Baseline Function Predicts Gains in Both Speech Recognition and Quality of Life After Cochlear Implantation

Abstract

Objective:

To characterize patterns of improvement in speech recognition and quality of life (QOL) after cochlear implantation and examine how baseline performance, device use, and candidacy status (“borderline,” qualifying in noise only vs. “traditional,” qualifying in quiet) predict postoperative gains in both domains.

Study design:

Retrospective cohort study.

Setting:

Tertiary referral center.

Patients:

One hundred one adult unilateral CI recipients implanted between 2019 and 2023 with complete preoperative and postoperative (6 or 12 mo) speech recognition and QOL data.

Intervention(s):

Unilateral cochlear implantation.

Main outcome measure(s):

AzBio sentence recognition in quiet (AzBioQ) and Cochlear Implant Quality of Life (CIQOL) Global scores. Clinically meaningful improvement was defined using validated thresholds for each measure.

Results:

Of 101 patients, 97 demonstrated clinically meaningful improvement in at least one domain: 44 improved in both speech recognition and QOL (dual improvers), 43 in speech recognition only, 10 in QOL-only, and 4 showed no significant improvements. Patients with lower baseline scores in a given domain demonstrated greater gains in that domain. Greater daily device use (“datalogging”) was associated with larger speech-recognition gains (r=0.34, P<0.01) but not QOL improvements (r=0.12, P=0.24). Borderline candidates (n=24) achieved QOL gains comparable to traditional candidates (n=77) despite smaller speech-recognition improvements.

Conclusions:

The vast majority of CI recipients demonstrate meaningful improvements in speech recognition, QOL, or both domains. Baseline performance predicts domain-specific gains, and borderline candidates achieve meaningful QOL benefits despite smaller speech-recognition gains. These findings provide further support for incorporating patient-reported outcome measures into candidacy evaluations and postoperative assessments.

Introduction

Cochlear implants (CIs) provide access to sound for individuals with moderate-to-profound sensorineural hearing loss and typically result in significant improvements in outcome domains of both speech recognition and quality of life (QOL)^1–3. However, outcomes vary widely across patients, and the association between speech-recognition performance and hearing-related QOL using patient-reported outcome measures (PROMs) is inconsistent^2,4,5. Nevertheless, CI candidacy criteria and assessments of outcomes continue to rely heavily on speech-recognition abilities, which may not adequately represent everyday communication difficulties. A growing emphasis on PROMs aligns with evolving FDA guidance and clinical trial design standards, where PROMs are recognized as primary endpoints to capture real-world impact of interventions⁶.

Although CI candidacy is still frequently determined using speech recognition in quiet (eg, CNC words, AzBio sentences), many CI centers now incorporate testing in noise to better reflect real-world listening challenges⁷. Some patients demonstrate borderline candidacy performance in quiet—such as CNC word recognition around 50% to 60%—yet meet candidacy criteria when tested using sentences in noise^7–9. These “borderline” candidates who only qualify in noise fall into a gray zone, and there is less evidence describing their postoperative CI benefits compared with “traditional” CI candidates who qualify in quiet. In one study, Schauwecker et al.¹⁰ reported that there was a significant difference in the proportions of borderline versus traditional CI candidates who demonstrated improvements in speech recognition in the CI-only testing condition, but not in their bilateral best-aided (“everyday listening”) condition. Although the findings of that study suggest that everyday listening speech-recognition benefits are similar between groups, the degree to which borderline CI candidates improve in their perceived real-world communication abilities and QOL after implantation remains unclear. This topic is clinically important, as demonstrated by a large multicenter study using a registry (n=1611) with data from 12 clinical sites across 9 states¹¹. That study, which is now almost a decade old, showed that even then up to one-third of adult CI candidates qualified only in noise (10% at +10 dB signal-to-noise ratio, SNR; 27% at +5 dB SNR). Although that study found overall postoperative improvements in speech recognition and self-reported outcomes, approximately one-quarter of patients did not demonstrate meaningful gains. Direct comparison between noise-qualified and quiet-qualified (borderline vs. traditional) candidates was not performed, underscoring the need to clarify how these groups differ in real-world CI benefits.

