 months (SD = 5.8). Twenty infants, 29% of those identified by NHS, were identified to have connexin-related hearing loss. Thus, connexin-related hearing loss occurs in a significant proportion of infants who fail NHS.

Twelve infants with hearing loss passed NHS (Table 4). In this group, five had some hearing thresholds in the mild range in at least one ear. Of these five infants not detected by the NHS process, one was found to be homozygous for the V37I Cx26 hearing loss-associated variant. This infant was screened by ABR and his hearing loss was diagnosed at about 21 months of age. The V37I variant is the result of a missense mutation and data suggest that this variant produces a mild to moderate range hearing loss.15,31,32 Hence, infants with this Cx26 variant may be more likely to pass NHS as a result of a milder audiologic phenotype than infants with protein truncating variants.

Table 4 Types of hearing loss by genotype
Full size table

Seven of the 12 infants who passed the NHS were found to demonstrate bilateral severe to profound hearing loss. The chance of passing the screening in both ears for an infant with severe, bilateral hearing loss is very small. Of these seven, two (unrelated) infants were known to display biallelic Cx26 variants with one having an older sibling with hearing loss. In both cases, the infants were compound heterozygotes for 35delG and 167delT. One infant, with an older sibling with hearing loss, was screened with OAE and was identified at approximately 2 months of age with profound hearing loss. The other infant was screened with ABR and was identified about approximately 1 year of age with a severe hearing loss. As there was no evidence of hearing loss progression (data available upon request), that is, no change in hearing levels over time once detected, in these infants, we must consider the possibility that this genotype predisposes to delayed onset of hearing loss. Our data suggests that nonpenetrance of connexin-related hearing loss at birth is 8%.

Table 4 shows the types of hearing loss in our sample separated by genotype. For infants with connexin-related hearing loss, all infants had bilateral sensorineural hearing loss with severity ranging from mild to profound (Tables 4 and 5). For infants who did not have connexin-related hearing loss, the hearing loss phenotypes were more variable and included unilateral, mixed, conductive, and auditory neuropathy. These data show that infants without connexin-related hearing loss have a broader range of phenotypes than infants with connexin-related hearing loss. The difference in PTA as a function of number of hearing loss-associated variants among the 85 infants with bilateral sensorineural hearing loss was significant (Kruskal-Wallis F(2,82) = 3.56, P = 0.03). The median PTA (Table 5) from infants without connexin-related hearing loss was more severe.

Table 5 Severity of bilateral SNHL by genotype based on PTA from better performing ear
Full size table

We examined the rates of neonatal risk factors as a function of genotype (supplemental Table 3, available online only). These analyses focused on infants with either two or zero connexin-related hearing loss-associated variants. We found no statistical difference in gestational age or birth weight or differences in Apgar scores, aminoglycoside exposure, hyperbilirubinemia, sepsis, or meningitis between groups. Our sample size provided 70–100% power to detect differences on the quantitative variables as small as 0.67 SD, and 70% power to detect a difference between proportions of at least 20%.

We studied the association of family history with number of allelic variants (Table 6). Family history of hearing loss has been listed as one of the risk factors for hearing loss.27 Although nearly all infants in this study were born to hearing parents, through obtaining a three-generation pedigree on each participating family, we identified 17 families with a history of childhood hearing loss in at least one family member. Taken in aggregate, the presence of either first, second, or third degree relatives in the family history was significantly correlated with the presence of biallelic variants in Cx26/30 (FE, P = 0.003). Thus, presence of childhood hearing loss in a family member increased the likelihood of connexin-related hearing loss caused by biallelic hearing loss-associated variants in Cx26/30 (odds ratio = 6.8, 95% confidence interval = [2.37–17.7]). In the scenario of a positive family history, notably a second degree relative with hearing loss, the index of suspicion for connexin-related hearing loss should be increased.

