Prediction models for clinical outcome after cochlear implantation: a systematic review

Objectives: Cochlear implants (CIs) are implantable hearing devices with a wide variation in clinical outcome between patients. We aim to provide an overview of the literature on prediction models and their performance for clinical outcome after cochlear implantation in bilateral hearing loss or deafness. Study Design and Setting: In this systematic review, studies describing the development or external validation of a multivariable model for predicting clinical CI outcome were eligible for selection. Results: A total of 4,042 references were screened. We included nine development studies and one external validation study. The outcome measure of all development studies was speech perception performance after cochlear implantation. The most commonly used model predictors were duration of hearing loss or deafness (n = 7), different types of preoperative measurements (n = 5), and etiology (n = 3). In three studies, crucial information to enable the model to be used for individual risk prediction was missing. One study performed internal validation,two models were externally validated. One study reported specific discrimination or calibration performance measures. Conclusion: Although many articles describe development studies of prediction models for speech perception performance after cochlear implantation, the value of most of these models for their application in clinical practice remains unclear. Therefore, research should focus on increasing the clinical relevance of existing CI outcome prediction models