Unlike speech-recognition measures in the booth, PROMs capture broader domains of communication, engagement, and well-being, offering a more patient-centered assessment of benefits. However, patient-reported QOL outcomes in borderline CI candidates remain poorly characterized. For example, Schauwecker et al.¹⁰ demonstrated that borderline CI candidates achieved smaller CI-only gains but equivalent everyday listening speech recognition, and the groups showed comparable final scores on patient-reported QOL. However, that study did not consider the impact of baseline QOL on post-CI QOL changes. Thus, one objective of this study is to investigate whether QOL changes differ between borderline and traditional CI recipients, using the Cochlear Implant Quality of Life (CIQOL), a validated CI-specific PROM.

Prior studies typically evaluated speech recognition and QOL independently, yet real-world benefits may be better captured by examining patterns of improvement across both domains. PROMs quantify perceived hearing-related functioning, listening effort, and social participation, which together complement behavioral test performance. By evaluating both domains concurrently, clinicians can identify whether improvements in speech recognition translate into meaningful gains in daily communication and QOL, or whether patients benefit in one domain but not the other¹². A second objective is to characterize patterns of improvement: dual improvement, improvement in speech recognition only, improvement in QOL-only, or lack of improvement in either domain. Stratifying patients based on clinically meaningful changes from pre-CI to post-CI, using the minimal detectable change for the CIQOL Global and test-retest reliability for AzBio, may provide more comprehensive guidance for clinicians and patients^3,13,14. In addition to preoperative performance measures, postoperative listening behavior (captured through device datalogging) offers another dimension of outcome assessment. Greater CI daily use has been associated with better speech recognition and auditory acclimatization, yet its relationship with QOL remains less clear¹⁵. Evaluating how datalogging relates to both speech recognition and patient-reported outcomes may provide a more comprehensive understanding of real-world CI benefit. Prior work has demonstrated baseline-dependent speech-recognition improvements and associations between device use and speech outcomes, but whether these patterns extend to QOL remains unclear^10,14,16.

The primary aim of this study was to determine whether borderline CI candidates—those who qualify for implantation based on testing in noise but with relatively preserved performance in quiet—achieve QOL gains comparable to traditional candidates, despite potentially smaller speech-recognition improvements. Secondary aims were to validate and expand upon expected patterns within our cohort: specifically, that baseline performance predicts domain-specific gains and that device use relates to postoperative outcomes in both speech recognition and QOL domains.

Methods

Study design and participants

After local Institutional Review Board (IRB: 240876) approval, a retrospective chart review identified adults who underwent cochlear implantation from 2019 to 2023. The cohort study included 101 adult CI recipients evaluated and implanted at a tertiary referral CI center. All included participants had complete speech recognition and CIQOL Global data (defined below) both preoperatively and at either 6 or 12 months post-CI. Patients without complete outcome data at these timepoints were excluded. Postoperative data were analyzed from the most recent follow-up at 6 or 12 months, as outcomes typically plateau by 6 months after implantation^2,17. Of the 101 patients, 41 (40.6%) were evaluated at 6 months and 60 (59.4%) at 12 months post-implantation. Inclusion criteria were age older than or equal to 18 years, postlingual onset of deafness, and having undergone unilateral cochlear implantation. Revision cases, patients with auditory nerve tumors (eg, acoustic neuromas), and patients with single-sided deafness (SSD) were excluded.

