Is COVID‐19 to Blame for Sensorineural Hearing Deterioration? A Pre/Post COVID‐19 Hearing Evaluation Study

Abstract

Objectives

Here, we aimed to (a) determine whether a clinically significant sensorineural hearing loss (SNHL) change could be detected in post‐coronavirus disease (COVID‐19) hearing levels on comparing them with pre‐infection hearing levels after controlling for the effect of age and (b) to identify risk factors, such as hypertension, diabetes, and smoking, which increase the likelihood of hearing loss in COVID‐19 patients.

Methods

We retrospectively analyzed hearing thresholds in unvaccinated patient's pre‐ and post‐COVID‐19 infection. Thresholds were controlled for age and the duration between the pre‐ and post‐COVID‐19 hearing evaluations. Correlations between additional COVID‐19‐related symptoms, hypertension, diabetes, and smoking and hearing threshold changes were analyzed.

Results

A significant (but not clinical) threshold elevation was found post‐COVID‐19 infection. However, on controlling for age and the duration between the pre‐ and post‐COVID‐19 hearing evaluations, no significant threshold elevation was found. No significant correlation was found between hearing threshold changes and additional COVID‐19‐related symptoms, hypertension, diabetes, or smoking.

Conclusion

COVID‐19 did not lead to a significant hearing threshold elevation in our cohort, even among patients with additional COVID‐19 symptoms, hypertension, or diabetes mellitus or among those who smoked.

Level of Evidence

3. nonrandomized controlled cohort, follow‐up study Laryngoscope, 2022

Our study aimed to evaluate the associations between COVID‐19 infection and deterioration in sensorineural hearing thresholds among a cohort of patients from Meuhedet healthcare services (the third largest of four public healthcare provider organizations in Israel) that had undergone hearing evaluations pre‐ and post‐COVID‐19. Our findings suggest that COVID‐19 does not appear to be associated with deterioration of the sensorineural hearing among unvaccinated patients who had known hearing loss pre‐infection, after correcting for age‐related hearing loss. Moreover, our findings do not validate previous reports of a greater deterioration in the hearing of infected patients who possess background risk factors such as smoking, hypertension, and diabetes, suggesting no interaction between risk factors and hearing deterioration post‐infection.

INTRODUCTION

The coronavirus disease (COVID‐19) pandemic emerged in 2019 and led to many cases of severe acute respiratory syndrome, which was the main cause of death in millions of patients. However, it has been reported that people infected with COVID‐19 may also suffer from additional neurological and sensory deficits. ¹ , ² Among these sensory deficits, it has been hypothesized that similar to other viruses, such as cytomegalovirus and rubella, ⁵ severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) may lead to hearing loss due to damage to the inner ear cells. ³ , ⁴ This hypothesis has been further supported by the finding indicating that receptors mediating the entry of SARS‐CoV‐2 ⁶ are expressed in human inner ear tissue, which leads to the possibility of infection in the inner ear cells.

Several case series and case–control studies have provided evidence for the possible effects of COVID‐19 on the hearing system. For example, one study found that among 223 patients infected with COVID‐19, 2% reported experiencing hearing loss, and these patients were categorized as having severe to critical COVID‐19. ⁷ In case–control studies, it was demonstrated that patients with COVID‐19 had poorer high‐frequency hearing thresholds (≥4000 Hz) ⁸ , ⁹ and lower otoacoustic emission (OAE) signal‐to‐noise ratios (SNRs) ³ , ⁹ than COVID‐19‐negative controls, which supports the notion of possible inner ear damage. Other studies, however, failed to demonstrate any difference in hearing thresholds or OAE SNRs among COVID‐19 patients compared to controls ¹⁰ ; the possible reasons for this contradictory result might be the small patient groups and difficulty in finding a control group with matching background characteristics to those of the patient group.

