Sensorineural hearing loss after cardiac surgery: a systematic review

Sensorineural hearing loss (SNHL) may occur following cardiac surgery. Although preventing post‐operative complications is vitally important in cardiac surgery, there are few guidelines regarding this issue. This review aimed to characterize SNHL after cardiac surgery.


Introduction
In cardiac surgery, guidelines for perioperative care must consider potential post-operative complications to optimize patient outcomes. 1he prevention of post-operative complications is essential given the strains placed on healthcare systems associated with the COVID-19 pandemic. 2,3In addition to mortality and the factors that affect it, [4][5][6] current international guidelines for care after cardiac surgery focus on strategies to manage glycaemia, pain, delirium, persistent hypothermia, chest tube patency, thromboprophylaxis, acute kidney injury and anaemia. 1,7Hearing loss has also been associated with cardiac surgery, but there are few guidelines regarding this issue.
Hearing loss is an uncommon complication after non-otological surgery. 8The phenomenon has been reported in the cardiac surgery literature for over 40 years 9 ; however, the overall prevalence across the global evidence base is unknown.Accordingly, it is not a significant priority in current perioperative guidelines. 1 Therefore, to inform international cardiac surgery clinical care, we performed a systematic review to characterize sensorineural hearing loss (SNHL) after cardiac surgery.

Methods
The methods for this study were established within a protocol developed before commencement of the review.We prospectively registered this protocol with PROSPERO (number CRD42020177047), and followed the preferred reporting items for systematic reviews and meta-analyses 2020 (PRISMA 2020) 10 reporting guidelines (Appendix S1).

Search strategy and selection criteria
The population, intervention, comparator group, outcome (PICO) framework was used to develop the research questions and inclusion criteria.The population comprised patients undergoing cardiac surgery who experienced post-operative SNHL.The intervention was any cardiac surgical procedure.There was no comparator group.The primary outcome was post-operative SNHL, measured by pure-tone audiometry, distortion-product otoacoustic emissions (DPOAE), or auditory brainstem responses (ABRs).Study designs that did not report these observational data, such as reviews, editorials, perspectives, letters or conference abstracts were excluded.Studies, such as review articles, that did not present primary data were excluded.
PubMed (incorporating MEDLINE), Embase and the Cochrane Library were searched from database inception to 14 April 2021 for studies of any design and in any setting that reported observational data regarding SNHL after cardiac surgery.The search strategies used can be found in the Appendix S2.Searches were supplemented by consultation of current contents, reviews and original research relating to SNHL after cardiac surgery identified through targeted searches of Google Scholar and PubMed.Searches were not limited by language; no publication restrictions were implemented.During the process of searching, seven full-texts could not be obtained.

Data extraction and analysis
Two reviewers (J.D. and J.M.G.) independently screened titles and abstracts, reviewed full texts and extracted data using a standard extraction form.Screening of titles and abstracts was facilitated through the use of a web application (Rayyan, Qatar Computing Research Institute, Ar-Rayyan, Qatar). 11Disagreements were resolved by consensus.Extracted data included research design, study setting, population characteristics, intervention characteristics, comparator characteristics, timeframe for follow-up, quantitative and qualitative outcomes, source of funding and reported conflicts of interest, methodological quality information and other information relevant to the review questions.Data were synthesized in narrative and tabular formats.The primary outcome was postoperative sensorineural hearing function.The secondary outcome of interest was the grade of hearing impairment according to global guidelines developed by the World Health Organization. 12Qualitative data relevant to the study outcomes were organized and synthesized according to thematic commonality.Quantitative data pertinent to study outcomes were summarized to determine effect sizes across the included studies.Meta-analysis was precluded due to heterogeneity in reported data across the included studies.
Two reviewers (J.D. and J.M.G.) independently assessed methodological quality using validated tools.The Downs and Black checklist 13 was used for risk of bias assessment for included nonrandomized studies of interventions (NRSIs).The Joanna Briggs Institute Checklist for Case Reports was used to evaluate methodological quality for included case reports. 14The Cochrane Collaboration's tool 15 for assessing risk of bias in randomized controlled trials (RCTs) would have been utilized; however, no eligible RCTs were identified.