Data cleaning and derived variables

To construct the final analytical cohort, we only included those with complete preoperative AzBio sentence recognition in quiet (AzBioQ) and CIQOL Global data. Two final outcome variables were created: Final AzBioQ (the most recent postoperative AzBioQ score at 6 or 12 mo post-CI) and Final CIQOL Global outcome score (converted from the most recent raw CIQOL Global score at 6 or 12 mo post-CI)³. Pre-CI to post-CI change scores (ΔAzBioQ, ΔCIQOL) were also computed, and clinical significance of change was determined using validated thresholds (see below). Patients were then characterized for speech-recognition improvement or QOL improvement using binary indicators (yes or no for each outcome domain). This classification was based on established clinically meaningful thresholds, allowing outcomes to be interpreted in terms of whether a patient experienced a meaningful benefit¹³.

Outcome measures

Speech recognition was assessed using AzBio sentences presented in sound field in quiet (AzBioQ) at 60 dB SPL and in multi-talker babble noise (AzBioN) with speech at 65 dB SPL and noise at 60 dB SPL (ie, +5 dB SNR), with the speaker at zero degrees azimuth. Testing was performed in the to-be-implanted ear, best-aided pre-CI, and in CI-alone condition after implantation⁷.

Hearing-related QOL was measured using the CIQOL-10 Global (which is a subset of 10 items from the CIQOL-35 Profile, and referred to as CIQOL Global hereafter), a validated PROM with raw scores ranging from 10 to 50, which are then converted to standardized outcome measure scores ranging from 0 to 100. Pre-CI and post-CI outcome measure scores were used to calculate change (ΔCIQOL). The CIQOL Global captures facets of communication, such as listening effort and social participation, and can be used both preoperatively to establish baseline performance and postoperatively to characterize CI-related improvements in QOL¹⁶. The CIQOL Global has also been used in clinical settings to set expectations and guide counseling, complementing speech-recognition scores².

Clinically meaningful improvement in CIQOL Global was defined based on the minimal detectable change reported by McRackan et al.¹³ Significant improvement in speech recognition, interpreted as clinically meaningful improvement, was determined based on a change greater than the test-retest reliability of 95% CIs, as described by Thornton and Raffin and more recently employed in a similar study by Dornhoffer and colleagues^18–20.

Group classification

Patients were categorized into 4 outcome groups based on whether they demonstrated clinically meaningful gains in speech recognition and/or QOL. “Dual improvers” improved in both AzBioQ and CIQOL Global. “Speech-only improvers” improved in AzBioQ only. “QOL-only improvers” improved in CIQOL Global only. “Non-improvers” demonstrated no significant improvement in either domain.

A subgroup of 24 patients who qualified for implantation based on AzBio sentence testing in noise (AzBioN) at +5 dB SNR (<40% in the CI ear) but scored ≥40% in AzBioQ were classified as borderline candidates. Their outcomes were compared with the 77 patients who met traditional candidacy criteria based on <40% AzBioQ preoperative performance in the implanted ear. These criteria may be more conservative than contemporary practice: in 2022, the Centers for Medicare and Medicaid Services (CMS) revised its National Coverage Determination to expand implantation criteria from <40% to <60% sentence recognition scores, with broad clinical adoption occurring over 2023-2024^21,22. Nonetheless, these criteria were used to characterize our borderline candidates in this study based on inclusion of patients from our center implanted between December 2019 and October 2023.

Predictor variables

Demographic characteristics (age, sex), hearing history (duration of deafness, defined as patient-reported years without benefit from traditional amplification), and device use (average daily hours from device datalogging) were extracted from clinical records and programming software. Device use was collected from the timepoint (either 6 or 12 mo post-CI) at which post-CI speech recognition and QOL data were taken.

Statistical analysis

One-way ANOVA was used to compare continuous variables such as age, duration of deafness, daily device use, and baseline AzBioQ and CIQOL Global scores across the 4 outcome groups (dual improvers, speech-only improvers, QOL-only improvers, and non-improvers). When overall group differences were significant, post hoc comparisons were conducted using Tukey honestly significant difference (HSD) test. Bivariate Pearson correlation coefficients were calculated to assess unadjusted associations between continuous predictors (eg, daily device use, age) and change scores (ΔAzBio, ΔCIQOL). Independent-samples t tests were used for comparisons in gains in speech recognition and QOL between borderline and non-borderline candidates. All statistical tests were 2-tailed with significance set at α=0.05.

