Factors Influencing Time to Cochlear Implantation

Abstract

Objective:

To determine demographic and audiologic factors associated with time to treatment with cochlear implantation.

Methods:

Retrospective review of a prospectively maintained adult cochlear implant database. A total of 492 patients were implanted from 2012 to 2017. Time to implantation, pre-implantation audiologic outcomes, and demographic data were collected. Multivariate analysis was undertaken to establish demographic/audiologic factors that predict time to cochlear implantation.

Results:

Using multivariate analysis, non-white race (HR 0.157, p=0.038) and increased age (HR 0.970, p=0.038) were associated with increased time to cochlear implantation. Non-white patients had significantly higher pure-tone averages and lower speech recognition scores (CNC words and AzBio sentences in quiet) and were less likely to use hearing aids as compared to white patients (all p<0.001). Sex (p=0.188), health insurance type (p=0.255), pre-operative hearing aid use (p=0.174), and audiologic outcomes were not significant predictors of time to implantation.

Conclusion:

Non-white patients have poorer preoperative hearing and speech recognition and lower hearing aid use and are at risk for delay in referral and treatment for severe to profound sensorineural hearing loss. Other demographic factors, notably health insurance status, did not significantly predict time to cochlear implantation. Given the observed hearing healthcare disparities, special outreach programs may be needed to ensure timely cochlear implantation and effective hearing screening and rehabilitation.

INTRODUCTION

Approximately 38 million Americans 12 years of age or over have some form of bilateral hearing loss. Of these, approximately 2 million adults have severe to profound bilateral sensorineural hearing loss and may be candidates for cochlear implantation.¹ Meta-analyses have shown consistent improvement in quality of life^2,3 and speech recognition following cochlear implantation. Multiple cost-effectiveness studies have cited implantation as having a positive benefit-to-cost ratio.^4–7 Paralleling these results and technological innovations, including the approval of electric-acoustic stimulation (EAS) and hybrid cochlear implants, the United States (US) Food and Drug Administration candidacy criteria for traditional implantation have expanded from total deafness in the 1980s to the current criteria of <50% open-set sentence recognition with properly fitted hearing aids, and 10–60% CNC aided word recognition for EAS and hybrid implantation.^6,7 Nevertheless, cochlear implantation in the US remains underutilized compared to similar industrialized nations, with studies citing approximately 6–10% usage in the candidate population.^8,9

Reasons for this underutilization are unclear and may include limited knowledge of candidacy criteria by primary care providers, poor patient understanding of expected cochlear implant outcomes, long travel distances to cochlear implant centers, and/or financial burdens related to device and other implantation-related costs. Disparity in reimbursement for implantation is a commonly cited concern, with reports of hospital reimbursement by Medicaid as low as 10% of device cost, and hospitals reporting losses of up to $10,000 per implant with Medicaid coverage.^8–10 As such, demographic and socioeconomic factors can impact hearing healthcare access and the ability to provide timely cochlear implantation.

Delay in implantation not only prolongs patient disability but also may predict poorer outcomes. Early intervention with cochlear implants is one of the few consistent predictors of improvements in speech recognition following implantation.^11–14 Thus, delays in cochlear implantation may represent a modifiable risk factor for cochlear implant candidates. The aims of the current study are to identify demographic and audiologic factors that may be associated with increased time before cochlear implantation. We hypothesize that demographic factors such as age, sex, and race; audiologic factors, such as magnitude of hearing loss; and socioeconomic factors, such as health insurance status, predict longer delays in cochlear implantation.

MATERIALS AND METHODS

Patients

The present study was approved by our Institutional Review Board. A retrospective review of a prospectively maintained database was performed for adult patients undergoing unilateral cochlear implantation for bilateral sensorineural hearing loss between January 2012 and August 2017. Evaluation for candidacy and surgery was performed at a tertiary university-based otology practice. Inclusion criteria included documented history of post-lingual onset of hearing loss and age ≥18 years. Exclusion criteria were initial cochlear implant surgery at another institution (thus, history prior to cochlear implantation was not available), incomplete audiometric data, revision cochlear implantation, second-sided cochlear implantation, and implantation for single-sided deafness. Table 1 includes the demographics of the 492 patients who met inclusion/exclusion criteria.