Hearing loss data analysis
In an attempt to standardize results, the World Health Organisation (WHO) classification of hearing loss was applied to every audiometric value in the identified studies.Although the classification of SNHL requires the average pure tone audiometry of the better ear, the audiometric results of the non-affected (better) ear were often not provided.Thus, all pure tone audiometry values for the effected ear were collected and the severity of hearing loss was graded using the WHO criteria across all included studies. 16Findings of DPOEA and transient otoacoustic emissions (TOEA) were also noted; however, they were not included in the final assessment of hearing loss if audiometry results were unavailable to clarify the severity of the loss.

Results
Our search identified a total of 1051 records (1005 unique reports), from which 49 full-text articles were retrieved.Twenty-three of these studies were included in the systematic review (Fig. 1).A list of studies excluded at full-text review, with a justification of exclusion for each study, can be found in the Appendix S3.The characteristics of the included studies are outlined in Table 1.
Included studies used different definitions and standards for the degree of hearing loss.Some of the earlier studies, including Shapiro et al. qualify significant hearing loss as any deficit exceeding 10 dB, whereas others use WHO criteria.There is considerable diversity not only in the definition of hearing loss but also in how to scale the severity of the hearing impairment itself.The two prevailing classification systems of SNHL used most commonly across the literature are the World Health Organisation (WHO) classification, as well as the newer Global Burden of Diseases (GBD) classification.

Post-operative sensorineural hearing loss incidence
There were approximately 133 cases of SNHL after cardiac surgery that were identified in the literature across both adult and paediatric populations.Of these 45 were adult cases, and 88 were paediatric.Out of the six cohort studies in adult populations, there was a total of 36 hearing losses defined by losses greater than 10 dB on a postoperative audiogram, out of a total of 7135 patients studied, giving an incidence of 4.75 cases per 1000 operations.Plasse et al. in 1980 estimated an incidence of 1 in 1000 (0.001) of this relatively new phenomenon, describing a total of seven patients with a hearing loss out of 7000 who underwent cardiac surgery. 9Shapiro et al. showed 9 of their 68 with greater than 10 dB losses. 17This was in contrast to Brownson et al. who showed no hearing loss in a cohort of 50 patients undergoing cardiopulmonary bypass surgery (CPBS).This was corroborated by Ness et al., who conducted a prospective cohort analysis in 1993, also showing no objective hearing loss in their 145 patients cohort as well. 18They did, however identify four cases of subjective hearing loss with standard audiometry.The other four cohort studies; Shapiro, Phillipps, Aytacoglu and Plasse all found some level of hearing loss within their cohorts. 9,17,19,20hillipps et al. showed four individuals out of a cohort of 20. 19 A more recent 32 analysis conducted by Aytacoglu et al. demonstrated a surprising 12 out of 37 patients with 34 hearing loss after undergoing cardiac surgery, both with and without extracorporeal circulation (ECC). 20][23][24][25][26][27] There were multiple variables across these studies, including follow-up times, presence of genetic abnormalities, and type of cardiac surgery, with a far more significant proportion of patients undergoing surgery for congenital cardiac surgery such as the Norwood procedure for Hypoplastic left heart syndrome.Overall, the incidence of hearing loss across all papers was 97 out of 1342 (65.6 per 1000 operations).

