Article Text

Download PDFPDF

Hearing outcome of infants with congenital cytomegalovirus and hearing impairment
  1. Efraim Bilavsky1,2,
  2. Keren Shahar-Nissan1,2,
  3. Joseph Pardo2,3,
  4. Joseph Attias4,5,
  5. Jacob Amir1,2
  1. 1Department of Pediatrics C, Schneider Children's Medical Center, Petah Tiqva, Israel
  2. 2Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel
  3. 3Department of Gynecology and Obstetrics, Rabin Medical Center, Beilinson Hospital, Petah Tiqva, Israel
  4. 4Institute of Audiology and Clinical Neurophysiology, Schneider Children's Medical Center, Petah Tiqva, Israel
  5. 5Department of Communication Sciences & Disorders, University of Haifa, Haifa, Israel
  1. Correspondence to Dr Efraim Bilavsky, Department of Pediatrics C, Schneider Children's Medical Center of Israel, Petah Tiqva 49202, Israel; yoji{at}


Background Congenital cytomegalovirus (cCMV) is the most common non-genetic cause of childhood sensorineural hearing loss. Antiviral treatment has been shown to prevent hearing deterioration in these infants. However, studies focused on infants with hearing impairment at birth and on the specific degree of impairment and further improvement or deterioration are lacking.

Objective To investigate the relationship between hearing status at birth and any change in hearing status at the end of a prolonged follow-up period, after receiving 12 months of antiviral treatment in children born with hearing impairment due to congenital cCMV.

Methods Clinical, laboratory, radiological and audiological data of all infants with cCMV infection followed in our centre between 2005 and 2013 were reviewed. Treatment with antiviral medication for hearing impairment found during the neonatal period was12 months of gan/valganciclovir. Hearing studies were performed only on infants who had been followed up for more than 1 year after treatment.

Results Hearing impairment at birth was found in 54 (36.2%) of the 149 infants diagnosed with symptomatic cCMV, and found in 77 affected ears; unilateral in 31 (57.4%) and bilateral in 23 (42.6%). After 1 year of antiviral treatment and a long-term follow-up of the 77 affected ears at baseline, 50 (64.9%) had improved, 22 (28.6%) remained unchanged and 5 (6.5%) had deteriorated. Most improved ears (38/50=76%) returned to normal hearing. Improvement was most likely to occur in infants born with mild or moderate hearing loss and less in those with severe impairment.

Conclusions We found that infants born with cCMV and hearing impairment, receiving 12 months of antiviral treatment, showed significant improvement in hearing status. The probability of hearing improvement seems inversely related to the severity of the impairment at birth.

  • Audiology
  • Congenital Abnorm

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

What is already known on this topic

  • Congenital cytomegalovirus is the most common non-genetic cause of childhood sensorineural hearing loss.

  • Antiviral treatment has been shown to prevent hearing deterioration in these infants.

  • The exact relationship between hearing impairment at birth and hearing improvement or deterioration after antiviral treatment has never been reported in the past.

What this study adds

  • Most of the affected ears will improve after prolonged (1 year) antiviral treatment.

  • Most improved ears will return to normal hearing.

  • The probability of hearing improvement is inversely related to the severity of the impairment at birth.


Congenital cytomegalovirus (cCMV) is the leading cause of congenital infections, affecting about 1% of all live births worldwide.1 It is also acknowledged as an important cause of neurodevelopmental delay and the most common non-genetic cause of childhood sensorineural hearing loss (SNHL).2

A recent systematic review attempted to determine the prevalence of cCMV associated SNHL, using studies of universal newborn screening for cCMV and found that SNHL occurred at birth in 32.8% of symptomatic cases.2 Fourteen other cohort studies assessed the nature of cCMV associated SNHL and found bilateral hearing loss in 71.2% and unilateral in 28.8% of cases. Most cases were severe to profound, necessitating hearing amplification and rehabilitation.2