Additional evidence supporting a possible effect of COVID‐19 on the hearing system hails from several studies showing that among other complaints, people infected with COVID‐19 also subjectively report hearing loss, tinnitus, and dizziness. ⁷ , ¹¹ Among three meta‐analyses published over the last 2 years, the prevalence of reported hearing loss symptoms after COVID‐19 infection ranged from 3% to 7%. ¹² , ¹³ , ¹⁴ Nevertheless, in most studies, data collection was retrospective and relied on subjective self‐reports and recall of participants. Indeed, Sauders et al. (2022) demonstrated that when participants infected with COVID‐19 were asked about the impact of the infection on their health, they reported on experiencing a larger number of symptoms that are currently recognized to be related to COVID‐19 and symptoms not known to be related to the disease (e.g., toothache). Moreover, the number of reported symptoms (whether related or unrelated to COVID‐19) was larger among participants who also reported on facing a larger number of psychological challenges during the pandemic. The authors concluded that the subjective participant reports were inconsistent, included possible recall bias, and may have been biased due to a nocebo effect. ¹⁵ Thus, the limitations of case, case–control, and subjective studies have raised the need for additional audiological measurements and comparisons among COVID‐19 patients to further investigate the possible effect of the infection on the hearing system.

A possible way to overcome the abovementioned limitations is to compare the audiological data of individual patients before and after COVID‐19 infection. However, this comparison is not typically practical because of a lack of routine testing; accordingly, even if a hearing test is to be performed during or after COVID infection, a pre‐COVID‐19 audiogram is often not available for comparison. Thus, in the present retrospective study, we aimed to compare the pre‐ and post‐COVID‐19 infection audiograms of unvaccinated adult patients for the first time. Our data were collected during the first and second COVID‐19 waves before the emergence of SARS‐CoV‐2 variants and before the availability any COVID‐19 vaccine. The cohort comprised patients who were insured by Meuhedet Health Services, the third largest health organization in Israel, which provided a unique opportunity to shed light on the possible causal connection between COVID‐19 and hearing system changes. Specifically, we aimed to answer two questions. (a) The first question was as follows: among COVID‐19 patients, is there greater sensorineural hearing loss (SNHL) post‐COVID‐19 compared with the pre‐infection hearing status after controlling for age effects? (b) It has been shown that hypertension, diabetes mellitus, and smoking are risk factors for hearing loss. ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ Thus, we aimed to address an additional question in this study: do these risk factors also increase the likelihood of hearing deterioration in COVID‐19 patients?

MATERIALS AND METHODS

Study Sample and Design

The current study data were extracted from the electronic health records (EHRs) of “Meuhedet Healthcare Services” (hereafter “Meuhedet”), the third largest of four integrated health care organizations in Israel, which provides insurance services to 14% of the total population with nationwide health care coverage. Ethical approval for this study was obtained from the Helsinki Institutional Review Board of Meuhedet. The eligibility criteria included age ≥18 years; a previously documented positive SARS‐CoV‐2 polymerase chain reaction (PCR) test performed between March 1, 2020, and December 15, 2020, before the implementation of COVID‐19 vaccines in Israel; and hearing evaluation conducted both pre‐ and post‐COVID‐19 infection. We chose this specific timeframe because it was prior to vaccine availability. Therefore, being vaccinated was not a confounding factor. It should be noted that in Israel, the PCR test was mandatory for any individual suspected of having been exposed to a patient infected with COVID‐19. There was no lack of PCR test kits during this time period; therefore, anyone who needed or wanted to be tested underwent the test. The exclusion criteria included hearing evaluation conducted only pre‐ or post‐COVID‐19, missing auditory data, or evidence of a significant conductive component (air‐bone gap ≥20 dB HL) in the pre‐COVID‐19 test. In addition, participants with known ear diseases or chronic conditions, such as otosclerosis or cholesteatoma, which could affect hearing, were excluded. Figure 1 illustrates the cohort selection process. All patients who met all the inclusion criteria (N = 83) had some level of pre‐COVID‐19 SNHL: ears with sloping curves (N = 54), high tone loss (N = 43), flat curves (N = 30), notched audiograms (N = 15), to normal hearing to slight hearing loss (N = 16), u‐shaped audiogram (N = 3), picking audiogram (N = 4), and corner audiogram (N = 1).