To further examine these relationships while adjusting for multiple predictors simultaneously, separate multivariable linear regression with ordinary least squares estimation analyses were performed with ΔAzBioQ and ΔCIQOL as dependent variables. Predictors included baseline AzBioQ, baseline CIQOL Global, age at implantation, and duration of deafness. All variables were entered simultaneously using the enter method. Analyses were conducted using SPSS version 31 (IBM Corp, Armonk, NY).

Results

Participant characteristics

The final cohort included 101 adult CI recipients (mean age=64.3 y, SD=14.9), with 65 males and 36 females. The mean duration of deafness was 5.4 years (SD=8.0), and average daily device use was 11.2 hours (SD=4.1). Of the 101 patients, 41 (40.6%) were evaluated at 6 months and 60 (59.4%) at 12 months post-implantation. For the subset of 37 patients (36.6%) with CIQOL Global data at both timepoints, scores remained stable between 6 and 12 months (6-mo: 52.0±6.8; 12-moh: 49.1±6.7; mean difference: 2.9±4.9 points, P=0.051). Mean AzBioQ sentence recognition improved from 22% (SD=24.8) preoperatively to 68% (SD=25.2) postoperatively in the CI ear. Mean CIQOL Global scores improved from 39.4 (SD=7.3) to 50.1 (SD=9.7). Details regarding the participants can be found in Table 1. For the subset of patients with available postoperative AzBioN data (n=67), mean AzBioN sentence recognition in the CI ear improved from 6.8% (SD=12.8) to 49.0% (SD=28.6).

Table 1.

Participant characteristics

Characteristics
n	101
Age, mean (SD) [range], y	64.3 (14.9) [20.0-88.0]
Duration of deafness, mean (SD) [range], y	5.4 (8.0) [0-50]
Daily device use, mean (SD) [range], h	11.2 (4.1) [0.1-17.3]
Sex, n (%)	Male: 65 (64), female: 36 (36)
Race, n (%)	White: 98 (97), Black: 2 (2), Hispanic: 1 (1)
Hearing loss etiology, n (%)	Idiopathic: 25 (25), history of noise exposure: 22 (22), genetic/family history: 12 (12), Ménière disease: 8 (8), infection (meningitis/labyrinthitis): 6 (6), otosclerosis: 2 (2), trauma: 4 (4), other (medication, radiation, congenital, etc.): 5 (5)
Laterality, n (%)	Left: 63 (62), right: 38 (38)
Manufacturer, n (%)	Cochlear: 69 (68), MED-EL: 19 (19), Advanced Bionics: 13 (13)

Patients could endorse more than one hearing loss etiology. Therefore, the sum of counts for each row in the table may exceed the total number of patients.

Outcome classification

Of the 101 patients with complete data for both AzBioQ and CIQOL Global, 97 (96.0%) demonstrated clinically meaningful improvement in at least one outcome measure. Patients were classified into the 4 outcome groups as described above: dual improvers (n=44, 43.6%) who improved in both AzBioQ and CIQOL Global, speech-only improvers (n=43, 42.6%) who improved in AzBioQ only, QOL-only improvers (n=10, 9.9%) who improved in CIQOL Global only, and non-improvers (n=4, 4.0%) who showed no significant improvement in either domain. Data for these 4 groups can be seen in Figures 1 and 2.

Figure 1.

