Validation of the Ménière’s Disease Patient-Oriented Symptom-Severity Index

Objective  To validate a disease-specific, health-related quality-of-life instrument for people with active, unilateral, cochleovestibular Ménière’s disease (MD).

Design  Prospective, blinded, randomized cohort study using a 16-item instrument, the Ménière’s Disease Patient-Oriented Symptom-Severity Index (MDPOSI), version 2.

Setting  Four clinical sites, including 3 university hospitals and 1 private practice.

Participants  Sixty-one adult participants in a clinical trial of a new treatment modality (the Meniett device) at 4 medical centers.

Intervention  The MDPOSI was administered at baseline and at 2 and 4 months after the use of the Meniett device or the placebo.

Main Outcome Measures  Each item, domain, and overall MDPOSI score was compared with self-reported cumulative frequency of definitive attacks of vertigo at baseline and 2 and 4 months later.

Results  The MDPOSI global score, domain scores, and 13 of 16 items were significantly associated with the frequency of vertigo (gold standard) as noted on the daily symptom report card.

Conclusions  The MDPOSI is a valid and reliable instrument to evaluate the impact of MD on patients’ disease-related quality of life, and it is responsive to change in the status of the disorder. The MDPOSI is suitable for use in clinical practice and clinical research in people with MD.

Measuring the effects of Ménière’s disease (MD) on patients is difficult because MD is an intermittent disorder with diverse symptoms that vary in intensity from day to day and change in character over time. The spectrum of symptoms tends to reflect the stage of the disorder. In the early stages, vertigo attacks are typically infrequent but severe, and the hearing returns toward normal after the vertigo spell subsides. In the later stages, vertigo intensity decreases, hearing remains poor, and unsteadiness increases. Moreover, there is great variability in this typical clinical picture between patients, with some having prompt remission, while others have a progressively worsening experience with unrelenting vertigo.

Increasingly, physicians are using standardized questionnaires to evaluate new patients and monitor therapy. However, in patients with MD, physicians have had to rely on patients’ overall assessment of their general condition to select, monitor, and adjust therapy. For example, if the patient feels that the vertigo is worsening, the physician may recommend an escalation of treatment. However, such “feelings” may be influenced by non-MD considerations, such as depression or stress, which may affect well-being but do not adequately describe the status of the primary disorder.

To probe the psychosocial manifestations of MD, a 23-item instrument was designed, the Ménière’s Disease Patient-Oriented Symptom-Severity Index (MDPOSI),¹ to assess the disease-specific quality of life of people with MD. The original instrument assessed the severity of symptoms (balance, hearing, mentality, and daily activities), both during and between attacks, and probed other important aspects of the patient’s life (social, interpersonal, employment, and outlook). The instrument also asked the respondent to rate the importance of each item on a separate scale. (The actual MDPOSI instrument is available from the corresponding author).

After evaluation, the original instrument was revised by deleting 7 items that either did not correlate with symptoms or that appeared redundant.² The hearing question was replaced with a query about ear pressure and tinnitus during attacks. The resulting instrument (version 2) was administered as a disease-specific, health-related quality-of-life measure to participants in a clinical trial of a new treatment modality, the Meniett device.³ Herein we report the results of that analysis in terms of reliability, construct validity, and sensitivity to change.

The details of the Meniett device clinical trial have been reported elsewhere.³ In short, 67 people with typical unilateral cochleovestibular MD confirmed by abnormal electrocochleography findings were randomly assigned to receive an active or a placebo device. In general, participants in both groups noted improvement over time, but the improvement noted by the active treatment group was significantly greater. As part of the assessment of outcomes, the MDPOSI was administered along with 3 other quality-of-life instruments (the Vertigo Symptom Scale,⁴ the World Health Organization Quality of Life–Abbreviated scale,⁵ and the Sense of Coherence scale⁶) at baseline and again at 2 and 4 months after the start of the trial. Of the 62 patients who completed the clinical trial, 61 had complete MDPOSI data from all 3 times, and these made up the analysis cohort. The present analysis is based on the vertigo experience of the entire cohort, not on treatment assignment. The protocol was approved by each site’s institutional review board, and informed consent was obtained by the study coordinators for every participant.

Vertigo intensity and frequency were recorded by each participant each day by noting the presence or absence of a vertigo attack on a symptom report card using a scale of 0 to 4. A definitive attack of vertigo (level 2) was defined according to the American Academy of Otolaryngology–Head and Neck Surgery Foundation criteria, namely one that lasted 20 minutes or longer.⁷ A noncounting attack of vertigo (level 1) lasted less than 20 minutes and was mild in intensity. Level 3 vertigo attacks were considered major spells because of associated symptoms of nausea and/or vomiting and the need to stop activity for a protracted period. A level 4 attack was defined as the worst attack to date. Days without any vertigo were scored as vertigo free (level 0).

A summary statistic, the cumulative proportion of days with definitive vertigo (level 2 or higher), was calculated for each participant for each month. (For example, a patient report of 2 vertigo attack days per month [2/30] would correspond to 0.06 or 6% of the days during that month with significant vertigo.) During the last prerandomization month, the mean vertigo frequency was 0.23 (range, 0.06-0.62), which indicates that significant vertigo occurred on an average of 7 (23%) of 30 days.