Table 1:

Patient Demographic Data

	All	White	Non-White†	p value
N	492	391	101
Age in years (SD)	61.2 (18.1)	63.9 (17.1)	50.7 (18.2)	<0.001
Sex				0.014
Male (%)	260 (52.8)	218 (55.8)	42 (41.6)
Female (%)	232 (47.2)	173 (44.3)	59 (58.4)
Health Insurance				0.091
Private (%)	152 (30.9)	128 (32.7)	24 (23.8)
Public*(%)	340 (69.1)	263 (67.3)	77 (76.2)

^*Public insurance includes Medicare, Medicaid, VA, or Tricare plans

^†Non-White cohort includes African American patients (89.1%), and Hispanic, Asian, and Native American Patients

Bold indicates significant relationships

SD-standard deviation

Data Acquisition

The following data were extracted from our adult cochlear implant database: age at implantation, sex, race/ethnicity, history of hearing aid use, health insurance provider, time to implantation, and pre-operative audiometric data. Audiometric data were obtained for the best aided ear and included aided pure-tone average (PTA) using thresholds at 500, 1000, and 2000 Hz, speech recognition threshold (SRT), and best aided speech recognition scores using Consonant-Nucleus-Consonant (CNC) Words, AzBio sentences in quiet, and AzBio sentences in noise at a +10 dB signal-to-noise ratio (SNR) (AzBio +10).^15–17 AzBio +10 was used for patients obtaining ≥50% AzBio quiet scores. Pre-implantation speech recognition was measured with hearing aids (personal or stock aids) fitted to NAL-RL targets.¹⁸ All speech recognition testing was performed in a sound-treated room in the sound field with speech presented at 60 dB SPL (0 degrees azimuth).

Time to implantation was defined by self-report as the number of years with hearing loss prior to implantation. Pre-operative hearing aid use was defined as the patient’s self-reported active hearing aid use at the time of the cochlear implant evaluation (yes/no). Race/ethnicity was defined as white and non-white, with white including non-Hispanic white patients. Due to small numbers of patients in certain racial/ethnic groups, non-white included African-American, Hispanic, Asian, and Native American patients (89.1% of patients in the non-white group were African American). For insurance status, patients were grouped into two insurance provider categories, private and public. Private insurance included patients enrolled in any privately held or employee-acquired healthcare plan or pension. Public insurance included patients enrolled in Medicaid, Medicare, VA, or Tricare plans.

Data Analysis

Analyses were performed with SPSS 25.0 (IBMCorp., Armonk, NY). Continuous variables were tested for normal distribution as determined by the Kolmogorov-Smirnov test. Nominal variables (sex, racial group, hearing aid use, insurance) were summarized by frequency, percentage, and/or range. Continuous variables (age at implantation, time to implantation, audiometric outcomes) were summarized by mean (standard deviation; SD) where appropriate. Analysis between racial groups was performed using Chi-square analysis or Fisher exact test for nominal data and a student’s t test or Mann-Whitney U test for quantitative data.

A Cox regression was performed for multivariate analysis of all demographic and audiometric covariates to determine the independent relationship of the variable with time to implantation, while controlling for possible cofounding effects. For nominal covariates the largest cohort was designated as the reference category. A p-value of ≤0.05 was used as the measure for statistical significance for all statistical tests.

RESULTS

Patient Demographics

Demographics of the 492 patients included in this study are included in Table 1. Mean age at implantation was 61.2 years ± 18.1 years; 52.8% of patients were male. Most patients were non-Hispanic white (79.5%) or African American (18.3%). Hispanic (1.0%), Asian (1.0%), and Native American (0.2%) patients made of the remainder of the study population. A majority of patients were enrolled in some form of public health insurance (69.1%). Mean time with hearing loss before implantation was 24.2 ± 17.2 years and most patients (61.8%) were using hearing aids at the time of the cochlear implant evaluation. Audiological data are detailed in Table 2.

Table 2:

Patient Audiologic Data

	All Patients	White	Non-White†	p value
Hearing Aid Use				<0.001
Yes (%)	304 (61.8)	264 (67.5)	40 (39.6)
No (%)	179 (36.4)	121 (31.0)	58 (57.4)
Unknown (%)	9 (1.8)	6 (1.5)	3 (3.0)
Audiometric Outcomes(Aided)
PTA-dB HL (SD)	39.0 (15.7)	36.3 (11.1)	41.9 (17.3)	0.007
SRT- dB HL (SD)	35.8 (10.5)	35.7 (10.5)	36.2 (11.0)	0.728
CNC-W - %Correct (SD)	17.4 (20.1)	19.3 (21.0)	9.4 (13.2)	<0.001
AzBio Quiet - %Correct (SD)	20.3 (23.2)	22.9 (24.2)	11.4 (16.4)	<0.001
AzBio Quiet - %Correct for those undergoing AzBio +10 testing (SD)	61.6 (22.4)	61.5 (22.9)	63.5 (6.4)	0.756
AzBio +10 - %Correct (SD)	35.1 (21.1)	35.0 (21.5)	38.2 (9.3)	0.511