Level of hearing impairment
Audiogram results were compiled where available, to provide analysis of the severity of hearing loss in each of these cases.The majority of cases (34/45, 75.6%) were unilateral in nature.There was often limited pre-operative audiometric data particularly in observational case studies.In all cases, there was a subjective hearing loss by patients who had no known pre-existing hearing pathology and were not under ongoing otological care.With the application of the WHO criteria for measuring the degree of hearing loss, the majority of hearing loss in effected ears were classified as mild in severity (27/45, 60% of cases) with hearing thresholds greater than or equal to 10 dB and less than 25 dB in the affected ear at all frequencies measured.11.1% (5/45) of all cases were moderate in severity (41-55 dB).Severe hearing loss (55-90 dB) was present in 17.8% (8/145) of those with hearing loss.Profound hearing loss, defined as hearing threshold greater than 90 dB, in the affected ear was present in 11.1% (5/45).
There were notable similarities and differences between the adult and paediatric populations.Amongst the paediatric population, only 25.7% (25/97) of cases were mild in severity where as 75.3% (73/97) were moderate severity and over.Interestingly amongst the paediatric population, hearing loss was far more often bilateral when compared with the adult population where hearing loss was usually unilateral post-surgery.Of the papers which provided such information 80.1% of cases (42/52) showed bilateral hearing loss.It was difficult to determine improvement in hearing losses over time due to minimal testing beyond the acute post operative period.Plasse et al., one of the few studies to have serial testing showed that four out of seven patients showed some level of improvement 4 months later. 9As did Millen et al. showing mild improvement in their one case.Walstead et al. showed no improvement across months of their case series. 28Generally, studies focussing on adults usually had audiometry within the first 10 days, with some cases extending to 30-40 days post procedure.Paediatric studies had a far greater variability for time to first audiometric with one study measuring hearing even 4 years after surgery.

High versus low frequency
High frequency on audiometry is generally considered to be frequencies above 2000 Hz, values below this, commonly; 250, 500 and 1000 Hz are deemed to be low.High frequencies were the most frequently affected range of hearing in patients with SNHL following cardiac surgery.This frequency of hearing loss was the case with Arenberg's initial observed case. 19In a more recent cohort study, Aytacoglu et al. showed the largest changes in pre-and postoperative audiometry values from 4000 to 8000 Hz. 20 Furthermore, the 8000 Hz frequency showed a mean threshold level of 29 dB and qualifies in itself for mild hearing impairment of the affected ear.In case reports by Millen et al., Bruschi et al. and Carre et al., the greatest hearing loss occurred at 6000 and 8000 Hz. 20,21,26 Phillipps and Thornton also demonstrated this pattern in their cohort of 40 patients, in whom post-operative audiograms showed an increase in hearing threshold by at least 10 dB in the 6000 and 8000 Hz frequencies. 19 Sensorineural hearing loss after cardiac surgery frequencies above 8000 Hz. 29 This study found that as the frequency increased to values above 8000 Hz, the hearing thresholds increased in all patients with hearing loss in a statistically significant manner.Shapiro found that 13.2% of their cohort of 68 patients identified hearing loss at high frequencies. 17Not only was there a proportionately larger number of patients with hearing loss in the higher frequencies, but there was also more severe hearing loss at frequencies above 6000 Hz in particular.Munjal et al. further strengthens this argument, finding highly significant differences in hearing thresholds at frequencies of 10 000, 12 000 and 16 000 Hz (P < 0.0001). 30n paediatric studies, the majority of cases also showed highfrequency losses.In the Grasty et al. study looking at hearing loss after open-heart surgery, eight individuals had high-frequency hearing loss. 26Bork found that in 20/29 bilateral hearing loss cases there were hearing deficits at higher frequencies. 22In the El Ganzoury et al. study, there was no objective hearing loss found in paediatric patients undergoing cardiac surgery on audiometry. 24n this study, it was noted that DPOAEs, and also later TEOAEs, of the hypothermic group showed a statistically significant reduction of amplitude at high frequencies.

CPBS with extracorporeal circulation
The included studies described evidence that SNHL associated with cardiac surgery may be associated with ECC.A study done by Phillipps and Thornton in 1996 was the first to show a statistically significant difference in hearing changes in patients who were undergoing CPBS, and hence requiring ECC, when compared to a control group undergoing thoracotomy. 19A more recent study by Aytacoglu et al. further supported this finding, showing a statistically significant post-operative hearing loss in patients undergoing CABG with ECC as compared to patients undergoing CABG without the use of ECC (P = 0.0426). 20The findings of Casale et al. did not confirm these findings in a small sample size of 10 patients, which showed no significant changes in hearing thresholds after CPBS. 31orin et al. used DPOAE measurements to assess cochlear dysfunction. 32This study found moderate hypothermia during heart surgery with ECC resulted in a statistically significant decrease in the amplitude of otoacoustic emissions in a 20 patient cohort (P < 0.05).El Ganzoury et al. in 2012 compared normothermic and hypothermic ECC to assess hearing outcomes. 24In this study, there was no overall significant change in audiometry.However, DPOAEs of the hypothermic group showed a significant reduction of amplitude, as was confirmed by TEOAEs that showed a 75% partial pass response.The hypothermic group had a significant reduction in the overall amplitude of TEOAEs at 2000-4000 Hz.Munjal et al. also supported this finding, showing a highly significant decrease in the signal to noise ratio at high frequencies in patients undergoing moderate hypothermia ECC CPBS (P = 0.0001). 30