A categorical score for level of vertigo control (excellent, good, or fair-poor) was then assigned to each subject during the 4 months of follow-up, regardless of treatment assignment. Subjects with complete or nearly complete vertigo control after randomization (≤1 definitive attack/mo) were given a score of excellent (n = 15); those with 50% vertigo control (2-3 definitive attacks/mo), good (n = 19); and those who showed no improvement or worsened (≥4 definitive attacks/mo), fair to poor (n = 27). The mean proportion of days (range) with definitive vertigo for excellent vertigo control was 0.009 (0.000-0.036); for good, 0.064 (0.040-0.101); and for fair to poor, 0.215 (0.105-0.512). Thus, the vertigo control categorization is a convenient summary statistic of the cohort’s vertigo experience.

Questionnaires were mainly completed by the participant at each study center and checked for completeness by the study coordinator. Some forms were mailed or faxed in. The forms were then transferred to a data coordination center and optically scanned and verified. Stata software, version 8 (StataCorp LP, College Station, Tex), was used for data storage and analysis.

Thirty-four (1.3%) of the severity items were not completed. The missing values ranged from 0 to 6 per item (mean, 2) and occurred with equal frequency at each of the 3 measurement times. The missing values were assigned the mean score for that item by the rest of the cohort at that time. This strategy returns the same group means as with the item dropped but is preferable because it includes all the cases. (Statistical programs usually omit cases with missing values from group summaries and analyses.)

Each item on the MDPOSI was evaluated statistically using repeated measures analysis of variance with vertigo control category and time as covariates. In this way, the separate effects of vertigo or time could be estimated while adjusting for the level of the other variable and for variation among individual subjects. In the exploratory phase of analysis, other independent variables were added 1 at a time to the model to determine any effects. These variables were age, sex, and treatment group. After these additional variables were found to have no effect on the model, they were dropped from the final analysis. Because of the multiple comparisons and correlated data, the level of significance was set at P<.01.

Test-retest reliability was determined by comparing the responses on the importance scales and the overall MDPOSI score for subjects whose vertigo did not change from baseline to 2 months. The importance scores were assessed by 1-way analysis of variance to determine constancy of reporting. Traditional repeated administration of the severity questions was not deemed appropriate because MD symptoms vary from day to day and there is no independent objective method to assure constancy of disease status. Internal consistency, which was found to be satisfactory in the prior report,² was reassessed using the Cronbach α statistic.

The standardized response mean (SRM) was calculated by dividing the mean difference in the total MDPOSI score (items 1-16) from baseline to the 4-month measurement time by the standard deviation of the difference. An SRM of 0.2 indicates small responsiveness to clinical change; 0.5, moderate responsiveness; and 0.8 or greater, large responsiveness.⁸

The demographic aspects of the subjects are summarized in Table 1 by vertigo control category. There were no notable demographic differences between the subjects categorized by symptom severity except for treatment group.

The mean score on each item on the MDPOSI at baseline, 2 months, and 4 months was compared with vertigo control category (excellent, good, or fair-poor) in a repeated measures analysis of variance. These results are listed in Table 2. The mean scores for each item are noted for both vertigo control category and time of measurement. In general, those items that showed a gradient across the vertigo categories or time periods also were found to be statistically significant. The gradient gives a view of the magnitude of the change in the MDPOSI items over time.

There was no significant variation in the importance scores by analysis of variance at the 3 measurement times (F_2,61= 1.39; P = .25). Twelve subjects who experienced little change in vertigo score on the diaries from baseline to 2 months (mean ± SD difference, 0.10 ± 0.15; 95% confidence interval, 0.06-0.14) were used to further assess test-retest reliability. There was no significant difference in the MDPOSI score from baseline to 2 months for these subjects (mean ± SD difference, 3.17 ± 8.07; 95% confidence interval, −1.96 to 8.29) as well as no change in the importance scores. Therefore, test-retest reliability was deemed adequate. The Cronbach αvalue was 0.888 overall, which indicates excellent interitem reliability.

The items that varied by vertigo category (items 4 and 8) indicate the impact of vertigo on self-perceived well-being and psychosocial status. These items related to daily activities during and between attacks. The effects on social life, closeness, mood, memory, travel, outlook for the future, job loss, and disability retirement were small and did not vary by vertigo category, even though the scores on all 4 of these items improved significantly from baseline.

We asked participants to rate the importance of each item to gain additional insight and to determine possible incongruity of participant perceptions of each aspect of their disorder. In general, the importance score varied with the symptom score, with correlations ranging from 0.16 to 0.5. The highest symptom scores related to balance and activity level during attacks, and these were given the greatest importance score as well. Memory received the third highest importance score even though memory loss received the lowest severity score among all the health and function items. Concerns about employment were scored low both from the severity and importance aspects.

The overall SRM comparing the change in the total MDPOSI score from baseline to the 4-month measurement was 0.80. Moreover, when calculated within each vertigo category, the SRM was 1.26 for the excellent responders, 1.02 for the good responders, and 0.45 for the fair-poor responders. Thus, the total MDPOSI score was highly responsive to the status and the change in status of the clinical condition.