^†Non-White cohort includes African American patients (89.1%), and Hispanic, Asian, and Native American Patients

Bold indicates significant relationships

SD-standard deviation, PTA-pure tone threshold, SRT-speech reception threshold, CNCw-consonant-nucleus-consonant words

Multivariate Analysis of Demographic and Audiometric Covariates

In order to assess the effects of multiple covariates on time before implantation, a Cox regression was performed and results are included in Table 3. Hazard ratio (HR) denotes a relative incidence of implantation at any point in time as modified by each covariate. A HR less than one represents a decreased incidence at that point in time, and, in the case of this study, a delay in cochlear implantation. The lower the HR, the longer the delay in implantation. While controlling for effects of all other variables, non-white race was associated with a delay in cochlear implantation (HR=0.157 (0.027–0.904), p=0.038), meaning that white patients were approximately 6 times more likely to undergo implantation during each year of hearing loss as compared to non-white patients. Increased age was also significantly associated with a delay in implantation (HR 0.970 (0.942–0.998), p=0.038); although statistically significant, HRs near 1 represent very minor associations. That is, with each one-year increase in age, the likelihood of obtaining a cochlear implant increased by only 3% over the study timeline. The remaining demographic and audiometric covariates were not significantly associated with time to cochlear implantation (all p>0.05). Of note, several large HR ranges were observed, particularly in the demographic variables. This variance likely reflects differences in the size of the categorical variable groups. This is contrasted to HR intervals for continuous variables such as age or audiologic outcomes, which can be compared among the entire sample without concern for intragroup variance.

Table 3:

Demographic and Audiological Factors Influencing Time to Cochlear Implantation as Indicated by Hazard Ratios

Covariate	Hazard Ratio (95% Confidence Interval)	p-value
Demographics
Age	0.97 (0.94–0.99)	0.038
Race (Non-White/White)	0.16 (0.03–0.90)	0.038
Sex (Female/Male)	0.59 (0.27–1.30)	0.188
Health Insurance Category (Public/Private)	1.80 (0.77–4.18)	0.255
Audiological Data
Hearing Aid Use (Use/No Use)	1.60 (0.71–3.57)	0.174
PTA	0.96 (0.90–1.04)	0.329
SRT	0.99 (0.94–1.04)	0.756
CNC-W	0.99 (0.96–1.01)	0.270
AzBio Quiet	1.00 (0.97–1.02)	0.688
AzBio +10	1.00 (0.98–1.02)	0.782

Bold indicates significant Hazard Ratios

PTA-pure tone threshold, SRT-speech reception threshold, CNCw-consonant-nucleus-consonant words

Additional analyses were performed to explore further racial/ethnic differences that may influence time to cochlear implantation (Tables 1 and 2). On average, non-white patients were younger than white patients (50.7 ± 18.2 vs 63.9 ± 17.1; p<0.001) and had significantly higher PTAs (41.9 ± 17.3 vs 36.3 ± 11.1, p=0.007) and lower word recognition scores in quiet (CNC: 9.4 ± 13.2 vs. 19.3 ± 21.0; AzBio: 11.4 ± 16.4 vs 22.9 ± 24.2; all p<0.001). Only patients with AzBio in quiet scores greater than 50% underwent AzBio +10 testing. As such, these borderline patients were fewer in number had better hearing scores in a much narrower range. Although their hearing was poorer, non-white patients reported a lower percentage of hearing aid use at the time of implantation evaluation than white patients (39.6 vs 67.5; p<0.001). Unadjusted years to implantation was shorter in non-white patients (17.3 (15.9) vs 26.0 (17.4); p<0.001); when adjusted using multivariate statistical methods for age at implantation, audiologic outcomes, and hearing aid usage, time to implantation was longer in non-white than white patients. No other significant differences were noted between the two racial/ethnic groups.

DISCUSSION

Cochlear implantation is a safe and effective treatment for moderate to profound sensorineural hearing loss in adults, although more inclusive candidacy criteria has not increased utilization in the US.^8,9 Possible reasons include centralization of cochlear implant centers in urban areas leading to long travel distances for some patients, poor primary care awareness of candidacy criteria, and inadequate insurance reimbursement, but little evidence is available to explain low utilization or access.^8,9 To better understand hearing healthcare access and timely utilization by adult cochlear implant users, we assessed factors that may contribute to time with sensorineural hearing loss prior to cochlear implantation. We found a significant independent association between non-white race and increased time to implantation.