Table 6 Family history by genotype
Full size table

DISCUSSION

We analyzed data from a longitudinal study to determine prevalence of connexin-related hearing loss in an ethnically diverse group of infants and toddlers under 3 years of age confirmed to have hearing loss and correlated the prevalence of hearing loss-associated variants with ethnic origin, NHS status, hearing loss severity, family history, and newborn risk factors.

Participants entered the study via parental interest from brochures placed in nurseries, clinics, and education centers or from recommendations of our study from audiologists, otolaryngologists, and deaf educators. In this way, our sample may be biased toward infants with more severe hearing loss or parents who were more motivated to participate in our study.

In this diverse group of infants, ∼25% were confirmed to have connexin-related hearing loss. This prevalence falls within the range of previously published prevalence of Cx26/30 hearing loss-associated variants of 12–16% in a predominantly pediatric age group of mixed hearing loss type33–35 to 40–50% in populations of adults and children with apparently nonsyndromic sensorineural hearing loss.6,11 However, as our study group was relatively homogenous in the age of ascertainment and type of hearing loss but diverse in ethnic representation, the prevalence of connexin-related hearing loss differed predominantly by ethnicity and family history.

Patients of Hispanic origin and non-Hispanic whites were nearly equally represented in our study population, (37% and 36%, respectively). Participation of families of Asian origin, 14%, was comparable to the American Community Survey.36 However, few participants of African American origin entered the study (1%) when compared with the Los Angeles demographic of 7.8%.

Our genetic testing strategy revealed hearing loss-associated variants in ∼33% of white infants and ∼46% of Asian infants. However, there was a surprisingly low rate of connexin-related hearing loss, ∼9%, in Hispanic infants. The low rates of connexin-related hearing loss in a cohort of Hispanic infants primarily of Mexican, Central and South American origin has not been previously tabulated but has been implied in other studies.29,35 Low rates of connexin-related hearing loss have been tabulated in populations of African American and Caribbean Hispanic origin in New York City.37 This suggests there are other causes contributing to hearing loss in this predominantly Mexican and Central American Hispanic population requiring further study.

Data from the “Hispanic Health and Nutrition Examination Survey: 1982–1984” showed the prevalence of hearing loss >15 dB in children of Mexican American heritage to be 27.6 per 1000 compared with non-Hispanic white children with a prevalence of 15.5 per 1000 suggesting that the prevalence of hearing loss among Hispanic children is more common than non-Hispanic white children. Despite the low prevalence of connexin-related hearing loss among children of Hispanic heritage, there is a higher rate of hearing loss. This observation suggests a pressing need to identify the major causes of hearing loss in the Mexican American population as they will be underserved by any effort to include genetic testing into NHS protocols.

For 10.8% of the infants with hearing loss, only a single hearing loss-associated variant was identified. This observed frequency of single alleles is greater than the observed carrier frequency of 3.01% in the US Midwestern population.11 It is very unlikely that the single alleles could be dominant, as the identified sequence variants have been published previously as recessive and because in our patients we had no indication of a clear autosomal dominant family history in any of the subjects. Additionally, two of the alleles, S129N and Y152X, are rarely identified,15 suggesting that they are unlikely to be polymorphisms and have a role in the pathogenesis of hearing loss. Other groups identified a higher than expected number of single hearing loss variants.38 Thus, the presence of a single recessive allele in an individual with hearing loss may not simply represent a carrier state and should be interpreted cautiously when providing genetic counseling. Other factors, such as infection, may be co-occurring with mutations in Cx26/3039 warranting further investigation.