Preoperative (A) and postoperative (B) AzBio sentence recognition in quiet (AzBioQ) scores across outcome groups, stratified by candidacy status. Boxplots display distributions of AzBioQ scores at preoperative (A) and postoperative (B) timepoints for each outcome group: dual Improvers (n=44), speech-only improvers (n=43), quality of life (QOL)-only improvers (n=10), and non-improvers (n=4). Within each outcome group, scores are stratified by candidacy status (borderline candidates, n=24; traditional candidates, n=77). Boxes represent the interquartile range (IQR), horizontal lines indicate the median, whiskers extend to the minimum and maximum values within 1.5 times the IQR, and points represent outliers.

Figure 2.

Preoperative (A) and postoperative (B) Cochlear Implant Quality of Life (CIQOL) Global scores across outcome groups, stratified by candidacy status. Boxplots display distributions of CIQOL Global scores at preoperative (A) and postoperative (B) timepoints for each outcome group: dual Improvers (n=44), speech-only improvers (n=43), QOL-only improvers (n=10), and non-improvers (n=4). Within each outcome group, scores are stratified by candidacy status (borderline candidates, n=24; traditional candidates, n=77). Boxes represent the interquartile range (IQR), horizontal lines indicate the median, whiskers extend to the minimum and maximum values within 1.5 times the IQR, and points represent outliers.

Baseline predictors of improvement

We next examined baseline predictors of improvement among the 101 patients. Analysis using 1-way ANOVA revealed that baseline AzBioQ scores differed significantly across outcome groups, F(3, 97)=10.37, P<0.001, η²=0.24. Post hoc testing demonstrated that non-improvers had significantly higher baseline AzBioQ, averaging 47.7 percentage points above speech-only improvers (P<0.001) and 38.7 points above dual improvers (P=0.006). QOL-only improvers also demonstrated higher baseline AzBioQ than dual improvers by 23.4 points (P=0.016). No significant differences were detected in baseline AzBioQ between non-improvers and QOL-only improvers (P=0.614) or between speech-only and dual improvers (P=0.226).

Baseline CIQOL Global differences among the 4 outcome groups were smaller than those of their AzBioQ scores but remained statistically significant, F(3, 97)=3.40, P=0.021, η²=0.10. Post hoc testing showed that dual improvers had lower baseline CIQOL Global than speech-only improvers (P=0.057), although this did not reach statistical significance. No other pairwise comparisons were significant (p ≥0.186).

Multivariable linear regression analyses demonstrated that lower baseline performance predicted greater postoperative gains within and across outcome domains. In the model predicting change in speech recognition (ΔAzBioQ), lower baseline AzBioQ significantly predicted greater postoperative improvement (β=–0.58, P <0.001). Lower baseline CIQOL Global showed a nonsignificant trend toward greater improvement (β=−0.18, P=0.066). Age at implantation and duration of deafness were not significant predictors (P=0.613 and 0.454, respectively). The overall model accounted for ~36% of the variance in ΔAzBioQ and was statistically significant (R²=0.36, F(4, 72)=9.94, P <0.001).

In the multivariable linear regression model predicting change in QOL (ΔCIQOL), lower baseline CIQOL Global showed a nonsignificant trend toward predicting greater improvement (β=−0.20, P=0.075). Baseline AzBioQ, age, and duration of deafness were not significant (P=0.152, 0.169, and 0.848, respectively). The overall model accounted for ~10% of the variance in ΔCIQOL but was not statistically significant (R²=0.10, F(4, 72)=2.02, P=0.10).

When considered together, these findings indicate that patients with poorer preoperative speech recognition—and to a lesser and not statistically significant extent those with lower baseline QOL—generally tend to experience the largest postoperative gains in the corresponding outcome domain, even after controlling for demographic and clinical covariates.