Risk of bias
The included studies were of moderate methodological quality upon critical appraisal using the Downs and Black checklist. 13verall percentages on risk of bias assessments for the included studies can be found in Table 2. Mean scores, representing the  mean of the average of the two reviewer scores, were calculated for each category.Calculated: the total mean score was 22.1 out of 32 (range 18.5-26.5;Table 3).

Discussion
Hearing loss following cardiac surgery is a rare but potentially debilitating consequence of surgery.In an adult population, the available evidence suggests that this complication is most often unilateral and affects higher hearing frequencies.It appears that the issue may be more common in a paediatric population, although there are multiple confounding factors in this population.There is a possible association between ECC for CBPS and hearing loss.Historically, the accepted incidence of SNHL post-cardiac surgery is estimated to be 1 in 1000 (0.1%), as established in an evaluation of 7000 patients by Plasse et al. in 1980. 9Subsequent cohort studies have had substantially smaller sample sizes.Given the changes in cardiac surgery since 1980, further epidemiological studies to investigate the modern incidence of cardiac surgery are indicated.The paediatric incidence from cohort studies was elevated compared with the adult populations.Neonatal hypoxia has previously been implicated in brain injury and effect on hearing. 25,26In most studies, patients with pre-existing hearing loss were not included in the study to better isolate the surgery as a risk factor for hearing loss.However, it is difficult to interpret the described rates of hearing loss in the paediatric cohort due to the high proportion of patients with concurrent risk factors such as genetic abnormalities and intensive care unit admissions that may independently contribute to hearing loss.
4][35][36] The most discussed relate to hypoperfusion.Performing cardiac surgery may produce microemboli of any form that could occlude aspects of the vasculature associated with hearing.Similarly, hypoperfusion of elements within the cerebral vasculature as a result of cardiac surgery could produce similar results, particularly if prolonged hypotensive states occur intraoperatively.
The organ of hearing being an end-organ pertaining to vascular flow is potentially an important factor in the aetiology of this condition.4][35][36] When an offpump approach is utilized for coronary artery bypass surgery, manipulating the beating heart creates considerable haemodynamic fluctuations that could lead to similar issues with perfusion. 24The use of ECC may be accompanied by a hypothermic state that could also be associated. 24The available literature suggests an association between ECC and hearing loss following cardiac surgery. 20However, hearing loss has also been described in cases that did not involve ECC.There is a wide range in the timelines over which reported post-cardiac surgery hearing loss occurs, other mechanisms may also be implicated.Pharmacological intervention with ototoxic adverse effects may produce a post-operative hearing loss in some cases.Infection or autoimmune conditions may produce inflammatory states that could lead to vasoconstriction and hypoperfusion via a mechanism comparable to that discussed above.The intraoperative use of nitric oxide as a selective pulmonary vasodilator may simultaneously increase middle ear pressure and compromise the membrane, creating perilymphatic fistulae. 36he consequences of hearing loss can be significant for both adult and paediatric populations.In the adult population, hearing loss results in impaired communication, which can affect relationships and employment.Accordingly, hearing loss can have significant psychosocial and economic impacts for adults. 37Strategies can be employed to mitigate these effects of hearing loss, including communication aids and modification communication strategies.In children, hearing loss can result in significant developmental delays.These delays may impact upon education have significant long-term effects. 38Once detected, treatment strategies are available to help mitigate the impact that hearing loss has on the education and development of children.Accordingly, an argument could be made that, particularly for paediatric populations, it may be beneficial to have routine audiometric testing post-operatively so that SNHL can be identified early.This would enable the earlier implementation of strategies to prevent developmental and educational consequences of hearing loss.The health economic considerations of such a strategy would have to evaluate the frequency with which SNHL occurs, and the cost of screening, against the potential productivity lost over the life-course if significant educational and developmental delays are suffered.
This study has multiple limitations.Due to heterogeneity in reported data across the included studies, meta-analysis was precluded.However, quantitative and qualitative data relevant to the study outcomes were able to be synthesized and analysed so that useful clinical takeaways were able to be derived.The included studies were of low to moderate overall risk of bias, which may affect the reliability and clinical translatability of our results.Within the included studies, most reported solely post-operative values, so pre-existing hearing conditions were unable to be determined and potential confounding effects unable to be accounted for.Similarly, some studies did not report data regarding hearing loss at higher frequencies, which is where other included studies stated that the greater degree of post-operative hearing loss occurred.However, these data were valuable in informing our characterization of prevalence and severity of SNHL after cardiac surgery.The majority of studies were of smaller sample size and thus may not be accurately representative of real-life incidence rates, however, this was somewhat overcome by the very large study populations in multiple other included studies.Our data synthesis found variability to be present regarding definitions of hearing loss, techniques for measuring sensorineural hearing function, decibel thresholds for gradations of severity, and post-operative timepoints when measurements of hearing function were taken; however, is reflective of modern clinical practice and the limitations of research tools that are used to identify degrees of hearing loss.Few studies controlled for extraotological pathology that may impact hearing function, which would potentially add confounding to these reported data.Within the included studies, none reported both paediatric and adult populations, so direct comparisons were unable to be conducted.However, the clinical implications from our findings are likely to be different between paediatric and adult populations given the role bi-directional relationship between biopsychosocial development and sensorineural hearing function.
Further research in this area may target the epidemiology, pathophysiology, prevention and treatment of cardiac surgery-associated hearing loss. 39In over 40 years since this issue was first described, Plasse et al. has been the only large-scale study looking at the incidence of this phenomenon. 9In view of the changes in surgical procedures over the last 40 years, further studies are required to examine the incidence of hearing loss in the modern setting.Future studies that examine the frequency of hearing loss in the setting of cardiac surgery should employ internationally recognized criteria for hearing loss.Studies incorporating recurrent testing to better delineate long-term hearing outcomes will provide a more wholistic understanding on patient impact.Additional research to identify which patients are most at risk of developing this complication is required.Pragmatic studies are required to evaluate the effectiveness of screening strategies for SNHL following cardiac surgery.The need for studies evaluating screening for SNHL following cardiac surgery is strongest in the paediatric population, in whom hearing loss may be more difficult to detect and where it may influence developmental outcomes, and in diverse geographically and socioeconomic patient populations. 40,41Such studies may investigate the use of audiometry in combination with TEAOA or DPEOA, particularly for hearing loss at higher frequencies.

