The Failure Index as Marker of the Neural Health Profile in MED-EL CI Users: A Retrospective Analysis.

It is a retrospective look at clinical records. No one was randomised, and the authors could not control the stimulation settings used for each recording (phase durations of 40, 50 or 60 µs and two different inter-phase gaps), who was tested with which speech test, or how old the users were. The software's quality check threw away 55% of the nerve-response recordings, and weak recordings were the most likely to be discarded. Contacts and ears with genuinely poor nerve function may have been dropped, which could make the measure look better than it is. People with the lowest speech scores are underrepresented in the sample, so the relationship between the measure and speech was worked out mostly on better performers. There is no direct measure of how many nerve fibres survive to compare the Failure Index against. The clearest supporting evidence is 7 children with cochlear nerve hypoplasia. The strongest speech result rests on small groups: 27 to 41 ears for the extreme-value analysis, and 28 ears for the model combining the measure with age and duration of hearing loss. Everything comes from one centre and one manufacturer's implants and fitting software, so it is unknown whether the numbers transfer to other devices or clinics. The proposed cut-off is described by the authors themselves as exploratory and not clinically validated.

Funded by the MHH-plus foundation, which the authors say had no involvement in the study. Eight of the authors are employed by MED-EL, the company that makes the implants and the fitting software the recordings came from; they state their roles are scientific, with no marketing or sales involvement. All other authors report no conflict of interest.

Do not treat the Failure Index as a ready-made clinical test, and do not read the reported cut-off as a validated threshold. The authors call it exploratory and say it needs validation in independent cohorts before it can stratify patients. The striking speech result came from 28 ears using the single worst contact along the array; averaged across 145 ears the same measure was only weakly related to speech scores. Nothing in this study shows that switching off electrodes improves anyone's speech.