In 2003, Kimberlin et al3 reported the effect of antiviral therapy on hearing in symptomatic cCMV infants. Treatment with intravenous ganciclovir for 6 weeks was found to prevent hearing deterioration at 6 months and possibly prevent hearing deterioration at ≥1 year.3 Moreover, their new Phase III randomised and blinded investigation showed that a prolonged course of oral valganciclovir for 6 months further improved audiological outcomes up to at least 2 years of age.4

However, while a third of neonates with symptomatic cCMV were born with some degree of SNHL, the benefit of antiviral treatment on affected ears is unclear. No study to date has tried to explore how the effect of valganciclovir/ganciclovir differs across different degrees of hearing loss and no study has focused on hearing outcomes of neonates born with different degrees of SNHL. Thus, the answer to the question whether the degree of SNHL at birth may be a predictor of further hearing deterioration (or improvement) at ≥1 year in these infants is still uncertain.

The aim of the current study was to investigate the relationship between hearing status at birth and any change in hearing status at the end of a prolonged follow-up period, after receiving 12 months of antiviral treatment in a large cohort of infants with cCMV.


Study setting and population

This is a retrospective study of a cohort of patients conducted at Schneider Children's Medical Center of Israel, the largest paediatric hospital in Israel. Clinical, radiological and audiological data of all infants with symptomatic cCMV infection followed in our paediatric clinic between January 2005 and October 2013 were reviewed. In Israel routine screening for cCMV is not done. However, our clinic is a dedicated reference clinic for infants who are suspected or identified as having cCMV. Most of our referred infants are born to mothers with a known primary infection during pregnancy. Also evaluated in our clinic are those infants with signs or symptoms of cCMV after birth, including those who failed routine hearing screening test after birth.

The diagnosis of primary maternal infection with CMV was established according to the following criteria and as previously reported:5 seroconversion from negative IgG for CMV to positive IgG for CMV during pregnancy, or if previous serology for CMV was not available—low CMV IgG avidity during pregnancy in the presence of specific positive IgM for CMV and a significant rise of IgG and avidity at later test. Periconceptional CMV infection was defined as primary CMV infection occurring from 4 weeks before the last menstrual period and up to 3 weeks of gestation.

cCMV infection was diagnosed in infants with a positive urine culture (shell vial) or PCR taken during the first 2 weeks of life. Additional studies of infants diagnosed with cCMV immediately after birth included a complete physical examination, including head circumference; a blood test including complete blood count, liver and kidney function tests; a funduscopy performed by a paediatric ophthalmologist; a brain ultrasound (US) performed by a paediatric radiologist and a hearing assessment.

An infant was diagnosed with symptomatic cCMV infection if any sign of central nervous system involvement were identified including: (1) microcephaly, head circumference <3%, (2) hearing impairment detected by the brainstem evoked audiometry (BERA) test; (3) chorioretinitis or (4) abnormal findings on brain US compatible with cCMV (eg, calcifications, periventricular hyperechosity, ventricular dilatation, pseudocysts and lenticulostriate vasculopathy).

For the purpose of the current study we enrolled all infants who were diagnosed with symptomatic cCMV, who started antiviral treatment during the first 4 weeks of life and had at least 1 year of follow-up.

Hearing assessment

All infants diagnosed with cCMV infection were studied using BERA within 4 weeks after birth. The audiologists were unaware of the infants’ US and treatment status. Auditory evoked potentials were recorded in response to rarefaction clicks of 100 μs duration presented at a rate of 13/s through inserted earphone transducers attached to a plastic tube, providing an acoustic delay of 80 ms (Eartone 5A; Aearo Company, Indianapolis, Illinois, USA) and calibrated according to ANSI S3.7-1995 (American National Standard Institute; Headquarters: 1819L Street, NW, Washington DC 20036, USA). Rarefaction clicks (main frequency energy ranges between 2000 Hz and 4000 Hz) were used in order to enable recording of cochlear microphonics (CM) and to identify cases of neural hearing loses (auditory neuropathy). It is important to indicate that all the infants born in our medical centre, including those who participated in this study, were screened by otoacoustic emission (OAE) at birth. As stated before, in our centre, infants that repeatedly fail OAE screening are tested for possible cCMV infection. Moreover, failure of OAEs at birth with absent CM and elevated BERA thresholds are indications for SNHL.