Conclusion
SNHL is a relatively uncommon but potentially serious complication after cardiac surgery.This hearing loss affects both paediatric and adult populations and may have significant long-term impacts.To inform clinical guidelines, future research should investigate methods of identifying patients at greater risk of SNHL after cardiac surgery and identifying patients who have suffered this complication early in the post-operative course.methodology; supervision; validation; visualization; writingreview and editing.Joseph N. Hewitt: Supervision; writingreview and editing.Raguwinder S. Sahota: Methodology; resources; supervision; validation; visualization; writingreview and editing.Justin Chan: Investigation; methodology; supervision; validation; visualization; writingreview and editing.Fabio Ramponi: Supervision; visualization; writingreview and editing.Giri Krishnan: Investigation; methodology; resources; supervision; validation; visualization; writingreview and editing.Aashray K. Gupta: Conceptualization; data curation; investigation; methodology; project administration; resources; supervision; validation; writingoriginal draft; writingreview and editing.

Table 1
Iriz et al. was the only group to measure Characteristics of included studies ANZ Journal of Surgery published by John Wiley & Sons Australia, Ltd on behalf of Royal Australasian College of Surgeons.

Table 2
Risk of bias assessment of included studies

Table 3
Risk of bias assessment of included studies JBI case report critical appraisal