A comparison of water and air caloric responses and their ability to distinguish between patients with normal and impaired ears.

OBJECTIVES: The caloric test is a mainstay of modern vestibular assessment. Yet caloric test methods have not been well standardized, and normal response values have not been universally agreed upon. The air caloric test has been particularly problematic. In this article, we present our efforts to establish a population-based description of the caloric response evoked by water and air stimuli at both cool and warm temperatures.
DESIGN: Data were collected from a retrospective record review of patients who underwent caloric testing at Mayo Clinic Jacksonville between 2002 and 2006. Two subgroups were identified. One group was found to have no vestibulopathy after comprehensive medical investigation. The second group was found to have severe bilateral vestibular weakness; this diagnosis was based on medical evaluation and objective test results. Caloric response distributions and associated probability estimates were developed from each group.
RESULTS: A total of 2587 medical records were found to contain caloric response data. Of these, 693 patients met the criteria to be classified as having no identifiable vestibulopathy (otologically normal patients with normal caloric responses). Sixty-eight patients met the criteria for bilateral vestibular weakness (reduced or absent rotatory chair responses). Our analysis yielded the following results: (1) there were differences between nystagmus distributions across stimuli. On average, the magnitude of cool water (30 degrees C) maximum slow-phase velocities was smaller than those from warm water (44 degrees C). Maximum slow-phase velocity distributions from cool (21 degrees C) and warm (51 degrees C) air stimuli were more similar to each other than were responses to water stimuli and fell between the water distributions. (2) Combined metrics (combined eye speed and total eye speed) were comparable for water and air stimuli. (3) Response distributions from otologically normal patients were different from those of patients with bilateral vestibular weakness. (4) Derived probability estimates allowed for quantification of caloric response normal limits, sensitivity, specificity, and error rates.
CONCLUSIONS: Current bithermal test methods assume an equivalence of caloric response strength from warm and cool stimuli. Our results show standard cool and warm water stimuli provoke substantially different response magnitudes, with warm stimuli provoking stronger responses. When calibrated as described herein, air stimuli perform comparably with water stimuli for bithermal caloric test purposes, with more uniform and less variable response distributions. Both air- and water-based tests were able to distinguish between normal and abnormally weak ears with sensitivity and specificity values between 0.82 and 0.84. We advocate for the calibration of all caloric stimuli based on the test's statistical performance and not arbitrary assumptions about stimulus equivalence.