Identification of predictors and development of a screening protocol for cerebello-pontine lesions in patients presenting with audio-vestibular dysfunction.

Objectives:  Magnetic resonance imaging (MRI) scanning in the investigation of cerebellopontine angle lesions represents a finite resource, the use of which needs to be carefully rationalised. Our aim was to identify predictive factors that can distinguish between patients with and without cerebellopontine angle lesions, and develop a screening protocol which could be useful in the clinical setting as an aid to clinical judgment. Design:  Case–control study. Setting:  Secondary care. Participants:  Audio-vestibular features were collated on 136 patients (M : F 1.39 : 1) and 288 controls (M : F 1 : 1.1). Intervention:  Diagnostic by analysis of symptoms and audiometric data using logistic regression, receiver-operator characteristic curves and backwards elimination. Main outcome measures:  Development of a predictive algorithm comprising those factors which are most strongly predictive of the presence of a cerebellopontine angle lesion. Results:  Positive predictors of a cerebellopontine angle lesion include the interaural threshold difference at 1 (P = 0.044) and 4 kHz (P = 0.034). The threshold in the better hearing ear at 0.25 kHz exerts a negative predictive (i.e. protective) effect (P = 0.005). The presence of tinnitus does not appear to influence the outcome on logistic regression. Although vertigo does exert an influence on the overall model, its impact is highly equivocal. Conclusions:  We have identified audiometric factors which exert a positive and negative predictive influence on the presence of a cerebellopontine angle lesion, and audiovestibular symptoms which appear to exert little effect on the model. Our predictive equation represents a user-friendly standardised method of risk-stratification of patients within a general otolaryngology clinic.