Fig. 1.

Flow chart of the cohort selection process. COVID‐19, coronavirus disease.

Each patient's pre‐ and post‐COVID‐19 pure‐tone hearing thresholds were recorded at each frequency for air and bone conduction; their speech reception threshold (SRT) and phonemically balanced (PB) word identification scores were also recorded.

The following demographic background information was collected. The documented reasons for hearing evaluation postinfection, including routine follow‐up (N = 37), self‐reported hearing deterioration (N = 26), and worsening of symptoms/appearance of additional symptoms (N = 18), were recorded. Medical diagnoses and clinical characteristics, such as hypertension, diabetes mellitus, smoking status, and occurrence of symptoms during the COVID‐19 infection (i.e., symptomatic/asymptomatic), were recorded (Table I). The presence of additional symptoms, such as tinnitus (N = 26) and vertigo (N = 17), was recorded. Finally, the age at and duration between each patient's last hearing evaluation pre‐COVID‐19 and first hearing evaluation post‐COVID‐19 (Table II) were used as independent variables that could affect the results. The cohort included 10, 16, 22, 21, and 15 patients aged between 20–39, 40–54, 55–64, 65–75, and 75–91 years, respectively. Most patients (58/83) were >55 years old.

TABLE I.

Demographic and Clinical Characteristics of the Study Cohort.

Characteristics	Patients (N)
N	83
Age (mean, SD), years	60.1 (15.8)
Biological sex
Female	38
Male	45
Population sector
General Jewish	41
Ultraorthodox Jewish	27
Arab	15
Duration between hearing evaluations (mean, SD), years	2.98 (2.01)
COVID‐19 symptoms
Yes	41
No	39
Missing data	2
Hypertension
Yes	41
No	42
Smoking
Yes	26
No	57
Diabetes mellitus
Yes	18
No	65

COVID‐19, coronavirus disease; SD, standard deviation.

TABLE II.

Duration Between Hearing Evaluations.

Duration	Patients (N)
<1 year	14
1–2 years	18
2–3 years	16
3–4 years	13
4–5 years	10
>5 years	13

Statistical Analyses

According to the power analysis calculation, with a moderate effect size of 0.25, α of 0.05, and power of 0.80, a sample size of least 65 subjects was required. Our study sample comprised 83 patients, which was larger than the size required to reduce the influence of the effect size.

Because there was no reason to assume that the virus would attack a specific ear and as previous studies have shown similar effects in both ears, ³ , ⁸ , ⁹ , ¹⁰ the results in the right and left ears of each participant were averaged. To answer our research questions, the air conduction thresholds were divided into three frequency ranges: low = 250–500 Hz, mid = 1000–3000 Hz, and high = 4000–8000 Hz. To determine whether a deterioration in hearing thresholds occurred between the pre‐ and post‐COVID‐19 hearing evaluations, a one‐way repeated measures multivariate analysis of variance (MANOVA) was performed to identify changes in the averages of the three frequency range thresholds (low, mid, and high), SRTs, and PB word identification scores. Subsequently, a one‐way repeated measures multivariate analysis of covariance (MANCOVA) was performed, in which the age of the participants and duration between the two hearing evaluations were considered as covariants to ensure that the threshold changes were not attributable to these two variables.

Three Pearson's correlation coefficients were computed to assess the linear relationship between health risk factors (hypertension, diabetes mellitus, smoking, and occurrence of symptoms during COVID‐19 infection) and each frequency range. Two paired sample t‐tests were performed to identify possible differences in the self‐reports of tinnitus and vertigo pre‐ and post‐COVID‐19.