We calculated the allele frequency for hearing loss-associated variants and found seven alleles at a frequency of >1% and these included 35delG, V37I, 167delT, 235delC, del(GJB6-D13S1830), V27I;E114G (in cis), and M34T. Skewing of allele prevalence by ethnicity was observed and these observations will be important for development of genetic testing panels, much like those developed for cystic fibrosis.40

In this study, we identified 12 infants with hearing loss who passed NHS. Screening results were not associated with number of connexin-related hearing loss associated variants or with screening methods in this sample, as both ABR and OAE revealed pass results in some cases, and so other reasons should be explored to explain this pass rate. It is not unexpected that five infants with mild/moderate hearing loss within this group were not identified by NHS. Data from Norton et al.41 demonstrated that fail rates for all basic types of hearing screening including ABR, and distortion product or transient OAEs, do not approach 100% until the degree of hearing loss reaches or exceeds 40 dB. In other words, infants with any hearing thresholds of 40 dB or better may be expected to pass the NHS. As one of the infants in this group was a V37I homozygote, we must assume that the degree of hearing loss was better than 40 dB shortly after birth. Additionally, within this group of 12 infants, three had connexin-related hearing loss and had hearing loss in the severe to profound range. Not all infants with connexin-related hearing loss may be detected by bedside NHS and for some infants, the onset of severe hearing loss may be later than in the immediate newborn period or so mild in some frequency regions that they would go undetected. We estimate the rate of nonpenetrance of connexin-related hearing loss to be 8% at birth. These data collected in a systematic manner support previous case based reports that some infants with connexin-related hearing loss will pass NHS.23–25,42

In this study, we compared hearing loss severity between groups of infants with and without hearing loss-associated variants. In infants with connexin-related hearing loss, SNHL was the only phenotype. There were significant differences between groups of infants on the basis of number of hearing loss associated variants, with those who had connexin-related hearing loss in aggregate showing better PTA than those without. However, given the spread of phenotypes associated with truncating and missense variants,15 we must caution that severity of hearing loss alone cannot be used to predict which infant should be offered genetic testing for Cx26/30 although the type of hearing loss, SNHL, would be more predictive of connexin-related hearing loss.

We wanted to determine whether the rate of neonatal complications differed as a function of number of Cx26/30 hearing loss-associated variants. We evaluated neonatal risk factors in our population and compared them with the presence or absence of hearing loss-associated variants. Neonatal risk factors were observed in infants with and without connexin hearing loss-associated variants. However, we found no significant difference between groups as a function of the number of alleles and neonatal factors such as birth weight, gestational age, Apgar scores, hyperbilirubinemia, sepsis, meningitis, and aminoglycoside exposure. We conclude that genetic testing for hearing loss should not be deferred based on the presence of neonatal complications. Additional studies with larger samples and additional factors are warranted to examine associations between hearing loss genotypes and neonatal risk factors.

One factor that significantly correlates with connexin-related hearing loss is the presence of other family members with hearing loss, most notably, second degree relatives such as grandparents and parental siblings. We found that the presence of a positive family history increased the odds of having connexin-related hearing loss approximately 7-fold and contrasts strongly with the infants who did not have connexin-related hearing loss. Although this seems intuitively obvious, in nonconsanguineous families, the proband with the recessive condition is frequently the first person in the family with the condition. Again, we must caution that absence of a family history should not preclude genetic testing for hearing loss, however, a family history of hearing loss increases suspicion.

In summary, connexin-related hearing loss can be identified in nearly a quarter of infants with hearing loss from a diverse US population residing in Southern California. Biallelic hearing loss-associated variants confirming the diagnosis of autosomal recessive connexin-related hearing loss are more prevalent in infants of non-Hispanic white and Asian origin (33.3% and 46.2%, respectively) with a significantly lower rate among infants of Hispanic origin (9%). Identification of a single hearing loss-associated variant is common in infants with hearing loss (10.8%) and significantly greater than the established US carrier frequency of 3% strongly suggesting that identification of a single hearing loss-associated variant in an individual with hearing loss should be interpreted with caution and may not simply represent a carrier state. Although most infants with connexin-related hearing loss will be identified by NHS, a nontrivial subset of genetically predisposed infants will pass screening and remain at risk for hearing loss. Family history of childhood hearing loss increases suspicion that an infant will have connexin-related hearing loss. The presence of neonatal complications does not preclude the possibility that an infant will have connexin-related hearing loss.