The relationship of patient race to time to implantation is likely multifactorial and may be related to availability of hearing and healthcare services in minority communities and general socioeconomic considerations.^19,20 A 2002 Institute of Medicine report showed that minority populations received substantially lower rates of routine care and procedures than white populations.²¹ Similarly, rates of aural rehabilitation in minority populations lag behind that of majority populations, with roughly 5% of African Americans with severe hearing loss pursuing treatment compared to 40% of white patients with similar hearing loss.^22,23 Such underutilization of hearing health care in minority populations may explain lower rates of hearing aid utilization in non-white patients in the current study. If minority patients bypass hearing aids, they may seek cochlear implantation at earlier ages, which may explain their earlier age at implantation. However, further research is needed to confirm this association. Studies have also shown some circumstantial evidence that hearing aid use before cochlear implantation may slow degradation of central auditory pathways. As such, lower rates of hearing aid prior to cochlear implantation use may partially explain poorer audiologic outcomes in non-white patients.^11,12

Commonly considered causes for discrepancies in hearing health care utilization are lack of routine hearing healthcare and education in minority communities, economic factors, and lack of minority physicians and providers.^19–21 Geographic considerations may also act as barriers to timely cochlear implantation. Hixon et al.²⁴ previously reported that rural cochlear implant recipients in Kentucky had lower income levels, were largely covered by public insurance, and had significantly longer commutes to cochlear implantation centers. In addition, this population had significantly longer times before implantation.

Another frequently considered barrier to cochlear implantation is public versus private health insurance. A common assumption is that public insurance presents a barrier to implantation through both lower physician compensation and more stringent candidacy criteria.^8–10,20,25 Such barriers may exacerbate racial disparities in healthcare accessibility. In the current study, private/public insurance proportions were not significantly different for white and non-white patients (Table 1), which may not be the case nationally. Accordingly, race, and not insurance type, was significantly associated with time before cochlear implantation (Table 3). Fortnum et al.²⁶ showed affluent children in the United Kingdom were more likely to obtain cochlear implants than less affluent children, despite a social medical system with high public funding. As such, factors beyond access to a certain type of health insurance, such as personal, financial, and other resources, may contribute to timely cochlear implantation.

The results of this study demonstrate that, despite similar insurance coverage, non-white patients underutilized hearing aids and/or cochlear implants. On average, non-white patients were younger than white patients at implantation, had poorer hearing at the time of the cochlear implant candidacy evaluation, but also reported lower rates of hearing aid use. This shows a potentially disturbing trend toward overall underutilization of or lack of access to hearing healthcare by non-white patients. Effective screening, education, and more rapid intervention in minority communities are needed to achieve timely cochlear implantation.

Our study also showed a very small but significant association of increasing age and increased time to cochlear implantation, with each one-year increase in age increasing by 3% the likelihood of obtaining a cochlear implant over the study timeline. Thus, although statistically significant, this result is unlikely to be of clinical significance, because a small change in age will yield only a negligible change in likelihood of implantation.

The current study failed to show any significant association between health insurance type and time to cochlear implantation. Socioeconomic status and health insurance coverage remain major issues in healthcare policy. We have insufficient information to discuss the impact of specific private insurance policies on hearing healthcare access, nor can we claim that our results reflect the overall population of adult cochlear implant candidates in the United States. In 2016, 67.5% of Americans held some form of private insurance and 37.3% were insured through government policies.²⁷ The South Carolina adult population, from which the study data was drawn, held private and government policies at 61% and 38.5%, respectively.²⁸ These figures differ from our study population, with 69.1% of patients holding government insurance policies and only 30.9% holding private insurance (Table 1). However, we note that the average age of our patient population is 61 years, close to the eligible age for Medicare of 65. Roughly 34% of Medicare beneficiaries enroll in private health insurance, which more closely matchesmay explain the ratio of private to public insurance seen in this study.²⁹ As noted earlier, these ratios did not differ by race (Table 1).

In this regard, one limitation of the current study is the single university–based sample, which may limit the generalizability of the results. Including community and private practices and a larger sample in future studies address these limitations. Another limitation is that 89.1% of our non-white patients were African-American with numbers of other non-white races too small to provide sufficient statistical power for sub-analyses. A larger and more diverse sample is necessary to analyze the independent impact of other non-white patients on time to implantation. An additional limitation is the reliance on self-report estimates of patients’ time to implantation, which were used due to the absence of serial audiograms. Nevertheless, patient-reported estimates of onset of hearing loss is commonly used and often reported in the literature. ^15,30,31

CONCLUSION

Cochlear implantation is an effective therapeutic option for patients with moderate to profound sensorineural hearing loss, but utilization remains low. In the setting of a tertiary university-based otology practice, results from the current study suggest that non-white patients may be at risk for delay in referral for cochlear implantation, leading to increased time without treatment. Additional research is needed to investigate these findings in larger sample sizes and in community and private practice settings to determine the generalizability of these results.