The data were analyzed using IBM SPSS Statistics (Statistical Package for Social Sciences) version 24 (Armonk, NY: IBM Corp). A p value of <0.05 was considered statistically significant.

RESULTS

The one‐way repeated measures MANOVA performed to determine whether a deterioration in hearing thresholds occurred between the pre‐ and post‐COVID‐19 hearing evaluations revealed a significant difference (F (5,74) = 2.720, p = 0.026, η ² = 0.155) in the low‐, mid‐, and high‐frequency ranges, all of which were higher post‐infection (F (1,78) = 5.152, p = 0.026, η ² = 0.062; F (1,78) = 13.064, p = 0.001, η ² = 0.143; F (1,78) = 10.064, p = 0.002, η ² = 0.114, respectively). The SRTs did not differ significantly (F (1,78) = 2.443, p = 0.122, η ² = 0.030), nor were there significant differences in the PB word identification percentage (F (1,78) = 0.575, p = 0.451, η ² = 0.007) pre‐ and post‐infection. Figure 2 presents the average pure‐tone hearing thresholds pre‐ and post‐COVID‐19.

Fig. 2.

Changes in low‐, mid‐, and high‐frequency hearing thresholds in the left (A) and (B) right ears. COVID‐19, coronavirus disease. [Color figure can be viewed in the online issue, which is available at www.laryngoscope.com.]

However, when a one‐way repeated measures MANCOVA was performed with participant age at and duration between the last hearing evaluation pre‐COVID‐19 and first hearing evaluation post‐COVID‐19 as covariants, no significant difference was found (F (5,72) = 1.973, p = 0.093, η ² = 0.120).

Pearson's correlations showed no linear connection between the health risk factors and hearing threshold changes in each of the three frequency ranges. As no correlations were found, no additional statistical analysis was performed.

Moreover, two paired sample t‐tests were performed to determine whether there was a larger number of self‐reports of tinnitus and vertigo post‐COVID‐19. The analysis showed no significant difference in the self‐reports of either tinnitus (t(38) = −1.00, p = 0.324) or vertigo (t(36) = −1.405, p = 0.169) pre and postinfection.

DISCUSSION

This study aimed to evaluate the association between COVID‐19 infection and deterioration in sensorineural hearing thresholds in a cohort of patients insured by Meuhedet Health care Services, which maintained records of data on hearing evaluations performed pre‐ and post‐COVID‐19. All patients were infected during the first two waves of COVID‐19 before the availability of vaccines and emergence of variants. We divided the hearing thresholds into three frequency ranges (low, mid, and high) and compared them with SRTs and word identification scores pre and postinfection. Subsequently, we performed the analysis again, controlling for the patient's age at and duration between hearing evaluations. Furthermore, we tested the association between the appearance of tinnitus and vertigo and deterioration of thresholds. Finally, we correlated threshold changes to background characteristics, such as hypertension, diabetes mellitus, and smoking, which are all known risk factors for hearing loss.

Our first research question was regarding whether deterioration in hearing thresholds was associated with COVID‐19. Indeed, we found significant differences between the pre‐ and post‐COVID‐19 thresholds at all three (low, mid, and high) frequency ranges. However, although these differences were significant, their clinical importance was very small (mean of 3–4 dB across the frequency ranges). In addition, when we analyzed the effects of age and duration between the two hearing evaluations, the results were no longer significant. These findings may reflect the expected age‐related hearing threshold shift. Indeed, most of our patients were older than 55 years of age, and as reported in previous cross‐sectional and longitudinal studies, the average rate of change in pure‐tone thresholds is approximately 1 dB per year. ²¹ This is in line with our findings, as the average duration between the hearing evaluations was approximately 4 years, and the threshold shift was between 3–4 dB, reflecting a possible change rate of approximately 1 dB per year. Moreover, as all the patients in our cohort exhibited SNHL during their baseline hearing evaluation, the threshold shift may be attributed to the natural progression of their hearing loss.

