Abstract
The slow-phase velocity of nystagmus is one of the most sensitive parameters of vestibular function and is currently the standard for evaluating the caloric test. However, the assessment of this parameter requires recording the response by using nystagmography. The aim of this study was to evaluate whether frequency and duration of the caloric nystagmus, as measured by using a clinical test with Frenzel glasses, could predict the result of the recorded test. The retrospective analysis of 222 caloric test results recorded by means of electronystagmography has shown a good association between the 3 parameters for unilateral weakness. The asymmetry observed in the velocity can be predicted by a combination of frequency and duration. On the other hand, no relationship was observed between the parameters for directional preponderance. These results indicate that a clinical caloric test with frequency and duration as parameters can be used to predict the unilateral weakness, which would be obtained by use of nystagmography. We propose an evaluation of the caloric test on the basis of diagrams combining the 3 response parameters.
The caloric test is one of the most useful clinical methods for evaluating vestibular function. The caloric response can be quantified by using various nystagmus parameters. Among these parameters, the most studied include the total duration, the maximum frequency at culmination (Fmax), and the maximum slow-phase velocity (Vmax). 1-10 Although a controversy exists as to which parameter is the most appropriate, Vmax appears to be a highly sensitive indicator of vestibular response. 6-10 Today, most computerized nystagmography devices provide results from caloric tests calculated with Vmax.
Regardless of the intrinsic quality of each parameter in evaluating the vestibular response, it is a fact that estimating Vmax inevitably requires recording the response and using sophisticated and expensive instruments, yielding electronystagmography (ENG) or videonystagmography. On the other hand, vestibular tests, especially ENG, require much time, and in practice it is not possible to perform this test on all patients. However, the total duration of the response and the maximum frequency of the nystagmus can easily be estimated by the observer in a clinical caloric test with Frenzel glasses. Considering time and cost savings, it would be most advantageous to have a reliable clinical caloric test that would permit dispensing with the recorded test in certain cases. Indeed, the caloric test with Frenzel glasses is accessible to all ear, nose, and throat specialists and could be used as a screening test before referring patients to a neuro-otology center.
The goal of this study was to evaluate whether the clinical caloric test with Frenzel glasses, with duration, frequency, or both as parameters, could predict the result of the recorded test on the basis of Vmax. For this purpose, we have analyzed these 3 caloric response parameters and compared the clinical assessments obtained from each parameter. The analysis is made on the basis of the recorded caloric test result, which is the only method allowing an objective and simultaneous assessment of the different parameters. Even though the clinical caloric test is less accurate than a nystagmographic recording, the evaluation of duration and frequency of nystagmus is comparable with the two techniques for a well-trained observer.
METHODS AND MATERIAL
In our clinic we perform the alternate, binaural, bithermal caloric test popularized by Fitzgerald and Hallpike. 11 When compared with the original technique, we use cool water at 22°C and warm water at 44°C, and we administer 20 mL of water by means of a syringe during 20 seconds. The irrigation order is as follows: cool water in the right ear, cool water in the left ear, warm water in the right ear, and warm water in the left ear. After the end of the response, a break of at least 5 minutes is made between each stimulation. The patient is placed in the dorsal decubitus position, with the head and trunk tilted at 30°. The caloric test is recorded by means of computerized ENG (Nystar Plus). The examination is carried out with the eyes open and in total darkness, and an ocular calibration is undertaken before each irrigation. The recording commences at the end of the syringing. Before the caloric test, a complete vestibular examination is performed, including oculomotor tests (smooth pursuit, saccades, optokinesis, and gaze-evoked nystagmus), a sinusoidal rotational examination (0.05 Hz, maximal velocity of 60°/s) with the visual suppression test, and the test for spontaneous or positional nystagmus. For the caloric test, the Nystar Plus program automatically calculates Vmax during a period of 10 seconds at the culmination of each response. The complete test is summarized as a 2-part outline (Fig 1): the top part shows the tracing of each response during the 10 seconds of culmination (windows), and the lower part shows a diagram where the 4 responses are given in their entirety with the coordinates of the slow-phase velocity as a function of time. The different parameters of caloric nystagmus are easily noted from the outline: Fmax is calculated by counting the number of saccades in each window, Vmax is given by the computer, and total duration (in seconds) is estimated on the diagram at the end of the response. For this last parameter, we have decided to establish the end of the reaction as the last recorded observation; the device rejects all nystagmus of less than 2°/s. Finally, we have considered the time at the culmination, which is located at the midpoint of each window (indicated by an arrow on the summary diagram). All these data are treated by using the Stata Statistical Software 5.0 (StataCorp 1997, College Station, TX).
