Why Do Cochlear Implant Candidates Defer Surgery? A Retrospective Case-Control Study

Abstract

Objective(s):

Despite undergoing thorough cochlear implant (CI) candidacy evaluation and counseling, some patients ultimately elect against implantation. This study sought to identify patient-related and socioeconomic factors predicting CI deferral.

Methods:

A retrospective study of adult (≥18 years old) CI candidates presenting between 2007 and 2021 at a tertiary academic CI center was performed. The primary outcome was device implantation. Data collected included age, gender, hearing status, race, zip code of residence, median family income (MFI), distance traveled from the CI center, marital status, employment status and insurance status. Multivariable binary logistic regression was performed to identify predictors of implantation.

Results:

A total of 200 patients qualifying for cochlear implantation were included, encompassing 77 adults deferring surgery (CI-deferred) and 123 consecutive adults electing for surgery (CI-pursued). Age, gender, hearing status, insurance type, employment status, distance from the implant center, and MFI were comparable between the groups (p>0.05). Compared to CI-pursued patients, CI-deferred patients were more likely to be non-Caucasian (24.7% vs 9.8%, p=0.015) and unmarried (55.8% vs 38.2%, p=0.015). On multivariable logistic regression, older age (OR 0.981, 0.964–0.998, p=0.027), African American race (OR 0.227, 0.071–0.726, p=0.012), and unmarried status (OR 0.505, 0.273–0.935, p=0.030) were independent predictors of implant deferral.

Conclusion:

This study demonstrates that increasing age at evaluation, African American race, and unmarried status are predictors for deferring CI surgery despite being implant candidates. These patients may benefit from increased outreach in the form of counseling, education and social support prior to undergoing CI surgery.

Lay Summary:

Cochlear implants are hearing restoration devices that greatly benefit hard-of-hearing adults. However, not all who qualify ultimately choose to receive the implant. This study suggests older, African American, and unmarried patients are less likely to undergo implantation.

Level of Evidence:

3 – retrospective study with internal control group

Introduction

It is estimated that 1.2 million adults currently live in the United States (US) with severe to profound sensorineural hearing loss (SNHL) that would benefit from cochlear implantation (CI). Despite this, the rate of CI in adults remains extremely low at 10–12%^1–3. Significant contributors to this issue include: 1) lack of well-established screening strategies for adult hearing loss, 2) low referral volumes, and 3) lack of clinician familiarity with implant criteria^2,4,5. Furthermore, once patients qualify for implantation, up to 20–50% ultimately elect against surgery^6–8. This occurs despite mounting evidence linking hearing loss to cognitive decline and dementia, social isolation and depression, decreased employability, and increased risk in falls and hospitalizations^9–11.

The economic impact of hearing loss in the US is significant, with excess medical costs ranging between $3.3–12.8 billion¹⁰. On an individual level, adults with untreated hearing loss in the US could face a near-50% increase in total healthcare expenditure over a 10-year period compared to those with treated hearing loss⁹. Studies also show CI surgery carries a favorable cost-utility and effectiveness profile for adults with severe to profound SNHL in the US and other developed countries^13–16. Subsequently, increasing CI adoption is necessary to curb financial implications of adult hearing loss, particularly in the context of a growing elderly population.

Therefore, identifying patient-specific factors predicting pursuit of CI surgery is important in determining ways to improve uptake. Previous studies have found longer distances from CI centers, older patient age, racial minorities, and unmarried status negatively affect the likelihood of undergoing implantation^7,17,18. This study therefore sought to validate and investigate other socioeconomic and patient-related factors, such as income and employment status, that may impact patients’ decision-making in undergoing CI surgery. In addition, given the higher need for implantation in the elderly population, these factors were reanalyzed in a subset of patients aged 65 years and older to identify changes in predictor profile.

Methods

The current study was approved by the institutional review board (IRB) of The Ohio State University Wexner Medical Center (IRB# 2020H0457), a tertiary high-volume CI referral center. A retrospective review of the electronic medical charts was performed to identify all adults (≥18 years old) who have undergone formal CI evaluations between the years of 2007–2021. Patients that qualified for CIs following evaluation and ultimately underwent CI surgery were categorized as “CI-pursued”, whereas patients who elected to defer surgery were categorized as “CI-deferred”. Of note, the CI-pursued group was composed of a series of consecutive patients by date of CI evaluation between a narrower duration of 2016–2018 given their significantly higher representation compared to the CI-deferred group and temporal association with the 2018 median family income (MFI) data at hand.

