Three Years of Vestibular Infant Screening in Infants With Sensorineural Hearing Loss Free

The vestibular screening was offered to all Flemish infants with permanent hearing loss around the age of 6 months in 1 of the 25 reference centers involved in the UNHSP. As the preliminary VIS-Flanders study showed normal screening results in infants with permanent conductive hearing loss, only infants with sensorineural hearing loss were included.²⁶  This multicenter crosssectional study was approved by the leading Ethical Committee of the Ghent University Hospital and the ethical committees of all participating centers (Belgian registration number: B670201835971, ClinicalTrials.gov registration number: NCT05061069). In accordance with the ethical standards of the Helsinki Declaration, written informed consents of parents were obtained.

The vestibular screening protocol and its addition to the existing Flemish neonatal hearing screening protocol are reported by Martens et al.^25,26  The cVEMP test was performed with the commercial Neuro-Audio equipment (Neurosoft, version 2010, Ivanovo, Russia). Bone-conducted stimuli (RadioEar B71W, Middelfart, Denmark) (59 dB nHL, 129 dB FL) were presented at the ipsilateral mastoid. More details about the applied stimulus parameters, software algorithm, recording parameters, electrode configuration, and test setup can be consulted in Martens et al.²⁶  Electromyographic background activity was automatically quantified by the software (ie, accepted range of mean rectified voltage: 80–250 µV) and displayed on a screen.²⁶  At least 2 trials were recorded on each side to check waveform reproducibility. The averaged rectified interpeak amplitude was calculated from the 2 trials with equivalent sternocleidomastoid muscle tension (ie, averaged electromyographic differences ≤30 µV). Only final screening results were used for further analyses. More specifically, if the child needed a retest (ie, within 3 months after the first screening, thus before the age of 10 months, prior to possible CI-surgery) to confirm the first screening results (eg, in case of inconclusive results, see Martens et al²⁶ ), only the second screening results were included. Results were considered as normal if 2 reproducible biphasic P1-N1 waveforms were recorded with an averaged rectified interpeak amplitude ≥1.3 (ie, based on normative data of the Ghent University Hospital in 34 control subjects (mean age = 7.6 months; standard deviation [SD] = 1.5 months).²⁶  Abnormal screening results included inconclusive responses (ie, insufficient sternocleidomastoid muscle tension) during the retest, absent responses (ie, no reproducible waveforms), and decreased responses (ie, reproducible waveforms with an averaged rectified interpeak amplitude <1.3). Since presence and amplitude of cVEMP responses strongly correlate with the child’s motor performance,^9,11  these criteria are clinically relevant to decide if referral for motor assessment is needed.

All centers collected results of transient evoked or distortion product otoacoustic emissions, high frequency tympanometry (1000 Hz), and click-evoked ABR. These results were used to determine possible predisposing factors for abnormal vestibular screening results (ie, further described as “risk factors”). The International Bureau of Audiophonology³²  criteria were considered to categorize the degree of hearing loss. Onset of hearing loss was grouped as congenital (ie, abnormal hearing screening after birth) or early-onset (ie, normal hearing screening after birth, but permanent hearing loss detected before 10 months of age as a cVEMP retest is advised before this age). Additionally, risk factors related to hearing loss etiology included congenital cytomegalovirus (cCMV)-status, results of Connexin 26 (Cx26 or GJB2, autosomal recessive deafness type 1 [DFNB1]) mutation analysis, and the presence of perinatal factors that are associated with an increased risk for hearing loss. Infants with sensorineural hearing loss were standardly tested for cCMV within 3 weeks after birth by performing virus isolation or polymerase chain reaction in urine or saliva, or after this period with polymerase chain reaction on neonatal dried blot spots (ie, Guthrie card). cCMV definitions were applied as described in the European Consensus Statement of 2017.³³  Perinatal factors were considered as present in case of prematurity (ie, gestational age < 36 weeks), low birth weight (ie, birth weight < 2500g),³⁴  or hospitalization longer than 5 days at a NICU. Other perinatal factors, such as hyperbilirubinemia and ototoxic medications, were not included as they are not provided by all centers. All aforementioned risk factors, including Cx26- and cCMV-status, were standardly known at the age of screening, whereas more advanced etiological work-up, such as imaging and genetic testing by means of next generation sequencing technology, was not always allowed by the parents or available at the moment of screening. Therefore, underlying etiology of hearing loss was reported descriptively to estimate the most important groups at risk for abnormal screening results. Etiological work-up results were classified as genetic nonsyndromic; genetic syndromic; TORCHES infections (ie, toxoplasmosis, other infections such as syphilis, varicella-zoster, parvovirus B19, rubella, cytomegalovirus, and herpes); meningitis; cochleovestibular anomalies (ie, cochleovestibular nerve aplasia or inner ear malformations confirmed by MRI or CT that could not be classified into 1 of the first 4 categories); or an unknown etiology.