Correlates of improvement in AzBioQ and CIQOL

In our analysis of daily device use, we found that datalogging was positively correlated with ΔAzBioQ scores (r=0.34 [95% CI: 0.15-0.51]), indicating that greater device use was associated with larger postoperative speech-recognition gains. In contrast, datalogging was not significantly correlated with ΔCIQOL (r=0.12 [95% CI: –0.09 to 0.33]). Improvement in speech recognition and QOL were only weakly related (r=0.13 [95% CI: –0.07 to 0.32]). Neither age at implantation (r=–0.08 [95% CI –0.28 to 0.13]) nor duration of deafness (r=0.07 [95% CI: –0.15 to 0.28)) showed a reliable association with either outcome domain.

Device use did not significantly differ across the 4 outcome groups. Neither age at implantation nor duration of deafness was related to outcome group membership or magnitude of improvement, consistent with their lack of significant correlations with either ΔAzBioQ or ΔCIQOL.

Borderline versus traditional subgroup analysis

Data for the borderline versus traditional groups can be found in Figure 3. The 24 borderline patients (qualified by scoring <40% AzBioN but ≥40% AzBioQ) demonstrated comparable QOL gains to the 77 traditional patients. That is, ΔCIQOL did not significantly differ between groups, t(99)=–0.77, P=0.44, d=–0.18 (95% CI: –0.64-0.28). In contrast, speech-recognition improvement was markedly smaller in the borderline subgroup, t(99)=5.12, P<0.001, d=1.19 (95% CI: 0.71-1.68). In summary, despite achieving smaller gains in AzBio sentence recognition, borderline patients experienced QOL improvements comparable to those of traditional candidates.

Figure 3.

Change in AzBio sentence recognition in quiet (AzBioQ) and Cochlear Implant Quality of Life (CIQOL) Global scores for borderline versus traditional candidates. Boxplots display distributions of change scores (postoperative – preoperative) in AzBioQ and CIQOL Global for borderline (n=24) and traditional (n=77) candidates. Boxes represent the interquartile range (IQR), horizontal lines indicate the median, whiskers extend to the minimum and maximum values within 1.5 times the IQR, and points represent outliers.

Discussion

In this cohort of 101 adult CI recipients, 96% demonstrated improvement in at least one outcome domain, with over 40% demonstrating gains in both speech recognition and QOL. Regarding our primary aim, borderline candidates achieved QOL gains comparable in magnitude to traditional candidates despite smaller speech-recognition gains. Our secondary aims were also supported: patients starting with lower baseline scores in a given domain showed the greatest gains, with non-improvers having the highest preoperative AzBioQ and dual improvers having the lowest baseline CIQOL Global. In addition, greater daily device use through datalogging was associated with greater gains in speech recognition but not with greater gains in QOL.

These findings align with prior reports that speech recognition and QOL outcomes are only weakly correlated⁵. Supporting this dissociation, 43% of patients demonstrated significant improvement in speech recognition only, whereas 10% showed the opposite pattern. This means that sentence recognition tests alone may underestimate the benefits many patients experience in daily life. Our results suggest a practical counseling framework: patients with relatively high preoperative AzBioQ but low CIQOL Global may still realize meaningful benefit, whereas those presenting with both high baseline scores may be less likely to demonstrate measurable improvement, warranting careful counseling and expectations setting.

The magnitude of QOL improvements observed in this study, while appearing modest on an absolute scale, warrants careful interpretation. These modest gains reflect the CIQOL Global’s design to capture complex listening situations and psychosocial domains that improve more gradually than speech recognition. In addition, QOL is influenced by multiple factors beyond auditory performance, as evidenced by the weak correlation (r=0.13) between speech recognition and QOL improvements.

Notably, QOL improvements did not demonstrate the same robust baseline-dependent pattern observed for speech recognition. Although patients with poorer preoperative AzBioQ showed larger speech-recognition gains, baseline speech recognition was not a significant predictor of QOL change. As shown in Supplementary Table 1, Supplemental Digital Content 1, http://links.lww.com/MAO/C470, patients across all baseline AzBioQ performance categories—from profound hearing loss (≤10%) to expanded criteria (>60%)—demonstrated similar magnitudes of CIQOL improvement (range: 8.8 to 9.6 points), despite markedly different baseline speech-recognition abilities.