BERA test was acquired via an active electrode attached to the vertex or forehead, referred to an electrode at the ipsilateral mastoid. A third (ground) electrode was placed on the contralateral mastoid. Amplification band pass filters were set at 30–3000 Hz; the analysis period was set at 15.36 ms for two replications of 1024 sweeps each. The artefact rejection level is set adaptively and manually during the recording session. Based on noise associating the recording conditions, the artefact rejection was set from ±6 µV in the quiet condition up to ±10 µV in the severe noise conditions. Monaural BERA tests were recorded in response to air conduction decreasing stimulus levels from 90 dBnHL with 20 dB steps up to thresholds where steps were reduced to 5 dB. Following determination of the absolute latencies of waves I, III and V at 90 dBnHL, threshold was determined by the presence of wave V which was determined as the largest amplitude deflection at a time window beyond 5 ms, with clear repeatability in two sequential traces at the same stimulus level and no recordable wave V response at 5 dB less. The normative data used are based on previous reports by Gorga et al.6 ,7 Specifically, the system has a utility programme (CALIB) that sets and adjusts the intensity levels for evoked potentials-generated auditory stimulation through insert phones or bone oscillators. The calibration files store threshold intensities for click, noise and tone stimuli. These threshold intensities are referenced to nHL when auditory stimuli are delivered. In addition, an annual calibration test of the BERA system is routinely performed to check the calibration file. Stimulus intensities are measured with the Bruel & Kjaer 112 613 amplifier (Naerum, Denmark) and a 2120 frequency processor coupled to a 4153 microphone.

If air conduction thresholds were elevated beyond 25–30 dBnHL, bone conduction (BC) click-stimuli were applied. Due to variability in BC thresholds and following previous clinical recommendations, below those thresholds, the children were considered as a ‘normal hearing group’.8 The auditory responses were digitised at a sampling rate of 10 kHz with 12-bit accuracy using the Bio-Logic Explorer System (Bio-Logic Systems, Mundelein, Illinois, USA). Corrected BERA thresholds to behavioural thresholds (embedded as part of the system calibration) were categorised as normal thresholds (<25 dBHL); mild hearing loss (25–44 dBHL); moderate hearing loss (45–69 dBHL) and severe hearing loss (≥70 dBHL). In cases of air-bone gap thresholds, only BC results were reported.

BERA (in children aged ≤2 years) or behavioural hearing test (in children aged >2 years) was performed on all children during the neonatal period and at follow-up every 4–6 months until age 4 years. BERA test of children beyond 5 months was done during natural sleep or under sedation by 50 mg/kg oral Triclonam (Elixir) with the same parameters as for newborns’ assessment, except the latency norms which were matched to the age-appropriate norm values. Behavioural hearing tests of children included visual reinforcement audiometry for developmental ages of 6 months to 2 years and conditioned play audiometry or conventional audiometry for older children. In any case, the exact type of the hearing test was adjusted to the developmental age of the child. When necessary, in the BERA and behavioural tests masking noise to the contralateral ear was applied according to the audiological clinical guidelines of Hood.9 Only thresholds that were reliably masked as required were included in the analysis. The average of the behavioural thresholds across 1000 kHz, 2000 kHz and 4000 kHz were used for comparison with the BERA tests. Behavioural audiological responses were measured by calibrated GSI-61 audiometers and speakers (Grason Stadler, Madison, Wisconsin, USA) in a double-wall sound-attenuated room (Industrial Acoustics, Winchester, Hampshire, UK) that meets the standard of the ambient noise levels for audiometric test rooms (ANSI S3.1-1991, American National Standard Institute Headquarters: 1819L Street, NW, Washington DC 20036, USA). In addition, standard tympanometry was performed using an electroacoustic admittance instrument (AZ26 Middle Ear Analyzer; Interacoustic, Assens, Denmark). A high-frequency (1 kHz) probe tone was used for children under 6 months of age and a 226 Hz probe frequency was used in older children. Patients in whom middle ear dysfunction was suspected underwent a detailed otological inspection. All auditory threshold tests were administered by certified and skilled paediatric audiologists and were considered as reliable thresholds.