Statistical analysis was completed with SPSS software (IBM, version 27.0, Armonk, NY). On subject level, abnormal screening results indicated abnormal responses in at least 1 ear, and the degree of hearing loss was categorized according to the worst ear in case of bilateral hearing loss. The two-tailed Fisher’s Exact test was used to evaluate the association between screening results and possible predisposing factors. On ear level, data were analyzed more in-depth by means of generalized estimating equations, which takes the clustered data structure (ie, 2 ears within 1 child) into account and provides a robust estimator of the covariance matrix. Results of etiological work-up of hearing loss were not included as a predictor to avoid multicollinearity and because the underlying etiology was not known in all infants. Included predictors were on ear level (ie, degree and onset of hearing loss) or on subject level (ie, gender, laterality of hearing loss, cCMV-status, Cx26-status, and presence of perinatal factors). Odds ratios (OR) with 95% confidence intervals (95% CI) were reported. The significance level (ie, two-tailed) was set at P < .01 to correct for multiple testing.

Overall, 301 hearing-impaired infants were screened, of which 47 infants were excluded (Fig 1). All 22 excluded infants with permanent conductive hearing loss showed normal vestibular screening results. In total, 254 infants (ie, 125 boys and 129 girls) with sensorineural hearing loss were included (ie, 508 ears). The mean age during the final screening test was 7.4 months (SD = 2.4 months). Table 1 displays the distribution of hearing loss characteristics and etiology. The majority showed unilateral or bilateral severe to profound hearing loss (60.6%, 154 of 254). Six infants had normal hearing at birth but developed early-onset hearing loss (2.4%, 6 of 254), in 3 children it was caused by meningitis, and in 3 children it was caused by cCMV (symptomatic: n = 1 and asymptomatic: n = 2). Genetic (ie, nonsyndromic and syndromic) hearing loss was found in 29.9% (76 of 254) of the infants, whereas 17.0% (43 of 254) had acquired hearing loss (ie, TORCHES and meningitis). Most TORCHES infections were caused by cCMV (97.5%, 39 of 40). DFNB1 (Cx26) was the leading cause for genetic nonsyndromic hearing loss (66.7%, 32 of 48), followed by DFNB3 (MYO15A) (6.3%, 3 of 48), DFNB16 (STRC) (6.3%, 3 of 48), and DFNB12 (CDH23) (4.1%, 2 of 48). The most common syndromic causes were Waardenburg syndrome (10.7%, 3 of 28), CHARGE syndrome (7.1%, 2 of 28), Usher type 1 syndrome (7.1%, 2 of 28), Usher type 2 syndrome (7.1%, 2 of 28), Down syndrome (7.1%, 2 of 28), and Pendred syndrome (7.1%, 2 of 28).