Data Collection

Age at evaluation, gender, race, zip code, insurance status, marital status, and employment status were collected. Race was categorized as Caucasian, African American, or other (e.g. Hispanic, Asian). Insurance status was categorized as with or without insurance. Marital status was categorized into two categories, one collective category for single, divorced, or separated and the other for married. Employment status was categorized as disabled, unemployed/retired, or employed. The primary outcome was undergoing CI surgery.

Other socioeconomic factors such as distance from the CI center, MFI in US dollars (USD), and MFI percentile ranking were derived separately. Patient distance from the CI center was calculated by the distance between zip codes of the CI center (43212) and patient residence by latitude and longitudinal coordinates. MFI and percentile ranking were obtained by zip code through the 2018 census zip code tabulation areas (ZCTAs)¹⁹.

Preoperative hearing data was queried wherever available. This included unaided pure tone averages (PTA, decibels [dB HL]), word recognition scores (WRS, %), and AzBio sentence (in quiet and noise, %), Hearing in Noise Test (HINT, %), and Consonant-Nucleus-Consonant (CNC, %) word test scores. In the setting of missing data, two assumptions were made to increase yield: 1) If the binaural AzBio score was 0% then individually tested ears were also assumed to score 0%; 2) If the ear of interest was known to be of poor hearing (e.g. implanted ear in the CI-pursued group) and not tested in the available audiogram, the unaided PTA was assumed to be in the profound level (≥90 dB HL). In the CI-deferred group, candidate ears were defined as the poorer performing ear by audiologic measures, and these ears were grouped with the known implanted ears in the CI-pursued group. Contralateral ears were therefore defined as the opposite ear and grouped together.

Statistics

Statistical analyses were performed utilizing SPSS statistical software (Version 28.01.1; Armonk, NY). Variables were presented through summary statistics. Continuous variables were tested for normal distribution by the Kolmogorov-Smirnov and Shapiro-Wilk tests and presented by means or medians as appropriate. Categorical variables were presented as frequencies and percentages. Comparison of continuous variables were performed by an independent t- or Mann-Whitney U test as appropriate. Comparison of categorical variables were performed utilizing a Fischer exact or chi-squared tests.

Predictors of undergoing CI surgery were identified through multivariable binary logistic regression with an enter likelihood ratio method (threshold of p<0.15). Age at evaluation, sex, race, distance from CI center, MFI (USD and percentile), insurance status, marital status, and employment status were included in the model. Effect size was expressed as odds ratios (OR) accompanied by 95% confidence intervals.

Results

There were 77 adults who qualified for CI but ultimately deferred surgery (“CI-deferred”). This group was compared to 123 consecutive adults who were evaluated at this center between the years 2016–2018 and underwent implantation (“CI-pursued”).

Patient demographics, including socioeconomic characteristics, were summarized in Table 1. Median age at evaluation was not statistically significant between CI-deferred and CI-pursued groups (64 vs 64 years, p=0.178). Statistically significant differences were observed in race and marital status. Compared to CI-pursued patients, CI-deferred patients were more often African American (13.0% vs 4.1%, p=0.015) or other race (11.7% vs 5.7%, p=0.015) and unmarried (55.8% vs 38.2%, p=0.015). Patients in the CI-deferred group had similar insurance rates compared to the CI-pursued group (93.5% vs 98.4%, p=0.110).

Table 1.

Patient Demographics. SD – standard deviation, MFI – median family income

		CI-deferred (N = 77)	CI-pursued (N = 123)	P
Age, years, median (range)		64 (25–94)	64 (18–89)	0.178
Gender				0.402
	Male	36 (46.7%)	65 (52.8%)
	Female	41 (53.2%)	58 (47.2%)
Insurance Status				0.110
	Uninsured	5 (6.5%)	2 (1.6%)
	Insured	72 (93.5%)	121 (98.4%)
Marital Status				0.015*
	Single, divorced, separated, widowed	43 (55.8%)	47 (38.2%)
	Married	34 (44.2%)	76 (61.8%)
Race				0.015*
	Caucasian	58 (75.3%)	111 (90.2%)
	African American	10 (13.0%)	5 (4.1%)
	Others	9 (11.7%)	7 (5.7%)
Employment Status				0.584
	Disabled	9 (11.7%)	10 (8.1%)
	Unemployed/Retired	55 (71.4%)	87 (70.7%)
	Employed	13 (16.9%)	26 (21.1%)
Distance from CI center, miles, median (range)		30.3 (0–406.0)	38.6 (1.7–1644.2)	0.491
MFI, USD, median (range)		$64,626 (23,125–143,421)	$68,438 (23,125–127,661)	0.583
MFI percentile, median (range)		47 (0–96)	55 (0–95)	0.778

