Development of a novel screening tool for predicting Cochlear implant candidacy

Estimating the United States Patient Population Size Meeting Audiologic Candidacy for Cochlear Implantation

OBJECTIVE

This study aimed to estimate the size of the United States candidacy pool meeting expanded Center for Medicare Services criteria for cochlear implantation.

STUDY DESIGN

Retrospective cross-sectional.

SETTING

Tertiary care center.

METHODS

Preimplantation audiometric data from 486 patients seen at a single academic medical center were collected retrospectively and used to generate a predictive model of AzBio score based on audiometric pure tone thresholds. This model was then used to estimate nationally representative cochlear implantation (CI)-candidacy using pure tone averages included in the National Health and Nutrition Examination Survey. Qualitative and quantitative analyses were performed.

RESULTS

We find that the estimated prevalence of CI candidacy in individuals 65 years of age or older is expected to more than double with a change in the CI candidacy criteria from ≤40% to ≤60% (from 1.42%, 95% confidence interval [1.33, 1.63] to 3.73% [2.71, 6.56]) on speech testing. We also found the greatest absolute increase in candidacy in the 80+ age group, increasing from 4.14% [3.72, 5.1] of the population meeting the ≤40% criteria to 12.12% [9.19, 18.35] meeting the ≤60% criteria.

CONCLUSION

The United States population size meeting expanded CMS audiologic criteria for cochlear implantation is estimated to be 2.5 million adults and 2.1 million age 65 or older. Changing the CI candidacy criteria from ≤40% to ≤60% on CI testing has the greatest effect on the eligible patient population in the >65-year-old age group. The determination of utilization rates in newly eligible patients will require further study.

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.

External Validation of Cochlear Implant Screening Tools Demonstrates Modest Generalizability

OBJECTIVE

To assess the clinical application of five recently published cochlear implant (CI) candidacy evaluation (CICE) referral screening tools through external validation.

STUDY DESIGN

Retrospective cohort study.

SETTING

Tertiary otology/neurotology practice.

PATIENTS

Adults who underwent CICE between December 2020 and September 2021.

INTERVENTIONS

CICE referral screening tools versus CI candidacy criteria.

MAIN OUTCOME MEASURES

CICE screening tool performance, based on the ability to identify patients who met the CI candidacy criteria, was evaluated. CI candidacy criteria were defined as best-aided AzBio sentences at +10 signal-to-noise ratio and either 60% or less accuracy to reflect traditional criteria used in clinical settings or 40% or less accuracy (only patients 65 years or older) to reflect Medicare-eligible criteria.

RESULTS

Screening criteria of proposed CICE referral tools vary widely across pure-tone average and word recognition scores. When screened by traditional criteria, the sensitivities and specificities of these referral tools varied from 40 to 77% and from 22 to 86%, respectively. When screened by Medicare-eligible criteria, sensitivities and specificities varied from 41 to 81% and from 24 to 91%, respectively. The screening tool proposed by Zwolan et al. ( Otol Neurotol 2020;41(7):895-900) demonstrated the best overall performance for traditional (Youden's J , 0.37; sensitivity, 62%; specificity, 75%) and Medicare-eligible patients (Youden's J , 0.44; sensitivity, 66%; specificity, 78%). All screening tools performed worse on the validation cohort compared with their respective development cohorts.

CONCLUSIONS

Current tools for determining CICE referral have diverse screening criteria. These combinations of pure-tone average and word recognition score are modestly successful at identifying CI candidates.