The diagnostic value of imaging the patient with dizziness - A Bayesian approach

Background: Imaging studies are routinely used in the evaluation of patients with dizziness. A principal concern of the ordering physician is to rule out a cerebellopontine angle (CPA) mass. The incidence of such masses in patients presenting with dizziness is quite low, however, raising the question of the value of imaging this population. Objective: To calculate the probability, using Bayes theorem, that a given patient with dizziness has a CPA mass. Design: Meta-analysis of epidemiological data on CPA masses and of studies reporting the incidence of otologic symptoms in patients with these masses. We also conducted a study of consecutive patients with dizziness to determine the frequency of asymmetric hearing loss in this population. These data were combined in applications of Bayes theorem to calculate disease probabilities. Results: The probability that a patient with dizziness has a CPA mass is 0.0004, indicating that 2500 imaging studies would have to be performed to identify 1 CPA mass. If patients with subjectively normal hearing are investigated (ie, those with isolated dizziness), the probability is 0.000107, indicating that 9307 scans would have to be performed to identify 1 CPA mass. If the search is restricted to those patients with dizziness and asymmetric hearing loss (the patients usually felt to be high risk), the probability is 0.00156, indicating that 638 scans would have to be performed to identify 1 CPA mass. Conclusions: Even when studying patients with dizziness and asymmetric hearing loss, the probability of identifying a CPA mass is sufficiently low that we do not feel imaging is generally warranted. When faced with a patient with dizziness, we recommend a careful neurologic and otologic examination. If abnormalities are detected on examination that suggest central nervous system disease or invasive otologic disease, imaging should be pursued as appropriate. In cases of acute vertigo, if the patient is at high risk for cerebrovascular disease by virtue of age and additional risk factors, imaging should probably be pursued. For the remainder of patients, if progression of hearing loss is not documented, we do not believe imaging is warranted. Progressive hearing loss with abnormal speech reception thresholds probably warrants a magnetic resonance imaging scan of the internal auditory canals.