It is generally observed that the symptoms of MD improve with time,⁹ and the present report affirms this observation. The improvement in vertigo frequency noted herein reflects the use of the Meniett device plus the natural history of the disorder.³ Our analysis provides an excellent opportunity to determine the attributes of the MDPOSI in relation to a changing disorder. This situation is different from most outcomes studies that focus on a progressively worsening disorder. Thus, MD poses unique problems in assessment.

Disease-specific health-related outcomes instruments have potential uses in (1) defining and understanding the clinical pathophysiology of disease, particularly in the realm of so-called soft outcomes, (2) diagnosis and evaluation of patients, and (3) monitoring of treatment. The MDPOSI was developed with all 3 goals in mind. Given the large variation in diagnostic criteria, staging, and evaluation of treatment results in people with MD, it seemed that such an instrument, once validated, might be useful in patient care and clinical trials.

The present report demonstrates an important attribute of evaluative instruments, namely responsiveness to change over time in relation to an independent measure of disease severity (vertigo category). The SRM of the total MDPOSI score is highly responsive to the status of the patients’ symptoms as measured by the daily vertigo diary. An overall SRM of 0.8, which was seen with the MDPOSI, is considered excellent. Moreover, the SRM varied appropriately and substantially by relief of vertigo category, indicating excellent responsiveness.

Version 2 of the MDPOSI demonstrated the same level of internal consistency and construct validity as version 1, which suggests that the dropped items did not affect the internal consistency and construct validity of the second version of the instrument. Moreover, the test-retest reliability of version 2 has been established as part of this report.

As expected, the 4 items labeled “during,” which reflect symptoms during attacks, showed the greatest change over time: on average, 1 full point on the 0-to-5 scale. They also showed the largest differences across the 3 vertigo categories. The “between” items also showed change but of slightly lower magnitude. Hearing showed little subjective change, whereas activity showed the greatest change.

Vertigo is the chief symptom for which people with MD seek relief, and its effect on balance is a key concern for patients because they are unable to function normally in their daily activities. Thus, the finding of high importance attached to balance and daily activities (including travel) during attacks was not unexpected. Given that many people develop unilateral loss of vestibular function as their vertigo subsides with time, we anticipated greater concern about balance between attacks. It may be the case that because 50% of the participants had already sustained vestibular loss in the affected ear at baseline, some degree of compensation had already occurred, and thus balance between attacks, mitigated by the degree of compensation, may not have posed a great problem to this cohort.

The 3 hearing variables (hearing during, hearing between, and tinnitus/pressure) did not discriminate among levels of disease severity based on vertigo. The vast majority of these subjects had fixed unilateral hearing loss with little variation associated with vertigo spells. Thus, the responses to hearing questions probably reflect the poor but stable hearing in the affected ear. One would suspect that the responses to these questions would be quite different in people with bilateral disease.

Many people with vestibular disorders report difficulty with thinking, concentrating, and short-term memory. However, studies to substantiate a specific deficit have been inconclusive.¹⁰ The MDPOSI demonstrates self-reported difficulty with concentration and memory that improves over time. However, the level of difficulty in these 2 areas is considerably below that experienced in other areas affected by MD.

The items related to the psychosocial domain were not affected by vertigo level even though they improved with time. The level of concern in this domain was moderate at worst, which suggests that the disorder had little impact on the subjects’ personal lives.

The low scores on the 2 employment items (items 15 and 16) were unexpected, given the disruptive effects of MD on daily activity and the importance of work to most patients with MD. This suggests that these participants were quite secure in their employment situation and that they were not worried about job loss or reassignment.

Importance ratings had a narrow range, between 3.3 and 3.9, with the exception of balance during attacks and hearing between attacks, for both of which the average ranking was 4.4 of a possible 5. Given the high importance ascribed to all items on the MDPOSI and the narrow range of responses, this variable added little to the severity information provided by the instrument. We originally planned to compute a value score (severity × importance) when the instrument was designed. However, because the importance correlated significantly with the severity responses, this additional statistic seems redundant and will be dropped from version 3.

In conclusion, the MDPOSI, version 2, is a sensitive and responsive disease-specific quality-of-life instrument for people with classic MD and is suitable for use in clinical practice and clinical research.

Correspondence: George A. Gates, MD, Virginia Merrill Bloedel Hearing Research Center, University of Washington, Box 357923, Seattle, WA 98195-7923 (ggates@u.washington.edu).

Submitted for Publication: December 29, 2004; final revision received April 8, 2005; accepted May 10, 2005.

Financial Disclosure: None.

Funding/Support: The data accrual in this report was funded by an unrestricted grant from Medtronics Xomed Inc, Jacksonville, Fla. Development of the MDPOSI was aided by a grant from the Deafness Research Foundation, McLean, Va. Preparation of this report was supported by the Virginia Merrill Bloedel Hearing Research Foundation at the University of Washington, Seattle.

Acknowledgment: We thank Doug Green, MD, Steve Telian, MD, and Debara Tucci, MD, for their help with data accrual.