^*- statistically significant by Mann-Whitney U or Fischer Exact/𝜒² test (p < 0.05)

Preoperative hearing status are shown in Table 2. Unaided PTAs were categorized as normal hearing (0–24dB), mild (25–40dB), moderate (41–55dB), moderately severe (56–70dB), severe (71–90dB), and profound SNHL/not tested (90+dB). AzBio sentence scores in quiet, including binaural, were stratified into categories of >60%, 41–60%, 21–40%, and 0–20%, accounting for historical thresholds qualifying for CI. No difference in preoperative hearing performance between the two groups based on these measures were observed (p>0.05 for all). None of the audiometric variables were found significant on univariable analysis (p>0.05 for all analyses).

Table 2.

Baseline/pre-operative hearing status. PTA – pure tone average

		CI-deferred (N=77)	CI-pursued (N=123)	P
Candidate/Implanted Ear PTA				0.500
	Normal hearing	0 (0%)	0 (0%)
	Mild hearing loss	0 (0%)	3 (2.5%)
	Moderate hearing loss	1 (1.5%)	5 (4.1%)
	Moderately severe hearing loss	14 (21.5%)	17 (14.1%)
	Severe hearing loss	20 (30.8%)	40 (33.1%)
	Profound hearing loss/Not tested	30 (46.2%)	56 (46.3%)
Contralateral Ear PTA				0.710
	Normal hearing	1 (1.5%)	5 (4.3%)
	Mild hearing loss	0 (0%)	4 (3.4%)
	Moderate hearing loss	5 (7.7%)	11 (9.5%)
	Moderately severe hearing loss	18 (27.7%)	29 (25.0%)
	Severe hearing loss	22 (33.8%)	36 (31.0%)
	Profound hearing loss/Not tested	19 (29.2%)	31 (26.7%)
Binaural AzBio Score in Quiet				0.746
	>60%	17 (32.1%)	34 (32.7%)
	41–60%	6 (11.3%)	14 (13.5%)
	21–40%	10 (18.9%)	25 (24.0%)
	0–20%	20 (37.7%)	31 (29.8%)
Candidate/Implanted Ear AzBio Score in Quiet				0.403
	>60%	3 (5.8%)	6 (4.9%)
	41–60%	5 (9.6%)	5 (4.1%)
	21–40%	8 (15.4%)	11 (8.9%)
	0–20%	36 (69.2%)	86 (69.9%)
Contralateral Ear AzBio Score in Quiet				0.297
	>60%	7 (13.5%)	18 (17.1%)
	41–60%	3 (5.8%)	13 (12.4%)
	21–40%	9 (17.3%)	24 (22.9%)
	0–20%	33 (63.5%)	50 (47.6%)

A summary of the multivariable binary logistic regression analysis is displayed in Table 3. Age at evaluation, marital status, race, and insurance status were included in the regression model, for which all but insurance status were significant independent predictors for undergoing surgery. Every 5-year increase in age lead to a 9% decrease in relative likelihood of undergoing surgery. This conferred an 18% and 32% decrease in relative likelihood of implantation for every 10- and 20-year increase in age, respectively. Unmarried patients were half as likely to undergo surgery (OR 0.505, p=0.030). Finally, African Americans were approximately four-times less likely to undergo CI surgery compared to Caucasian patients (OR 0.227, p=0.015). Insurance status approached significance, suggesting lacking insurance would decrease the likelihood of undergoing CI surgery (OR 0.220, p=0.089).

Table 3.

Patient-related and socioeconomic predictors of cochlear implant pursuit by multivariable binary logistic regression

		Odds Ratio (95% CI)	P
Age at evaluation (years)		0.981 (0.964–0.998)	0.027*
Marital Status
	Married	1 (ref)	–
	Single/Divorced/Widowed	0.505, 0.273–0.935	0.030*
Race
	Caucasian	1 (ref)	–
	African American	0.227 (0.071–0.726)	0.012*
	Other race	0.343 (0.116–1.018)	0.054
Insurance Status
	Insured	1 (ref)	–
	Uninsured	0.220 (0.038–1.26)	0.089

^*- statistically significant by Mann-Whitney U or Fischer Exact/𝜒² test (p < 0.05)

Lastly, a sub-analysis was performed on an elderly subgroup (age ≥65 years) with identical variables. There were 40 and 61 patients within this age group in the CI-deferred and CI-pursued cohorts, respectively. Increasing age was the only persisting independent predictor of CI deferral. Its effect size in this elderly sample was slightly greater (OR 0.942, p=0.043) than when including adults of all ages (OR 0.981).