At the subject level, abnormal results were found in 13.8% of the infants (35 of 254) (Table 5). Unilateral abnormal results were found in 8.3% (21 of 254), whereas bilateral abnormal results were seen in 5.5% (14 of 254). The latter group contained 6 infants with bilateral inconclusive results, in which a bilateral vestibular deficit was confirmed in 5 infants during vestibular follow-up at a later age (Table 2). Abnormal results were found significantly more often in infants with unilateral or bilateral severe to profound hearing loss (20.8%, 32 of 154) compared with unilateral or bilateral mild-moderate hearing loss (3.0%, 3 of 100) (P < .001) (Fig 2). Moreover, abnormal results were more frequently noticed in early-onset hearing loss (ie, only caused by meningitis and cCMV) (50.0%, 3 of 6) compared with congenital hearing loss (12.9%, 32 of 248) (P = .036) (Fig 2). cCMV-positive infants more often showed abnormal results (20.5%, 8 of 39) compared with infants without cCMV (12.6%, 27 of 215) (P = .207) (Fig 3). No major differences in screening results were noticed between infants with and without perinatal factors (P = .840) (Fig 3). In Cx26-infants, abnormal results were less common (3.1%, 1 of 32) compared with infants without Cx26 (15.3%, 34 of 222) (P = .095) (Fig 3). cCMV detection was negative on the Guthrie card of the only Cx26-infant with abnormal results. In respect to the advanced etiological work-up, most abnormal results were found if the hearing loss was caused by meningitis (66.7%, 2 of 3), followed by syndromic hearing loss (ie, especially Usher type 1 and CHARGE syndrome) (28.6%, 8 of 28), hearing loss caused by TORCHES infections (ie, only in cases with cCMV) (20.0%, 8 of 40), and cochleovestibular anomalies (19.2%, 5 of 26) (Fig 4).

On the ear level, abnormal vestibular screening results were found in 9.6% (49 of 508) of the ears (Table 3). More in-depth analysis on ear level by means of generalized estimating equations modeling demonstrated that the odds ratios of an abnormal result were only significant for the degree of hearing loss (Table 4). More specifically, the odds of an abnormal result were about 9 times higher (P < .001, OR = 9.16, 95% CI = 2.75–30.58) for ears with severe to profound hearing loss compared with the ears with normal hearing, for equal values on the other risk factors. No significantly increased risk (P = .76, OR = 0.83, 95% CI = 0.25–2.80) was found between ears with mild-moderate hearing loss and normal hearing. In 1 of the 3 normal hearing ears with abnormal results, the vestibular screening was normal for the contralateral hearing-impaired ear (ie, cCMV-positive). Moreover, the odds ratio of an abnormal result was almost 13 times higher but was on the border of statistical significance (P = .01, 95% CI = 1.80–89.92) for the confined group of ears with early-onset hearing loss (ie, solely caused by cCMV and meningitis) in comparison with congenital hearing loss, for equal values on the other risk factors.

The authors of this pioneer study aimed to map the occurrence of and to identify the most important risk factors for abnormal vestibular screening results in a large group of infants with congenital or early-onset sensorineural hearing loss. The cVEMP was selected as the vestibular screening tool because saccular function is essential for the development of head control and gross motor milestones in young children.⁹  Abnormal cVEMP results were found in 14% of all included infants, and in 21% of infants with severe to profound hearing loss, whereas higher percentages of abnormal cVEMP results (ie, between 17% and 91%) were reported in literature.^{6,10,16,35–45}  However, various test protocols and cut-off criteria have been used to evaluate cVEMP abnormality in literature, resulting in a wide range of cVEMP abnormalities.^8,25  Additionally, previous studies mainly focused on older children with severe to profound sensorineural hearing loss, and were mostly conducted in specialized CI centers, therefore dealing with a population of children more at risk for vestibular deficits. Since hearing losses with various degrees and etiologies were included in the current study, the results are more representative for the whole group of infants who fail the neonatal hearing screening. As only the cVEMP was used to screen the vestibular function, which only assesses 1 part of the vestibular system, ongoing longitudinal vestibular research at the Ghent University Hospital will map the sensitivity of the vestibular screening to evaluate the overall vestibular function in hearing-impaired infants.