Our findings are supported by those of a few previous studies. Bhatta et al. (2021) found no differences in hearing thresholds between 331 COVID‐19 patients and 80 controls. Moreover, in 3.2% of the cases in which a hearing change was observed during the first evaluation, the hearing loss had resolved by the time of the follow‐up evaluation and was attributed to a conductive component rather than to SNHL. ²² Dror et al., also did not find any differences in hearing thresholds, OAEs, and auditory brainstem responses between COVID‐19 patients and controls. ¹⁰ Similarly, Gallus et al., compared hearing test results between 48 COVID‐19 patients and 28 controls and found that both groups had normal hearing. ²³

Our second research question was regarding whether the risk factors associated with hearing loss, such as smoking, hypertension, and diabetes mellitus, along with the severity of the COVID‐19 infection would exacerbate the deterioration in hearing thresholds post‐COVID‐19. Among the aforementioned risk factors, evidence has been accumulated regarding the relationship between smoking and hearing loss in both cross‐sectional ¹⁶ and longitudinal studies. ¹⁷ Furthermore, although diabetes mellitus is considered a risk factor for hearing loss, some studies have reported that no such relationship exists when the associations are adjusted for age, biological sex, and hypertension. ¹⁸ However, others have found that diabetes is an independent risk factor for hearing impairment. ¹⁹ In addition, studies have shown a relationship between hypertension and hearing impairment. For example, a recent study found a hearing loss prevalence of 37.4% in adults with hypertension aged 18–55 years compared with that of 14.1% in those without hypertension. ²⁰ Although these factors are associated with hearing loss, our findings indicate no correlation between them and a hearing threshold shift post‐COVID‐19. Therefore, no synergetic effects were found.

The present study has several strengths. First, owing to the comparison of objective measurements pre‐ and post‐COVID‐19, subjective patient reports, which have been demonstrated to be influenced by recall bias and nocebo effects, were avoided. ¹⁵ Second, the within‐subject comparison of hearing thresholds pre‐ and post‐COVID‐19 did not rely on a matched control group, for which participants need to be selected carefully such that their background factors match those of the patients. Third, because it has been shown that the BNT162b2 messenger ribonucleic acid (RNA) COVID‐19 vaccine raised the incidence of sudden SNHL, ²² the timing of the retrospective analysis was chosen such that it was prior to the availability of any COVID‐19 vaccine and emergence of any variant. Fourth, the use of a relatively large number of patients from Meuhedet EHRs enabled the performance of a correlation analysis between threshold changes and background factors and of statistical adjustments for age and duration between the hearing tests.

Despite the novelty and strengths of this study, some limitations should be considered. First, the study cohort consisted of COVID‐19 infected patients, all of whom had experienced SHNL before contracting COVID‐19. Therefore, the generalizability of the results to people reporting changes in hearing as a new symptom post‐infection is limited. Thus, in future studies, the incidence of new hearing loss in a patient cohort post‐COVID‐19 should be evaluated in comparison with that in a similarly large cohort (matched for age and background characteristics) comprising individuals not infected with COVID‐19. Second, our sample included unvaccinated patients infected during the first two waves of COVID‐19. However, as there have been some reports on the effect of vaccines on hearing, ²⁴ , ²⁵ future studies should also examine vaccinated patients. Finally, our study tested only short‐term effects. Future studies should also aim to test the long‐term effects of COVID‐19 on hearing.

CONCLUSION

The current study adds important knowledge to the literature, as it has elucidated the effect of COVID‐19 on hearing among patients with a baseline (preinfection) audiogram. Our conclusion is that COVID‐19 does not appear to be associated with a deterioration in sensorineural hearing in unvaccinated cohort of patients with known hearing loss after correcting for age‐related hearing loss. Moreover, our findings do not validate the finding indicating greater deterioration in the hearing of COVID‐19 patients with background risk factors, such as smoking, hypertension, and diabetes, thereby suggesting that there is no interaction between risk factors and hearing deterioration post‐infection.