The primary study end point was hearing deterioration/improvement, measured using BERA or behavioural study between the first and the last hearing evaluations. The last hearing evaluation was performed at a follow-up of at least 1 year. As stated before, only those infants who started antiviral treatment during the first 4 weeks of life were considered eligible for our study. Data was also collected only for those followed for ≥1 year. There were no other exclusion criteria.

Hearing deterioration/improvement was defined as an increase/decrease of ≥10 dB in the auditory threshold on consecutive hearing assessments and a change in hearing category.

Audiological analyses were performed on total evaluable ears (‘biological’ assessment) and best evaluate ear (‘functional’ assessment) as previously described.4 The total ear assessment evaluated the biological effects of the antiviral treatment and the best-ear assessment correlated functional hearing impairment in daily living. For example, if an infant had mild hearing loss in the left ear and severe hearing loss in the right ear, the best-ear classification was mild hearing loss.


During the study period, all infants with symptomatic cCMV infection as defined earlier, were treated with one of two protocols: (1) intravenous ganciclovir 5 mg/kg/dose in two daily doses for 6 weeks followed by oral valganciclovir (Valcyte, Hoffmann-La Roche, Basel, Switzerland) 17 mg/kg/dose in two daily doses for another 6 weeks and then one daily dose until 1 year of age or (2) oral valganciclovir 17 mg/kg/dose in two daily doses for 12 weeks, then one daily dose until 1 year of age, as previously described.9

Statistical analysis

Categorical variables (eg, severity of hearing deterioration) were compared using a standard χ2 test. Proportion of patients with hearing improvement or worsening of hearing were compared using the χ2 test. Fisher’s exact method was used to calculate CIs of 95% for improvement and worsening rates (for each ear and for the best ear). Mantel test for trend was used to compare trends in improvement and deterioration rates according to initial severity of hearing loss. All analyses were performed by SPSS (SPSS V.15.0, Chicago, Illinois, USA).


During the study period, 205 infants diagnosed with cCMV in our clinic received antiviral treatment. As shown in figure 1, 149 of them were eligible for the current study. Median time of follow-up in the 149 enrolled infants was 40.2 months (range 13.1–117.7 months). Of these infants, maternal primary infection with CMV was diagnosed in 131 (87.9%) (first and second trimesters in 69 (52.7%) and 47 (35.9%), and the exact trimester of the maternal primary infection could not be determined in 15 (11.4%)) cases; non-primary maternal infection was diagnosed in 4 (2.7%) cases. In 14 (9.4%) cases, the maternal primary/non-primary infection could not be determined.

Figure 1

Infants with a congenital cytomegalovirus infection treated with antiviral treatment in our clinic between January 2005 and October 2013.

Some degree of hearing impairment at birth was found in 77 (25.8%) ears and in 54 (36.2%) of the enrolled infants. Among the 54 infants born with hearing impairment, an abnormal brain US was found in most (40/54=74.1%). Among the 14 infants with normal brain US and hearing impairment at birth, 1 child had microcephalus and 1 chorioretinitis. Thus, hearing impairment was the only indication for antiviral treatment in a minority of our enrolled infants (12/149=8.1%). Of the 54 infants born with hearing impairment, 31 (57.4%) had unilateral impairment and 23 (42.6%) bilateral.

Results of the last hearing assessment of each treated child (after at least 1 year of follow-up) are shown in tables 1 and 2. As seen in tables 1 and 2, of the 77 affected ears at baseline, 50 (64.9%) improved and 5 (6.5%) deteriorated. Most of the improved ears (38/50=76%) returned to normal hearing. On a best ear analysis, as seen in table 2, among the 24 infants with abnormal hearing at baseline, 17 (70.8%) improved and only 1 (4.2%) deteriorated. Most infants (13/17=76.5%) with improved hearing returned to normal. As seen in both tables, the chance of improving is inversely related to the severity of the impairment at birth.

