A Multi-Level Modeling Approach of Speech Perception after Cochlear Implantation

Abstract

Objective: To evaluate the long-term development of patients with a cochlear implant of the University Medical Centre in Utrecht; with regard to three potential predictive factors: severity of preoperative hearing loss, duration of preoperative deafness and cause of deafness (the ‘bony disorders’ meningitis and otosclerosis vs. all other causes of deafness). Study design: Retrospective longitudinal clinical study. Predictors of speech perception, after cochlear implantation surgery, included preoperative hearing loss, duration of deafness and effect of a bony disorder as cause of deafness (meningitis or otosclerosis); with use of Multi-Level Modeling analysis. Patients: 247 adult patients with a cochlear implant. Interventions: Unilateral multichannel cochlear implantation. Main outcome measures: Postoperative speech perception (CVC) scores. Results and conclusion: Perception of CVC words after cochlear implantation is significantly predicted by duration of deafness, preoperative hearing loss and presence or absence of a bony disorder as cause of deafness. There is no effect of interaction of these prediction variables, nor among themselves nor with the time predictors (=duration of implant use). The development of speech perception over time is best described by a linear and a negative quadratic growth model. As Multi-Level Modeling has been demonstrated in previous studies to be more powerful in hypothesis testing than other analysis tools, our unexpected result of cause of deafness being a significant predictor of speech perception might be due to the sensitivity of the Multi-Level Modeling method.