Summary of the bithermal caloric test (Nystar Plus).
Calculations of unilateral weakness (UW) and directional preponderance (DP) are made for each parameter of nystagmus (Vmax,Fmax, and duration) with Jongkees' classical formulae 8 :
Detection of pathologic UW of Vmax by using the UW of Fmax and duration (see text for details).
where R indicates the right ear and L indicates the left ear.
With the Vmax parameter being the reference, the comparative study of the results respects the normal asymmetry limits provided by Nystar Plus for this parameter (ie, 22% for UW and 28% for DP). The generally accepted asymmetry limit for Vmax varies between 20% and 25% for UW and between 20% and 30% for DP. 12-15
Predicting the UW measured with Vmax, starting with the UW observed with Fmax and duration, was done by using a logistic regression. The studied outcome is that the UW of Vmax is greater than 22%, which is predicted by using absolute values (ie, ignoring direction) for UW of Fmaxand duration, as well as the combined effect (interaction) of these two factors. This interaction gives a curvilinear shape for the reference lines in Fig 2. The upper panel shows the relationship obtained between the bithermal UW of Vmax and the other bithermal UWs. The lower panel shows the relationship obtained for the bithermal UW of Vmax and monothermal UWs (cool) of Fmax and duration.
Relationship between the UW index (
Data used for this study originated from 222 caloric test results recorded on 200 patients (111 men and 89 women; average age, 49 years; range, 14-85 years) examined for dizziness in our clinic between June 1994 and June 1997. Sixteen patients were examined several times. Inclusion criteria for the caloric test were as follows. First, the presence of a caloric response for the 4 irrigations was required. Test results showing a complete unilateral areflexia have been excluded on the one hand because all parameters give the same result (bias of their relative sensitivity) and on the other because the bithermal response of only one functioning ear may produce a bias of directional preponderance caused by a temperature effect (magnitude difference between cool and warm response). 10 Other requirements were a reliable reading quality for the 3 nystagmus parameters and the absence of a spontaneous nystagmus (error in estimating the duration and slope of the DP).
We have classified the clinical diagnoses of the 200 patients into 5 categories: group 1, Ménière's disease (35 patients); group 2, labyrinthitis with vestibular or cochleovestibular deficit (56 patients); group 3, benign paroxysmal positional vertigo and posttraumatic vestibulopathy (29 patients); group 4, dizziness of tumoral, vascular, or degenerative origin (33 patients); and group 5, vertigo of unknown origin (47 patients).
Average and SD of the 222 responses
R, Right ear; L, left ear; Vmax , maximum slow-phase velocity (in degrees per second); Fmax maximum frequency (number of saccades per 10 seconds); TD, total duration (in seconds); CT, culmination time (in seconds).
Add 20 seconds to the TD and CT for literature comparison purposes.
RESULTS
The observed values for each variable are given in Table 1.
Unilateral Weakness
The comparison of the UW index as a function of Vmax,Fmax, and duration of the nystagmus shows a good correlation between the 3 parameters. The correlation coefficients between the parameters (ie, Vmax and duration, Vmax and Fmax, and duration and Fmax) are 0.81, 0.77, and 0.72, respectively. These values are similar in the 5 different groups of diseases. The top panel of Fig 3 shows the relationship between the 3 UWs, with a slightly sigmoid shape for Fmax and duration curves. The vestibular asymmetry appears globally more marked with Vmax than with the other two parameters. A linear regression of the results shows that 22% of the accepted normal asymmetry for Vmax corresponds to a tolerance of 19% for Fmax and 12% for duration. The top panel of Fig 2 shows the quality of detection for Vmax by means of the other two parameters. Results are given as a function of Vmax: the plus signs (81 observations) represent the pathologic results (UW of Vmax > 22%), and the dots (141 observations) are normal results. The vertical line defines the normal limit of duration (12%), and the horizontal line defines that of Fmax (19%). To detect a pathologic result for Vmax on the basis of a single other parameter, Fmax (UW > 19%) and duration (UW > 12%), respectively, show a sensitivity of 53% and 64%, a specificity of 97% and 94%, and a correct classification of 81% and 83%. If the criteria of the two parameters are considered together (the lower left rectangle defines the normal), the sensitivity reaches 75%, specificity reaches 92%, and correct classification reaches 86%. The predictive quality is further improved if Fmax and duration are combined by means of logistic regression (see Methods and Material section). This association is expressed by probability levels for finding a UW of Vmax of greater than 22% when the other two parameters are combined. The region within the 25th percentile contains 16% (13/81) false-negative results; nevertheless, the percentage of pathologic UW in this area is 10% (13/125), giving a negative predictive value of 90%. The region above the 75th percentile contains 1.5% (2/141) false-positive results.
