External Validation of Cochlear Implant Screening Tools Demonstrates Modest Generalizability

Qualifying Cochlear Implant Candidates-Does it Matter How Patients Are Qualified?

OBJECTIVE

Evaluate variable qualification criteria for cochlear implant (CI) recipients and 12-month speech perception outcomes.

STUDY DESIGN

Retrospective cohort study.

SETTING

HERMES national database and nonoverlapping single-institution CI database.

PATIENTS

A total of 2,124 adult unilateral CI recipients categorized by qualifying status: AzBio in quiet (n = 1,239), +10 dB SNR (but not in quiet; n = 519), +5 dB SNR (but not in quiet or +10 dB SNR; n = 366); CNC ≤40% (n = 1,037), CNC 41% to 50% (n = 31), and CNC 51% to 60% (n = 20).

INTERVENTIONS

CI.

MAIN OUTCOME MEASURES

Pre- and 12-month postoperative speech perception performance. Clinically significant improvement was defined as ≥15% gain.

RESULTS

Quiet qualifiers experienced improvement in all listening conditions, whereas +10 dB SNR and +5 dB SNR qualifiers only improved in their qualifying condition and implanted ear CNC. When stratified by expanded Medicare criteria (binaural AzBio ≤60% correct), patients that qualified in quiet experienced improvements regardless of qualifying threshold or background noise. However, those that qualified in noise and AzBio ≤60% experienced mixed results in quiet and limited gain in background noise. When ≤60% criteria was applied to CNC of the worse ear, ≤40% qualifiers experienced large improvements in all tested conditions, but those who qualified by 41% to 50% or 51% to 60% only demonstrated modest improvements in AzBio sentence testing.

CONCLUSIONS

Quiet qualifiers improved in all testing conditions, while those qualifying in noise improved in their qualifying condition. Patients who qualified by expanded Medicare criteria (≤60%) showed improvement when qualifying with AzBio in quiet, but should be used with caution when qualifying patients in background noise or CNC due to more limited gains in performance.

A Consensus to Revise the Minimum Speech Test Battery-Version 3

PURPOSE

The Minimum Speech Test Battery (MSTB) for adults was introduced in 1996 (Nilsson et al., 1996) and subsequently updated in 2011 (Advanced-Bionics et al., 2011). The MSTB has been widely used by clinicians as a guide for cochlear implant (CI) candidacy evaluations and to document post-operative speech recognition performance. Due to changes in candidacy over the past 10 years, a revision to the MSTB was needed.

METHOD

In 2022, the Institute for Cochlear Implant Training (ICIT) recruited a panel of expert CI audiologists to update and revise the MSTB. This panel utilized a modified Delphi consensus process to revise the test battery and to improve its applicability considering recent changes in CI care.

RESULTS

This resulted in the MTSB-Version 3 (MSTB-3), which includes test protocols for identifying not only traditional CI candidates but also possible candidates for electric-acoustic stimulation and patients with single-sided deafness and asymmetric hearing loss. The MSTB-3 provides information that supplements the earlier versions of the MSTB, such as recommendations of when to refer patients for a CI, recommended patient-reported outcome measures, considerations regarding the use of cognitive screeners, and sample report templates for clinical documentation of pre- and post-operative care. Electronic versions of test stimuli, along with all the materials described above, will be available to clinicians via the ICIT website (https://www.cochlearimplanttraining.com).

CONCLUSION

The goal of the MSTB-3 is to be an evidence-based test battery that will facilitate a streamlined standard of care for adult CI candidates and recipients that will be widely used by CI clinicians.

Individual Patient Comorbidities and Effect on Cochlear Implant Performance

OBJECTIVE

To examine the association between preoperative comorbidities and cochlear implant speech outcomes.

STUDY DESIGN

Retrospective cohort.

SETTING

Tertiary referral center.

PATIENTS

A total of 976 patients who underwent cochlear implantation (CI) between January 2015 and May 2022. Adult patients with follow-up, preoperative audiologic data, and a standardized anesthesia preoperative note were included.

EXPOSURE

Adult Comorbidity Evaluation 27 (ACE-27) based on standardized anesthesia preoperative notes.

MAIN OUTCOME MEASURES

Postoperative change in consonant-nucleus-consonant (CNC) score, AzBio Sentence score in quiet, and AzBio + 10 dB signal-to-noise ratio (SNR). Sentence score of the implanted ear at 3, 6, and 12 months.