The current study showed that infants with severe to profound sensorineural hearing loss are the most important group at risk for abnormal cVEMP results, thereby confirming the preliminary VIS-Flanders results.²⁶  Consistent with these findings, Maes et al⁴³  reported significantly higher cVEMP abnormality rates in children with profound hearing loss compared with those with nonprofound hearing loss. Also the systematic review of Verbecque et al⁸  concluded that pediatric vestibular loss was reported more frequently with increasing degrees of hearing loss as the auditory and vestibular end organs are closely related. Screening results did not differ significantly according to the laterality of hearing loss, which is supported by previous studies that described vestibular deficits in unilaterally hearing-impaired children.^44,46  In early-onset hearing loss, screening results seemed to be more abnormal compared with congenital hearing loss. However, only 6 infants with early-onset hearing loss could be included in the current study since infants were screened at an early age. In all infants with early-onset hearing loss, the underlying etiology was meningitis or cCMV, which are 2 etiologies repeatedly associated with a high risk for vestibular deficits in literature.^42,47–52  Besides cCMV and meningitis, abnormal screening results were noticed more often in infants with syndromic hearing loss, such as Usher type 1 and CHARGE syndrome, and cochleovestibular anomalies, which is in agreement with findings of previous studies and the recent review of Hazen et al.^42,53–57  No significant odds ratio for abnormal screening results was found in the group of cCMV-positive infants, because the group without cCMV also consisted of infants at risk for vestibular deficits, such as infants with meningitis, syndromic hearing loss, and cochleovestibular anomalies. Furthermore, infants with genetic nonsyndromic hearing loss were only at risk for abnormal vestibular screening results if the hearing loss was severe to profound (Table 5). Similar to the current findings, only a few pediatric cases with genetic nonsyndromic hearing loss and concurrent severe vestibular deficits have been described in literature.^10,38,42  However, a major limitation of the current study was that advanced etiological work-up was not performed in all included infants, resulting in a large group of infants with an unknown underlying etiology of hearing loss. Nevertheless, only a minority of the latter group showed abnormal screening results, which were mostly found in infants with severe to profound hearing loss (Table 5). Finally, abnormal screening results occurred approximately equally in hearing-impaired infants with and without perinatal factors. In line with the results of Zagólski et al,⁵⁸  abnormal cVEMP results of hearing-impaired infants with perinatal factors were mainly found in cases of severe to profound hearing loss (Table 5). Thus, it seems that perinatal factors measured in this study are only relevant in the light of predisposing the infant to hearing loss, because it is the degree of hearing loss that determines the risk for vestibular deficits.

This is the first study to report results after 3 years of vestibular screening in a group of infants with congenital or early-onset sensorineural hearing loss. Based on the current results, the authors highly recommend early vestibular screening for all infants with unilateral or bilateral severe to profound sensorineural hearing loss, and additionally for hearing-impaired infants with meningitis, syndromes, cCMV, and cochleovestibular anomalies. Hereby, vestibular deficits can be detected at a young age, which enables early referral for motor assessment and rehabilitation if needed. Future research should map the sensitivity of the cVEMP as a vestibular screening tool in order to fine-tune this vestibular screening protocol.

Special thanks go to all participating otorhinolaryngology departments, all members of the project’s advisory committee, the infant welfare agency of the Flemish government “Child and Family,” the parent organization “VLOK-CI,” the Federation of Centers for Ambulatory Rehabilitation, and all participating rehabilitation centers and private physiotherapists across Flanders. The Unit of Biostatistics at the Faculty of Medicine and Health Sciences of the Ghent University, especially Dr Ineke van Gremberghe, is thanked for the professional support during statistical analyses. The authors would also like to thank all infants and their parents who participated in this study.

95% CI

95% confidence intervals

ABR

auditory brainstem responses

cCMV

congenital cytomegalovirus

CI

cochlear implant

cVEMP

cervical Vestibular Evoked Myogenic Potentials

Cx26

Connexin 26

DFNB

autosomal recessive deafness

OR

odds ratios

TORCHES

toxoplasmosis, other infections such as syphilis, varicella-zoster, parvovirus B19, rubella, cytomegalovirus, and herpes

UNHSP

Universal Newborn Hearing Screening Program

VIS

Vestibular Infant Screening