Establishing a method to predict the outcome of vestibular schwannoma surgery based on one's own results

Objectives: To establish a prognostication method based on our own results of vestibular schwannoma surgery. Design: Retrospective data analysis. Settings: Tertiary referral centre. Participants: 141 ambulatory patients operated for unilateral vestibular schwannoma by the translabyrinthine approach in the period 1996--2003. Main outcome measures: Facial impairment defined by House Brackmann grade III-VI, and the relation with tumor size. Results: For our institution we found that in a range of tumor sizes the tumor size of 17.5 mm was the cut-off point associated with highest sensitivity and specificity values available concerning the prediction of facial impairment, these were 0.86 and 0.61. Compared to a random cut-off point (11 mm), this leads to 36% more accurate predictions. Conclusions: By establishing sensitivity and specificity values of predictions, one is aware of the rate of false predictions. By Receiver Operating Curve analysis the rate of false predictions can be minimised. In vestibular schwannoma surgery this leads to more precise predictions concerning outcome, as we have demonstrated for the facial function.