Relationship between the UWs of monothermal and bithermal tests
r Correlation coefficient; CC, correct classification.
Directional Preponderance
No relationship is observed between the 3 parameters for the DP index (Fig 3, bottom). The correlation coefficients between the parameters (ie, Vmax and duration, Vmax and Fmax, duration and Fmax) are 0.53, 0.27, and 0.43, respectively. The absence of association between the DP values is independent of the clinical diagnosis. These weak correlations show that the DP index does not reflect the vestibular status of the patient.
Monothermal Caloric Test
Because the clinical interest in the DP was not demonstrated, we investigated whether monothermal stimulation (warm or cool) could predict a UW with the same quality as a complete test. For each parameter, we compared the asymmetry index, as well as the percentage of patients labeled as healthy or diseased, on the basis of only one temperature (according to the following simplified formula: [R - L]/[R + L]) with those thus classified by means of the bithermal test. The results are presented in Table 2. Globally, the monothermal test shows a good agreement with the bithermal test. Nevertheless, the best characteristics are observed with the cool stimulation. The bottom panel of Fig 2 shows the combination of Fmax and duration obtained by a cool monothermal test for predicting a bithermal UW of Vmax of greater than 22%. The region within the 25th percentile contains 25% (20/81) false-negative results, but the percentage of abnormal UWs in this area is 18% (20/112), giving a negative predictive value of 82%. The region above the 75th percentile contains 2% (3/141) false-positive results.
DISCUSSION
Although there is no general consensus regarding the methodology for the caloric test, the alternate, binaural, bithermal, caloric stimulation introduced by Fitzgerald and Hallpike 11 is one of the most commonly used techniques. These authors proposed a procedure (injection of 250 mL of water at 30°C and 44°C during 40 seconds) that has been largely adopted, with some minor variations. In our clinic, for convenience, we have changed the technical character on two points: first, we perform a cool stimulation at 22°C because this temperature corresponds to the ambient temperature of the ENG room, which is air conditioned and therefore does not require special preparation (only one water tray with thermostat is used for warm water at 44°C); and second, we have chosen to use a small irrigation volume (20 mL during 20 seconds) to decrease the patient's discomfort and to allow a simple syringing without additional equipment. The mean values of the various response parameters that we obtain with this method (Table 1) are comparable with those reported by authors using the original technique, 2,7,9,10,16 taking into account that our recording begins at the end of the irrigation and that it is necessary to add 20 seconds to the total duration and to the response culmination time. The only significant difference is that with our process, we observe a similar temperature effect for all parameters (cool response greater than warm response), which contrasts with the classical method (cool response greater than warm response for duration and the opposite for velocity and frequency). However, as noted by Aschan et al, 2 we also believe that the magnitude difference in the parameters between cool and warm responses are of little importance, but rather the primary aim of the caloric test is to demonstrate whether the response to the same stimulus is symmetric.
The disagreement persists as to which parameter is the most useful for quantifying the caloric response. The total duration of the nystagmus was initially used because it was the response measure most easily identifiable. With the development of recording techniques, other parameters have been studied, and Fmax 4,5 or Vmax 6–10 appear to be more sensitive indicators of vestibular function. Today, most American authors 12–15 propose Vmax for evaluating caloric nystagmus. On the other hand, Freyss et al17 recommend Fmax, which still remains a widely used parameter, especially in France.
In this study we have compared these different parameters to be able to know to which degree they could be used together for evaluating the caloric test. We have given no qualitative preference to one or the other of the measurements; if we have chosen Vmax as a standard, it is because it is accessible with nystagmography alone.