RESULTS

A total of 560 patients met inclusion criteria; 112 patients (20%) had no comorbidity, 204 patients (36.4%) had mild comorbidities, 161 patients (28.8%) had moderate comorbidities, and 83 patients (14.8%) had severe comorbidities. Mixed model analysis revealed all comorbidity groups achieved a clinically meaningful improvement in all speech outcome measures over time. This improvement was significantly different between comorbidity groups over time for AzBio Quiet ( p = 0.045) and AzBio + 10 dB SNR ( p = 0.0096). Patients with severe comorbidities had worse outcomes. From preop to 12 months, the estimated marginal mean difference values (95% confidence interval) between the no comorbidity group and the severe comorbidity group were 52.3 (45.7-58.9) and 32.5 (24.6-40.5), respectively, for AzBio Quiet; 39.5 (33.8-45.2) and 21.2 (13.6-28.7), respectively, for AzBio + 10 dB SNR; and 43.9 (38.7-49.0) and 31.1 (24.8-37.4), respectively, for CNC.

CONCLUSIONS

Comorbidities as assessed by ACE-27 are associated with CI performance. Patients with more severe comorbidities have clinically meaningful improvement but have worse outcome compared to patients with no comorbidities.

Machine Learning Approach for Screening Cochlear Implant Candidates: Comparing With the 60/60 Guideline

OBJECTIVE

To develop a machine learning-based referral guideline for patients undergoing cochlear implant candidacy evaluation (CICE) and to compare with the widely used 60/60 guideline.

STUDY DESIGN

Retrospective cohort.

SETTING

Tertiary referral center.

PATIENTS

772 adults undergoing CICE from 2015 to 2020.

INTERVENTIONS

Variables included demographics, unaided thresholds, and word recognition score. A random forest classification model was trained on patients undergoing CICE, and bootstrap cross-validation was used to assess the modeling approach's performance.

MAIN OUTCOME MEASURES

The machine learning-based referral tool was evaluated against the 60/60 guideline based on ability to identify CI candidates under traditional and expanded criteria.

RESULTS

Of 587 patients with complete data, 563 (96%) met candidacy at our center, and the 60/60 guideline identified 512 (87%) patients. In the random forest model, word recognition score; thresholds at 3000, 2000, and 125; and age at CICE had the largest impact on candidacy (mean decrease in Gini coefficient, 2.83, 1.60, 1.20, 1.17, and 1.16, respectively). The 60/60 guideline had a sensitivity of 0.91, a specificity of 0.42, and an accuracy of 0.89 (95% confidence interval, 0.86-0.91). The random forest model obtained higher sensitivity (0.96), specificity (1.00), and accuracy (0.96; 95% confidence interval, 0.95-0.98). Across 1,000 bootstrapped iterations, the model yielded a median sensitivity of 0.92 (interquartile range [IQR], 0.85-0.98), specificity of 1.00 (IQR, 0.88-1.00), accuracy of 0.93 (IQR, 0.85-0.97), and area under the curve of 0.96 (IQR, 0.93-0.98).

CONCLUSIONS

A novel machine learning-based screening model is highly sensitive, specific, and accurate in predicting CI candidacy. Bootstrapping confirmed that this approach is potentially generalizable with consistent results.

Responsible Imputation of Missing Speech Perception Testing Data & Analysis of 4,739 Observations and Predictors of Performance

OBJECTIVE

To address outcome heterogeneity in cochlear implant (CI) research, we built imputation models using multiple imputation by chained equations (MICEs) and K-nearest neighbors (KNNs) to convert between four common open-set testing scenarios: Consonant-Nucleus-Consonant word (CNCw), Arizona Biomedical (AzBio) in quiet, AzBio +5, and AzBio +10. We then analyzed raw and imputed data sets to evaluate factors affecting CI outcome variability.

STUDY DESIGN

Retrospective cohort study of a national CI database (HERMES) and a nonoverlapping single-institution CI database.

SETTING

Multi-institutional (32 CI centers).

PATIENTS

Adult CI recipients (n = 4,046 patients).

MAIN OUTCOME MEASURES

Mean absolute error (MAE) between imputed and observed speech perception scores.

RESULTS

Imputation models of preoperative speech perception measures demonstrate a MAE of less than 10% for feature triplets of CNCw/AzBio in quiet/AzBio +10 (MICE: MAE, 9.52%; 95% confidence interval [CI], 9.40-9.64; KNN: MAE, 8.93%; 95% CI, 8.83-9.03) and AzBio in quiet/AzBio +5/AzBio +10 (MICE: MAE, 8.85%; 95% CI, 8.68-9.02; KNN: MAE, 8.95%; 95% CI, 8.74-9.16) with one feature missing. Postoperative imputation can be safely performed with up to four of six features missing in a set of CNCw and AzBio in quiet at 3, 6, and 12 months postcochlear implantation using MICE (MAE, 9.69%; 95% CI, 9.63-9.76). For multivariable analysis of CI performance prediction, imputation increased sample size by 72%, from 2,756 to 4,739, with marginal change in adjusted R2 (0.13 raw, 0.14 imputed).

CONCLUSIONS

Missing data across certain sets of common speech perception tests may be safely imputed, enabling multivariate analysis of one of the largest CI outcomes data sets to date.