This dissociation reflects several important factors. First, our data do not demonstrate a ceiling effect in CIQOL scores, as evidenced by the broad distribution of postoperative scores (range: 23.2 to 69.5, mean: 50.1±9.7)^2,3. Second, although our cohort’s mean preoperative CIQOL score of 39.4 leaves substantial room for improvement on the 0 to 100 scale, postoperative scores reached only 50.1 on average, suggesting that substantial improvements in auditory function may not translate proportionally into QOL gains due to persistent challenges in complex listening environments, ongoing listening effort, or non-auditory psychosocial factors. Third, patients with severe-to-profound hearing loss may have developed compensatory strategies and social adaptations over time, resulting in relatively good baseline QOL despite poor speech recognition. These findings suggest that predicting individual QOL outcomes remains more challenging than predicting speech-recognition improvements.

Our findings build on the recent work of Schauwecker et al.¹⁰, who reported that although borderline candidates achieved smaller speech-recognition gains, they attained similar bilateral best-aided speech recognition compared with traditional candidates. Our study extends these findings by demonstrating equivalent QOL improvements in borderline candidates, underscoring the importance of PROMs in capturing benefit for groups whose candidacy is less certain based on traditional metrics.

Greater datalogging was associated with larger speech-recognition gains, but this relationship should be interpreted cautiously given the observational design and possible bidirectional association. In contrast, QOL outcomes were not associated with device use time, suggesting these outcomes may be shaped more by psychosocial factors than auditory input alone. Incorporating broader psychological and social measures (eg, resilience, optimism, social support) into future assessments may clarify this variability. Our findings have implications for preoperative counseling. Clinicians should consider both patients’ baseline speech-recognition scores and PROM scores for more personalized outcome expectations. Patients with high baseline performance in one domain but not the other may be counseled that improvements might occur selectively rather than uniformly across both domains.

This study has several limitations. First, the retrospective single-center design limits generalizability. Of approximately 412 postlingually deafened adult patients receiving a unilateral CI who were implanted during the study period, excluding those with retrocochlear pathology or known cognitive impairments, only 101 (41.7%) had complete speech recognition and CIQOL Global data at baseline and follow-up. Incomplete data, particularly missing CIQOL Global assessments (as systematic PROM administration was implemented gradually during the study period), missed follow-up appointments, and limited cohort size. Patients with complete data may represent a more engaged population with better follow-up, potentially introducing selection bias. Second, postoperative outcomes were assessed at either 6 months (n=41) or 12 months (n=60). Although prior literature suggests outcomes typically plateau by 6 months, some patients may continue to improve beyond this timepoint, particularly in speech-recognition performance. Third, postoperative speech recognition in noise was available for only 67 patients and we did not analyze bilateral/bimodal listening conditions, which may provide complementary information about real-world benefit. The CIQOL Global partially addresses this, but future studies should incorporate systematic testing in real-world conditions. Fourth, the modest borderline sample (n=24) may limit power. Fifth, CMS revised candidacy criteria in 2022 from ≤40% to ≤60% sentence recognition²². Our borderline classification reflects standards during the study period and may be more conservative than contemporary practice.

In conclusion, 96% of CI recipients demonstrated meaningful improvement in at least one domain, although magnitude varies with baseline performance. Borderline candidates achieved comparable QOL benefits despite smaller speech-recognition gains, supporting the inclusion of PROMs in both candidacy evaluation and outcome assessment. Future research should validate these findings in larger multicenter cohorts and investigate how psychosocial predictors shape long-term benefits.

Sources of support

This study was supported by NIDCD R01DC019088 to A.C.M., NIDCD R21DC019382 to T.N.T., and NIA R01AG089200 to T.N.T.

Supplementary Material

mao-47-0985-s001.docx ^{(15.5KB, docx)}