Unilateral Weakness
The comparison of the UW index shows a good correlation between the 3 parameters, and this result is independent of the vestibular pathology. It is interesting to note that our results demonstrate the best correlation between Vmax and duration, whereas others 18 have observed a poor association between these two parameters. These results show that it is possible to evaluate the UW by using one or the other of the response measurements. Nevertheless, vestibular asymmetry appears globally more pronounced with Vmax than with the other two parameters, and Vmax may be accepted as a more sensitive indicator of a labyrinthine lesion. By using an asymmetry limit of 22% for Vmax, we obtain a tolerance corresponding to 19% for Fmax and 12% for duration. In the literature the asymmetry limits accepted for these two parameters vary from 15% to 25% 17,19 and from 10% to 14%, 8,19 respectively. Aiming to detect a pathologic result of Vmax with the other measurements, the relative sensitivity of duration and Fmax is satisfactory, but upon combining the two parameters, the predictive quality is substantially improved (Fig 2, top).
Directional Preponderance
The DP should reflect an imbalance in the vestibular system that causes the caloric nystagmus to be more pronounced in one direction than in the other, regardless of the ear stimulated. This alteration in nystagmic function is attributed to an asymmetry of central vestibular balance, which would result from a decrease in the labyrinthine activity or from a primary central dysfunction. In labyrinthine pathologies the DP would especially allow one to follow the status of the central compensation. Our results show a poor correlation between the DP index of the 3 parameters, which means that each measurement gives a different value. The observation does apply to all types of pathologies. This weak association supports the argument that the DP does not reflect the status of the patient and cannot be taken as an indicator of vestibular dysfunction. Other authors 4,10,14,15,18,20 have already discussed the poor clinical value of the DP, which appears to be a nonspecific measurement most often correlated with a spontaneous or positional nystagmus.
Monothermal Stimulation
Except in cases of spontaneous nystagmus, the DP has a limited clinical significance and is difficult to interpret. Therefore it is no longer necessary to irrigate each ear twice to evaluate a normal response or a UW. The concept of a caloric test with a sole stimulation is not new, and several authors have already studied the predictive value of the monothermal test. 20–24 Except for Barber et al 20 and Jacobson and Means, 21 these authors observe a poor sensitivity of the warm or cool monothermal test in detecting the results of the bithermal test and conclude that the monothermal test cannot be used for clinical screening. Our results show a good correlation between the monothermal and bithermal tests for each temperature and the 3 response parameters, indicating that the monothermal test globally offers the same predictive quality for UW as does a complete test. However, we observe a better percentage of correct classifications with the cool stimulation (>85%), whereas a better accuracy with warm water is generally reported in the literature. 22–24 This observation suggests that the temperature providing the best accuracy corresponds to the most intense stimulus of the bithermal test (warm in a test with 30°C and 44°C and cool in a test with 22°C and 44°C). To predict a bithermal UW of Vmax of greater than 22% with the other two parameters, the cool monothermal test is slightly poorer than the bithermal test, as is seen by the comparison of the top and bottom panels of Fig 2. In the region within the 25th percentile, where the UW of Fmax and duration are normal, the probability of having a pathologic Vmax response nevertheless remains moderate at 18% (10% with the bithermal test). We believe that this risk of error is acceptable for a caloric evaluation, considering the savings of time with the monothermal test.
From a practical standpoint, we propose the following evaluation of the caloric test, on the basis of Fig 2. A clinical examination with Frenzel glasses should be performed first by using frequency and duration. To estimate Fmax, we recommend counting the number of nystagmus during 30 seconds, starting 40 seconds after the beginning of the stimulation, to include the culmination time. The test begins with the cool stimulation (22°C) used as a monothermal screening test. If the UW index of the two parameters falls within the region of the 25th percentile or above that of the 75th percentile (Fig 2, bottom), the probability of finding, respectively, a normal or pathologic UW of Vmax is very high, and the examination is ended (the 3 parameters give the same information). If Fmax and duration indicate the intermediate region (between the 25th and 75th per-centile range), the test is completed with the warm water stimulation, and the bithermal result is interpreted as in the top panel of Fig 2. If the final result is still within the intermediate region, the UW index of Vmax is doubtful, and a recorded bithermal test is required to clarify the vestibular asymmetry. We believe that this method has great advantages if the observer is well trained in the clinical caloric test with Frenzel glasses.
CONCLUSION
A caloric test with Frenzel glasses can be used to predict the result of a recorded bithermal test. With a bit of practice, the clinical test is easy to perform and has a good screening value in vestibular diagnosis. Furthermore, it enables one to identify patients who do not require expensive nystagmography.