Table 1

Total evaluated ears (‘Biological’ assessment) at baseline and at last follow-up—298 ears

Table 2

Best ear evaluation (‘Functional’ assessment) at baseline and at last follow-up—149 children

As seen in table 1, the results of the first hearing test serves as a strong predictor for further improvement in hearing assessment; 77.8% of the ears with mild SNHL at birth improved, of those with moderate SNHL 68.4% improved and of those with severe involvement only 40.9% improved. Deterioration rate among patients with moderate hearing loss at birth (5.6%) was significantly higher compared with patients with normal hearing at birth (0.5%) (p=0.03). OR for improvement in patients with severe hearing loss at birth compared with patients with mild hearing loss (table 1) was 0.20 (95% CI 0.05 to 0.72, p=0.01). Among the study group, seven children had a surgically implanted electronic device.

Among the 12 infants with only hearing impairment as an indication for antiviral treatment, improvement occurred in 11 (91.7%), while hearing of 1 (8.3%) infant deteriorated.

As seen in figure 1, 43 infants were not eligible for the current study due to the fact that antiviral treatment was started after 4 weeks of age. It is worth mentioning that 32 of these infants started treatment between 4 weeks and 12 weeks of age. Some degree of hearing impairment at the first assessment was found in 15 (34.9%) of these 45 children (22 ears). At the final follow-up, improvement was observed in 15 (68.2%) of these ears.

The main side effects of the treatment in the 149 treated infants were neutropenia, defined as absolute neutrophil count (ANC) ≤1000/mm3. Forty-one episodes of neutropenia occurred in 33 infants (20 in the intravenous group and 13 in the only oral treatment group) and were only observed during the first 3 months of treatment, mainly in infants who were on intravenous antiviral treatment during the first 6 weeks. Two of these infants had severe neutropenia defined as ANC<500 mm3. Antiviral treatment was stopped for 2–3 days until the ANC returned to normal. All other cases required no change of treatment, only repeated blood counts.


Herein, we report on the relationship between hearing impairment status at birth and hearing improvement/loss at the end of a prolonged follow-up period and after antiviral treatment in infants with cCMV. Our findings show that improvement occurred in most of the affected children (and ears) diagnosed with SNHL at birth. The probability of hearing improvement is inversely related to the severity of the impairment at birth.

Goderis et al's2 recent comprehensive systemic review2 found that some degree of SNHL occurred in 32.8% of symptomatic cases at birth. Our cohort of children reported a similar finding in 36.2% of the enrolled infants born with hearing impairment. On the other hand, a significant difference between our results and theirs is the fact that among our affected children at birth, unilateral hearing loss was more frequent than bilateral (57.4%/42.6%) while they reported the opposite (28.8%/71.2%). We could not find an explanation for this difference.

Many studies have focused on hearing impairment at birth in infants with cCMV; however, our study is the first to focus on the relationship of long-term hearing assessment in a relatively large cohort of treated infants with cCMV born with hearing impairment.

In our cohort of children, the results were surprising. We found that in most ears with mild or moderate hearing loss at birth, hearing improved, returning to normal. Even in cases of severe hearing loss at birth, 40% of ears benefited from the antiviral treatment.

We compared our current results with two pivotal studies by Kimberlin et al. The study published in 2003 assessed the effect of 6 weeks antiviral treatment with ganciclovir on the hearing of children with cCMV.3 Improved hearing was observed at 6 months of age in 22% and 24% of total ear assessment and best ear analysis, respectively. After 1 year of age, improvement was reported in 25% and 17% of children, respectively. However, while no case of deterioration occurred at 6 months of age, after >1 year of follow up, 21% deteriorated (best ear and total ears).3 Comparing our current results with the report from 2003 shows significantly higher rates of improvement and on the other hand, lower rates of deterioration. Unfortunately, specific results of the relationship between the degree of hearing loss at birth and improvement/deterioration were not given in the 2003 report.