Discussion

The adult cochlear implant (CI) uptake and utilization rate is exceedingly low in the US despite a tremendous need, currently estimated at 10–12%^1,2. Existing literature examines the underlying roots of this problem, which is multifaceted and span from the challenge of identifying adults with hearing loss to educating clinicians on appropriate CI referrals⁵. Survey studies show that only 15–60% of primary care physicians routinely screen adults for hearing loss^2,4 and with inconsistent protocols. As little as 26% have ever referred patients for CI evaluation with unfamiliarity of CI candidacy being the most common reason⁴. As a result, only 3% of CI recipients obtain initial implantation referrals from their primary care providers². Once patients have undergone CI evaluation and qualify for surgery, as many as 20–50% ultimately decide against implantation, further limiting implantation rates^6–8. This study aimed to gain insight on patients at risk of deferring CI surgery in the context of socioeconomic factors. We found that age, race, and marital status were critical factors predicting surgery after being deemed CI candidates.

Relative to the growing evidence supporting implantation in older patients, surgical safety and device benefits may be understated when counseling older patients^20–22. CI surgery is established to be safe in the elderly^20,22–24. Even in very elderly patients (≥85 years), Spitzer et al found comparable complication rates to younger adults, at 5% or less, particularly when performed under local anesthesia²³. Furthermore, once implanted, the elderly patients enjoy considerable benefit from CIs by measures of device usage, hearing, and quality of life (QoL) outcomes^20–21,23. Spitzer et al found 71% of their very elderly cohort were wearing their device full-time²³ and Wichova et al also demonstrated that their elderly patients utilized the device as much as the younger population²¹. In a recent multi-institutional prospective study, Wick et al demonstrated clinically meaningful benefit of CIs in audiometric and QoL measures in patients ≥65 years²⁰. Given this, a patient should not feel or be deemed “unfit” for implantation simply due to their age.

Despite this, lower implantation rates continue to be observed in the elderly^7,18. A study by Henkin et al noted patients that pursued surgery were significantly younger than those that deferred surgery by almost a decade (47.3 vs 57.9 years). Although the median age of our CI-pursued and CI-deferred patient cohorts were both 64 years, we found an incremental decrease in likelihood of pursuing CI surgery with increasing age. For every additional year in age, the likelihood of undergoing surgery declined by 2%. This extrapolates to an 18% and 32% decrease in likelihood of implantation for every10- and 20-year increase in age, respectively. Tolisano et al showed a comparable decrease in likelihood as patients aged (3% per year, OR 0.97)⁷. The significance of age is further emphasized after examining a subset of the current cohort aged ≥65 years. On multivariable analysis, age was the only persisting independent predictor and its influence more pronounced: there was a 6% incremental decrease in likelihood of undergoing surgery for every year increase beyond 65 years. For the same 10- and 20-year increase in age, this calculates to a 46% and 71% decrease in implantation likelihood, respectively. We hypothesize that older patients may view the risk-benefit ratio of surgery as unfavorable, thereby leading to implant deferral. Therefore, educating both patients and healthcare professionals on the safety and benefits of implantation is critical from stages of patient referral through CI evaluation and implantation. Surely, additional studies are necessary to understand the most effective approach to increase device uptake in the elderly.

The impact of race is described broadly within healthcare literature, and adult hearing rehabilitation is no exception with racial minorities suffering poorer outcomes^25–27. When controlling for age and severity of hearing loss, African Americans are roughly 60% less likely to be hearing aid users compared to Caucasians, and Hispanics are nearly 80% less likely²⁷. Tolisano et al observed that minorities were only half as likely to undergo CI surgery compared to Caucasian patients (OR 0.47)⁷. Once electing for a CI, minorities could experience a six-fold delay in receiving an implant compared to Caucasians¹². Remarkably, we found that African Americans are four-times less likely to undergo surgery compared to Caucasians. This negative impact is consistent with findings by Tolisano et al⁷. However, the group did not examine African American race as a subcategory, which may explain the higher magnitude in our study. While minority race is often associated with lower socioeconomic standing in terms of income and insurance status, these variables were not significant in the current study. This may be secondary to the geographic location (Central Ohio) of our CI center.

