Preoperative versus postoperative Frequency Allocation of Cochlear Implant Electrodes: A Retrospective Cohort Study

Background/objectives. The mismatch between the predicted and preset electrode frequencies and the actual postoperative position of the electrode array may influence the cochlear implant performance. This study aimed to evaluate the accuracy of prediction and preset values and examine their interactions with anatomical factors and hearing outcomes in CI recipients. Study design. This retrospective cohort study was conducted in accordance with the STROBE guidelines. Methods. Sixty-nine postlingually deafened patients implanted with lateral wall electrode arrays at a tertiary hospital underwent OTOPLAN analysis using pre- and postoperative computed tomography. The patients were categorized as fully inserted (n= 49) or incompletely inserted (n = 20). Cochlear duct length (CDL), electrode angular insertion depth (AID), age at implantation, years of severe hearing loss, surgical approach to the cochlea, and frequency-to-place mismatch (prediction and fitting software preset vs. postoperative position), expressed in Hertz and semitones, were evaluated in relation to hearing outcomes (sound field audiometry, speech audiometry, and monosyllabic word test in noise) 12 months after implantation. The analyses included Wilcoxon tests, Kruskal–Wallis tests, correlation analyses, and multivariable linear regression. Results. The predicted electrode frequencies were significantly higher than the postoperative measurements for E12–E2 (all p < 0.001). Postoperative frequencies exceeded MAESTRO central frequency presets for all electrodes (all p < 0.001), with the largest mismatch basally (E12 median 3199 Hz, IQR 2706–4743 Hz) and the largest semitone mismatch apically (E1 median 18.3 semitones, IQR 13.7–21.9 semitones). Preoperative evaluation using the Organ of Corti (OC) map showed less mismatch than that using the Spiral Ganglion (SG) map (p < 0.001). AID strongly predicted semitone mismatch (β = –0.958, p < 0.001) and positional prediction error (β = 0.436, p = 0.008), whereas CDL, age, and surgical approach did not. Only years of severe hearing loss 6correlated with monosyllabic word recognition (ρ = –0.293, p = 0.041) in noise. Conclusion. There was a systematic overestimation of electrode AID and frequency by OTOPLAN, both in preoperative predictions and preset values, particularly in the basal and middle cochlear regions. Frequency-to-place mismatch depended on insertion depth, not cochlear size. Hearing outcomes were unaffected by anatomical or mismatch variables, except for the duration of auditory deprivation. Further research is required to refine the prediction models and fitting strategies.