The second current work by the same group, assessed 6 months versus 6 weeks of oral valganciclovir in infants with symptomatic cCMV.4 In children who were treated with 6 months of antiviral treatment, improvement was observed in 7.3–8.6%, depending on the specific time period of follow-up.4 Hearing deterioration was noted in 11.3% after 24 months of follow-up.4 However, there were no explicit data focusing on children with a specific degree of hearing loss at birth.4

Our study raises a simple, yet interesting question, as to how the antiviral treatment generates improvement rather than prevent deterioration of SNHL. No answer has as yet been found in the literature. However, our hypothesis is that an ongoing, dynamic process of infection/inflammation due to the virus continues to occur even after birth. Administering antiviral treatment affects the balance needed to reduce this harmful process. We believe that if given at the right time (mild SNHL), hearing loss might be reversible. Our study supports this hypothesis in a way that when SNHL is severe (vs lower degree of hearing loss), the chances of improvement are lower.

Similar to other studies, we analysed only those infants who started treatment from birth until 4 weeks of age. However, another interesting subgroup of infants, seen on a daily basis, is of those who for one reason or another came to our clinic only after 4 weeks of age with cCMV and hearing loss. Our results showed that many of these children can benefit from the antiviral treatment. Clinicians should be better informed that antiviral treatment can improve hearing, even in older infants with cCMV and hearing loss. The specific time frame as to when those presenting late will still benefit from the treatment is still undetermined.

Our protocol of treatment includes 1 year antiviral treatment. During the past few years, researchers have concluded that a more prolonged antiviral treatment in children with symptomatic cCMV is needed.4 ,10 ,11 Our group previously reported, in a preliminary retrospective study, that a 1 year course of treatment in these children is safe and appears to lead to a better auditory outcome than short-term treatment.12

The main limitations of our study were the retrospective methodology and relatively small number of infants with SNHL. However, it is still the largest cohort of infants born with cCMV and SNHL. Another limitation is that our group of reported infants might not fully represent regular epidemiology of cCMV; in Israel, many physicians recommend maternal screening for CMV infection during pregnancy, which has led to the identification of primary maternal infections on one hand, but on the other hand, occasionally indicating a termination of pregnancy. Our clinic also serves as a referral centre and thus, a selection bias for more symptomatic infants with cCMV can occur. However, this fact does not interfere with the current reported results which focused on those who were already born with hearing impairment. Another deviation from the classic report of hearing evaluation in cCMV3 ,4 was our definition of hearing deterioration or improvement. We defined it as a change in hearing category, and as a combination of increase/decrease of ≥10 dB in the auditory threshold and a change in hearing category. Due to the well known hearing fluctuations in infants with cCMV, and due to the fact that minor changes from one BERA test to another can occur including in cases of intertester variability, we felt that our stricter definition would give a more precise picture of this important issue. Even after using our firmer definition, our results are surprisingly encouraging. Nevertheless, our broader range of definition may also contribute to the significant change in hearing categories we found.

Another important limitation of our study, as well as of any other future study in this field, is the lack of control group with hearing follow-up without treatment. While fluctuations in hearing may occur in infants with cCMV,2 one might think that some children with a mild degree of hearing impairment at birth might improve spontaneously, even without treatment.

In conclusion, the results of this study indicate a favourable hearing outcome in infants born with cCMV and SNHL treated with antiviral medication. While improvement will occur in most of the affected ears/children, the probability of hearing improvement is inversely related to the severity of the impairment at birth. Our results can assist paediatricians, infectious diseases experts and auditory specialists who treat infants with cCMV in ascertaining a future prognosis and management planning for infants born with hearing impairment. Larger, long-term prospective studies are needed to report the long-term stability of the hearing improvement post treatment, to further characterise this specific group of infants and to find the most efficient treatment protocol for infants born with SNHL due to cCMV.


The authors thank Mrs Phyllis Curchack Kornspan for her editorial services.



  • Contributors EB conceived the study and designed it, collected data and drafted the manuscript. KS-N contributed substantially to its revision. JP contributed substantially to the manuscript and its revision. JAt contributed critical remarks to the manuscript and its revision. JAm conceived the study and designed it and contributed substantially to its revision.

  • Competing interests None declared.

  • Ethics approval The study was approved by the institutional Helsinki Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • Atoms
    Nick Brown