Exploiting routine clinical measures to inform strategies for better hearing performance in cochlear implant users

Neuroprostheses designed to interface with the nervous system to replace injured or missing senses can significantly improve a patient’s quality of life. The challenge remains to provide implants that operate optimally over several decades. Changes in the implant-tissue interface may precede performance problems. Tools to identify and characterise such changes using existing clinical measures would be highly valuable. Modern cochlear implant systems allow easy and regular measurements of electrode impedance (EI). This measure is routinely performed as a hardware integrity test, but it also allows a level of insight into the immune mediated response to the implant, which is associated with performance outcomes. This study is a 5-year retrospective investigation of MED-EL CI users at the University of Southampton Auditory Implant Service including 176 adult ears (18-91) and 74 paediatric ears (1-17). The trend in EI in adults showed a decrease at apical electrodes. An increase was seen at the basal electrodes which are closest to the surgery site. The trend in the paediatric cohort was increasing EI over time for nearly all electrode positions, although this group showed greater variability and had a smaller sample size. We applied an outlier-labelling rule to statistically identify individuals that exhibit raised impedance. This highlighted 14 adult ears (8%) and 3 paediatric ears (5%) with impedance levels that deviated from the group distribution. The slow development of EI suggests intra-cochlear fibrosis and/or osteogenesis as the underlying mechanism. The usual clinical intervention for extreme impedance readings is to deactivate the relevant electrode. Our findings highlight some interesting clinical contradictions: some cases with raised (but not extreme) impedance had not prompted an electrode deactivation; and many cases of electrode deactivation had been informed by subjective patient reports. This emphasises the need for improved objective evidence to inform electrode deactivations in borderline cases, for which our outlier-labelling approach is a promising candidate. A data extraction and analysis protocol that allows ongoing and automated statistical analysis of routinely-collected data could benefit both the cochlear implant and wider neuroprosthetics communities. Our approach provides new tools to inform practice and to improve the function and longevity of neuroprosthetic devices.