Discrepancies in treatment pursuit likely also hinge on broader social issues, including racial biases and mistrust in healthcare. For example, 32% of African Americans and 20% of Hispanics in the US report discrimination in clinical encounters, leading to 22% and 17% avoiding seeking medical attention due to fears of discrimination, respectively^25,28. However, one might also hypothesize that non-English speaking status, poorer health literacy, and lower education could also be confounding factors ^29–31. Nevertheless, increased outreach efforts, as in a pilot outreach study by Sims et al³², may be required to improve implantation rates in minority populations. The group partnered with local organizations to improve education, develop support groups, and work on local policy to improve CI access. The group also suggested implementing a CI telemedicine clinic to increase access of marginalized minorities³². Diversifying the hearing healthcare workforce has also been suggested to strengthen outcomes of hearing health³³.

Familial support for hearing-impaired individuals positively impacts efforts towards hearing rehabilitation. Studies show that family involvement is associated with seeking treatment and adopting hearing aids^34–36. These observations are often driven by underlying third-party disability, or disability of a patient extending to and impacting family members, eroding the quality of life of all individuals involved^37,38. In the context of CI adoption, familial support is also valued as a motivator in a patients’ decision-making process³⁹. Involvement of significant others may help instill patient self-confidence to reach post-operative goals⁴⁰. In our study, unmarried individuals were twice as likely to defer CI surgery compared to married patients. To combat this, encouraging involvement of family members or even connecting with a CI recipient support group may serve as meaningful tactics for improving CI uptake.

It is critical to note that in the current study, hearing status did not influence the outcome of undergoing surgery. Between those who deferred versus pursued surgery, the hearing loss severity profiles demonstrated no differences (Table 2). Interestingly, this was irrespective of hearing abilities binaurally, in the candidate/implanted or contralateral ears. This result suggests that once patients are considered implant candidates, hearing status largely escapes the decision-making process to proceed with surgery. This underscores accounting for socioeconomic factors when counseling potential CI candidates. Of note, SAT/SRT and WRS were inconsistently documented in available audiograms (<50% of cases) and therefore excluded from further analyses. Similarly, AzBio in noise, CNC, and HINT scores were rarely tested and therefore excluded.

The primary limitation to this study is its retrospective nature. The quality of pre-operative counseling was not controlled for consistency. This point is important when considering counseling minorities for which language, medical literacy, and cultural differences would stand as considerable barriers to make an informed decision. In addition, variables such as distance from CI center and MFI were estimates and are not accurate reflections of true values. For instance, income likely varies significantly within a particular zip code of residence, which may explain why the variable was found non-significant on statistical analyses. Other limitations are the absence of medical comorbidity and comprehensive hearing data. Medical comorbidities are potential confounders to undergoing surgery and affect different demographics at varying rates. Regarding hearing, one can hypothesize that candidates with residual hearing and some benefit from hearing aid use may forego CI surgery, whereas patients with no hearing and no benefit from hearing aids would favor surgery. Though data from the current study refutes this hypothesis, assumptions to hearing performance were needed to supplant the available data. Also, sparsely available variables outside of unaided PTAs and AzBio sentence scores that were excluded in the analyses could carry unfounded influence on undergoing CI surgery. Future studies examining these audiometric components and their interplay with socioeconomic factors are necessary.

Finally, regarding the study design, it is critical to note that the study dates (2007–2021) were primarily selected to raise the power for candidates deferring surgery (“CI-deferred”), as this cohort was significantly smaller than the cohort that underwent surgery (“CI-pursued”). The CI-pursued group was a representative cohort of consecutive candidates by date of evaluation between years 2016–2018 to best coincide with that of the CI-deferred group and available 2018 MFI data. Not all patients that pursued surgery within all years of study were included as adequate power was achieved, and thus, limited statistical benefit is expected by further increasing the number of patients within this group without a corresponding increase in its counterpart. The true rate of implantation at our institution is therefore not reflected in the current study sample.

Conclusion

This study demonstrates that increasing age at evaluation, African American race, and unmarried status are significant predictors of deferring CI surgery despite their candidacy. These particularly vulnerable patient cohorts may benefit from increased healthcare provider outreach in the form of additional counseling, targeted education, and increased social support. Further research is necessary to determine exact pre-operative needs to maximize implant utilization in these cohorts.